Split (1)

train · 107k rows

id stringlengths 8 78	source stringclasses 743 values	chunk_id int64 1 5.05k	text stringlengths 593 49.7k
sql-reference-009	sql-reference.pdf	9	month. select date '2023-03-31' - interval '4 months'; ?column? --------------------- 2022-11-30 00:00:00 Interval data types and literals You can use an interval data type to store durations of time in units such as, seconds, minutes, hours, days, months, and years. Interval data types and literals can be used in datetime calculations, such as, adding intervals to dates and timestamps, summing intervals, and subtracting an interval from a date or timestamp. Interval literals can be used as input values to interval data type columns in a table. Syntax of interval data type To specify an interval data type to store a duration of time in years and months: INTERVAL year_to_month_qualifier To specify an interval data type to store a duration in days, hours, minutes, and seconds: INTERVAL day_to_second_qualifier [ (fractional_precision) ] Datetime types 27 AWS Clean Rooms Syntax of interval literal SQL Reference To specify an interval literal to deﬁne a duration of time in years and months: INTERVAL quoted-string year_to_month_qualifier To specify an interval literal to deﬁne a duration in days, hours, minutes, and seconds: INTERVAL quoted-string day_to_second_qualifier [ (fractional_precision) ] Arguments quoted-string Speciﬁes a positive or negative numeric value specifying a quantity and the datetime unit as an input string. If the quoted-string contains only a numeric, then AWS Clean Rooms determines the units from the year_to_month_qualiﬁer or day_to_second_qualiﬁer. For example, '23' MONTH represents 1 year 11 months, '-2' DAY represents -2 days 0 hours 0 minutes 0.0 seconds, '1-2' MONTH represents 1 year 2 months, and '13 day 1 hour 1 minute 1.123 seconds' SECOND represents 13 days 1 hour 1 minute 1.123 seconds. For more information about output formats of an interval, see Interval styles. year_to_month_qualiﬁer Speciﬁes the range of the interval. If you use a qualiﬁer and create an interval with time units smaller than the qualiﬁer, AWS Clean Rooms truncates and discards the smaller parts of the interval. Valid values for year_to_month_qualiﬁer are: • YEAR • MONTH • YEAR TO MONTH day_to_second_qualiﬁer Speciﬁes the range of the interval. If you use a qualiﬁer and create an interval with time units smaller than the qualiﬁer, AWS Clean Rooms truncates and discards the smaller parts of the interval. Valid values for day_to_second_qualiﬁer are: • DAY • HOUR • MINUTE Datetime types 28 AWS Clean Rooms • SECOND • DAY TO HOUR • DAY TO MINUTE • DAY TO SECOND • HOUR TO MINUTE • HOUR TO SECOND • MINUTE TO SECOND SQL Reference The output of the INTERVAL literal is truncated to the smallest INTERVAL component speciﬁed. For example, when using a MINUTE qualiﬁer, AWS Clean Rooms discards the time units smaller than MINUTE. select INTERVAL '1 day 1 hour 1 minute 1.123 seconds' MINUTE The resulting value is truncated to '1 day 01:01:00'. fractional_precision Optional parameter that speciﬁes the number of fractional digits allowed in the interval. The fractional_precision argument should only be speciﬁed if your interval contains SECOND. For example, SECOND(3) creates an interval that allows only three fractional digits, such as 1.234 seconds. The maximum number of fractional digits is six. The session conﬁguration interval_forbid_composite_literals determines whether an error is returned when an interval is speciﬁed with both YEAR TO MONTH and DAY TO SECOND parts. Interval arithmetic You can use interval values with other datetime values to perform arithmetic operations. The following tables describe the available operations and what data type results from each operation. Note Operations that can produce both date and timestamp results do so based on the smallest unit of time involved in the equation. For example, when you add an interval to a date the result is a date if it is a YEAR TO MONTH interval, and a timestamp if it is a DAY TO SECOND interval. Datetime types 29 AWS Clean Rooms SQL Reference Operations where the ﬁrst operand is an interval produce the following results for the given second operand: Operator - + * / Date N/A Date N/A N/A Timestamp Interval Numeric N/A Interval Date/Timestamp Interval N/A N/A N/A N/A N/A N/A Interval Interval Operations where the ﬁrst operand is a date produce the following results for the given second operand: Operator Date Timestamp Interval Numeric - + Numeric Interval Date/Timestamp Date N/A N/A N/A N/A Operations where the ﬁrst operand is a timestamp produce the following results for the given second operand: Operator Date Timestamp Interval Numeric Numeric Interval Timestamp Timestamp N/A N/A N/A N/A - + Interval styles • postgres – follows PostgreSQL style. This is the default. • postgres_verbose – follows PostgreSQL verbose style. • sql_standard – follows the SQL standard interval literals style. Datetime types 30 AWS Clean Rooms SQL Reference The following command sets the interval style to sql_standard. SET IntervalStyle to 'sql_standard'; postgres output format The following is the output format for postgres interval style. Each numeric value can
sql-reference-010	sql-reference.pdf	10	N/A Operations where the ﬁrst operand is a timestamp produce the following results for the given second operand: Operator Date Timestamp Interval Numeric Numeric Interval Timestamp Timestamp N/A N/A N/A N/A - + Interval styles • postgres – follows PostgreSQL style. This is the default. • postgres_verbose – follows PostgreSQL verbose style. • sql_standard – follows the SQL standard interval literals style. Datetime types 30 AWS Clean Rooms SQL Reference The following command sets the interval style to sql_standard. SET IntervalStyle to 'sql_standard'; postgres output format The following is the output format for postgres interval style. Each numeric value can be negative. '<numeric> <unit> [, <numeric> <unit> ...]' select INTERVAL '1-2' YEAR TO MONTH::text varchar --------------- 1 year 2 mons select INTERVAL '1 2:3:4.5678' DAY TO SECOND::text varchar ------------------ 1 day 02:03:04.5678 postgres_verbose output format postgres_verbose syntax is similar to postgres, but postgres_verbose outputs also contain the unit of time. '[@] <numeric> <unit> [, <numeric> <unit> ...] [direction]' select INTERVAL '1-2' YEAR TO MONTH::text varchar ----------------- @ 1 year 2 mons select INTERVAL '1 2:3:4.5678' DAY TO SECOND::text varchar Datetime types 31 AWS Clean Rooms SQL Reference --------------------------- @ 1 day 2 hours 3 mins 4.56 secs sql_standard output format Interval year to month values are formatted as the following. Specifying a negative sign before the interval indicates the interval is a negative value and applies to the entire interval. '[-]yy-mm' Interval day to second values are formatted as the following. '[-]dd hh:mm:ss.ffffff' SELECT INTERVAL '1-2' YEAR TO MONTH::text varchar ------- 1-2 select INTERVAL '1 2:3:4.5678' DAY TO SECOND::text varchar --------------- 1 2:03:04.5678 Examples of interval data type The following examples demonstrate how to use INTERVAL data types with tables. create table sample_intervals (y2m interval month, h2m interval hour to minute); insert into sample_intervals values (interval '20' month, interval '2 days 1:1:1.123456' day to second); select y2m::text, h2m::text from sample_intervals; y2m \| h2m ---------------+----------------- 1 year 8 mons \| 2 days 01:01:00 Datetime types 32 AWS Clean Rooms SQL Reference update sample_intervals set y2m = interval '2' year where y2m = interval '1-8' year to month; select * from sample_intervals; y2m \| h2m ---------+----------------- 2 years \| 2 days 01:01:00 delete from sample_intervals where h2m = interval '2 1:1:0' day to second; select * from sample_intervals; y2m \| h2m -----+----- Examples of interval literals The following examples are run with interval style set to postgres. The following example demonstrates how to create an INTERVAL literal of 1 year. select INTERVAL '1' YEAR intervaly2m --------------- 1 years 0 mons If you specify a quoted-string that exceeds the qualiﬁer, the remaining units of time are truncated from the interval. In the following example, an interval of 13 months becomes 1 year and 1 month, but the remaining 1 month is left out because of the YEAR qualiﬁer. select INTERVAL '13 months' YEAR intervaly2m --------------- 1 years 0 mons If you use a qualiﬁer lower than your interval string, leftover units are included. select INTERVAL '13 months' MONTH Datetime types 33 AWS Clean Rooms intervaly2m --------------- 1 years 1 mons SQL Reference Specifying a precision in your interval truncates the number of fractional digits to the speciﬁed precision. select INTERVAL '1.234567' SECOND (3) intervald2s -------------------------------- 0 days 0 hours 0 mins 1.235 secs If you don't specify a precision, AWS Clean Rooms uses the maximum precision of 6. select INTERVAL '1.23456789' SECOND intervald2s ----------------------------------- 0 days 0 hours 0 mins 1.234567 secs The following example demonstrates how to create a ranged interval. select INTERVAL '2:2' MINUTE TO SECOND intervald2s ------------------------------ 0 days 0 hours 2 mins 2.0 secs Qualiﬁers dictate the units that you're specifying. For example, even though the following example uses the same quoted-string of '2:2' as the previous example, AWS Clean Rooms recognizes that it uses diﬀerent units of time because of the qualiﬁer. select INTERVAL '2:2' HOUR TO MINUTE intervald2s ------------------------------ 0 days 2 hours 2 mins 0.0 secs Abbreviations and plurals of each unit are also supported. For example, 5s, 5 second, and 5 seconds are equivalent intervals. Supported units are years, months, hours, minutes, and seconds. Datetime types 34 AWS Clean Rooms SQL Reference select INTERVAL '5s' SECOND intervald2s ------------------------------ 0 days 0 hours 0 mins 5.0 secs select INTERVAL '5 HOURS' HOUR intervald2s ------------------------------ 0 days 5 hours 0 mins 0.0 secs select INTERVAL '5 h' HOUR intervald2s ------------------------------ 0 days 5 hours 0 mins 0.0 secs Examples of interval literals without qualiﬁer syntax Note The following examples demonstrate using an interval literal without a YEAR TO MONTH or DAY TO SECOND qualiﬁer. For information about using the recommended interval literal with a qualiﬁer, see Interval data types and literals. Use an interval literal to identify speciﬁc periods of time, such as 12 hours or 6 months. You can use these interval literals in conditions and
sql-reference-011	sql-reference.pdf	11	secs select INTERVAL '5 HOURS' HOUR intervald2s ------------------------------ 0 days 5 hours 0 mins 0.0 secs select INTERVAL '5 h' HOUR intervald2s ------------------------------ 0 days 5 hours 0 mins 0.0 secs Examples of interval literals without qualiﬁer syntax Note The following examples demonstrate using an interval literal without a YEAR TO MONTH or DAY TO SECOND qualiﬁer. For information about using the recommended interval literal with a qualiﬁer, see Interval data types and literals. Use an interval literal to identify speciﬁc periods of time, such as 12 hours or 6 months. You can use these interval literals in conditions and calculations that involve datetime expressions. An interval literal is expressed as a combination of the INTERVAL keyword with a numeric quantity and a supported date part, for example INTERVAL '7 days' or INTERVAL '59 minutes'. You can connect several quantities and units to form a more precise interval, for example: INTERVAL '7 days, 3 hours, 59 minutes'. Abbreviations and plurals of each unit are also supported; for example: 5 s, 5 second, and 5 seconds are equivalent intervals. If you don't specify a date part, the interval value represents seconds. You can specify the quantity value as a fraction (for example: 0.5 days). Datetime types 35 AWS Clean Rooms SQL Reference The following examples show a series of calculations with diﬀerent interval values. The following adds 1 second to the speciﬁed date. select caldate + interval '1 second' as dateplus from date where caldate='12-31-2008'; dateplus --------------------- 2008-12-31 00:00:01 (1 row) The following adds 1 minute to the speciﬁed date. select caldate + interval '1 minute' as dateplus from date where caldate='12-31-2008'; dateplus --------------------- 2008-12-31 00:01:00 (1 row) The following adds 3 hours and 35 minutes to the speciﬁed date. select caldate + interval '3 hours, 35 minutes' as dateplus from date where caldate='12-31-2008'; dateplus --------------------- 2008-12-31 03:35:00 (1 row) The following adds 52 weeks to the speciﬁed date. select caldate + interval '52 weeks' as dateplus from date where caldate='12-31-2008'; dateplus --------------------- 2009-12-30 00:00:00 (1 row) The following adds 1 week, 1 hour, 1 minute, and 1 second to the speciﬁed date. select caldate + interval '1w, 1h, 1m, 1s' as dateplus from date Datetime types 36 AWS Clean Rooms SQL Reference where caldate='12-31-2008'; dateplus --------------------- 2009-01-07 01:01:01 (1 row) The following adds 12 hours (half a day) to the speciﬁed date. select caldate + interval '0.5 days' as dateplus from date where caldate='12-31-2008'; dateplus --------------------- 2008-12-31 12:00:00 (1 row) The following subtracts 4 months from February 15, 2023 and the result is October 15, 2022. select date '2023-02-15' - interval '4 months'; ?column? --------------------- 2022-10-15 00:00:00 The following subtracts 4 months from March 31, 2023 and the result is November 30, 2022. The calculation considers the number of days in a month. select date '2023-03-31' - interval '4 months'; ?column? --------------------- 2022-11-30 00:00:00 Boolean type Use the BOOLEAN data type to store true and false values in a single-byte column. The following table describes the three possible states for a Boolean value and the literal values that result in that state. Regardless of the input string, a Boolean column stores and outputs "t" for true and "f" for false. Boolean type 37 AWS Clean Rooms SQL Reference State True False Valid literal values Storage TRUE 't' 1 byte 'true' 'y' 'yes' '1' FALSE 'f' 1 byte 'false' 'n' 'no' '0' Unknown NULL 1 byte You can use an IS comparison to check a Boolean value only as a predicate in the WHERE clause. You can't use the IS comparison with a Boolean value in the SELECT list. Examples You can use a BOOLEAN column to store an "Active/Inactive" state for each customer in a CUSTOMER table. select * from customer; custid \| active_flag -------+-------------- 100 \| t In this example, the following query selects users from the USERS table who like sports but do not like theatre: select firstname, lastname, likesports, liketheatre from users where likesports is true and liketheatre is false order by userid limit 10; firstname \| lastname \| likesports \| liketheatre ----------+------------+------------+------------- Alejandro \| Rosalez \| t \| f Akua \| Mansa \| t \| f Arnav \| Desai \| t \| f Boolean type 38 AWS Clean Rooms SQL Reference Carlos \| Salazar \| t \| f Diego \| Ramirez \| t \| f Efua \| Owusu \| t \| f John \| Stiles \| t \| f Jorge \| Souza \| t \| f Kwaku \| Mensah \| t \| f Kwesi \| Manu \| t \| f (10 rows) The following example selects users from the USERS table for whom is it unknown whether they like rock music. select firstname, lastname, likerock from users where likerock is unknown order by userid limit 10; firstname \| lastname \| likerock ----------+----------+---------- Alejandro \| Rosalez \|
sql-reference-012	sql-reference.pdf	12	AWS Clean Rooms SQL Reference Carlos \| Salazar \| t \| f Diego \| Ramirez \| t \| f Efua \| Owusu \| t \| f John \| Stiles \| t \| f Jorge \| Souza \| t \| f Kwaku \| Mensah \| t \| f Kwesi \| Manu \| t \| f (10 rows) The following example selects users from the USERS table for whom is it unknown whether they like rock music. select firstname, lastname, likerock from users where likerock is unknown order by userid limit 10; firstname \| lastname \| likerock ----------+----------+---------- Alejandro \| Rosalez \| Carlos \| Salazar \| Diego \| Ramirez \| John \| Stiles \| Kwaku \| Mensah \| Martha \| Rivera \| Mateo \| Jackson \| Paulo \| Santos \| Richard \| Roe \| Saanvi \| Sarkar \| (10 rows) The following example returns an error because it uses an IS comparison in the SELECT list. select firstname, lastname, likerock is true as "check" from users order by userid limit 10; [Amazon](500310) Invalid operation: Not implemented The following example succeeds because it uses an equal comparison ( = ) in the SELECT list instead of the IS comparison. Boolean type 39 AWS Clean Rooms SQL Reference select firstname, lastname, likerock = true as "check" from users order by userid limit 10; firstname \| lastname \| check ----------+-----------+------ Alejandro \| Rosalez \| Carlos \| Salazar \| Diego \| Ramirez \| true John \| Stiles \| Kwaku \| Mensah \| true Martha \| Rivera \| true Mateo \| Jackson \| Paulo \| Santos \| false Richard \| Roe \| Saanvi \| Sarkar \| Boolean literals The following rules are for working with Boolean literals that are supported by AWS Clean Rooms Spark SQL. Use a Boolean literal to specify a Boolean value, such as TRUE or FALSE. Syntax TRUE \| FALSE Example The following example shows a column with a speciﬁed value of TRUE . SELECT TRUE AS col; +----+ \| col\| +----+ \|true\| +----+ Boolean type 40 AWS Clean Rooms Binary type SQL Reference Use the BINARY data type to store and manage ﬁxed-length, uninterpreted binary data, providing eﬃcient storage and comparison capabilities for speciﬁc use cases. The BINARY data type stores a ﬁxed number of bytes, regardless of the actual length of the data being stored. The maximum length is typically 255 bytes. BINARY is used to store raw, uninterpreted binary data, such as images, documents, or other types of ﬁles. The data is stored exactly as it is provided, without any character encoding or interpretation. Binary data stored in BINARY columns is compared and sorted byte-by-byte, based on the actual binary values, rather than any character encoding or collation rules. The following example query shows the binary representation of the string "abc". Each character in the string is represented by its ASCII code in hexadecimal format: "a" is 0x61, "b" is 0x62, and "c" is 0x63. When combined, these hexadecimal values form the binary representation "616263". SELECT 'abc'::binary; binary --------- 616263 Nested type AWS Clean Rooms supports queries involving data with nested data types, speciﬁcally the AWS Glue STRUCT, ARRAY, and MAP column types. Only the custom analysis rule supports nested data types. Notably, nested data types don't conform to the rigid, tabular structure of the relational data model of SQL databases. Nested data types contains tags that reference distinct entities within the data. They can contain complex values such as arrays, nested structures, and other complex structures that are associated with serialization formats, such as JSON. Nested data types support up to 1 MB of data for an individual nested data type ﬁeld or object. Topics • ARRAY type • MAP type Binary type 41 AWS Clean Rooms • STRUCT type • Examples of nested data types ARRAY type SQL Reference Use the ARRAY type to represent values comprising a sequence of elements with the type of elementType. array(elementType, containsNull) Use containsNull to indicate if elements in an ARRAY type can have null values. MAP type Use the MAP type to represent values comprising a set of key-value pairs. map(keyType, valueType, valueContainsNull) keyType: the data type of keys valueType: the data type of values Keys aren't allowed to have null values. Use valueContainsNull to indicate if values of a MAP type value can have null values. STRUCT type Use the STRUCT type to represent values with the structure described by a sequence of StructFields (ﬁelds). struct(name, dataType, nullable) StructField(name, dataType, nullable): Represents a ﬁeld in a StructType. dataType: the data type a ﬁeld name: the name of a ﬁeld Use nullable to indicate if values of these ﬁelds can have null values. Nested type 42 AWS Clean Rooms SQL Reference Examples of nested data types For the struct<given:varchar, family:varchar> type, there are two attribute names: given, and family, each corresponding
sql-reference-013	sql-reference.pdf	13	values. Use valueContainsNull to indicate if values of a MAP type value can have null values. STRUCT type Use the STRUCT type to represent values with the structure described by a sequence of StructFields (ﬁelds). struct(name, dataType, nullable) StructField(name, dataType, nullable): Represents a ﬁeld in a StructType. dataType: the data type a ﬁeld name: the name of a ﬁeld Use nullable to indicate if values of these ﬁelds can have null values. Nested type 42 AWS Clean Rooms SQL Reference Examples of nested data types For the struct<given:varchar, family:varchar> type, there are two attribute names: given, and family, each corresponding to a varchar value. For the array<varchar> type, the array is speciﬁed as a list of varchar. The array<struct<shipdate:timestamp, price:double>> type refers to a list of elements with struct<shipdate:timestamp, price:double> type. The map data type behaves like an array of structs, where the attribute name for each element in the array is denoted by key and it maps to a value. Example For example, the map<varchar(20), varchar(20)> type is treated as array<struct<key:varchar(20), value:varchar(20)>>, where key and value refer to the attributes of the map in the underlying data. For information about how AWS Clean Rooms enables navigation into arrays and structures, see Navigation. For information about how AWS Clean Rooms enables iteration over arrays by navigating the array using the FROM clause of a query, see Unnesting queries. Type compatibility and conversion The following topics describe how type conversion rules and data type compatibility work in AWS Clean Rooms Spark SQL. Topics • Compatibility • General compatibility and conversion rules • Implicit conversion types Compatibility Data type matching and matching of literal values and constants to data types occurs during various database operations, including the following: Type compatibility and conversion 43 AWS Clean Rooms SQL Reference • Data manipulation language (DML) operations on tables • UNION, INTERSECT, and EXCEPT queries • CASE expressions • Evaluation of predicates, such as LIKE and IN • Evaluation of SQL functions that do comparisons or extractions of data • Comparisons with mathematical operators The results of these operations depend on type conversion rules and data type compatibility. Compatibility implies that a one-to-one matching of a certain value and a certain data type is not always required. Because some data types are compatible, an implicit conversion, or coercion, is possible. For more information, see Implicit conversion types. When data types are incompatible, you can sometimes convert a value from one data type to another by using an explicit conversion function. General compatibility and conversion rules Note the following compatibility and conversion rules: • In general, data types that fall into the same type category (such as diﬀerent numeric data types) are compatible and can be implicitly converted. For example, with implicit conversion you can insert a decimal value into an integer column. The decimal is rounded to produce a whole number. Or you can extract a numeric value, such as 2008, from a date and insert that value into an integer column. • Numeric data types enforce overﬂow conditions that occur when you attempt to insert out- of-range values. For example, a decimal value with a precision of 5 does not ﬁt into a decimal column that was deﬁned with a precision of 4. An integer or the whole part of a decimal is never truncated. However, the fractional part of a decimal can be rounded up or down, as appropriate. However, results of explicit casts of values selected from tables are not rounded. • Diﬀerent types of character strings are compatible. VARCHAR column strings containing single- byte data and CHAR column strings are comparable and implicitly convertible. VARCHAR strings that contain multibyte data are not comparable. Also, you can convert a character string to a date, time, timestamp, or numeric value if the string is an appropriate literal value. Any leading or trailing spaces are ignored. Conversely, you can convert a date, time, timestamp, or numeric value to a ﬁxed-length or variable-length character string. Type compatibility and conversion 44 AWS Clean Rooms Note SQL Reference A character string that you want to cast to a numeric type must contain a character representation of a number. For example, you can cast the strings '1.0' or '5.9' to decimal values, but you can't cast the string 'ABC' to any numeric type. • If you compare DECIMAL values with character strings, AWS Clean Rooms attempts to convert the character string to a DECIMAL value. When comparing all other numeric values with character strings, the numeric values are converted to character strings. To enforce the opposite conversion (for example, converting character strings to integers, or converting DECIMAL values to character strings), use an explicit function, such as CAST function. • To convert 64-bit DECIMAL or NUMERIC values to a higher precision, you must use an explicit conversion function
sql-reference-014	sql-reference.pdf	14	'5.9' to decimal values, but you can't cast the string 'ABC' to any numeric type. • If you compare DECIMAL values with character strings, AWS Clean Rooms attempts to convert the character string to a DECIMAL value. When comparing all other numeric values with character strings, the numeric values are converted to character strings. To enforce the opposite conversion (for example, converting character strings to integers, or converting DECIMAL values to character strings), use an explicit function, such as CAST function. • To convert 64-bit DECIMAL or NUMERIC values to a higher precision, you must use an explicit conversion function such as the CAST or CONVERT functions. Implicit conversion types There are two types of implicit conversions: • Implicit conversions in assignments, such as setting values in INSERT or UPDATE commands • Implicit conversions in expressions, such as performing comparisons in the WHERE clause The following table lists the data types that can be converted implicitly in assignments or expressions. You can also use an explicit conversion function to perform these conversions. From type BIGINT To type BOOLEAN CHAR DECIMAL (NUMERIC) DOUBLE PRECISION (FLOAT8) INTEGER REAL (FLOAT4) Type compatibility and conversion 45 AWS Clean Rooms From type CHAR DATE SQL Reference To type SMALLINT or SHORT VARCHAR VARCHAR CHAR VARCHAR TIMESTAMP TIMESTAMPTZ DECIMAL (NUMERIC) BIGINT or LONG CHAR DOUBLE PRECISION (FLOAT8) INTEGER INT) REAL (FLOAT4) SMALLINT or SHORT VARCHAR DOUBLE PRECISION (FLOAT8) BIGINT or LONG CHAR DECIMAL (NUMERIC) INTEGER (INT) REAL (FLOAT4) SMALLINT or SHORT Type compatibility and conversion 46 AWS Clean Rooms From type SQL Reference To type VARCHAR INTEGER (INT) BIGINT or LONG BOOLEAN CHAR DECIMAL (NUMERIC) DOUBLE PRECISION (FLOAT8) REAL (FLOAT4) SMALLINT or SHORT VARCHAR BIGINT or LONG CHAR DECIMAL (NUMERIC) INTEGER (INT) SMALLINT or SHORT VARCHAR BIGINT or LONG BOOLEAN CHAR DECIMAL (NUMERIC) DOUBLE PRECISION (FLOAT8) REAL (FLOAT4) SMALLINT Type compatibility and conversion 47 AWS Clean Rooms From type TIME Note SQL Reference To type INTEGER (INT) REAL (FLOAT4) VARCHAR VARCHAR TIMETZ Implicit conversions between DATE, TIME, TIMESTAMP_LTZ, TIMESTAMP_NTZ, or character strings use the current session time zone. The VARBYTE data type can't be implicitly converted to any other data type. For more information, see CAST function. AWS Clean Rooms Spark SQL commands The following SQL commands are supported in AWS Clean Rooms Spark SQL: Topics • SELECT SELECT The SELECT command returns rows from tables and user-deﬁned functions. The following SELECT SQL commands, clauses, and set operators are supported in AWS Clean Rooms Spark SQL: Topics • SELECT list • WITH clause SQL commands 48 SQL Reference AWS Clean Rooms • FROM clause • JOIN clause • WHERE clause • VALUES clause • GROUP BY clause • HAVING clause • Set operators • ORDER BY clause • Subquery examples • Correlated subqueries The syntax, arguments, and some examples come from the Apache Spark SQL Reference. SELECT list The SELECT list names the columns, functions, and expressions that you want the query to return. The list represents the output of the query. Syntax SELECT [ TOP number ] [ DISTINCT ] \| expression [ AS column_alias ] [, ...] Parameters TOP number TOP takes a positive integer as its argument, which deﬁnes the number of rows that are returned to the client. The behavior with the TOP clause is the same as the behavior with the LIMIT clause. The number of rows that is returned is ﬁxed, but the set of rows is not ﬁxed. To return a consistent set of rows, use TOP or LIMIT in conjunction with an ORDER BY clause. DISTINCT Option that eliminates duplicate rows from the result set, based on matching values in one or more columns. SELECT 49 AWS Clean Rooms expression SQL Reference An expression formed from one or more columns that exist in the tables referenced by the query. An expression can contain SQL functions. For example: coalesce(dimension, 'stringifnull') AS column_alias AS column_alias A temporary name for the column that is used in the ﬁnal result set. The AS keyword is optional. For example: coalesce(dimension, 'stringifnull') AS dimensioncomplete If you don't specify an alias for an expression that isn't a simple column name, the result set applies a default name to that column. Note The alias is recognized right after it is deﬁned in the target list. You cannot use an alias in other expressions deﬁned after it in the same target list. Usage notes TOP is a SQL extension. TOP provides an alternative to the LIMIT behavior. You can't use TOP and LIMIT in the same query. WITH clause A WITH clause is an optional clause that precedes the SELECT list in a query. The WITH clause deﬁnes one or more common_table_expressions. Each common table expression (CTE) deﬁnes a temporary table, which is similar to a view deﬁnition. You can reference these temporary tables in the FROM
sql-reference-015	sql-reference.pdf	15	right after it is deﬁned in the target list. You cannot use an alias in other expressions deﬁned after it in the same target list. Usage notes TOP is a SQL extension. TOP provides an alternative to the LIMIT behavior. You can't use TOP and LIMIT in the same query. WITH clause A WITH clause is an optional clause that precedes the SELECT list in a query. The WITH clause deﬁnes one or more common_table_expressions. Each common table expression (CTE) deﬁnes a temporary table, which is similar to a view deﬁnition. You can reference these temporary tables in the FROM clause. They're used only while the query they belong to runs. Each CTE in the WITH clause speciﬁes a table name, an optional list of column names, and a query expression that evaluates to a table (a SELECT statement). WITH clause subqueries are an eﬃcient way of deﬁning tables that can be used throughout the execution of a single query. In all cases, the same results can be achieved by using subqueries in SELECT 50 AWS Clean Rooms SQL Reference the main body of the SELECT statement, but WITH clause subqueries may be simpler to write and read. Where possible, WITH clause subqueries that are referenced multiple times are optimized as common subexpressions; that is, it may be possible to evaluate a WITH subquery once and reuse its results. (Note that common subexpressions aren't limited to those deﬁned in the WITH clause.) Syntax [ WITH common_table_expression [, common_table_expression , ...] ] where common_table_expression can be non-recursive. Following is the non-recursive form: CTE_table_name AS ( query ) Parameters common_table_expression Deﬁnes a temporary table that you can reference in the FROM clause and is used only during the execution of the query to which it belongs. CTE_table_name A unique name for a temporary table that deﬁnes the results of a WITH clause subquery. You can't use duplicate names within a single WITH clause. Each subquery must be given a table name that can be referenced in the FROM clause. query Any SELECT query that AWS Clean Rooms supports. See SELECT. Usage notes You can use a WITH clause in the following SQL statement: • SELECT, WITH, UNION, UNION ALL, INTERSECT, INTERSECT ALL, EXCEPT, or EXCEPT ALL If the FROM clause of a query that contains a WITH clause doesn't reference any of the tables deﬁned by the WITH clause, the WITH clause is ignored and the query runs as normal. A table deﬁned by a WITH clause subquery can be referenced only in the scope of the SELECT query that the WITH clause begins. For example, you can reference such a table in the FROM clause SELECT 51 AWS Clean Rooms SQL Reference of a subquery in the SELECT list, WHERE clause, or HAVING clause. You can't use a WITH clause in a subquery and reference its table in the FROM clause of the main query or another subquery. This query pattern results in an error message of the form relation table_name doesn't exist for the WITH clause table. You can't specify another WITH clause inside a WITH clause subquery. You can't make forward references to tables deﬁned by WITH clause subqueries. For example, the following query returns an error because of the forward reference to table W2 in the deﬁnition of table W1: with w1 as (select * from w2), w2 as (select * from w1) select * from sales; ERROR: relation "w2" does not exist Examples The following example shows the simplest possible case of a query that contains a WITH clause. The WITH query named VENUECOPY selects all of the rows from the VENUE table. The main query in turn selects all of the rows from VENUECOPY. The VENUECOPY table exists only for the duration of this query. with venuecopy as (select * from venue) select * from venuecopy order by 1 limit 10; venueid \| venuename \| venuecity \| venuestate \| venueseats ---------+----------------------------+-----------------+------------+------------ 1 \| Toyota Park \| Bridgeview \| IL \| 0 2 \| Columbus Crew Stadium \| Columbus \| OH \| 0 3 \| RFK Stadium \| Washington \| DC \| 0 4 \| CommunityAmerica Ballpark \| Kansas City \| KS \| 0 5 \| Gillette Stadium \| Foxborough \| MA \| 68756 6 \| New York Giants Stadium \| East Rutherford \| NJ \| 80242 7 \| BMO Field \| Toronto \| ON \| 0 8 \| The Home Depot Center \| Carson \| CA \| 0 9 \| Dick's Sporting Goods Park \| Commerce City \| CO \| 0 v 10 \| Pizza Hut Park \| Frisco \| TX \| 0 (10 rows) SELECT 52 AWS Clean Rooms SQL Reference The following example shows a WITH clause that produces two tables, named VENUE_SALES and TOP_VENUES. The second WITH query table selects from the ﬁrst.
sql-reference-016	sql-reference.pdf	16	Stadium \| Foxborough \| MA \| 68756 6 \| New York Giants Stadium \| East Rutherford \| NJ \| 80242 7 \| BMO Field \| Toronto \| ON \| 0 8 \| The Home Depot Center \| Carson \| CA \| 0 9 \| Dick's Sporting Goods Park \| Commerce City \| CO \| 0 v 10 \| Pizza Hut Park \| Frisco \| TX \| 0 (10 rows) SELECT 52 AWS Clean Rooms SQL Reference The following example shows a WITH clause that produces two tables, named VENUE_SALES and TOP_VENUES. The second WITH query table selects from the ﬁrst. In turn, the WHERE clause of the main query block contains a subquery that constrains the TOP_VENUES table. with venue_sales as (select venuename, venuecity, sum(pricepaid) as venuename_sales from sales, venue, event where venue.venueid=event.venueid and event.eventid=sales.eventid group by venuename, venuecity), top_venues as (select venuename from venue_sales where venuename_sales > 800000) select venuename, venuecity, venuestate, sum(qtysold) as venue_qty, sum(pricepaid) as venue_sales from sales, venue, event where venue.venueid=event.venueid and event.eventid=sales.eventid and venuename in(select venuename from top_venues) group by venuename, venuecity, venuestate order by venuename; venuename \| venuecity \| venuestate \| venue_qty \| venue_sales ------------------------+---------------+------------+-----------+------------- August Wilson Theatre \| New York City \| NY \| 3187 \| 1032156.00 Biltmore Theatre \| New York City \| NY \| 2629 \| 828981.00 Charles Playhouse \| Boston \| MA \| 2502 \| 857031.00 Ethel Barrymore Theatre \| New York City \| NY \| 2828 \| 891172.00 Eugene O'Neill Theatre \| New York City \| NY \| 2488 \| 828950.00 Greek Theatre \| Los Angeles \| CA \| 2445 \| 838918.00 Helen Hayes Theatre \| New York City \| NY \| 2948 \| 978765.00 Hilton Theatre \| New York City \| NY \| 2999 \| 885686.00 Imperial Theatre \| New York City \| NY \| 2702 \| 877993.00 Lunt-Fontanne Theatre \| New York City \| NY \| 3326 \| 1115182.00 Majestic Theatre \| New York City \| NY \| 2549 \| 894275.00 Nederlander Theatre \| New York City \| NY \| 2934 \| 936312.00 Pasadena Playhouse \| Pasadena \| CA \| 2739 \| 820435.00 Winter Garden Theatre \| New York City \| NY \| 2838 \| 939257.00 (14 rows) SELECT 53 AWS Clean Rooms SQL Reference The following two examples demonstrate the rules for the scope of table references based on WITH clause subqueries. The ﬁrst query runs, but the second fails with an expected error. The ﬁrst query has WITH clause subquery inside the SELECT list of the main query. The table deﬁned by the WITH clause (HOLIDAYS) is referenced in the FROM clause of the subquery in the SELECT list: select caldate, sum(pricepaid) as daysales, (with holidays as (select * from date where holiday ='t') select sum(pricepaid) from sales join holidays on sales.dateid=holidays.dateid where caldate='2008-12-25') as dec25sales from sales join date on sales.dateid=date.dateid where caldate in('2008-12-25','2008-12-31') group by caldate order by caldate; caldate \| daysales \| dec25sales -----------+----------+------------ 2008-12-25 \| 70402.00 \| 70402.00 2008-12-31 \| 12678.00 \| 70402.00 (2 rows) The second query fails because it attempts to reference the HOLIDAYS table in the main query as well as in the SELECT list subquery. The main query references are out of scope. select caldate, sum(pricepaid) as daysales, (with holidays as (select * from date where holiday ='t') select sum(pricepaid) from sales join holidays on sales.dateid=holidays.dateid where caldate='2008-12-25') as dec25sales from sales join holidays on sales.dateid=holidays.dateid where caldate in('2008-12-25','2008-12-31') group by caldate order by caldate; ERROR: relation "holidays" does not exist FROM clause The FROM clause in a query lists the table references (tables, views, and subqueries) that data is selected from. If multiple table references are listed, the tables must be joined, using appropriate SELECT 54 AWS Clean Rooms SQL Reference syntax in either the FROM clause or the WHERE clause. If no join criteria are speciﬁed, the system processes the query as a cross-join (Cartesian product). Topics • Syntax • Parameters • Usage notes • FROM TABLE VALUED FUNCTION Syntax FROM table_reference [, ...] where table_reference is one of the following: with_subquery_table_name \| table_name \| ( subquery ) [ [ AS ] alias ] table_reference [ NATURAL ] join_type table_reference [ USING ( join_column [, ...] ) ] table_reference [ INNER ] join_type table_reference ON expr Parameters with_subquery_table_name A table deﬁned by a subquery in the WITH clause. table_name Name of a table or view. alias Temporary alternative name for a table or view. An alias must be supplied for a table derived from a subquery. In other table references, aliases are optional. The AS keyword is always optional. Table aliases provide a convenient shortcut for identifying tables in other parts of a query, such as the WHERE clause. For example: select * from sales s, listing l where s.listid=l.listid SELECT 55 AWS Clean Rooms SQL Reference If you deﬁne a table alias is deﬁned,
sql-reference-017	sql-reference.pdf	17	expr Parameters with_subquery_table_name A table deﬁned by a subquery in the WITH clause. table_name Name of a table or view. alias Temporary alternative name for a table or view. An alias must be supplied for a table derived from a subquery. In other table references, aliases are optional. The AS keyword is always optional. Table aliases provide a convenient shortcut for identifying tables in other parts of a query, such as the WHERE clause. For example: select * from sales s, listing l where s.listid=l.listid SELECT 55 AWS Clean Rooms SQL Reference If you deﬁne a table alias is deﬁned, then the alias must be used to reference that table in the query. For example, if the query is SELECT "tbl"."col" FROM "tbl" AS "t", the query would fail because the table name is essentially overridden now. A valid query in this case would be SELECT "t"."col" FROM "tbl" AS "t". column_alias Temporary alternative name for a column in a table or view. subquery A query expression that evaluates to a table. The table exists only for the duration of the query and is typically given a name or alias. However, an alias isn't required. You can also deﬁne column names for tables that derive from subqueries. Naming column aliases is important when you want to join the results of subqueries to other tables and when you want to select or constrain those columns elsewhere in the query. A subquery may contain an ORDER BY clause, but this clause may have no eﬀect if a LIMIT or OFFSET clause isn't also speciﬁed. NATURAL Deﬁnes a join that automatically uses all pairs of identically named columns in the two tables as the joining columns. No explicit join condition is required. For example, if the CATEGORY and EVENT tables both have columns named CATID, a natural join of those tables is a join over their CATID columns. Note If a NATURAL join is speciﬁed but no identically named pairs of columns exist in the tables to be joined, the query defaults to a cross-join. join_type Specify one of the following types of join: • [INNER] JOIN • LEFT [OUTER] JOIN • RIGHT [OUTER] JOIN SELECT 56 AWS Clean Rooms • FULL [OUTER] JOIN • CROSS JOIN SQL Reference Cross-joins are unqualiﬁed joins; they return the Cartesian product of the two tables. Inner and outer joins are qualiﬁed joins. They are qualiﬁed either implicitly (in natural joins); with the ON or USING syntax in the FROM clause; or with a WHERE clause condition. An inner join returns matching rows only, based on the join condition or list of joining columns. An outer join returns all of the rows that the equivalent inner join would return plus non- matching rows from the "left" table, "right" table, or both tables. The left table is the ﬁrst- listed table, and the right table is the second-listed table. The non-matching rows contain NULL values to ﬁll the gaps in the output columns. ON join_condition Type of join speciﬁcation where the joining columns are stated as a condition that follows the ON keyword. For example: sales join listing on sales.listid=listing.listid and sales.eventid=listing.eventid USING ( join_column [, ...] ) Type of join speciﬁcation where the joining columns are listed in parentheses. If multiple joining columns are speciﬁed, they are delimited by commas. The USING keyword must precede the list. For example: sales join listing using (listid,eventid) Usage notes Joining columns must have comparable data types. A NATURAL or USING join retains only one of each pair of joining columns in the intermediate result set. A join with the ON syntax retains both joining columns in its intermediate result set. See also WITH clause. SELECT 57 AWS Clean Rooms SQL Reference FROM TABLE VALUED FUNCTION A Table Valued Function (TVF) is a special type of function that returns a table-like result set, rather than a single scalar value. The "FROM TABLE VALUED FUNCTION" syntax is used to call and utilize the output of a TVF within a larger SQL query. Table Valued Functions can be a powerful tool in SQL, as they allow you to encapsulate complex data retrieval and transformation logic, while also providing a ﬂexible and reusable way to incorporate that logic into your larger queries. Syntax: SELECT column1, column2, ..., columnn FROM table_valued_function(parameters) Parameters: table_valued_function(parameters The call to the Table Valued Function, which may require one or more input parameters. The result set returned by the TVF is treated as a temporary table that can be further queried, joined, or ﬁltered in the main query. In the following example, the GetCustomerOrders function is a Table Valued Function that takes a customer_id parameter and returns a table-like result set containing the orders for that customer. SELECT * FROM dbo.GetCustomerOrders(10) JOIN clause A SQL JOIN clause is used to
sql-reference-018	sql-reference.pdf	18	to incorporate that logic into your larger queries. Syntax: SELECT column1, column2, ..., columnn FROM table_valued_function(parameters) Parameters: table_valued_function(parameters The call to the Table Valued Function, which may require one or more input parameters. The result set returned by the TVF is treated as a temporary table that can be further queried, joined, or ﬁltered in the main query. In the following example, the GetCustomerOrders function is a Table Valued Function that takes a customer_id parameter and returns a table-like result set containing the orders for that customer. SELECT * FROM dbo.GetCustomerOrders(10) JOIN clause A SQL JOIN clause is used to combine the data from two or more tables based on common ﬁelds. The results might or might not change depending on the join method speciﬁed. Left and right outer joins retain values from one of the joined tables when no match is found in the other table. The combination of the JOIN type and the join condition determines which rows are included in the ﬁnal result set. The SELECT and WHERE clauses then control which columns are returned and how the rows are ﬁltered. Understanding the diﬀerent JOIN types and how to use them eﬀectively is SELECT 58 AWS Clean Rooms SQL Reference a crucial skill in SQL, because it allows you to combine data from multiple tables in a ﬂexible and powerful way. Syntax SELECT column1, column2, ..., columnn FROM table1 join_type table2 ON table1.column = table2.column; Parameters SELECT column1, column2, ..., columnN The columns you want to include in the result set. You can select columns from either or both of the tables involved in the JOIN. FROM table1 The ﬁrst (left) table in the JOIN operation. [JOIN \| INNER JOIN \| LEFT [OUTER] JOIN \| RIGHT [OUTER] JOIN \| FULL [OUTER] JOIN] table2: The type of JOIN to be performed. JOIN or INNER JOIN returns only the rows with matching values in both tables. LEFT [OUTER] JOIN returns all rows from the left table, with matching rows from the right table. RIGHT [OUTER] JOIN returns all rows from the right table, with matching rows from the left table. FULL [OUTER] JOIN returns all rows from both tables, regardless of whether there is a match or not. CROSS JOIN creates a Cartesian product of the rows from the two tables. ON table1.column = table2.column The join condition, which speciﬁes how the rows in the two tables are matched. The join condition can be based on one or more columns. SELECT 59 AWS Clean Rooms WHERE condition: SQL Reference An optional clause that can be used to ﬁlter the result set further, based on a speciﬁed condition. Example The following example is a join between two tables with the USING clause. In this case, the columns listid and eventid are used as the join columns. The results are limited to ﬁve rows. select listid, listing.sellerid, eventid, listing.dateid, numtickets from listing join sales using (listid, eventid) order by 1 limit 5; listid \| sellerid \| eventid \| dateid \| numtickets -------+----------+---------+--------+----------- 1 \| 36861 \| 7872 \| 1850 \| 10 4 \| 8117 \| 4337 \| 1970 \| 8 5 \| 1616 \| 8647 \| 1963 \| 4 5 \| 1616 \| 8647 \| 1963 \| 4 6 \| 47402 \| 8240 \| 2053 \| 18 Join types INNER This is the default join type. Returns the rows that have matching values in both table references. The INNER JOIN is the most common type of join used in SQL. It's a powerful way to combine data from multiple tables based on a common column or set of columns. Syntax: SELECT column1, column2, ..., columnn FROM table1 INNER JOIN table2 ON table1.column = table2.column; SELECT 60 AWS Clean Rooms SQL Reference The following query will return all the rows where there is a matching customer_id value between the customers and orders tables. The result set will contain the customer_id, name, order_id, and order_date columns. SELECT customers.customer_id, customers.name, orders.order_id, orders.order_date FROM customers INNER JOIN orders ON customers.customer_id = orders.customer_id; The following query is an inner join (without the JOIN keyword) between the LISTING table and SALES table, where the LISTID from the LISTING table is between 1 and 5. This query matches LISTID column values in the LISTING table (the left table) and SALES table (the right table). The results show that LISTID 1, 4, and 5 match the criteria. select listing.listid, sum(pricepaid) as price, sum(commission) as comm from listing, sales where listing.listid = sales.listid and listing.listid between 1 and 5 group by 1 order by 1; listid \| price \| comm -------+--------+-------- 1 \| 728.00 \| 109.20 4 \| 76.00 \| 11.40 5 \| 525.00 \| 78.75 The following example is an inner join with the ON clause. In this case, NULL rows are not returned. select listing.listid, sum(pricepaid) as price,
sql-reference-019	sql-reference.pdf	19	LISTID column values in the LISTING table (the left table) and SALES table (the right table). The results show that LISTID 1, 4, and 5 match the criteria. select listing.listid, sum(pricepaid) as price, sum(commission) as comm from listing, sales where listing.listid = sales.listid and listing.listid between 1 and 5 group by 1 order by 1; listid \| price \| comm -------+--------+-------- 1 \| 728.00 \| 109.20 4 \| 76.00 \| 11.40 5 \| 525.00 \| 78.75 The following example is an inner join with the ON clause. In this case, NULL rows are not returned. select listing.listid, sum(pricepaid) as price, sum(commission) as comm from sales join listing on sales.listid=listing.listid and sales.eventid=listing.eventid where listing.listid between 1 and 5 group by 1 order by 1; listid \| price \| comm -------+--------+-------- 1 \| 728.00 \| 109.20 4 \| 76.00 \| 11.40 5 \| 525.00 \| 78.75 SELECT 61 AWS Clean Rooms SQL Reference The following query is an inner join of two subqueries in the FROM clause. The query ﬁnds the number of sold and unsold tickets for diﬀerent categories of events (concerts and shows). The FROM clause subqueries are table subqueries; they can return multiple columns and rows. select catgroup1, sold, unsold from (select catgroup, sum(qtysold) as sold from category c, event e, sales s where c.catid = e.catid and e.eventid = s.eventid group by catgroup) as a(catgroup1, sold) join (select catgroup, sum(numtickets)-sum(qtysold) as unsold from category c, event e, sales s, listing l where c.catid = e.catid and e.eventid = s.eventid and s.listid = l.listid group by catgroup) as b(catgroup2, unsold) on a.catgroup1 = b.catgroup2 order by 1; catgroup1 \| sold \| unsold ----------+--------+-------- Concerts \| 195444 \|1067199 Shows \| 149905 \| 817736 LEFT [ OUTER ] Returns all values from the left table reference and the matched values from the right table reference, or appends NULL if there is no match. It's also referred to as a left outer join. It returns all the rows from the left (ﬁrst) table, and the matching rows from the right (second) table. If there is no match in the right table, the result set will contain NULL values for the columns from the right table. The OUTER keyword can be omitted, and the join can be written as simply LEFT JOIN. The opposite of a LEFT OUTER JOIN is a RIGHT OUTER JOIN, which returns all the rows from the right table and the matching rows from the left table. Syntax: SELECT column1, column2, ..., columnn FROM table1 LEFT [OUTER] JOIN table2 SELECT 62 AWS Clean Rooms SQL Reference ON table1.column = table2.column; The following query will return all the rows from the customers table, along with the matching rows from the orders table. If a customer has no orders, the result set will still include that customer's information, with NULL values for the order_id and order_date columns. SELECT customers.customer_id, customers.name, orders.order_id, orders.order_date FROM customers LEFT OUTER JOIN orders ON customers.customer_id = orders.customer_id; The following query is a left outer join. Left and right outer joins retain values from one of the joined tables when no match is found in the other table. The left and right tables are the ﬁrst and second tables listed in the syntax. NULL values are used to ﬁll the "gaps" in the result set. This query matches LISTID column values in the LISTING table (the left table) and the SALES table (the right table). The results show that LISTIDs 2 and 3 didn't result in any sales. select listing.listid, sum(pricepaid) as price, sum(commission) as comm from listing left outer join sales on sales.listid = listing.listid where listing.listid between 1 and 5 group by 1 order by 1; listid \| price \| comm -------+--------+-------- 1 \| 728.00 \| 109.20 2 \| NULL \| NULL 3 \| NULL \| NULL 4 \| 76.00 \| 11.40 5 \| 525.00 \| 78.75 RIGHT [ OUTER ] Returns all values from the right table reference and the matched values from the left table reference, or appends NULL if there is no match. It's also referred to as a right outer join. It returns all the rows from the right (second) table, and the matching rows from the left (ﬁrst) table. If there is no match in the left table, the result set will contain NULL values for the columns from the left table. The OUTER keyword can be omitted, and the join can be written as simply RIGHT JOIN. The opposite of a RIGHT OUTER JOIN is a LEFT OUTER JOIN, which returns all the rows from the left table and the matching rows from the right table. SELECT 63 AWS Clean Rooms Syntax: SELECT column1, column2, ..., columnn FROM table1 RIGHT [OUTER] JOIN table2 ON table1.column = table2.column; SQL Reference The following query will return all the rows from the
sql-reference-020	sql-reference.pdf	20	there is no match in the left table, the result set will contain NULL values for the columns from the left table. The OUTER keyword can be omitted, and the join can be written as simply RIGHT JOIN. The opposite of a RIGHT OUTER JOIN is a LEFT OUTER JOIN, which returns all the rows from the left table and the matching rows from the right table. SELECT 63 AWS Clean Rooms Syntax: SELECT column1, column2, ..., columnn FROM table1 RIGHT [OUTER] JOIN table2 ON table1.column = table2.column; SQL Reference The following query will return all the rows from the customers table, along with the matching rows from the orders table. If a customer has no orders, the result set will still include that customer's information, with NULL values for the order_id and order_date columns. SELECT orders.order_id, orders.order_date, customers.customer_id, customers.name FROM orders RIGHT OUTER JOIN customers ON orders.customer_id = customers.customer_id; The following query is a right outer join. This query matches LISTID column values in the LISTING table (the left table) and the SALES table (the right table). The results show that LISTIDs 1, 4, and 5 match the criteria. select listing.listid, sum(pricepaid) as price, sum(commission) as comm from listing right outer join sales on sales.listid = listing.listid where listing.listid between 1 and 5 group by 1 order by 1; listid \| price \| comm -------+--------+-------- 1 \| 728.00 \| 109.20 4 \| 76.00 \| 11.40 5 \| 525.00 \| 78.75 FULL [OUTER] Returns all values from both relations, appending NULL values on the side that doesn't have a match. It's also referred to as a full outer join. It returns all the rows from both the left and right tables, regardless of whether there is a match or not. If there is no match, the result set will contain NULL values for the columns from the table that doesn't have a matching row. The OUTER keyword can be omitted, and the join can be written SELECT 64 AWS Clean Rooms SQL Reference as simply FULL JOIN. The FULL OUTER JOIN is less commonly used than the LEFT OUTER JOIN or RIGHT OUTER JOIN, but it can be useful in certain scenarios where you need to see all the data from both tables, even if there are no matches. Syntax: SELECT column1, column2, ..., columnn FROM table1 FULL [OUTER] JOIN table2 ON table1.column = table2.column; The following query will return all the rows from both the customers and orders tables. If a customer has no orders, the result set will still include that customer's information, with NULL values for the order_id and order_date columns. If an order has no associated customer, the result set will include that order, with NULL values for the customer_id and name columns. SELECT customers.customer_id, customers.name, orders.order_id, orders.order_date FROM customers FULL OUTER JOIN orders ON customers.customer_id = orders.customer_id; The following query is a full join. Full joins retain values from the joined tables when no match is found in the other table. The left and right tables are the ﬁrst and second tables listed in the syntax. NULL values are used to ﬁll the "gaps" in the result set. This query matches LISTID column values in the LISTING table (the left table) and the SALES table (the right table). The results show that LISTIDs 2 and 3 didn't result in any sales. select listing.listid, sum(pricepaid) as price, sum(commission) as comm from listing full join sales on sales.listid = listing.listid where listing.listid between 1 and 5 group by 1 order by 1; listid \| price \| comm -------+--------+-------- 1 \| 728.00 \| 109.20 2 \| NULL \| NULL 3 \| NULL \| NULL 4 \| 76.00 \| 11.40 5 \| 525.00 \| 78.75 SELECT 65 AWS Clean Rooms SQL Reference The following query is a full join. This query matches LISTID column values in the LISTING table (the left table) and the SALES table (the right table). Only rows that do not result in any sales (LISTIDs 2 and 3) are in the results. select listing.listid, sum(pricepaid) as price, sum(commission) as comm from listing full join sales on sales.listid = listing.listid where listing.listid between 1 and 5 and (listing.listid IS NULL or sales.listid IS NULL) group by 1 order by 1; listid \| price \| comm -------+--------+-------- 2 \| NULL \| NULL 3 \| NULL \| NULL [ LEFT ] SEMI Returns values from the left side of the table reference that has a match with the right. It's also referred to as a left semi join. It returns only the rows from the left (ﬁrst) table that have a matching row in the right (second) table. It does not return any columns from the right table - only the columns from the left table. The LEFT SEMI JOIN is useful when you want to ﬁnd the
sql-reference-021	sql-reference.pdf	21	1 order by 1; listid \| price \| comm -------+--------+-------- 2 \| NULL \| NULL 3 \| NULL \| NULL [ LEFT ] SEMI Returns values from the left side of the table reference that has a match with the right. It's also referred to as a left semi join. It returns only the rows from the left (ﬁrst) table that have a matching row in the right (second) table. It does not return any columns from the right table - only the columns from the left table. The LEFT SEMI JOIN is useful when you want to ﬁnd the rows in one table that have a match in another table, without needing to return any data from the second table. The LEFT SEMI JOIN is a more eﬃcient alternative to using a subquery with an IN or EXISTS clause. Syntax: SELECT column1, column2, ..., columnn FROM table1 LEFT SEMI JOIN table2 ON table1.column = table2.column; The following query will return only the customer_id and name columns from the customers table, for the customers who have at least one order in the orders table. The result set won't include any columns from the orders table. SELECT customers.customer_id, customers.name FROM customers LEFT SEMI JOIN orders SELECT 66 AWS Clean Rooms SQL Reference ON customers.customer_id = orders.customer_id; CROSS JOIN Returns the Cartesian product of two relations. This means that the result set will contain all possible combinations of rows from the two tables, without any condition or ﬁlter applied. The CROSS JOIN is useful when you need to generate all possible combinations of data from two tables, such as in the case of creating a report that displays all possible combinations of customer and product information. The CROSS JOIN is diﬀerent from other join types (INNER JOIN, LEFT JOIN, etc.) because it doesn't have a join condition in the ON clause. The join condition isn't required for a CROSS JOIN. Syntax: SELECT column1, column2, ..., columnn FROM table1 CROSS JOIN table2; The following query will return a result set that contains all possible combinations of customer_id, customer_name, product_id, and product_name from the customers and products tables. If the customers table has 10 rows and the products table has 20 rows, the result set of the CROSS JOIN will contain 10 x 20 = 200 rows. SELECT customers.customer_id, customers.name, products.product_id, products.product_name FROM customers CROSS JOIN products; The following query is a cross join or Cartesian join of the LISTING table and the SALES table with a predicate to limit the results. This query matches LISTID column values in the SALES table and the LISTING table for LISTIDs 1, 2, 3, 4, and 5 in both tables. The results show that 20 rows match the criteria. select sales.listid as sales_listid, listing.listid as listing_listid from sales cross join listing where sales.listid between 1 and 5 and listing.listid between 1 and 5 order by 1,2; SELECT 67 AWS Clean Rooms SQL Reference sales_listid \| listing_listid -------------+--------------- 1 \| 1 1 \| 2 1 \| 3 1 \| 4 1 \| 5 4 \| 1 4 \| 2 4 \| 3 4 \| 4 4 \| 5 5 \| 1 5 \| 1 5 \| 2 5 \| 2 5 \| 3 5 \| 3 5 \| 4 5 \| 4 5 \| 5 5 \| 5 ANTI JOIN Returns the values from the left table reference that have no match with the right table reference. It's also referred to as a left anti join. The ANTI JOIN is a useful operation when you want to ﬁnd the rows in one table that don't have a match in another table. Syntax: SELECT column1, column2, ..., columnn FROM table1 LEFT ANTI JOIN table2 ON table1.column = table2.column; The following query will return all the customers who haven't placed any orders. SELECT customers.customer_id, customers.name FROM customers SELECT 68 AWS Clean Rooms LEFT ANTI JOIN orders ON customers.customer_id = orders.customer_id WHERE orders.order_id IS NULL; NATURAL SQL Reference Speciﬁes that the rows from the two relations will implicitly be matched on equality for all columns with matching names. It automatically matches columns with the same name and data type between the two tables. It doesn't require you to explicitly specify the join condition in the ON clause. It combines all the matching columns between the two tables into the result set. The NATURAL JOIN is a convenient shorthand when the tables you're joining have columns with the same names and data types. However, it's generally recommended to use the more explicit INNER JOIN ... ON syntax to make the join conditions more explicit and easier to understand. Syntax: SELECT column1, column2, ..., columnn FROM table1 NATURAL JOIN table2; The following example is a natural join between two tables, employees and departments, with the following columns: • employees table: employee_id, first_name, last_name,
sql-reference-022	sql-reference.pdf	22	join condition in the ON clause. It combines all the matching columns between the two tables into the result set. The NATURAL JOIN is a convenient shorthand when the tables you're joining have columns with the same names and data types. However, it's generally recommended to use the more explicit INNER JOIN ... ON syntax to make the join conditions more explicit and easier to understand. Syntax: SELECT column1, column2, ..., columnn FROM table1 NATURAL JOIN table2; The following example is a natural join between two tables, employees and departments, with the following columns: • employees table: employee_id, first_name, last_name, department_id • departments table: department_id, department_name The following query will return a result set that includes the ﬁrst name, last name, and department name for all matching rows between the two tables, based on the department_id column. SELECT e.first_name, e.last_name, d.department_name FROM employees e NATURAL JOIN departments d; The following example is a natural join between two tables. In this case, the columns listid, sellerid, eventid, and dateid have identical names and data types in both tables and so are used as the join columns. The results are limited to ﬁve rows. SELECT 69 AWS Clean Rooms SQL Reference select listid, sellerid, eventid, dateid, numtickets from listing natural join sales order by 1 limit 5; listid \| sellerid \| eventid \| dateid \| numtickets -------+-----------+---------+--------+----------- 113 \| 29704 \| 4699 \| 2075 \| 22 115 \| 39115 \| 3513 \| 2062 \| 14 116 \| 43314 \| 8675 \| 1910 \| 28 118 \| 6079 \| 1611 \| 1862 \| 9 163 \| 24880 \| 8253 \| 1888 \| 14 WHERE clause The WHERE clause contains conditions that either join tables or apply predicates to columns in tables. Tables can be inner-joined by using appropriate syntax in either the WHERE clause or the FROM clause. Outer join criteria must be speciﬁed in the FROM clause. Syntax [ WHERE condition ] condition Any search condition with a Boolean result, such as a join condition or a predicate on a table column. The following examples are valid join conditions: sales.listid=listing.listid sales.listid<>listing.listid The following examples are valid conditions on columns in tables: catgroup like 'S%' venueseats between 20000 and 50000 eventname in('Jersey Boys','Spamalot') year=2008 length(catdesc)>25 date_part(month, caldate)=6 SELECT 70 AWS Clean Rooms SQL Reference Conditions can be simple or complex; for complex conditions, you can use parentheses to isolate logical units. In the following example, the join condition is enclosed by parentheses. where (category.catid=event.catid) and category.catid in(6,7,8) Usage notes You can use aliases in the WHERE clause to reference select list expressions. You can't restrict the results of aggregate functions in the WHERE clause; use the HAVING clause for this purpose. Columns that are restricted in the WHERE clause must derive from table references in the FROM clause. Example The following query uses a combination of diﬀerent WHERE clause restrictions, including a join condition for the SALES and EVENT tables, a predicate on the EVENTNAME column, and two predicates on the STARTTIME column. select eventname, starttime, pricepaid/qtysold as costperticket, qtysold from sales, event where sales.eventid = event.eventid and eventname='Hannah Montana' and date_part(quarter, starttime) in(1,2) and date_part(year, starttime) = 2008 order by 3 desc, 4, 2, 1 limit 10; eventname \| starttime \| costperticket \| qtysold ----------------+---------------------+-------------------+--------- Hannah Montana \| 2008-06-07 14:00:00 \| 1706.00000000 \| 2 Hannah Montana \| 2008-05-01 19:00:00 \| 1658.00000000 \| 2 Hannah Montana \| 2008-06-07 14:00:00 \| 1479.00000000 \| 1 Hannah Montana \| 2008-06-07 14:00:00 \| 1479.00000000 \| 3 Hannah Montana \| 2008-06-07 14:00:00 \| 1163.00000000 \| 1 Hannah Montana \| 2008-06-07 14:00:00 \| 1163.00000000 \| 2 Hannah Montana \| 2008-06-07 14:00:00 \| 1163.00000000 \| 4 Hannah Montana \| 2008-05-01 19:00:00 \| 497.00000000 \| 1 Hannah Montana \| 2008-05-01 19:00:00 \| 497.00000000 \| 2 Hannah Montana \| 2008-05-01 19:00:00 \| 497.00000000 \| 4 (10 rows) SELECT 71 AWS Clean Rooms VALUES clause SQL Reference The VALUES clause is used to provide a set of row values directly in the query, without the need to reference a table. The VALUES clause can be used in the following scenarios: • You can use the VALUES clause in an INSERT INTO statement to specify the values for the new rows being inserted into a table. • You can use the VALUES clause on its own to create a temporary result set, or inline table, without the need to reference a table. • You can combine the VALUES clause with other SQL clauses, such as WHERE, ORDER BY, or LIMIT, to ﬁlter, sort, or limit the rows in the result set. This clause is particularly useful when you need to insert, query, or manipulate a small set of data directly in your SQL statement, without the need to create or reference a permanent table. It allows you to deﬁne the column names and the
sql-reference-023	sql-reference.pdf	23	a table. • You can use the VALUES clause on its own to create a temporary result set, or inline table, without the need to reference a table. • You can combine the VALUES clause with other SQL clauses, such as WHERE, ORDER BY, or LIMIT, to ﬁlter, sort, or limit the rows in the result set. This clause is particularly useful when you need to insert, query, or manipulate a small set of data directly in your SQL statement, without the need to create or reference a permanent table. It allows you to deﬁne the column names and the corresponding values for each row, giving you the ﬂexibility to create temporary result sets or insert data on the ﬂy, without the overhead of managing a separate table. Syntax VALUES ( expression [ , ... ] ) [ table_alias ] Parameters expression An expression that speciﬁes a combination of one or more values, operators and SQL functions that results in a value. table_alias An alias that speciﬁes a temporary name with an optional column name list. Example The following example creates an inline table, temporary table-like result set with two columns, col1 and col2. The single row in the result set contains the values "one" and 1, respectively. The SELECT 72 AWS Clean Rooms SQL Reference SELECT * FROM part of the query simply retrieves all the columns and rows from this temporary result set. The column names (col1 and col2) are automatically generated by the database system, because the VALUES clause doesn't explicitly specify the column names. SELECT * FROM VALUES ("one", 1); +----+----+ \|col1\|col2\| +----+----+ \| one\| 1\| +----+----+ If you want to deﬁne custom column names, you can do so by using an AS clause after the VALUES clause, like this: SELECT * FROM (VALUES ("one", 1)) AS my_table (name, id); +------+----+ \| name \| id \| +------+----+ \| one \| 1 \| +------+----+ This would create a temporary result set with the column names name and id, instead of the default col1 and col2. GROUP BY clause The GROUP BY clause identiﬁes the grouping columns for the query. Grouping columns must be declared when the query computes aggregates with standard functions such as SUM, AVG, and COUNT. If an aggregate function is present in the SELECT expression, any column in the SELECT expression that is not in an aggregate function must be in the GROUP BY clause. For more information, see AWS Clean Rooms SQL functions. Syntax GROUP BY group_by_clause [, ...] group_by_clause := { expr \| ROLLUP ( expr [, ...] ) \| } SELECT 73 AWS Clean Rooms Parameters expr SQL Reference The list of columns or expressions must match the list of non-aggregate expressions in the select list of the query. For example, consider the following simple query. select listid, eventid, sum(pricepaid) as revenue, count(qtysold) as numtix from sales group by listid, eventid order by 3, 4, 2, 1 limit 5; listid \| eventid \| revenue \| numtix -------+---------+---------+-------- 89397 \| 47 \| 20.00 \| 1 106590 \| 76 \| 20.00 \| 1 124683 \| 393 \| 20.00 \| 1 103037 \| 403 \| 20.00 \| 1 147685 \| 429 \| 20.00 \| 1 (5 rows) In this query, the select list consists of two aggregate expressions. The ﬁrst uses the SUM function and the second uses the COUNT function. The remaining two columns, LISTID and EVENTID, must be declared as grouping columns. Expressions in the GROUP BY clause can also reference the select list by using ordinal numbers. For example, the previous example could be abbreviated as follows. select listid, eventid, sum(pricepaid) as revenue, count(qtysold) as numtix from sales group by 1,2 order by 3, 4, 2, 1 limit 5; listid \| eventid \| revenue \| numtix -------+---------+---------+-------- 89397 \| 47 \| 20.00 \| 1 106590 \| 76 \| 20.00 \| 1 124683 \| 393 \| 20.00 \| 1 SELECT 74 AWS Clean Rooms SQL Reference 103037 \| 403 \| 20.00 \| 1 147685 \| 429 \| 20.00 \| 1 (5 rows) ROLLUP You can use the aggregation extension ROLLUP to perform the work of multiple GROUP BY operations in a single statement. For more information on aggregation extensions and related functions, see Aggregation extensions. Aggregation extensions AWS Clean Rooms supports aggregation extensions to do the work of multiple GROUP BY operations in a single statement. GROUPING SETS Computes one or more grouping sets in a single statement. A grouping set is the set of a single GROUP BY clause, a set of 0 or more columns by which you can group a query's result set. GROUP BY GROUPING SETS is equivalent to running a UNION ALL query on one result set grouped by diﬀerent columns. For example, GROUP BY GROUPING SETS((a), (b)) is equivalent to GROUP BY a UNION ALL GROUP
sql-reference-024	sql-reference.pdf	24	see Aggregation extensions. Aggregation extensions AWS Clean Rooms supports aggregation extensions to do the work of multiple GROUP BY operations in a single statement. GROUPING SETS Computes one or more grouping sets in a single statement. A grouping set is the set of a single GROUP BY clause, a set of 0 or more columns by which you can group a query's result set. GROUP BY GROUPING SETS is equivalent to running a UNION ALL query on one result set grouped by diﬀerent columns. For example, GROUP BY GROUPING SETS((a), (b)) is equivalent to GROUP BY a UNION ALL GROUP BY b. The following example returns the cost of the order table's products grouped according to both the products' categories and the kind of products sold. SELECT category, product, sum(cost) as total FROM orders GROUP BY GROUPING SETS(category, product); category \| product \| total ----------------------+----------------------+------- computers \| \| 2100 cellphones \| \| 1610 \| laptop \| 2050 \| smartphone \| 1610 \| mouse \| 50 (5 rows) SELECT 75 AWS Clean Rooms ROLLUP SQL Reference Assumes a hierarchy where preceding columns are considered the parents of subsequent columns. ROLLUP groups data by the provided columns, returning extra subtotal rows representing the totals throughout all levels of grouping columns, in addition to the grouped rows. For example, you can use GROUP BY ROLLUP((a), (b)) to return a result set grouped ﬁrst by a, then by b while assuming that b is a subsection of a. ROLLUP also returns a row with the whole result set without grouping columns. GROUP BY ROLLUP((a), (b)) is equivalent to GROUP BY GROUPING SETS((a,b), (a), ()). The following example returns the cost of the order table's products grouped ﬁrst by category and then product, with product as a subdivision of category. SELECT category, product, sum(cost) as total FROM orders GROUP BY ROLLUP(category, product) ORDER BY 1,2; category \| product \| total ----------------------+----------------------+------- cellphones \| smartphone \| 1610 cellphones \| \| 1610 computers \| laptop \| 2050 computers \| mouse \| 50 computers \| \| 2100 \| \| 3710 (6 rows) CUBE Groups data by the provided columns, returning extra subtotal rows representing the totals throughout all levels of grouping columns, in addition to the grouped rows. CUBE returns the same rows as ROLLUP, while adding additional subtotal rows for every combination of grouping column not covered by ROLLUP. For example, you can use GROUP BY CUBE ((a), (b)) to return a result set grouped ﬁrst by a, then by b while assuming that b is a subsection of a, then by b alone. CUBE also returns a row with the whole result set without grouping columns. GROUP BY CUBE((a), (b)) is equivalent to GROUP BY GROUPING SETS((a, b), (a), (b), ()). The following example returns the cost of the order table's products grouped ﬁrst by category and then product, with product as a subdivision of category. Unlike the preceding example for ROLLUP, the statement returns results for every combination of grouping column. SELECT 76 AWS Clean Rooms SQL Reference SELECT category, product, sum(cost) as total FROM orders GROUP BY CUBE(category, product) ORDER BY 1,2; category \| product \| total ----------------------+----------------------+------- cellphones \| smartphone \| 1610 cellphones \| \| 1610 computers \| laptop \| 2050 computers \| mouse \| 50 computers \| \| 2100 \| laptop \| 2050 \| mouse \| 50 \| smartphone \| 1610 \| \| 3710 (9 rows) HAVING clause The HAVING clause applies a condition to the intermediate grouped result set that a query returns. Syntax [ HAVING condition ] For example, you can restrict the results of a SUM function: having sum(pricepaid) >10000 The HAVING condition is applied after all WHERE clause conditions are applied and GROUP BY operations are completed. The condition itself takes the same form as any WHERE clause condition. Usage notes • Any column that is referenced in a HAVING clause condition must be either a grouping column or a column that refers to the result of an aggregate function. • In a HAVING clause, you can't specify: • An ordinal number that refers to a select list item. Only the GROUP BY and ORDER BY clauses accept ordinal numbers. SELECT 77 AWS Clean Rooms Examples SQL Reference The following query calculates total ticket sales for all events by name, then eliminates events where the total sales were less than $800,000. The HAVING condition is applied to the results of the aggregate function in the select list: sum(pricepaid). select eventname, sum(pricepaid) from sales join event on sales.eventid = event.eventid group by 1 having sum(pricepaid) > 800000 order by 2 desc, 1; eventname \| sum ------------------+----------- Mamma Mia! \| 1135454.00 Spring Awakening \| 972855.00 The Country Girl \| 910563.00 Macbeth \| 862580.00 Jersey Boys \| 811877.00 Legally Blonde \| 804583.00 (6 rows) The following query calculates a similar result
sql-reference-025	sql-reference.pdf	25	The following query calculates total ticket sales for all events by name, then eliminates events where the total sales were less than $800,000. The HAVING condition is applied to the results of the aggregate function in the select list: sum(pricepaid). select eventname, sum(pricepaid) from sales join event on sales.eventid = event.eventid group by 1 having sum(pricepaid) > 800000 order by 2 desc, 1; eventname \| sum ------------------+----------- Mamma Mia! \| 1135454.00 Spring Awakening \| 972855.00 The Country Girl \| 910563.00 Macbeth \| 862580.00 Jersey Boys \| 811877.00 Legally Blonde \| 804583.00 (6 rows) The following query calculates a similar result set. In this case, however, the HAVING condition is applied to an aggregate that isn't speciﬁed in the select list: sum(qtysold). Events that did not sell more than 2,000 tickets are eliminated from the ﬁnal result. select eventname, sum(pricepaid) from sales join event on sales.eventid = event.eventid group by 1 having sum(qtysold) >2000 order by 2 desc, 1; eventname \| sum ------------------+----------- Mamma Mia! \| 1135454.00 Spring Awakening \| 972855.00 The Country Girl \| 910563.00 Macbeth \| 862580.00 Jersey Boys \| 811877.00 Legally Blonde \| 804583.00 Chicago \| 790993.00 Spamalot \| 714307.00 SELECT 78 AWS Clean Rooms (8 rows) Set operators SQL Reference The set operators are used to compare and merge the results of two separate query expressions. AWS Clean Rooms Spark SQL supports the following set operators listed in the following table. Set operator INTERSECT INTERSECT ALL EXCEPT EXCEPT ALL UNION UNION ALL For example, if you want to know which users of a website are both buyers and sellers but their user names are stored in separate columns or tables, you can ﬁnd the intersection of these two types of users. If you want to know which website users are buyers but not sellers, you can use the EXCEPT operator to ﬁnd the diﬀerence between the two lists of users. If you want to build a list of all users, regardless of role, you can use the UNION operator. Note The ORDER BY, LIMIT, SELECT TOP, and OFFSET clauses can't be used in the query expressions merged by the UNION, UNION ALL, INTERSECT, and EXCEPT set operators. Topics • Syntax • Parameters • Order of evaluation for set operators SELECT 79 SQL Reference AWS Clean Rooms • Usage notes • Example UNION queries • Example UNION ALL query • Example INTERSECT queries • Example EXCEPT query Syntax subquery1 { { UNION [ ALL \| DISTINCT ] \| INTERSECT [ ALL \| DISTINCT ] \| EXCEPT [ ALL \| DISTINCT ] } subquery2 } [...] } Parameters subquery1, subquery2 A query expression that corresponds, in the form of its select list, to a second query expression that follows the UNION, UNION ALL, INTERSECT, INTERSECT ALL, EXCEPT, or EXCEPT ALL operator. The two expressions must contain the same number of output columns with compatible data types; otherwise, the two result sets can't be compared and merged. Set operations don't allow implicit conversion between diﬀerent categories of data types. For more information, see Type compatibility and conversion. You can build queries that contain an unlimited number of query expressions and link them with UNION, INTERSECT, and EXCEPT operators in any combination. For example, the following query structure is valid, assuming that the tables T1, T2, and T3 contain compatible sets of columns: select * from t1 union select * from t2 except select * from t3 UNION [ALL \| DISTINCT] Set operation that returns rows from two query expressions, regardless of whether the rows derive from one or both expressions. SELECT 80 AWS Clean Rooms INTERSECT [ALL \| DISTINCT] SQL Reference Set operation that returns rows that derive from two query expressions. Rows that aren't returned by both expressions are discarded. EXCEPT [ALL \| DISTINCT] Set operation that returns rows that derive from one of two query expressions. To qualify for the result, rows must exist in the ﬁrst result table but not the second. EXCEPT ALL doesn't remove duplicates from the result rows. MINUS and EXCEPT are exact synonyms. Order of evaluation for set operators The UNION and EXCEPT set operators are left-associative. If parentheses aren't speciﬁed to inﬂuence the order of precedence, a combination of these set operators is evaluated from left to right. For example, in the following query, the UNION of T1 and T2 is evaluated ﬁrst, then the EXCEPT operation is performed on the UNION result: select * from t1 union select * from t2 except select * from t3 The INTERSECT operator takes precedence over the UNION and EXCEPT operators when a combination of operators is used in the same query. For example, the following query evaluates the intersection of T2 and T3, then union the result with T1: select * from t1 union select * from t2 intersect select * from t3 By
sql-reference-026	sql-reference.pdf	26	is evaluated from left to right. For example, in the following query, the UNION of T1 and T2 is evaluated ﬁrst, then the EXCEPT operation is performed on the UNION result: select * from t1 union select * from t2 except select * from t3 The INTERSECT operator takes precedence over the UNION and EXCEPT operators when a combination of operators is used in the same query. For example, the following query evaluates the intersection of T2 and T3, then union the result with T1: select * from t1 union select * from t2 intersect select * from t3 By adding parentheses, you can enforce a diﬀerent order of evaluation. In the following case, the result of the union of T1 and T2 is intersected with T3, and the query is likely to produce a diﬀerent result. SELECT 81 AWS Clean Rooms SQL Reference (select * from t1 union select * from t2) intersect (select * from t3) Usage notes • The column names returned in the result of a set operation query are the column names (or aliases) from the tables in the ﬁrst query expression. Because these column names are potentially misleading, in that the values in the column derive from tables on either side of the set operator, you might want to provide meaningful aliases for the result set. • When set operator queries return decimal results, the corresponding result columns are promoted to return the same precision and scale. For example, in the following query, where T1.REVENUE is a DECIMAL(10,2) column and T2.REVENUE is a DECIMAL(8,4) column, the decimal result is promoted to DECIMAL(12,4): select t1.revenue union select t2.revenue; The scale is 4 because that is the maximum scale of the two columns. The precision is 12 because T1.REVENUE requires 8 digits to the left of the decimal point (12 - 4 = 8). This type promotion ensures that all values from both sides of the UNION ﬁt in the result. For 64-bit values, the maximum result precision is 19 and the maximum result scale is 18. For 128-bit values, the maximum result precision is 38 and the maximum result scale is 37. If the resulting data type exceeds AWS Clean Rooms precision and scale limits, the query returns an error. • For set operations, two rows are treated as identical if, for each corresponding pair of columns, the two data values are either equal or both NULL. For example, if tables T1 and T2 both contain one column and one row, and that row is NULL in both tables, an INTERSECT operation over those tables returns that row. Example UNION queries In the following UNION query, rows in the SALES table are merged with rows in the LISTING table. Three compatible columns are selected from each table; in this case, the corresponding columns have the same names and data types. SELECT 82 AWS Clean Rooms SQL Reference select listid, sellerid, eventid from listing union select listid, sellerid, eventid from sales listid \| sellerid \| eventid --------+----------+--------- 1 \| 36861 \| 7872 2 \| 16002 \| 4806 3 \| 21461 \| 4256 4 \| 8117 \| 4337 5 \| 1616 \| 8647 The following example shows how you can add a literal value to the output of a UNION query so you can see which query expression produced each row in the result set. The query identiﬁes rows from the ﬁrst query expression as "B" (for buyers) and rows from the second query expression as "S" (for sellers). The query identiﬁes buyers and sellers for ticket transactions that cost $10,000 or more. The only diﬀerence between the two query expressions on either side of the UNION operator is the joining column for the SALES table. select listid, lastname, firstname, username, pricepaid as price, 'S' as buyorsell from sales, users where sales.sellerid=users.userid and pricepaid >=10000 union select listid, lastname, firstname, username, pricepaid, 'B' as buyorsell from sales, users where sales.buyerid=users.userid and pricepaid >=10000 listid \| lastname \| firstname \| username \| price \| buyorsell --------+----------+-----------+----------+-----------+----------- 209658 \| Lamb \| Colette \| VOR15LYI \| 10000.00 \| B 209658 \| West \| Kato \| ELU81XAA \| 10000.00 \| S 212395 \| Greer \| Harlan \| GXO71KOC \| 12624.00 \| S 212395 \| Perry \| Cora \| YWR73YNZ \| 12624.00 \| B 215156 \| Banks \| Patrick \| ZNQ69CLT \| 10000.00 \| S 215156 \| Hayden \| Malachi \| BBG56AKU \| 10000.00 \| B SELECT 83 AWS Clean Rooms SQL Reference The following example uses a UNION ALL operator because duplicate rows, if found, need to be retained in the result. For a speciﬁc series of event IDs, the query returns 0 or more rows for each sale associated with each event, and 0 or 1 row for each listing of that event. Event IDs are unique to each row
sql-reference-027	sql-reference.pdf	27	S 212395 \| Perry \| Cora \| YWR73YNZ \| 12624.00 \| B 215156 \| Banks \| Patrick \| ZNQ69CLT \| 10000.00 \| S 215156 \| Hayden \| Malachi \| BBG56AKU \| 10000.00 \| B SELECT 83 AWS Clean Rooms SQL Reference The following example uses a UNION ALL operator because duplicate rows, if found, need to be retained in the result. For a speciﬁc series of event IDs, the query returns 0 or more rows for each sale associated with each event, and 0 or 1 row for each listing of that event. Event IDs are unique to each row in the LISTING and EVENT tables, but there might be multiple sales for the same combination of event and listing IDs in the SALES table. The third column in the result set identiﬁes the source of the row. If it comes from the SALES table, it is marked "Yes" in the SALESROW column. (SALESROW is an alias for SALES.LISTID.) If the row comes from the LISTING table, it is marked "No" in the SALESROW column. In this case, the result set consists of three sales rows for listing 500, event 7787. In other words, three diﬀerent transactions took place for this listing and event combination. The other two listings, 501 and 502, did not produce any sales, so the only row that the query produces for these list IDs comes from the LISTING table (SALESROW = 'No'). select eventid, listid, 'Yes' as salesrow from sales where listid in(500,501,502) union all select eventid, listid, 'No' from listing where listid in(500,501,502) eventid \| listid \| salesrow ---------+--------+---------- 7787 \| 500 \| No 7787 \| 500 \| Yes 7787 \| 500 \| Yes 7787 \| 500 \| Yes 6473 \| 501 \| No 5108 \| 502 \| No If you run the same query without the ALL keyword, the result retains only one of the sales transactions. select eventid, listid, 'Yes' as salesrow from sales where listid in(500,501,502) union select eventid, listid, 'No' from listing SELECT 84 SQL Reference AWS Clean Rooms where listid in(500,501,502) eventid \| listid \| salesrow ---------+--------+---------- 7787 \| 500 \| No 7787 \| 500 \| Yes 6473 \| 501 \| No 5108 \| 502 \| No Example UNION ALL query The following example uses a UNION ALL operator because duplicate rows, if found, need to be retained in the result. For a speciﬁc series of event IDs, the query returns 0 or more rows for each sale associated with each event, and 0 or 1 row for each listing of that event. Event IDs are unique to each row in the LISTING and EVENT tables, but there might be multiple sales for the same combination of event and listing IDs in the SALES table. The third column in the result set identiﬁes the source of the row. If it comes from the SALES table, it is marked "Yes" in the SALESROW column. (SALESROW is an alias for SALES.LISTID.) If the row comes from the LISTING table, it is marked "No" in the SALESROW column. In this case, the result set consists of three sales rows for listing 500, event 7787. In other words, three diﬀerent transactions took place for this listing and event combination. The other two listings, 501 and 502, did not produce any sales, so the only row that the query produces for these list IDs comes from the LISTING table (SALESROW = 'No'). select eventid, listid, 'Yes' as salesrow from sales where listid in(500,501,502) union all select eventid, listid, 'No' from listing where listid in(500,501,502) eventid \| listid \| salesrow ---------+--------+---------- 7787 \| 500 \| No 7787 \| 500 \| Yes 7787 \| 500 \| Yes 7787 \| 500 \| Yes 6473 \| 501 \| No 5108 \| 502 \| No SELECT 85 AWS Clean Rooms SQL Reference If you run the same query without the ALL keyword, the result retains only one of the sales transactions. select eventid, listid, 'Yes' as salesrow from sales where listid in(500,501,502) union select eventid, listid, 'No' from listing where listid in(500,501,502) eventid \| listid \| salesrow ---------+--------+---------- 7787 \| 500 \| No 7787 \| 500 \| Yes 6473 \| 501 \| No 5108 \| 502 \| No Example INTERSECT queries Compare the following example with the ﬁrst UNION example. The only diﬀerence between the two examples is the set operator that is used, but the results are very diﬀerent. Only one of the rows is the same: 235494 \| 23875 \| 8771 This is the only row in the limited result of 5 rows that was found in both tables. select listid, sellerid, eventid from listing intersect select listid, sellerid, eventid from sales listid \| sellerid \| eventid --------+----------+--------- 235494 \| 23875 \| 8771 235482 \| 1067 \| 2667 235479 \| 1589 \| 7303 235476 \| 15550 \|
sql-reference-028	sql-reference.pdf	28	INTERSECT queries Compare the following example with the ﬁrst UNION example. The only diﬀerence between the two examples is the set operator that is used, but the results are very diﬀerent. Only one of the rows is the same: 235494 \| 23875 \| 8771 This is the only row in the limited result of 5 rows that was found in both tables. select listid, sellerid, eventid from listing intersect select listid, sellerid, eventid from sales listid \| sellerid \| eventid --------+----------+--------- 235494 \| 23875 \| 8771 235482 \| 1067 \| 2667 235479 \| 1589 \| 7303 235476 \| 15550 \| 793 235475 \| 22306 \| 7848 SELECT 86 AWS Clean Rooms SQL Reference The following query ﬁnds events (for which tickets were sold) that occurred at venues in both New York City and Los Angeles in March. The diﬀerence between the two query expressions is the constraint on the VENUECITY column. select distinct eventname from event, sales, venue where event.eventid=sales.eventid and event.venueid=venue.venueid and date_part(month,starttime)=3 and venuecity='Los Angeles' intersect select distinct eventname from event, sales, venue where event.eventid=sales.eventid and event.venueid=venue.venueid and date_part(month,starttime)=3 and venuecity='New York City'; eventname ---------------------------- A Streetcar Named Desire Dirty Dancing Electra Running with Annalise Hairspray Mary Poppins November Oliver! Return To Forever Rhinoceros South Pacific The 39 Steps The Bacchae The Caucasian Chalk Circle The Country Girl Wicked Woyzeck Example EXCEPT query The CATEGORY table in the database contains the following 11 rows: catid \| catgroup \| catname \| catdesc -------+----------+-----------+-------------------------------------------- 1 \| Sports \| MLB \| Major League Baseball 2 \| Sports \| NHL \| National Hockey League 3 \| Sports \| NFL \| National Football League 4 \| Sports \| NBA \| National Basketball Association 5 \| Sports \| MLS \| Major League Soccer SELECT 87 AWS Clean Rooms SQL Reference 6 \| Shows \| Musicals \| Musical theatre 7 \| Shows \| Plays \| All non-musical theatre 8 \| Shows \| Opera \| All opera and light opera 9 \| Concerts \| Pop \| All rock and pop music concerts 10 \| Concerts \| Jazz \| All jazz singers and bands 11 \| Concerts \| Classical \| All symphony, concerto, and choir concerts (11 rows) Assume that a CATEGORY_STAGE table (a staging table) contains one additional row: catid \| catgroup \| catname \| catdesc -------+----------+-----------+-------------------------------------------- 1 \| Sports \| MLB \| Major League Baseball 2 \| Sports \| NHL \| National Hockey League 3 \| Sports \| NFL \| National Football League 4 \| Sports \| NBA \| National Basketball Association 5 \| Sports \| MLS \| Major League Soccer 6 \| Shows \| Musicals \| Musical theatre 7 \| Shows \| Plays \| All non-musical theatre 8 \| Shows \| Opera \| All opera and light opera 9 \| Concerts \| Pop \| All rock and pop music concerts 10 \| Concerts \| Jazz \| All jazz singers and bands 11 \| Concerts \| Classical \| All symphony, concerto, and choir concerts 12 \| Concerts \| Comedy \| All stand up comedy performances (12 rows) Return the diﬀerence between the two tables. In other words, return rows that are in the CATEGORY_STAGE table but not in the CATEGORY table: select * from category_stage except select * from category; catid \| catgroup \| catname \| catdesc -------+----------+---------+---------------------------------- 12 \| Concerts \| Comedy \| All stand up comedy performances (1 row) The following equivalent query uses the synonym MINUS. select * from category_stage minus SELECT 88 AWS Clean Rooms select * from category; SQL Reference catid \| catgroup \| catname \| catdesc -------+----------+---------+---------------------------------- 12 \| Concerts \| Comedy \| All stand up comedy performances (1 row) If you reverse the order of the SELECT expressions, the query returns no rows. ORDER BY clause The ORDER BY clause sorts the result set of a query. Note The outermost ORDER BY expression must only have columns that are in the select list. Topics • Syntax • Parameters • Usage notes • Examples with ORDER BY Syntax [ ORDER BY expression [ ASC \| DESC ] ] [ NULLS FIRST \| NULLS LAST ] [ LIMIT { count \| ALL } ] [ OFFSET start ] Parameters expression Expression that deﬁnes the sort order of the query result. It consists of one or more columns in the select list. Results are returned based on binary UTF-8 ordering. You can also specify the following: SELECT 89 AWS Clean Rooms SQL Reference • Ordinal numbers that represent the position of select list entries (or the position of columns in the table if no select list exists) • Aliases that deﬁne select list entries When the ORDER BY clause contains multiple expressions, the result set is sorted according to the ﬁrst expression, then the second expression is applied to rows that have matching values from the ﬁrst expression, and so on.
sql-reference-029	sql-reference.pdf	29	It consists of one or more columns in the select list. Results are returned based on binary UTF-8 ordering. You can also specify the following: SELECT 89 AWS Clean Rooms SQL Reference • Ordinal numbers that represent the position of select list entries (or the position of columns in the table if no select list exists) • Aliases that deﬁne select list entries When the ORDER BY clause contains multiple expressions, the result set is sorted according to the ﬁrst expression, then the second expression is applied to rows that have matching values from the ﬁrst expression, and so on. ASC \| DESC Option that deﬁnes the sort order for the expression, as follows: • ASC: ascending (for example, low to high for numeric values and 'A' to 'Z' for character strings). If no option is speciﬁed, data is sorted in ascending order by default. • DESC: descending (high to low for numeric values; 'Z' to 'A' for strings). NULLS FIRST \| NULLS LAST Option that speciﬁes whether NULL values should be ordered ﬁrst, before non-null values, or last, after non-null values. By default, NULL values are sorted and ranked last in ASC ordering, and sorted and ranked ﬁrst in DESC ordering. LIMIT number \| ALL Option that controls the number of sorted rows that the query returns. The LIMIT number must be a positive integer; the maximum value is 2147483647. LIMIT 0 returns no rows. You can use this syntax for testing purposes: to check that a query runs (without displaying any rows) or to return a column list from a table. An ORDER BY clause is redundant if you are using LIMIT 0 to return a column list. The default is LIMIT ALL. OFFSET start Option that speciﬁes to skip the number of rows before start before beginning to return rows. The OFFSET number must be a positive integer; the maximum value is 2147483647. When used with the LIMIT option, OFFSET rows are skipped before starting to count the LIMIT rows that are returned. If the LIMIT option isn't used, the number of rows in the result set is reduced by the number of rows that are skipped. The rows skipped by an OFFSET clause still have to be scanned, so it might be ineﬃcient to use a large OFFSET value. SELECT 90 AWS Clean Rooms Usage notes SQL Reference Note the following expected behavior with ORDER BY clauses: • NULL values are considered "higher" than all other values. With the default ascending sort order, NULL values sort at the end. To change this behavior, use the NULLS FIRST option. • When a query doesn't contain an ORDER BY clause, the system returns result sets with no predictable ordering of the rows. The same query run twice might return the result set in a diﬀerent order. • The LIMIT and OFFSET options can be used without an ORDER BY clause; however, to return a consistent set of rows, use these options in conjunction with ORDER BY. • In any parallel system like AWS Clean Rooms, when ORDER BY doesn't produce a unique ordering, the order of the rows is nondeterministic. That is, if the ORDER BY expression produces duplicate values, the return order of those rows might vary from other systems or from one run of AWS Clean Rooms to the next. • AWS Clean Rooms doesn't support string literals in ORDER BY clauses. Examples with ORDER BY Return all 11 rows from the CATEGORY table, ordered by the second column, CATGROUP. For results that have the same CATGROUP value, order the CATDESC column values by the length of the character string. Then order by columns CATID and CATNAME. select * from category order by 2, 1, 3; catid \| catgroup \| catname \| catdesc -------+----------+-----------+---------------------------------------- 10 \| Concerts \| Jazz \| All jazz singers and bands 9 \| Concerts \| Pop \| All rock and pop music concerts 11 \| Concerts \| Classical \| All symphony, concerto, and choir conce 6 \| Shows \| Musicals \| Musical theatre 7 \| Shows \| Plays \| All non-musical theatre 8 \| Shows \| Opera \| All opera and light opera 5 \| Sports \| MLS \| Major League Soccer 1 \| Sports \| MLB \| Major League Baseball 2 \| Sports \| NHL \| National Hockey League 3 \| Sports \| NFL \| National Football League 4 \| Sports \| NBA \| National Basketball Association (11 rows) SELECT 91 AWS Clean Rooms SQL Reference Return selected columns from the SALES table, ordered by the highest QTYSOLD values. Limit the result to the top 10 rows: select salesid, qtysold, pricepaid, commission, saletime from sales order by qtysold, pricepaid, commission, salesid, saletime desc salesid \| qtysold \| pricepaid \| commission \| saletime ---------+---------+-----------+------------+--------------------- 15401 \| 8 \| 272.00 \| 40.80 \|
sql-reference-030	sql-reference.pdf	30	1 \| Sports \| MLB \| Major League Baseball 2 \| Sports \| NHL \| National Hockey League 3 \| Sports \| NFL \| National Football League 4 \| Sports \| NBA \| National Basketball Association (11 rows) SELECT 91 AWS Clean Rooms SQL Reference Return selected columns from the SALES table, ordered by the highest QTYSOLD values. Limit the result to the top 10 rows: select salesid, qtysold, pricepaid, commission, saletime from sales order by qtysold, pricepaid, commission, salesid, saletime desc salesid \| qtysold \| pricepaid \| commission \| saletime ---------+---------+-----------+------------+--------------------- 15401 \| 8 \| 272.00 \| 40.80 \| 2008-03-18 06:54:56 61683 \| 8 \| 296.00 \| 44.40 \| 2008-11-26 04:00:23 90528 \| 8 \| 328.00 \| 49.20 \| 2008-06-11 02:38:09 74549 \| 8 \| 336.00 \| 50.40 \| 2008-01-19 12:01:21 130232 \| 8 \| 352.00 \| 52.80 \| 2008-05-02 05:52:31 55243 \| 8 \| 384.00 \| 57.60 \| 2008-07-12 02:19:53 16004 \| 8 \| 440.00 \| 66.00 \| 2008-11-04 07:22:31 489 \| 8 \| 496.00 \| 74.40 \| 2008-08-03 05:48:55 4197 \| 8 \| 512.00 \| 76.80 \| 2008-03-23 11:35:33 16929 \| 8 \| 568.00 \| 85.20 \| 2008-12-19 02:59:33 Return a column list and no rows by using LIMIT 0 syntax: select * from venue limit 0; venueid \| venuename \| venuecity \| venuestate \| venueseats ---------+-----------+-----------+------------+------------ (0 rows) Subquery examples The following examples show diﬀerent ways in which subqueries ﬁt into SELECT queries. See Example for another example of the use of subqueries. SELECT list subquery The following example contains a subquery in the SELECT list. This subquery is scalar: it returns only one column and one value, which is repeated in the result for each row that is returned from the outer query. The query compares the Q1SALES value that the subquery computes with sales values for two other quarters (2 and 3) in 2008, as deﬁned by the outer query. select qtr, sum(pricepaid) as qtrsales, (select sum(pricepaid) from sales join date on sales.dateid=date.dateid where qtr='1' and year=2008) as q1sales SELECT 92 AWS Clean Rooms SQL Reference from sales join date on sales.dateid=date.dateid where qtr in('2','3') and year=2008 group by qtr order by qtr; qtr \| qtrsales \| q1sales -------+-------------+------------- 2 \| 30560050.00 \| 24742065.00 3 \| 31170237.00 \| 24742065.00 (2 rows) WHERE clause subquery The following example contains a table subquery in the WHERE clause. This subquery produces multiple rows. In this case, the rows contain only one column, but table subqueries can contain multiple columns and rows, just like any other table. The query ﬁnds the top 10 sellers in terms of maximum tickets sold. The top 10 list is restricted by the subquery, which removes users who live in cities where there are ticket venues. This query can be written in diﬀerent ways; for example, the subquery could be rewritten as a join within the main query. select firstname, lastname, city, max(qtysold) as maxsold from users join sales on users.userid=sales.sellerid where users.city not in(select venuecity from venue) group by firstname, lastname, city order by maxsold desc, city desc limit 10; firstname \| lastname \| city \| maxsold -----------+-----------+----------------+--------- Noah \| Guerrero \| Worcester \| 8 Isadora \| Moss \| Winooski \| 8 Kieran \| Harrison \| Westminster \| 8 Heidi \| Davis \| Warwick \| 8 Sara \| Anthony \| Waco \| 8 Bree \| Buck \| Valdez \| 8 Evangeline \| Sampson \| Trenton \| 8 Kendall \| Keith \| Stillwater \| 8 Bertha \| Bishop \| Stevens Point \| 8 Patricia \| Anderson \| South Portland \| 8 (10 rows) SELECT 93 AWS Clean Rooms WITH clause subqueries See WITH clause. Correlated subqueries SQL Reference The following example contains a correlated subquery in the WHERE clause; this kind of subquery contains one or more correlations between its columns and the columns produced by the outer query. In this case, the correlation is where s.listid=l.listid. For each row that the outer query produces, the subquery is run to qualify or disqualify the row. select salesid, listid, sum(pricepaid) from sales s where qtysold= (select max(numtickets) from listing l where s.listid=l.listid) group by 1,2 order by 1,2 limit 5; salesid \| listid \| sum --------+--------+---------- 27 \| 28 \| 111.00 81 \| 103 \| 181.00 142 \| 149 \| 240.00 146 \| 152 \| 231.00 194 \| 210 \| 144.00 (5 rows) Correlated subquery patterns that are not supported The query planner uses a query rewrite method called subquery decorrelation to optimize several patterns of correlated subqueries for execution in an MPP environment. A few types of correlated subqueries follow patterns that AWS Clean Rooms can't decorrelate and doesn't support. Queries that contain the following correlation references return errors: • Correlation references that skip a query block, also known as "skip-level correlation references." For example, in the following query, the block containing the
sql-reference-031	sql-reference.pdf	31	\| 181.00 142 \| 149 \| 240.00 146 \| 152 \| 231.00 194 \| 210 \| 144.00 (5 rows) Correlated subquery patterns that are not supported The query planner uses a query rewrite method called subquery decorrelation to optimize several patterns of correlated subqueries for execution in an MPP environment. A few types of correlated subqueries follow patterns that AWS Clean Rooms can't decorrelate and doesn't support. Queries that contain the following correlation references return errors: • Correlation references that skip a query block, also known as "skip-level correlation references." For example, in the following query, the block containing the correlation reference and the skipped block are connected by a NOT EXISTS predicate: select event.eventname from event where not exists (select * from listing SELECT 94 AWS Clean Rooms where not exists (select * from sales where event.eventid=sales.eventid)); SQL Reference The skipped block in this case is the subquery against the LISTING table. The correlation reference correlates the EVENT and SALES tables. • Correlation references from a subquery that is part of an ON clause in an outer query: select * from category left join event on category.catid=event.catid and eventid = (select max(eventid) from sales where sales.eventid=event.eventid); The ON clause contains a correlation reference from SALES in the subquery to EVENT in the outer query. • Null-sensitive correlation references to an AWS Clean Rooms system table. For example: select attrelid from my_locks sl, my_attribute where sl.table_id=my_attribute.attrelid and 1 not in (select 1 from my_opclass where sl.lock_owner = opcowner); • Correlation references from within a subquery that contains a window function. select listid, qtysold from sales s where qtysold not in (select sum(numtickets) over() from listing l where s.listid=l.listid); • References in a GROUP BY column to the results of a correlated subquery. For example: select listing.listid, (select count (sales.listid) from sales where sales.listid=listing.listid) as list from listing group by list, listing.listid; • Correlation references from a subquery with an aggregate function and a GROUP BY clause, connected to the outer query by an IN predicate. (This restriction doesn't apply to MIN and MAX aggregate functions.) For example: select * from listing where listid in SELECT 95 AWS Clean Rooms SQL Reference (select sum(qtysold) from sales where numtickets>4 group by salesid); AWS Clean Rooms Spark SQL functions AWS Clean Rooms Spark SQL supports the following SQL functions: Topics • Aggregate functions • Array functions • Binary functions • Conditional expressions • Constructor functions • Data type formatting functions • Date and time functions • Encryption and decryption functions • Hash functions • Hyperloglog functions • JSON functions • Math functions • Scalar functions • String functions • Privacy-related functions • Window functions Aggregate functions Aggregate functions in AWS Clean Rooms Spark SQL are used to perform calculations or operations on a group of rows and return a single value. They are essential for data analysis and summarization tasks. AWS Clean Rooms Spark SQL supports the following aggregate functions: SQL functions 96 SQL Reference AWS Clean Rooms Topics • ANY_VALUE function • APPROX COUNT_DISTINCT function • APPROX PERCENTILE function • AVG function • BOOL_AND function • BOOL_OR function • CARDINALITY function • COLLECT_LIST function • COLLECT_SET function • COUNT and COUNT DISTINCT functions • COUNT function • LISTAGG function • MAX function • MEDIAN function • MIN function • PERCENTILE function • PERCENTILE_CONT function • SKEWNESS function • STDDEV_SAMP and STDDEV_POP functions • SUM and SUM DISTINCT functions • VAR_SAMP and VAR_POP functions ANY_VALUE function The ANY_VALUE function returns any value from the input expression values nondeterministically. This function can return NULL if the input expression doesn't result in any rows being returned. Syntax ANY_VALUE (expression[, isIgnoreNull] ) Aggregate functions 97 AWS Clean Rooms Arguments expression SQL Reference The target column or expression on which the function operates. The expression is one of the following data types: isIgnoreNull A boolean that determines if the function should return only non-null values. Returns Returns the same data type as expression. Usage notes If a statement that speciﬁes the ANY_VALUE function for a column also includes a second column reference, the second column must appear in a GROUP BY clause or be included in an aggregate function. Examples The following example returns an instance of any dateid where the eventname is Eagles. select any_value(dateid) as dateid, eventname from event where eventname ='Eagles' group by eventname; Following are the results. dateid \| eventname -------+--------------- 1878 \| Eagles The following example returns an instance of any dateid where the eventname is Eagles or Cold War Kids. select any_value(dateid) as dateid, eventname from event where eventname in('Eagles', 'Cold War Kids') group by eventname; Following are the results. Aggregate functions 98 AWS Clean Rooms SQL Reference dateid \| eventname -------+--------------- 1922 \| Cold War Kids 1878 \| Eagles APPROX COUNT_DISTINCT function APPROX COUNT_DISTINCT provides an eﬃcient way to estimate
sql-reference-032	sql-reference.pdf	32	any dateid where the eventname is Eagles. select any_value(dateid) as dateid, eventname from event where eventname ='Eagles' group by eventname; Following are the results. dateid \| eventname -------+--------------- 1878 \| Eagles The following example returns an instance of any dateid where the eventname is Eagles or Cold War Kids. select any_value(dateid) as dateid, eventname from event where eventname in('Eagles', 'Cold War Kids') group by eventname; Following are the results. Aggregate functions 98 AWS Clean Rooms SQL Reference dateid \| eventname -------+--------------- 1922 \| Cold War Kids 1878 \| Eagles APPROX COUNT_DISTINCT function APPROX COUNT_DISTINCT provides an eﬃcient way to estimate the number of unique values in a column or dataset. Syntax approx_count_distinct(expr[, relativeSD]) Arguments expr The expression or column for which you want to estimate the number of unique values. It can be a single column, a complex expression, or a combination of columns. relativeSD An optional parameter that speciﬁes the desired relative standard deviation of the estimate. It is a value between 0 and 1, representing the maximum acceptable relative error of the estimate. A smaller relativeSD value will result in a more accurate but slower estimation. If this parameter isn't provided, a default value (usually around 0.05 or 5%) is used. Returns Returns the estimated cardinality by HyperLogLog++. relativeSD deﬁnes the maximum relative standard deviation allowed. Example The following query estimates the number of unique values in the col1 column, with a relative standard deviation of 1% (0.01). Aggregate functions 99 AWS Clean Rooms SQL Reference SELECT approx_count_distinct(col1, 0.01) The following query estimates that there are 3 unique values in the col1 column (the values 1, 2, and 3). SELECT approx_count_distinct(col1) FROM VALUES (1), (1), (2), (2), (3) tab(col1) APPROX PERCENTILE function APPROX PERCENTILE is used to estimate the percentile value of a given expression or column without having to sort the entire dataset. This function is useful in scenarios where you need to quickly understand the distribution of a large dataset or track percentile-based metrics, without the computational overhead of performing an exact percentile calculation. However, it's important to understand the trade-oﬀs between speed and accuracy, and to choose the appropriate error tolerance based on the speciﬁc requirements of your use case. Syntax APPROX_PERCENTILE(expr, percentile [, accuracy]) Arguments expr The expression or column for which you want to estimate the percentile value. It can be a single column, a complex expression, or a combination of columns. percentile The percentile value you want to estimate, expressed as a value between 0 and 1. For example, 0.5 would correspond to the 50th percentile (median). accuracy An optional parameter that speciﬁes the desired accuracy of the percentile estimate. It is a value between 0 and 1, representing the maximum acceptable relative error of the estimate. A smaller accuracy value will result in a more precise but slower estimation. If this parameter isn't provided, a default value (usually around 0.05 or 5%) is used. Aggregate functions 100 AWS Clean Rooms Returns SQL Reference Returns the approximate percentile of the numeric or ANSI interval column col which is the smallest value in the ordered col values (sorted from least to greatest) such that no more than percentage of col values is less than the value or equal to that value. The value of percentage must be between 0.0 and 1.0. The accuracy parameter (default: 10000) is a positive numeric literal which controls approximation accuracy at the cost of memory. Higher value of accuracy yields better accuracy, 1.0/accuracy is the relative error of the approximation. When percentage is an array, each value of the percentage array must be between 0.0 and 1.0. In this case, returns the approximate percentile array of column col at the given percentage array. Examples The following query estimates the 95th percentile of the response_time column, with a maximum relative error of 1% (0.01). SELECT APPROX_PERCENTILE(response_time, 0.95, 0.01) AS p95_response_time FROM my_table; The following query estimates the 50th, 40th, and 10th percentile values of the col column in the tab table. SELECT approx_percentile(col, array(0.5, 0.4, 0.1), 100) FROM VALUES (0), (1), (2), (10) AS tab(col) The following query estimates the 50th percentile (median) of the values in the col column. SELECT approx_percentile(col, 0.5, 100) FROM VALUES (0), (6), (7), (9), (10) AS tab(col) AVG function The AVG function returns the average (arithmetic mean) of the input expression values. The AVG function works with numeric values and ignores NULL values. Aggregate functions 101 AWS Clean Rooms Syntax AVG (column) Arguments column SQL Reference The target column that the function operates on. The column is one of the following data types: • SMALLINT • INTEGER • BIGINT • DECIMAL • DOUBLE • FLOAT Data types The argument types supported by the AVG function are SMALLINT, INTEGER, BIGINT, DECIMAL, and DOUBLE. The return types supported by the AVG function are: • BIGINT
sql-reference-033	sql-reference.pdf	33	(9), (10) AS tab(col) AVG function The AVG function returns the average (arithmetic mean) of the input expression values. The AVG function works with numeric values and ignores NULL values. Aggregate functions 101 AWS Clean Rooms Syntax AVG (column) Arguments column SQL Reference The target column that the function operates on. The column is one of the following data types: • SMALLINT • INTEGER • BIGINT • DECIMAL • DOUBLE • FLOAT Data types The argument types supported by the AVG function are SMALLINT, INTEGER, BIGINT, DECIMAL, and DOUBLE. The return types supported by the AVG function are: • BIGINT for any integer type argument • DOUBLE for a ﬂoating point argument • Returns the same data type as expression for any other argument type The default precision for an AVG function result with a DECIMAL argument is 38. The scale of the result is the same as the scale of the argument. For example, an AVG of a DEC(5,2) column returns a DEC(38,2) data type. Example Find the average quantity sold per transaction from the SALES table. select avg(qtysold) from sales; Aggregate functions 102 AWS Clean Rooms BOOL_AND function SQL Reference The BOOL_AND function operates on a single Boolean or integer column or expression. This function applies similar logic to the BIT_AND and BIT_OR functions. For this function, the return type is a Boolean value (true or false). If all values in a set are true, the BOOL_AND function returns true (t). If any value is false, the function returns false (f). Syntax BOOL_AND ( [DISTINCT \| ALL] expression ) Arguments expression The target column or expression that the function operates on. This expression must have a BOOLEAN or integer data type. The return type of the function is BOOLEAN. DISTINCT \| ALL With the argument DISTINCT, the function eliminates all duplicate values for the speciﬁed expression before calculating the result. With the argument ALL, the function retains all duplicate values. ALL is the default. Examples You can use the Boolean functions against either Boolean expressions or integer expressions. For example, the following query return results from the standard USERS table in the TICKIT database, which has several Boolean columns. The BOOL_AND function returns false for all ﬁve rows. Not all users in each of those states likes sports. select state, bool_and(likesports) from users group by state order by state limit 5; state \| bool_and ------+--------- AB \| f Aggregate functions 103 AWS Clean Rooms AK \| f AL \| f AZ \| f BC \| f (5 rows) BOOL_OR function SQL Reference The BOOL_OR function operates on a single Boolean or integer column or expression. This function applies similar logic to the BIT_AND and BIT_OR functions. For this function, the return type is a Boolean value (true, false, or NULL). If a value in a set is true, the BOOL_OR function returns true (t). If a value in a set is false, the function returns false (f). NULL can be returned if the value is unknown. Syntax BOOL_OR ( [DISTINCT \| ALL] expression ) Arguments expression The target column or expression that the function operates on. This expression must have a BOOLEAN or integer data type. The return type of the function is BOOLEAN. DISTINCT \| ALL With the argument DISTINCT, the function eliminates all duplicate values for the speciﬁed expression before calculating the result. With the argument ALL, the function retains all duplicate values. ALL is the default. Examples You can use the Boolean functions with either Boolean expressions or integer expressions. For example, the following query return results from the standard USERS table in the TICKIT database, which has several Boolean columns. The BOOL_OR function returns true for all ﬁve rows. At least one user in each of those states likes sports. select state, bool_or(likesports) from users Aggregate functions 104 AWS Clean Rooms SQL Reference group by state order by state limit 5; state \| bool_or ------+-------- AB \| t AK \| t AL \| t AZ \| t BC \| t (5 rows) The following example returns NULL. SELECT BOOL_OR(NULL = '123') bool_or ------ NULL CARDINALITY function The CARDINALITY function returns the size of an ARRAY or MAP expression (expr). This function is useful to ﬁnd the size or length of an array. Syntax cardinality(expr) Arguments expr An ARRAY or MAP expression. Returns Returns the size of an array or a map (INTEGER). The function returns NULL for null input if sizeOfNull is set to false or enabled is set to true. Otherwise, the function returns -1 for null input. With the default settings, the function returns -1 for null input. Aggregate functions 105 AWS Clean Rooms Example SQL Reference The following query calculates the cardinality, or the number of elements, in the given array. The array ('b', 'd', 'c', 'a') has 4 elements, so
sql-reference-034	sql-reference.pdf	34	ﬁnd the size or length of an array. Syntax cardinality(expr) Arguments expr An ARRAY or MAP expression. Returns Returns the size of an array or a map (INTEGER). The function returns NULL for null input if sizeOfNull is set to false or enabled is set to true. Otherwise, the function returns -1 for null input. With the default settings, the function returns -1 for null input. Aggregate functions 105 AWS Clean Rooms Example SQL Reference The following query calculates the cardinality, or the number of elements, in the given array. The array ('b', 'd', 'c', 'a') has 4 elements, so the output of this query would be 4. SELECT cardinality(array('b', 'd', 'c', 'a')); 4 COLLECT_LIST function The COLLECT_LIST function collects and returns a list of non-unique elements. This type of function is useful when you want to collect multiple values from a set of rows into a single array or list data structure. Note The function is non-deterministic because the order of the collected results depends on the order of the rows, which may be non-deterministic after a shuﬄe operation is performed. Syntax collect_list(expr) Arguments expr An expression of any type. Returns Returns an ARRAY of the argument type. The order of elements in the array is non-deterministic. NULL values are excluded. If DISTINCT is speciﬁed, the function collects only unique values and is a synonym for collect_set aggregate function. Aggregate functions 106 AWS Clean Rooms Example SQL Reference The following query collects all the values from the col column into a list. The VALUES clause is used to create an inline table with three rows, where each row has a single column col with the values 1, 2, and 1 respectively. The collect_list() function is then used to aggregate all the values from the col column into a single array. The output of this SQL statement would be the array [1,2,1], which contains all the values from the col column in the order they appeared in the input data. SELECT collect_list(col) FROM VALUES (1), (2), (1) AS tab(col); [1,2,1] COLLECT_SET function The COLLECT_SET function collects and returns a set of unique elements. This function is useful when you want to collect all the distinct values from a set of rows into a single data structure, without including any duplicates. Note The function is non-deterministic because the order of the collected results depends on the order of the rows, which may be non-deterministic after a shuﬄe operation is performed. Syntax collect_set(expr) Arguments expr An expression of any type except MAP. Returns Returns an ARRAY of the argument type. The order of elements in the array is non-deterministic. NULL values are excluded. Aggregate functions 107 AWS Clean Rooms Example SQL Reference The following query collects all the unique values from the col column into a set. The VALUES clause is used to create an inline table with three rows, where each row has a single column col with the values 1, 2, and 1 respectively. The collect_set() function is then used to aggregate all the unique values from the col column into a single set. The output of this SQL statement would be the set [1,2], which contains the unique values from the col column. The duplicate value of 1 is only included once in the result. SELECT collect_set(col) FROM VALUES (1), (2), (1) AS tab(col); [1,2] COUNT and COUNT DISTINCT functions The COUNT function counts the rows deﬁned by the expression. The COUNT DISTINCT function computes the number of distinct non-NULL values in a column or expression. It eliminates all duplicate values from the speciﬁed expression before doing the count. Syntax COUNT (DISTINCT column) Arguments column The target column that the function operates on. Data types The COUNT function and the COUNT DISTINCT function supports all argument data types. The COUNT DISTINCT function returns BIGINT. Examples Count all of the users from the state of Florida. select count (identifier) from users where state='FL'; Count all of the unique venue IDs from the EVENT table. Aggregate functions 108 AWS Clean Rooms SQL Reference select count (distinct venueid) as venues from event; COUNT function The COUNT function counts the rows deﬁned by the expression. The COUNT function has the following variations. • COUNT ( * ) counts all the rows in the target table whether they include nulls or not. • COUNT ( expression ) computes the number of rows with non-NULL values in a speciﬁc column or expression. • COUNT ( DISTINCT expression ) computes the number of distinct non-NULL values in a column or expression. Syntax COUNT( * \| expression ) COUNT ( [ DISTINCT \| ALL ] expression ) Arguments expression The target column or expression that the function operates on. The COUNT function supports all argument data types. DISTINCT \| ALL With the argument DISTINCT, the function eliminates
sql-reference-035	sql-reference.pdf	35	COUNT ( * ) counts all the rows in the target table whether they include nulls or not. • COUNT ( expression ) computes the number of rows with non-NULL values in a speciﬁc column or expression. • COUNT ( DISTINCT expression ) computes the number of distinct non-NULL values in a column or expression. Syntax COUNT( * \| expression ) COUNT ( [ DISTINCT \| ALL ] expression ) Arguments expression The target column or expression that the function operates on. The COUNT function supports all argument data types. DISTINCT \| ALL With the argument DISTINCT, the function eliminates all duplicate values from the speciﬁed expression before doing the count. With the argument ALL, the function retains all duplicate values from the expression for counting. ALL is the default. Return type The COUNT function returns BIGINT. Examples Count all of the users from the state of Florida: Aggregate functions 109 AWS Clean Rooms SQL Reference select count() from users where state='FL'; count ------- 510 Count all of the event names from the EVENT table: select count(eventname) from event; count ------- 8798 Count all of the event names from the EVENT table: select count(all eventname) from event; count ------- 8798 Count all of the unique venue IDs from the EVENT table: select count(distinct venueid) as venues from event; venues -------- 204 Count the number of times each seller listed batches of more than four tickets for sale. Group the results by seller ID: select count(), sellerid from listing where numtickets > 4 group by sellerid order by 1 desc, 2; count \| sellerid ------+---------- 12 \| 6386 Aggregate functions 110 AWS Clean Rooms 11 \| 17304 11 \| 20123 11 \| 25428 ... LISTAGG function SQL Reference For each group in a query, the LISTAGG aggregate function orders the rows for that group according to the ORDER BY expression, then concatenates the values into a single string. LISTAGG is a compute-node only function. The function returns an error if the query doesn't reference a user-deﬁned table or AWS Clean Rooms system table. Syntax LISTAGG( [DISTINCT] aggregate_expression [, 'delimiter' ] ) [ WITHIN GROUP (ORDER BY order_list) ] Arguments DISTINCT (Optional) A clause that eliminates duplicate values from the speciﬁed expression before concatenating. Trailing spaces are ignored, so the strings 'a' and 'a ' are treated as duplicates. LISTAGG uses the ﬁrst value encountered. For more information, see Signiﬁcance of trailing blanks. aggregate_expression Any valid expression (such as a column name) that provides the values to aggregate. NULL values and empty strings are ignored. delimiter (Optional) The string constant to separate the concatenated values. The default is NULL. AWS Clean Rooms supports any amount of leading or trailing whitespace around an optional comma or colon as well as an empty string or any number of spaces. Examples of valid values are: ", " ": " Aggregate functions 111 AWS Clean Rooms " " WITHIN GROUP (ORDER BY order_list) SQL Reference (Optional) A clause that speciﬁes the sort order of the aggregated values. Returns VARCHAR(MAX). If the result set is larger than the maximum VARCHAR size (64K – 1, or 65535), then LISTAGG returns the following error: Invalid operation: Result size exceeds LISTAGG limit Usage notes If a statement includes multiple LISTAGG functions that use WITHIN GROUP clauses, each WITHIN GROUP clause must use the same ORDER BY values. For example, the following statement will return an error. select listagg(sellerid) within group (order by dateid) as sellers, listagg(dateid) within group (order by sellerid) as dates from winsales; The following statements will run successfully. select listagg(sellerid) within group (order by dateid) as sellers, listagg(dateid) within group (order by dateid) as dates from winsales; select listagg(sellerid) within group (order by dateid) as sellers, listagg(dateid) as dates from winsales; Examples The following example aggregates seller IDs, ordered by seller ID. Aggregate functions 112 AWS Clean Rooms SQL Reference select listagg(sellerid, ', ') within group (order by sellerid) from sales where eventid = 4337; listagg ---------------------------------------------------------------------------------------------------------------------------------------- 380, 380, 1178, 1178, 1178, 2731, 8117, 12905, 32043, 32043, 32043, 32432, 32432, 38669, 38750, 41498, 45676, 46324, 47188, 47188, 48294 The following example uses DISTINCT to return a list of unique seller IDs. select listagg(distinct sellerid, ', ') within group (order by sellerid) from sales where eventid = 4337; listagg ------------------------------------------------------------------------------------------- 380, 1178, 2731, 8117, 12905, 32043, 32432, 38669, 38750, 41498, 45676, 46324, 47188, 48294 The following example aggregates seller IDs in date order. select listagg(sellerid) within group (order by dateid) from winsales; listagg ------------- 31141242333 The following example returns a pipe-separated list of sales dates for buyer B. select listagg(dateid,'\|') within group (order by sellerid desc,salesid asc) from winsales where buyerid = 'b'; listagg --------------------------------------- 2003-08-02\|2004-04-18\|2004-04-18\|2004-02-12 The following example returns a comma-separated list of sales IDs for each buyer ID. Aggregate functions 113 AWS Clean Rooms SQL Reference
sql-reference-036	sql-reference.pdf	36	group (order by sellerid) from sales where eventid = 4337; listagg ------------------------------------------------------------------------------------------- 380, 1178, 2731, 8117, 12905, 32043, 32432, 38669, 38750, 41498, 45676, 46324, 47188, 48294 The following example aggregates seller IDs in date order. select listagg(sellerid) within group (order by dateid) from winsales; listagg ------------- 31141242333 The following example returns a pipe-separated list of sales dates for buyer B. select listagg(dateid,'\|') within group (order by sellerid desc,salesid asc) from winsales where buyerid = 'b'; listagg --------------------------------------- 2003-08-02\|2004-04-18\|2004-04-18\|2004-02-12 The following example returns a comma-separated list of sales IDs for each buyer ID. Aggregate functions 113 AWS Clean Rooms SQL Reference select buyerid, listagg(salesid,',') within group (order by salesid) as sales_id from winsales group by buyerid order by buyerid; buyerid \| sales_id -----------+------------------------ a \|10005,40001,40005 b \|20001,30001,30004,30003 c \|10001,20002,30007,10006 MAX function The MAX function returns the maximum value in a set of rows. DISTINCT or ALL might be used but do not aﬀect the result. Syntax MAX ( [ DISTINCT \| ALL ] expression ) Arguments expression The target column or expression that the function operates on. The expression is any numerical data type. DISTINCT \| ALL With the argument DISTINCT, the function eliminates all duplicate values from the speciﬁed expression before calculating the maximum. With the argument ALL, the function retains all duplicate values from the expression for calculating the maximum. ALL is the default. Data types Returns the same data type as expression. Examples Find the highest price paid from all sales: Aggregate functions 114 AWS Clean Rooms SQL Reference select max(pricepaid) from sales; max ---------- 12624.00 (1 row) Find the highest price paid per ticket from all sales: select max(pricepaid/qtysold) as max_ticket_price from sales; max_ticket_price ----------------- 2500.00000000 (1 row) MEDIAN function Calculates the median value for the range of values. NULL values in the range are ignored. MEDIAN is an inverse distribution function that assumes a continuous distribution model. MEDIAN is a special case of PERCENTILE_CONT(.5). MEDIAN is a compute-node only function. The function returns an error if the query doesn't reference a user-deﬁned table or AWS Clean Rooms system table. Syntax MEDIAN ( median_expression ) Arguments median_expression The target column or expression that the function operates on. Usage notes If the median_expression argument is a DECIMAL data type deﬁned with the maximum precision of 38 digits, it is possible that MEDIAN will return either an inaccurate result or an error. If the return Aggregate functions 115 AWS Clean Rooms SQL Reference value of the MEDIAN function exceeds 38 digits, the result is truncated to ﬁt, which causes a loss of precision. If, during interpolation, an intermediate result exceeds the maximum precision, a numeric overﬂow occurs and the function returns an error. To avoid these conditions, we recommend either using a data type with lower precision or casting the median_expression argument to a lower precision. If a statement includes multiple calls to sort-based aggregate functions (LISTAGG, PERCENTILE_CONT, or MEDIAN), they must all use the same ORDER BY values. Note that MEDIAN applies an implicit order by on the expression value. For example, the following statement returns an error. select top 10 salesid, sum(pricepaid), percentile_cont(0.6) within group (order by salesid), median (pricepaid) from sales group by salesid, pricepaid; An error occurred when executing the SQL command: select top 10 salesid, sum(pricepaid), percentile_cont(0.6) within group (order by salesid), median (pricepaid) from sales group by salesid, pricepai... ERROR: within group ORDER BY clauses for aggregate functions must be the same The following statement runs successfully. select top 10 salesid, sum(pricepaid), percentile_cont(0.6) within group (order by salesid), median (salesid) from sales group by salesid, pricepaid; Examples The following example shows that MEDIAN produces the same results as PERCENTILE_CONT(0.5). select top 10 distinct sellerid, qtysold, percentile_cont(0.5) within group (order by qtysold), median (qtysold) from sales group by sellerid, qtysold; Aggregate functions 116 AWS Clean Rooms SQL Reference sellerid \| qtysold \| percentile_cont \| median ---------+---------+-----------------+------- 1 \| 1 \| 1.0 \| 1.0 2 \| 3 \| 3.0 \| 3.0 5 \| 2 \| 2.0 \| 2.0 9 \| 4 \| 4.0 \| 4.0 12 \| 1 \| 1.0 \| 1.0 16 \| 1 \| 1.0 \| 1.0 19 \| 2 \| 2.0 \| 2.0 19 \| 3 \| 3.0 \| 3.0 22 \| 2 \| 2.0 \| 2.0 25 \| 2 \| 2.0 \| 2.0 MIN function The MIN function returns the minimum value in a set of rows. DISTINCT or ALL might be used but do not aﬀect the result. Syntax MIN ( [ DISTINCT \| ALL ] expression ) Arguments expression The target column or expression that the function operates on. The expression is any numerical data type. DISTINCT \| ALL With the argument DISTINCT, the function eliminates all duplicate values from the speciﬁed expression before calculating the minimum. With the argument ALL, the function retains all duplicate values from the
sql-reference-037	sql-reference.pdf	37	2 \| 2.0 \| 2.0 25 \| 2 \| 2.0 \| 2.0 MIN function The MIN function returns the minimum value in a set of rows. DISTINCT or ALL might be used but do not aﬀect the result. Syntax MIN ( [ DISTINCT \| ALL ] expression ) Arguments expression The target column or expression that the function operates on. The expression is any numerical data type. DISTINCT \| ALL With the argument DISTINCT, the function eliminates all duplicate values from the speciﬁed expression before calculating the minimum. With the argument ALL, the function retains all duplicate values from the expression for calculating the minimum. ALL is the default. Data types Returns the same data type as expression. Examples Find the lowest price paid from all sales: Aggregate functions 117 AWS Clean Rooms SQL Reference select min(pricepaid) from sales; min ------- 20.00 (1 row) Find the lowest price paid per ticket from all sales: select min(pricepaid/qtysold)as min_ticket_price from sales; min_ticket_price ------------------ 20.00000000 (1 row) PERCENTILE function The PERCENTILE function is used to calculates the exact percentile value by ﬁrst sorting the values in the col column and then ﬁnding the value at the speciﬁed percentage. The PERCENTILE function is useful when you need to calculate the exact percentile value and the computational cost is acceptable for your use case. It provides more accurate results than the APPROX_PERCENTILE function, but may be slower, especially for large datasets. In contrast, the APPROX_PERCENTILE function is a more eﬃcient alternative that can provide an estimate of the percentile value with a speciﬁed error tolerance, making it more suitable for scenarios where speed is a higher priority than absolute precision. Syntax percentile(col, percentage [, frequency]) Arguments col The expression or column for which you want to calculate the percentile value. Aggregate functions 118 AWS Clean Rooms percentage SQL Reference The percentile value you want to calculate, expressed as a value between 0 and 1. For example, 0.5 would correspond to the 50th percentile (median). frequency An optional parameter that speciﬁes the frequency or weight of each value in the col column. If provided, the function will calculate the percentile based on the frequency of each value. Returns Returns the exact percentile value of numeric or ANSI interval column col at the given percentage. The value of percentage must be between 0.0 and 1.0. The value of frequency should be positive integral Example The following query ﬁnds the value that is greater than or equal to 30% of the values in the col column. Since the values are 0 and 10, the 30th percentile is 3.0, because it is the value that is greater than or equal to 30% of the data. SELECT percentile(col, 0.3) FROM VALUES (0), (10) AS tab(col); 3.0 PERCENTILE_CONT function PERCENTILE_CONT is an inverse distribution function that assumes a continuous distribution model. It takes a percentile value and a sort speciﬁcation, and returns an interpolated value that would fall into the given percentile value with respect to the sort speciﬁcation. PERCENTILE_CONT computes a linear interpolation between values after ordering them. Using the percentile value (P) and the number of not null rows (N) in the aggregation group, the function computes the row number after ordering the rows according to the sort speciﬁcation. This row number (RN) is computed according to the formula RN = (1+ (P(N-1)). The ﬁnal result of the aggregate function is computed by linear interpolation between the values from rows at row numbers CRN = CEILING(RN) and FRN = FLOOR(RN). The ﬁnal result will be as follows. Aggregate functions 119 AWS Clean Rooms SQL Reference If (CRN = FRN = RN) then the result is (value of expression from row at RN) Otherwise the result is as follows: (CRN - RN) (value of expression for row at FRN) + (RN - FRN) * (value of expression for row at CRN). PERCENTILE_CONT is a compute-node only function. The function returns an error if the query doesn't reference a user-deﬁned table or AWS Clean Rooms system table. Syntax PERCENTILE_CONT ( percentile ) WITHIN GROUP (ORDER BY expr) Arguments percentile Numeric constant between 0 and 1. Nulls are ignored in the calculation. WITHIN GROUP ( ORDER BY expr) Speciﬁes numeric or date/time values to sort and compute the percentile over. Returns The return type is determined by the data type of the ORDER BY expression in the WITHIN GROUP clause. Examples The following example shows that MEDIAN produces the same results as PERCENTILE_CONT(0.5). select top 10 distinct sellerid, qtysold, percentile_cont(0.5) within group (order by qtysold), median (qtysold) from sales group by sellerid, qtysold; sellerid \| qtysold \| percentile_cont \| median ---------+---------+-----------------+------- 1 \| 1 \| 1.0 \| 1.0 2 \| 3 \| 3.0 \| 3.0 Aggregate functions 120 AWS Clean Rooms SQL Reference 5 \| 2 \| 2.0 \| 2.0 9
sql-reference-038	sql-reference.pdf	38	date/time values to sort and compute the percentile over. Returns The return type is determined by the data type of the ORDER BY expression in the WITHIN GROUP clause. Examples The following example shows that MEDIAN produces the same results as PERCENTILE_CONT(0.5). select top 10 distinct sellerid, qtysold, percentile_cont(0.5) within group (order by qtysold), median (qtysold) from sales group by sellerid, qtysold; sellerid \| qtysold \| percentile_cont \| median ---------+---------+-----------------+------- 1 \| 1 \| 1.0 \| 1.0 2 \| 3 \| 3.0 \| 3.0 Aggregate functions 120 AWS Clean Rooms SQL Reference 5 \| 2 \| 2.0 \| 2.0 9 \| 4 \| 4.0 \| 4.0 12 \| 1 \| 1.0 \| 1.0 16 \| 1 \| 1.0 \| 1.0 19 \| 2 \| 2.0 \| 2.0 19 \| 3 \| 3.0 \| 3.0 22 \| 2 \| 2.0 \| 2.0 25 \| 2 \| 2.0 \| 2.0 SKEWNESS function The SKEWNESS function returns the skewness value calculated from values of a group. Skewness is a statistical measure that describes the asymmetry or lack of symmetry in a dataset. It provides information about the shape of the data distribution. This function can be useful in understanding the statistical properties of a dataset and informing further analysis or decision-making. Syntax skewness(expr) Arguments expr An expression that evaluates to a numeric. Returns Returns DOUBLE. If DISTINCT is speciﬁed, the function operates only on a unique set of expr values. Examples The following query calculates the skewness of the values in the col column. In this example, the VALUES clause is used to create an inline table with four rows, where each row has a single column col with the values -10, -20, 100, and 1000. The skewness() function is then used to calculate the skewness of the values in the col column. The result, 1.1135657469022011, represents the degree and direction of skewness in the data. A positive skewness value indicates that the data is Aggregate functions 121 AWS Clean Rooms SQL Reference skewed to the right, with the bulk of the values concentrated on the left side of the distribution. A negative skewness value indicates that the data is skewed to the left, with the bulk of the values concentrated on the right side of the distribution. SELECT skewness(col) FROM VALUES (-10), (-20), (100), (1000) AS tab(col); 1.1135657469022011 The following query calculates the skewness of the values in the col column. Similar to the previous example, the VALUES clause is used to create an inline table with four rows, where each row has a single column col with the values -1000, -100, 10, and 20. The skewness() function is then used to calculate the skewness of the values in the col column. The result, -1.1135657469022011, represents the degree and direction of skewness in the data. In this case, the negative skewness value indicates that the data is skewed to the left, with the bulk of the values concentrated on the right side of the distribution. SELECT skewness(col) FROM VALUES (-1000), (-100), (10), (20) AS tab(col); -1.1135657469022011 STDDEV_SAMP and STDDEV_POP functions The STDDEV_SAMP and STDDEV_POP functions return the sample and population standard deviation of a set of numeric values (integer, decimal, or ﬂoating-point). The result of the STDDEV_SAMP function is equivalent to the square root of the sample variance of the same set of values. STDDEV_SAMP and STDDEV are synonyms for the same function. Syntax STDDEV_SAMP \| STDDEV ( [ DISTINCT \| ALL ] expression) STDDEV_POP ( [ DISTINCT \| ALL ] expression) The expression must have numeric data type. Regardless of the data type of the expression, the return type of this function is a double precision number. Note Standard deviation is calculated using ﬂoating point arithmetic, which might result in slight imprecision. Aggregate functions 122 AWS Clean Rooms Usage notes SQL Reference When the sample standard deviation (STDDEV or STDDEV_SAMP) is calculated for an expression that consists of a single value, the result of the function is NULL not 0. Examples The following query returns the average of the values in the VENUESEATS column of the VENUE table, followed by the sample standard deviation and population standard deviation of the same set of values. VENUESEATS is an INTEGER column. The scale of the result is reduced to 2 digits. select avg(venueseats), cast(stddev_samp(venueseats) as dec(14,2)) stddevsamp, cast(stddev_pop(venueseats) as dec(14,2)) stddevpop from venue; avg \| stddevsamp \| stddevpop -------+------------+----------- 17503 \| 27847.76 \| 27773.20 (1 row) The following query returns the sample standard deviation for the COMMISSION column in the SALES table. COMMISSION is a DECIMAL column. The scale of the result is reduced to 10 digits. select cast(stddev(commission) as dec(18,10)) from sales; stddev ---------------- 130.3912659086 (1 row) The following query casts the sample standard deviation for the COMMISSION column as an integer. select cast(stddev(commission) as integer) from sales; stddev -------- 130 (1
sql-reference-039	sql-reference.pdf	39	of the result is reduced to 2 digits. select avg(venueseats), cast(stddev_samp(venueseats) as dec(14,2)) stddevsamp, cast(stddev_pop(venueseats) as dec(14,2)) stddevpop from venue; avg \| stddevsamp \| stddevpop -------+------------+----------- 17503 \| 27847.76 \| 27773.20 (1 row) The following query returns the sample standard deviation for the COMMISSION column in the SALES table. COMMISSION is a DECIMAL column. The scale of the result is reduced to 10 digits. select cast(stddev(commission) as dec(18,10)) from sales; stddev ---------------- 130.3912659086 (1 row) The following query casts the sample standard deviation for the COMMISSION column as an integer. select cast(stddev(commission) as integer) from sales; stddev -------- 130 (1 row) Aggregate functions 123 AWS Clean Rooms SQL Reference The following query returns both the sample standard deviation and the square root of the sample variance for the COMMISSION column. The results of these calculations are the same. select cast(stddev_samp(commission) as dec(18,10)) stddevsamp, cast(sqrt(var_samp(commission)) as dec(18,10)) sqrtvarsamp from sales; stddevsamp \| sqrtvarsamp ----------------+---------------- 130.3912659086 \| 130.3912659086 (1 row) SUM and SUM DISTINCT functions The SUM function returns the sum of the input column or expression values. The SUM function works with numeric values and ignores NULL values. The SUM DISTINCT function eliminates all duplicate values from the speciﬁed expression before calculating the sum. Syntax SUM (DISTINCT column ) Arguments column The target column that the function operates on. The column is any numeric data types. Examples Find the sum of all commissions paid from the SALES table. select sum(commission) from sales Find the sum of all distinct commissions paid from the SALES table. select sum (distinct (commission)) from sales Aggregate functions 124 AWS Clean Rooms SQL Reference VAR_SAMP and VAR_POP functions The VAR_SAMP and VAR_POP functions return the sample and population variance of a set of numeric values (integer, decimal, or ﬂoating-point). The result of the VAR_SAMP function is equivalent to the squared sample standard deviation of the same set of values. VAR_SAMP and VARIANCE are synonyms for the same function. Syntax VAR_SAMP \| VARIANCE ( [ DISTINCT \| ALL ] expression) VAR_POP ( [ DISTINCT \| ALL ] expression) The expression must have an integer, decimal, or ﬂoating-point data type. Regardless of the data type of the expression, the return type of this function is a double precision number. Note The results of these functions might vary across data warehouse clusters, depending on the conﬁguration of the cluster in each case. Usage notes When the sample variance (VARIANCE or VAR_SAMP) is calculated for an expression that consists of a single value, the result of the function is NULL not 0. Examples The following query returns the rounded sample and population variance of the NUMTICKETS column in the LISTING table. select avg(numtickets), round(var_samp(numtickets)) varsamp, round(var_pop(numtickets)) varpop from listing; avg \| varsamp \| varpop -----+---------+-------- 10 \| 54 \| 54 (1 row) Aggregate functions 125 AWS Clean Rooms SQL Reference The following query runs the same calculations but casts the results to decimal values. select avg(numtickets), cast(var_samp(numtickets) as dec(10,4)) varsamp, cast(var_pop(numtickets) as dec(10,4)) varpop from listing; avg \| varsamp \| varpop -----+---------+--------- 10 \| 53.6291 \| 53.6288 (1 row) Array functions This section describes the array functions for SQL supported in AWS Clean Rooms. Topics • ARRAY function • ARRAY_CONTAINS function • ARRAY_DISTINCT function • ARRAY_EXCEPT function • ARRAY_INSERT function • ARRAY_INTERSECT function • ARRAY_JOIN function • ARRAY_REMOVE function • ARRAY_UNION function • EXPLODE function • FLATTEN function ARRAY function Creates an array with the given elements. Syntax ARRAY( [ expr1 ] [ , expr2 [ , ... ] ] ) Array functions 126 AWS Clean Rooms Argument expr1, expr2 SQL Reference Expressions of any data type except date and time types. The arguments don't need to be of the same data type. Return type The array function returns an ARRAY with the elements in the expression. Example The following example shows an array of numeric values and an array of diﬀerent data types. --an array of numeric values select array(1,50,null,100); array ------------------ [1,50,null,100] (1 row) --an array of different data types select array(1,'abc',true,3.14); array ----------------------- [1,"abc",true,3.14] (1 row) ARRAY_CONTAINS function The ARRAY_CONTAINS function can be used to perform basic membership checks on array data structures. The ARRAY_CONTAINS function is useful when you need to check if a speciﬁc value is present within an array. Syntax array_contains(array, value) Array functions 127 AWS Clean Rooms Arguments array An ARRAY to be searched. value SQL Reference An expression with a type sharing a least common type with the array elements. Return type The ARRAY_CONTAINS function returns a BOOLEAN. If value is NULL, the result is NULL. If any element in array is NULL, the result is NULL if value is not matched to any other element. Examples The following example checks if the array [1, 2, 3] contains the value 4. Since the array [1, 2, 3] doesn't contain the value 4, the
sql-reference-040	sql-reference.pdf	40	present within an array. Syntax array_contains(array, value) Array functions 127 AWS Clean Rooms Arguments array An ARRAY to be searched. value SQL Reference An expression with a type sharing a least common type with the array elements. Return type The ARRAY_CONTAINS function returns a BOOLEAN. If value is NULL, the result is NULL. If any element in array is NULL, the result is NULL if value is not matched to any other element. Examples The following example checks if the array [1, 2, 3] contains the value 4. Since the array [1, 2, 3] doesn't contain the value 4, the array_contains function returns false. SELECT array_contains(array(1, 2, 3), 4) false The following example checks if the array [1, 2, 3] contains the value 2. Since the array [1, 2, 3] does contain the value 2, the array_contains function returns true. SELECT array_contains(array(1, 2, 3), 2); true ARRAY_DISTINCT function The ARRAY_DISTINCT function can be used to remove duplicate values from an array. The ARRAY_DISTINCT function is useful when you need to remove duplicates from an array and work with only the unique elements. This can be helpful in scenarios where you want to perform operations or analyses on a dataset without the interference of repeated values. Syntax array_distinct(array) Array functions 128 AWS Clean Rooms Arguments array An ARRAY expression. Return type SQL Reference The ARRAY_DISTINCT function returns an ARRAY that contains only the unique elements from the input array. Examples In this example, the input array [1, 2, 3, null, 3] contains a duplicate value of 3. The array_distinct function removes this duplicate value 3 and returns a new array with the unique elements: [1, 2, 3, null]. SELECT array_distinct(array(1, 2, 3, null, 3)); [1,2,3,null] In this example, the input array [1, 2, 2, 3, 3, 3] contains duplicate values of 2 and 3. The array_distinct function removes these duplicates and returns a new array with the unique elements: [1, 2, 3]. SELECT array_distinct(array(1, 2, 2, 3, 3, 3)) [1,2,3] ARRAY_EXCEPT function The ARRAY_EXCEPT function takes two arrays as arguments and returns a new array that contains only the elements that are present in the ﬁrst array but not the second array. The ARRAY_EXCEPT is useful when you need to ﬁnd the elements that are unique to one array compared to another. This can be helpful in scenarios where you need to perform set-like operations on arrays, such as ﬁnding the diﬀerence between two sets of data. Syntax array_except(array1, array2) Array functions 129 AWS Clean Rooms Arguments array1 SQL Reference An ARRAY of any type with comparable elements. array2 An ARRAY of elements sharing a least common type with the elements of array1. Return type The ARRAY_EXCEPT function returns an ARRAY of matching type to array1 with no duplicates. Examples In this example, the ﬁrst array [1, 2, 3] contains the elements 1, 2, and 3. The second array [2, 3, 4] contains the elements 2, 3, and 4. The array_except function removes the elements 2 and 3 from the ﬁrst array, since they're also present in the second array. The resulting output is the array [1]. SELECT array_except(array(1, 2, 3), array(2, 3, 4)) [1] In this example, the ﬁrst array [1, 2, 3] contains the elements 1, 2, and 3. The second array [1, 3, 5] contains the elements 1, 3, and 5. The array_except function removes the elements 1 and 3 from the ﬁrst array, since they're also present in the second array. The resulting output is the array [2]. SELECT array_except(array(1, 2, 3), array(1, 3, 5)); [2] ARRAY_INSERT function The ARRAY_INSERT function is useful when you need to add a new element to an existing array at a speciﬁc index position. This can be helpful in scenarios where you need to modify an array by inserting new elements at speciﬁc locations. Syntax array_insert(x, pos, val) Array functions 130 AWS Clean Rooms Arguments x An ARRAY. Array indices start at 1. pos SQL Reference A non-zero INTEGER expression specifying where to insert val. If pos is negative, val is inserted relative to the end of the array. val An expression of the same type as the elements of array. Return type The ARRAY_INSERT function returns an ARRAY of the same type as array. Example In this example, the ARRAY_INSERT function inserts the value 5 at the index position 5 (the 6th element) in the array [1, 2, 3, 4]. This results in the output array [1, 2, 3, 4, 5]. SELECT array_insert(array(1, 2, 3, 4), 5, 5); [1,2,3,4,5] ARRAY_INTERSECT function The ARRAY_INTERSECT function takes two arrays as arguments and returns a new array that contains the elements that are present in both input arrays. This function is useful when you need to ﬁnd the common elements between two arrays. This can be helpful in scenarios where you need to
sql-reference-041	sql-reference.pdf	41	same type as array. Example In this example, the ARRAY_INSERT function inserts the value 5 at the index position 5 (the 6th element) in the array [1, 2, 3, 4]. This results in the output array [1, 2, 3, 4, 5]. SELECT array_insert(array(1, 2, 3, 4), 5, 5); [1,2,3,4,5] ARRAY_INTERSECT function The ARRAY_INTERSECT function takes two arrays as arguments and returns a new array that contains the elements that are present in both input arrays. This function is useful when you need to ﬁnd the common elements between two arrays. This can be helpful in scenarios where you need to perform set-like operations on arrays, such as ﬁnding the intersection between two sets of data. Syntax array_intersect(array1, array2) Arguments array1 An ARRAY of any type with comparable elements. Array functions 131 AWS Clean Rooms array2 SQL Reference An ARRAY of elements sharing a least common type with the elements of array1. Return type The ARRAY_INTERSECT function returns an ARRAY of matching type to array1 with no duplicates and elements contained in both array1 and array2. Examples In this example, the ﬁrst array [1, 2, 3] contains the elements 1, 2, and 3. The second array [1, 3, 5] contains the elements 1, 3, and 5. The ARRAY_INTERSECT function identiﬁes the common elements between the two arrays, which are 1 and 3. The resulting output array is [1, 3]. SELECT array_intersect(array(1, 2, 3), array(1, 3, 5)); [1,3] ARRAY_JOIN function The ARRAY_JOIN function takes two arguments: The ﬁrst argument is the input array that will be joined. The second argument is the separator string that will be used to concatenate the array elements. This function is useful when you need to convert an array of strings (or any other data type) into a single concatenated string. This can be helpful in scenarios where you want to present an array of values as a single formatted string, such as for display purposes or for use in further processing. Syntax array_join(array, delimiter[, nullReplacement]) Arguments array Any ARRAY type, but its elements are interpreted as strings. delimiter A STRING used to separate the concatenated array elements. Array functions 132 AWS Clean Rooms nullReplacement A STRING used to express a NULL value in the result. Return type SQL Reference The ARRAY_JOIN function returns a STRING where the elements of array are separated by delimiter and null elements are substituted for nullReplacement. If nullReplacement is omitted, null elements are ﬁltered out. If any argument is NULL, the result is NULL. Examples In this example, the ARRAY_JOIN function takes the array ['hello', 'world'] and joins the elements using the separator ' ' (a space character). The resulting output is the string 'hello world'. SELECT array_join(array('hello', 'world'), ' '); hello world In this example, the ARRAY_JOIN function takes the array ['hello', null, 'world'] and joins the elements using the separator ' ' (a space character). The null value is replaced with the provided replacement string ',' (a comma). The resulting output is the string 'hello , world'. SELECT array_join(array('hello', null ,'world'), ' ', ','); hello , world ARRAY_REMOVE function The ARRAY_REMOVE function takes two arguments: The ﬁrst argument is the input array from which the elements will be removed. The second argument is the value that will be removed from the array. This function is useful when you need to remove speciﬁc elements from an array. This can be helpful in scenarios where you need to perform data cleaning or preprocessing on an array of values. Syntax array_remove(array, element) Array functions 133 AWS Clean Rooms Arguments array An ARRAY. element SQL Reference An expression of a type sharing a least common type with the elements of array. Return type The ARRAY_REMOVE function returns the result type matched the type of the array. If the element to be removed is NULL, the result is NULL. Examples In this example, the ARRAY_REMOVE function takes the array [1, 2, 3, null, 3] and removes all occurrences of the value 3. The resulting output is the array [1, 2, null]. SELECT array_remove(array(1, 2, 3, null, 3), 3); [1,2,null] ARRAY_UNION function The ARRAY_UNION function takes two arrays as arguments and returns a new array that contains the unique elements from both input arrays. This function is useful when you need to combine two arrays and eliminate any duplicate elements. This can be helpful in scenarios where you need to perform set-like operations on arrays, such as ﬁnding the union between two sets of data. Syntax array_union(array1, array2) Arguments array1 An ARRAY. array2 An ARRAY of the same type as array1. Array functions 134 AWS Clean Rooms Return type SQL Reference The ARRAY_UNION function returns an ARRAY of the same type as array. Example In this example, the ﬁrst array [1, 2, 3] contains the elements 1, 2, and 3. The second array [1, 3,
sql-reference-042	sql-reference.pdf	42	function is useful when you need to combine two arrays and eliminate any duplicate elements. This can be helpful in scenarios where you need to perform set-like operations on arrays, such as ﬁnding the union between two sets of data. Syntax array_union(array1, array2) Arguments array1 An ARRAY. array2 An ARRAY of the same type as array1. Array functions 134 AWS Clean Rooms Return type SQL Reference The ARRAY_UNION function returns an ARRAY of the same type as array. Example In this example, the ﬁrst array [1, 2, 3] contains the elements 1, 2, and 3. The second array [1, 3, 5] contains the elements 1, 3, and 5. The ARRAY_UNION function combines the unique elements from both arrays, resulting in the output array [1, 2, 3, 5]. T SELECT array_union(array(1, 2, 3), array(1, 3, 5)); [1,2,3,5] EXPLODE function The EXPLODE function is used to transform a single row with an array or map column into multiple rows, where each row corresponds to a single element from the array or map. Syntax explode(expr) Arguments expr An array expression or a map expression. Return type The EXPLODE function returns a set of rows, where each row represents a single element from the input array or map. The data type of the output rows depends on the data type of the elements in the input array or map. Examples The following example takes the single-row array [10, 20] and transforms it into two separate rows, each containing one of the array elements (10 and 20). Array functions 135 AWS Clean Rooms SQL Reference SELECT explode(array(10, 20)); In the ﬁrst example, the input array was directly passed as an argument to explode(). In this example, the input array is speciﬁed using the => syntax, where the column name (collection) is explicitly provided. SELECT explode(array(10, 20)); Both approaches are valid and achieve the same result, but the second syntax can be more useful when you need to explode a column from a larger dataset, rather than just a simple array literal. FLATTEN function The FLATTEN function is used to "ﬂatten" a nested array structure into a single ﬂat array. Syntax flatten(arrayOfArrays) Arguments arrayOfArrays An array of arrays. Return type The FLATTEN function returns an array. Example In this example, the input is a nested array with two inner arrays, and the output is a single ﬂat array containing all the elements from the inner arrays. The FLATTEN function takes the nested array [[1, 2], [3, 4]] and combines all the elements into a single array [1, 2, 3, 4]. SELECT flatten(array(array(1, 2), array(3, 4))); [1,2,3,4] Array functions 136 AWS Clean Rooms Binary functions SQL Reference AWS Clean Rooms Spark SQL supports the following BINARY functions. Topics • BINARY function BINARY function BINARY converts an expression to a binary value. The binary representation of a number is the way the number is stored in computer memory, using a series of 1s and 0s. The BINARY() function is useful when you need to work with the underlying binary representation of a value, for example, in certain data manipulation or comparison operations. This function is a synonym for CAST(expr AS BINARY). Syntax binary(expr) Arguments expr An expression to be converted to BINARY. Return type BINARY Example The following exxample converts the integer 2024 to its binary representation and displays the resulting binary string. SELECT binary(2024); [00 00 07 C0] Binary functions 137 AWS Clean Rooms SQL Reference Conditional expressions In SQL, conditional expressions are used to make decisions based on certain conditions. They allow you to control the ﬂow of your SQL statements and return diﬀerent values or perform diﬀerent actions based on the evaluation of one or more conditions. AWS Clean Rooms supports the following conditional expressions: Topics • CASE conditional expression • COALESCE expression • GREATEST and LEAST expression • IF expression • ISNULL expression • ISNOTNULL expression • NVL and COALESCE functions • NVL2 function • NULLIF function CASE conditional expression The CASE expression is a conditional expression, similar to if/then/else statements found in other languages. CASE is used to specify a result when there are multiple conditions. Use CASE where a SQL expression is valid, such as in a SELECT command. There are two types of CASE expressions: simple and searched. • In simple CASE expressions, an expression is compared with a value. When a match is found, the speciﬁed action in the THEN clause is applied. If no match is found, the action in the ELSE clause is applied. • In searched CASE expressions, each CASE is evaluated based on a Boolean expression, and the CASE statement returns the ﬁrst matching CASE. If no match is found among the WHEN clauses, the action in the ELSE clause is returned. Conditional expressions 138 SQL Reference AWS Clean Rooms Syntax Simple CASE statement used to
sql-reference-043	sql-reference.pdf	43	types of CASE expressions: simple and searched. • In simple CASE expressions, an expression is compared with a value. When a match is found, the speciﬁed action in the THEN clause is applied. If no match is found, the action in the ELSE clause is applied. • In searched CASE expressions, each CASE is evaluated based on a Boolean expression, and the CASE statement returns the ﬁrst matching CASE. If no match is found among the WHEN clauses, the action in the ELSE clause is returned. Conditional expressions 138 SQL Reference AWS Clean Rooms Syntax Simple CASE statement used to match conditions: CASE expression WHEN value THEN result [WHEN...] [ELSE result] END Searched CASE statement used to evaluate each condition: CASE WHEN condition THEN result [WHEN ...] [ELSE result] END Arguments expression A column name or any valid expression. value Value that the expression is compared with, such as a numeric constant or a character string. result The target value or expression that is returned when an expression or Boolean condition is evaluated. The data types of all the result expressions must be convertible to a single output type. condition A Boolean expression that evaluates to true or false. If condition is true, the value of the CASE expression is the result that follows the condition, and the remainder of the CASE expression is not processed. If condition is false, any subsequent WHEN clauses are evaluated. If no WHEN condition results are true, the value of the CASE expression is the result of the ELSE clause. If the ELSE clause is omitted and no condition is true, the result is null. Conditional expressions 139 AWS Clean Rooms Examples SQL Reference Use a simple CASE expression to replace New York City with Big Apple in a query against the VENUE table. Replace all other city names with other. select venuecity, case venuecity when 'New York City' then 'Big Apple' else 'other' end from venue order by venueid desc; venuecity \| case -----------------+----------- Los Angeles \| other New York City \| Big Apple San Francisco \| other Baltimore \| other ... Use a searched CASE expression to assign group numbers based on the PRICEPAID value for individual ticket sales: select pricepaid, case when pricepaid <10000 then 'group 1' when pricepaid >10000 then 'group 2' else 'group 3' end from sales order by 1 desc; pricepaid \| case ----------+--------- 12624 \| group 2 10000 \| group 3 10000 \| group 3 9996 \| group 1 9988 \| group 1 ... Conditional expressions 140 AWS Clean Rooms COALESCE expression SQL Reference A COALESCE expression returns the value of the ﬁrst expression in the list that is not null. If all expressions are null, the result is null. When a non-null value is found, the remaining expressions in the list are not evaluated. This type of expression is useful when you want to return a backup value for something when the preferred value is missing or null. For example, a query might return one of three phone numbers (cell, home, or work, in that order), whichever is found ﬁrst in the table (not null). Syntax COALESCE (expression, expression, ... ) Examples Apply COALESCE expression to two columns. select coalesce(start_date, end_date) from datetable order by 1; The default column name for an NVL expression is COALESCE. The following query returns the same results. select coalesce(start_date, end_date) from datetable order by 1; GREATEST and LEAST expression Returns the largest or smallest value from a list of any number of expressions. Syntax GREATEST (value [, ...]) LEAST (value [, ...]) Conditional expressions 141 AWS Clean Rooms Parameters expression_list SQL Reference A comma-separated list of expressions, such as column names. The expressions must all be convertible to a common data type. NULL values in the list are ignored. If all of the expressions evaluate to NULL, the result is NULL. Returns Returns the greatest (for GREATEST) or least (for LEAST) value from the provided list of expressions. Example The following example returns the highest value alphabetically for firstname or lastname. select firstname, lastname, greatest(firstname,lastname) from users where userid < 10 order by 3; firstname \| lastname \| greatest -----------+-----------+----------- Alejandro \| Rosalez \| Ratliff Carlos \| Salazar \| Carlos Jane \| Doe \| Doe John \| Doe \| Doe John \| Stiles \| John Shirley \| Rodriguez \| Rodriguez Terry \| Whitlock \| Terry Richard \| Roe \| Richard Xiulan \| Wang \| Wang (9 rows) IF expression The IF conditional function returns one of two values based on a condition. This function is a common control ﬂow statement used in SQL to make decisions and return diﬀerent values based on the evaluation of a condition. It's useful for implementing simple if-else logic within a query. Conditional expressions 142 AWS Clean Rooms Syntax if(expr1, expr2, expr3) Arguments expr1 SQL
sql-reference-044	sql-reference.pdf	44	Jane \| Doe \| Doe John \| Doe \| Doe John \| Stiles \| John Shirley \| Rodriguez \| Rodriguez Terry \| Whitlock \| Terry Richard \| Roe \| Richard Xiulan \| Wang \| Wang (9 rows) IF expression The IF conditional function returns one of two values based on a condition. This function is a common control ﬂow statement used in SQL to make decisions and return diﬀerent values based on the evaluation of a condition. It's useful for implementing simple if-else logic within a query. Conditional expressions 142 AWS Clean Rooms Syntax if(expr1, expr2, expr3) Arguments expr1 SQL Reference The condition or expression that is evaluated. If it is true, the function will return the value of expr2. If expr1 is false, the function will return the value of expr3. expr2 The expression that is evaluated and returned if expr1 is true. expr3 The expression that is evaluated and returned if expr1 is false. Returns If expr1 evaluates to true, then returns expr2; otherwise returns expr3. Example The following example uses the if() function to return one of two values based on a condition. The condition being evaluated is 1 < 2, which is true, so the ﬁrst value 'a' is returned. SELECT if(1 < 2, 'a', 'b'); a] ISNULL expression The ISNULL conditional expression is used to check if a value is null. This expression is a synonym for IS NULL. Syntax isnull(expr) Conditional expressions 143 AWS Clean Rooms Arguments expr An expression of any type. Returns SQL Reference The ISNULL conditional expression returns a Boolean. If expr1 is NULL, returns true, otherwise returns false. Examples The following example checks if the value 1 is null, and returns the boolean result true because 1 is a valid, non-null value. SELECT is not null(1); true The following example selects the id column from the squirrels table, but only for the rows where the age column is null. SELECT id FROM squirrels WHERE isnull(age) ISNOTNULL expression The ISNOTNULL conditional expression is used to check if a value is not null. This expression is a synonym for IS NOT NULL. Syntax isnotnull(expr) Arguments expr An expression of any type. Conditional expressions 144 AWS Clean Rooms Returns SQL Reference The ISNOTNULL conditional expression returns a Boolean. If expr1 is not NULL, returns true, otherwise returns false. Examples The following example checks if the value 1 is not null, and returns the boolean result true because 1 is a valid, non-null value. SELECT is not null(1); true The following example selects the id column from the squirrels table, but only for the rows where the age column is not null. SELECT id FROM squirrels WHERE isnotnull(age) NVL and COALESCE functions Returns the value of the ﬁrst expression that isn't null in a series of expressions. When a non-null value is found, the remaining expressions in the list aren't evaluated. NVL is identical to COALESCE. They are synonyms. This topic explains the syntax and contains examples for both. Syntax NVL( expression, expression, ... ) The syntax for COALESCE is the same: COALESCE( expression, expression, ... ) If all expressions are null, the result is null. These functions are useful when you want to return a secondary value when a primary value is missing or null. For example, a query might return the ﬁrst of three available phone numbers: cell, home, or work. The order of the expressions in the function determines the order of evaluation. Conditional expressions 145 AWS Clean Rooms Arguments expression SQL Reference An expression, such as a column name, to be evaluated for null status. Return type AWS Clean Rooms determines the data type of the returned value based on the input expressions. If the data types of the input expressions don't have a common type, then an error is returned. Examples If the list contains integer expressions, the function returns an integer. SELECT COALESCE(NULL, 12, NULL); coalesce -------------- 12 This example, which is the same as the previous example, except that it uses NVL, returns the same result. SELECT NVL(NULL, 12, NULL); coalesce -------------- 12 The following example returns a string type. SELECT COALESCE(NULL, 'AWS Clean Rooms', NULL); coalesce -------------- AWS Clean Rooms The following example results in an error because the data types vary in the expression list. In this case, there is both a string type and a number type in the list. Conditional expressions 146 AWS Clean Rooms SQL Reference SELECT COALESCE(NULL, 'AWS Clean Rooms', 12); ERROR: invalid input syntax for integer: "AWS Clean Rooms" NVL2 function Returns one of two values based on whether a speciﬁed expression evaluates to NULL or NOT NULL. Syntax NVL2 ( expression, not_null_return_value, null_return_value ) Arguments expression An expression, such as a column name, to be evaluated for null status. not_null_return_value The value returned if expression evaluates to NOT NULL.
sql-reference-045	sql-reference.pdf	45	error because the data types vary in the expression list. In this case, there is both a string type and a number type in the list. Conditional expressions 146 AWS Clean Rooms SQL Reference SELECT COALESCE(NULL, 'AWS Clean Rooms', 12); ERROR: invalid input syntax for integer: "AWS Clean Rooms" NVL2 function Returns one of two values based on whether a speciﬁed expression evaluates to NULL or NOT NULL. Syntax NVL2 ( expression, not_null_return_value, null_return_value ) Arguments expression An expression, such as a column name, to be evaluated for null status. not_null_return_value The value returned if expression evaluates to NOT NULL. The not_null_return_value value must either have the same data type as expression or be implicitly convertible to that data type. null_return_value The value returned if expression evaluates to NULL. The null_return_value value must either have the same data type as expression or be implicitly convertible to that data type. Return type The NVL2 return type is determined as follows: • If either not_null_return_value or null_return_value is null, the data type of the not-null expression is returned. If both not_null_return_value and null_return_value are not null: • If not_null_return_value and null_return_value have the same data type, that data type is returned. • If not_null_return_value and null_return_value have diﬀerent numeric data types, the smallest compatible numeric data type is returned. Conditional expressions 147 AWS Clean Rooms SQL Reference • If not_null_return_value and null_return_value have diﬀerent datetime data types, a timestamp data type is returned. • If not_null_return_value and null_return_value have diﬀerent character data types, the data type of not_null_return_value is returned. • If not_null_return_value and null_return_value have mixed numeric and non-numeric data types, the data type of not_null_return_value is returned. Important In the last two cases where the data type of not_null_return_value is returned, null_return_value is implicitly cast to that data type. If the data types are incompatible, the function fails. Usage notes For NVL2, the return will have the value of either the not_null_return_value or null_return_value parameter, whichever is selected by the function, but will have the data type of not_null_return_value. For example, assuming column1 is NULL, the following queries will return the same value. However, the DECODE return value data type will be INTEGER and the NVL2 return value data type will be VARCHAR. select decode(column1, null, 1234, '2345'); select nvl2(column1, '2345', 1234); Example The following example modiﬁes some sample data, then evaluates two ﬁelds to provide appropriate contact information for users: update users set email = null where firstname = 'Aphrodite' and lastname = 'Acevedo'; select (firstname + ' ' + lastname) as name, nvl2(email, email, phone) AS contact_info from users where state = 'WA' Conditional expressions 148 AWS Clean Rooms SQL Reference and lastname like 'A%' order by lastname, firstname; name contact_info --------------------+------------------------------------------- Aphrodite Acevedo (555) 555-0100 Caldwell Acevedo Nunc.sollicitudin@example.ca Quinn Adams vel@example.com Kamal Aguilar quis@example.com Samson Alexander hendrerit.neque@example.com Hall Alford ac.mattis@example.com Lane Allen et.netus@example.com Xander Allison ac.facilisis.facilisis@example.com Amaya Alvarado dui.nec.tempus@example.com Vera Alvarez at.arcu.Vestibulum@example.com Yetta Anthony enim.sit@example.com Violet Arnold ad.litora@example.comm August Ashley consectetuer.euismod@example.com Karyn Austin ipsum.primis.in@example.com Lucas Ayers at@example.com NULLIF function Compares two arguments and returns null if the arguments are equal. If they aren't equal, the ﬁrst argument is returned. Syntax The NULLIF expression compares two arguments and returns null if the arguments are equal. If they aren't equal, the ﬁrst argument is returned. This expression is the inverse of the NVL or COALESCE expression. NULLIF ( expression1, expression2 ) Arguments expression1, expression2 The target columns or expressions that are compared. The return type is the same as the type of the ﬁrst expression. Conditional expressions 149 AWS Clean Rooms Examples SQL Reference In the following example, the query returns the string first because the arguments are not equal. SELECT NULLIF('first', 'second'); case ------- first In the following example, the query returns NULL because the string literal arguments are equal. SELECT NULLIF('first', 'first'); case ------- NULL In the following example, the query returns 1 because the integer arguments are not equal. SELECT NULLIF(1, 2); case ------- 1 In the following example, the query returns NULL because the integer arguments are equal. SELECT NULLIF(1, 1); case ------- NULL In the following example, the query returns null when the LISTID and SALESID values match: select nullif(listid,salesid), salesid from sales where salesid<10 order by 1, 2 desc; listid \| salesid --------+--------- 4 \| 2 Conditional expressions 150 SQL Reference AWS Clean Rooms 5 \| 4 5 \| 3 6 \| 5 10 \| 9 10 \| 8 10 \| 7 10 \| 6 \| 1 (9 rows) Constructor functions A SQL constructor function is a function that is used to create new data structures, such as arrays or maps. They take some input values and return a new data structure object. Constructor functions are typically named after the data type they create, such as ARRAY or MAP. Constructor functions are diﬀerent
sql-reference-046	sql-reference.pdf	46	salesid<10 order by 1, 2 desc; listid \| salesid --------+--------- 4 \| 2 Conditional expressions 150 SQL Reference AWS Clean Rooms 5 \| 4 5 \| 3 6 \| 5 10 \| 9 10 \| 8 10 \| 7 10 \| 6 \| 1 (9 rows) Constructor functions A SQL constructor function is a function that is used to create new data structures, such as arrays or maps. They take some input values and return a new data structure object. Constructor functions are typically named after the data type they create, such as ARRAY or MAP. Constructor functions are diﬀerent from scalar functions or aggregate functions, which operate on existing data and return a single value. Constructor functions are used to create new data structures that can then be used in further data processing or analysis. AWS Clean Rooms supports the following constructor functions: Topics • MAP constructor function • NAMED_STRUCT constructor function • STRUCT constructor function MAP constructor function The MAP constructor function creates a map with the given key/value pairs. Constructor functions like MAP are useful when you need to create new data structures programmatically within your SQL queries. They allow you to build complex data structures that can be used in further data processing or analysis. Syntax map(key0, value0, key1, value1, ...) Constructor functions 151 AWS Clean Rooms Arguments key0 SQL Reference An expression of any comparable type. All key0 must share a least common type. value0 An expression of any type. All valueN must share a least common type. Returns The MAP function returns a MAP with keys typed as the least common type of key0 and values typed as the least common type of value0. Examples The following example creates a new map with two key-value pairs: The key 1.0 is associated with the value '2'. The key 3.0 is associated with the value '4'. The resulting map is then returned as the output of the SQL statement. SELECT map(1.0, '2', 3.0, '4'); {1.0:"2",3.0:"4"} NAMED_STRUCT constructor function The NAMED_STRUCT constructor function creates a struct with the given ﬁeld names and values. Constructor functions like NAMED_STRUCT are useful when you need to create new data structures programmatically within your SQL queries. They allow you to build complex data structures, such as structs or records, that can be used in further data processing or analysis. Syntax named_struct(name1, val1, name2, val2, ...) Arguments name1 A STRING literal naming ﬁeld 1. Constructor functions 152 AWS Clean Rooms val1 An expression of any type specifying the value for ﬁeld 1. Returns SQL Reference The NAMED_STRUCT function returns a struct with ﬁeld 1 matching the type of val1. Examples The following example creates a new struct with three named ﬁelds: The ﬁeld "a" is assigned the value 1. The ﬁeld "b" is assigned the value 2. The ﬁeld "c" is assigned the value 3. The resulting struct is then returned as the output of the SQL statement. SELECT named_struct("a", 1, "b", 2, "c", 3); {"a":1,"b":2,"c":3} STRUCT constructor function The STRUCT constructor function creates a struct with the given ﬁeld values. Constructor functions like STRUCT are useful when you need to create new data structures programmatically within your SQL queries. They allow you to build complex data structures, such as structs or records, that can be used in further data processing or analysis. Syntax struct(col1, col2, col3, ...) Arguments col1 A column name or any valid expression. Returns The STRUCT function returns a struct with ﬁeld1 matching the type of expr1. If the arguments are named references, the names are used to name the ﬁeld. Otherwise, the ﬁelds are named colN, where N is the position of the ﬁeld in the struct. Constructor functions 153 AWS Clean Rooms Examples SQL Reference The following example creates a new struct with three ﬁelds: The ﬁrst ﬁeld is assigned the value 1. The second ﬁeld is assigned the value 2. The third ﬁeld is assigned the value 3. By default, the ﬁelds in the resulting struct are named col1, col2, and col3, based on their position in the argument list. The resulting struct is then returned as the output of the SQL statement. SELECT struct(1, 2, 3); {"col1":1,"col2":2,"col3":3} Data type formatting functions Using a data type formatting function, you can convert values from one data type to another. For each of these functions, the ﬁrst argument is always the value to be formatted and the second argument contains the template for the new format. AWS Clean Rooms Spark SQL supports several data type formatting functions. Topics • BASE64 function • CAST function • CONVERT function • DECODE function • ENCODE function • HEX function • STR_TO_MAP function • TO_CHAR • TO_DATE function • TO_NUMBER • UNBASE64 function • UNHEX function • Datetime format strings • Numeric format strings Data type formatting functions 154 AWS Clean
sql-reference-047	sql-reference.pdf	47	type formatting function, you can convert values from one data type to another. For each of these functions, the ﬁrst argument is always the value to be formatted and the second argument contains the template for the new format. AWS Clean Rooms Spark SQL supports several data type formatting functions. Topics • BASE64 function • CAST function • CONVERT function • DECODE function • ENCODE function • HEX function • STR_TO_MAP function • TO_CHAR • TO_DATE function • TO_NUMBER • UNBASE64 function • UNHEX function • Datetime format strings • Numeric format strings Data type formatting functions 154 AWS Clean Rooms SQL Reference • Teradata-style formatting characters for numeric data BASE64 function The BASE64 function converts an expression to a base 64 string using RFC2045 Base64 transfer encoding for MIME. Syntax base64(expr) Arguments expr A BINARY expression or a STRING which the function will interpret as BINARY. Return type STRING Example To convert the given string input into its Base64 encoded representation. use the following example. The result is the Base64 encoded representation of the input string 'Spark SQL', which is 'U3BhcmsgU1FM'. SELECT base64('Spark SQL'); U3BhcmsgU1FM CAST function The CAST function converts one data type to another compatible data type. For instance, you can convert a string to a date, or a numeric type to a string. CAST performs a runtime conversion, which means that the conversion doesn't change a value's data type in a source table. It's changed only in the context of the query. The CAST function is very similar to the section called “CONVERT”, in that they both convert one data type to another, but they are called diﬀerently. Data type formatting functions 155 AWS Clean Rooms SQL Reference Certain data types require an explicit conversion to other data types using the CAST or CONVERT function. Other data types can be converted implicitly, as part of another command, without using CAST or CONVERT. See Type compatibility and conversion. Syntax Use either of these two equivalent syntax forms to cast expressions from one data type to another. CAST ( expression AS type ) Arguments expression An expression that evaluates to one or more values, such as a column name or a literal. Converting null values returns nulls. The expression can't contain blank or empty strings. type One of the supported Data types , except for BINARY and BINARY VARYING data types. Return type CAST returns the data type speciﬁed by the type argument. Note AWS Clean Rooms returns an error if you try to perform a problematic conversion, such as a DECIMAL conversion that loses precision, like the following: select 123.456::decimal(2,1); or an INTEGER conversion that causes an overﬂow: select 12345678::smallint; Examples The following two queries are equivalent. They both cast a decimal value to an integer: Data type formatting functions 156 AWS Clean Rooms SQL Reference select cast(pricepaid as integer) from sales where salesid=100; pricepaid ----------- 162 (1 row) select pricepaid::integer from sales where salesid=100; pricepaid ----------- 162 (1 row) The following produces a similar result. It doesn't require sample data to run: select cast(162.00 as integer) as pricepaid; pricepaid ----------- 162 (1 row) In this example, the values in a timestamp column are cast as dates, which results in removing the time from each result: select cast(saletime as date), salesid from sales order by salesid limit 10; saletime \| salesid -----------+--------- 2008-02-18 \| 1 2008-06-06 \| 2 2008-06-06 \| 3 2008-06-09 \| 4 2008-08-31 \| 5 2008-07-16 \| 6 2008-06-26 \| 7 2008-07-10 \| 8 2008-07-22 \| 9 Data type formatting functions 157 AWS Clean Rooms 2008-08-06 \| 10 (10 rows) SQL Reference If you didn't use CAST as illustrated in the previous sample, the results would include the time: 2008-02-18 02:36:48. The following query casts variable character data to a date. It doesn't require sample data to run. select cast('2008-02-18 02:36:48' as date) as mysaletime; mysaletime -------------------- 2008-02-18 (1 row) In this example, the values in a date column are cast as timestamps: select cast(caldate as timestamp), dateid from date order by dateid limit 10; caldate \| dateid --------------------+-------- 2008-01-01 00:00:00 \| 1827 2008-01-02 00:00:00 \| 1828 2008-01-03 00:00:00 \| 1829 2008-01-04 00:00:00 \| 1830 2008-01-05 00:00:00 \| 1831 2008-01-06 00:00:00 \| 1832 2008-01-07 00:00:00 \| 1833 2008-01-08 00:00:00 \| 1834 2008-01-09 00:00:00 \| 1835 2008-01-10 00:00:00 \| 1836 (10 rows) In a case like the previous sample, you can gain additional control over output formatting by using TO_CHAR. In this example, an integer is cast as a character string: select cast(2008 as char(4)); Data type formatting functions 158 AWS Clean Rooms bpchar -------- 2008 SQL Reference In this example, a DECIMAL(6,3) value is cast as a DECIMAL(4,1) value: select cast(109.652 as decimal(4,1)); numeric --------- 109.7 This example shows a more complex expression. The PRICEPAID column (a DECIMAL(8,2) column) in the SALES table is converted
sql-reference-048	sql-reference.pdf	48	\| 1833 2008-01-08 00:00:00 \| 1834 2008-01-09 00:00:00 \| 1835 2008-01-10 00:00:00 \| 1836 (10 rows) In a case like the previous sample, you can gain additional control over output formatting by using TO_CHAR. In this example, an integer is cast as a character string: select cast(2008 as char(4)); Data type formatting functions 158 AWS Clean Rooms bpchar -------- 2008 SQL Reference In this example, a DECIMAL(6,3) value is cast as a DECIMAL(4,1) value: select cast(109.652 as decimal(4,1)); numeric --------- 109.7 This example shows a more complex expression. The PRICEPAID column (a DECIMAL(8,2) column) in the SALES table is converted to a DECIMAL(38,2) column and the values are multiplied by 100000000000000000000: select salesid, pricepaid::decimal(38,2)*100000000000000000000 as value from sales where salesid<10 order by salesid; salesid \| value ---------+---------------------------- 1 \| 72800000000000000000000.00 2 \| 7600000000000000000000.00 3 \| 35000000000000000000000.00 4 \| 17500000000000000000000.00 5 \| 15400000000000000000000.00 6 \| 39400000000000000000000.00 7 \| 78800000000000000000000.00 8 \| 19700000000000000000000.00 9 \| 59100000000000000000000.00 (9 rows) CONVERT function Like the CAST function, the CONVERT function converts one data type to another compatible data type. For instance, you can convert a string to a date, or a numeric type to a string. CONVERT performs a runtime conversion, which means that the conversion doesn't change a value's data type in a source table. It's changed only in the context of the query. Data type formatting functions 159 AWS Clean Rooms SQL Reference Certain data types require an explicit conversion to other data types using the CONVERT function. Other data types can be converted implicitly, as part of another command, without using CAST or CONVERT. See Type compatibility and conversion. Syntax CONVERT ( type, expression ) Arguments type One of the supported Data types, except for VARBYTE, BINARY, and BINARY VARYING data types. expression An expression that evaluates to one or more values, such as a column name or a literal. Converting null values returns nulls. The expression can't contain blank or empty strings. Return type CONVERT returns the data type speciﬁed by the type argument. Note AWS Clean Rooms returns an error if you try to perform a problematic conversion, such as a DECIMAL conversion that loses precision, like the following: SELECT CONVERT(decimal(2,1), 123.456); or an INTEGER conversion that causes an overﬂow: SELECT CONVERT(smallint, 12345678); Examples The following query uses the CONVERT function to convert a column of decimals into integers Data type formatting functions 160 AWS Clean Rooms SQL Reference SELECT CONVERT(integer, pricepaid) FROM sales WHERE salesid=100; This example converts an integer into a character string. SELECT CONVERT(char(4), 2008); In this example, the current date and time is converted to a variable character data type: SELECT CONVERT(VARCHAR(30), GETDATE()); getdate --------- 2023-02-02 04:31:16 This example converts the saletime column into just the time, removing the dates from each row. SELECT CONVERT(time, saletime), salesid FROM sales order by salesid limit 10; The following example converts variable character data into a datetime object. SELECT CONVERT(datetime, '2008-02-18 02:36:48') as mysaletime; DECODE function The DECODE function is the counterpart to the ENCODE function, which is used to convert a string to a binary format using a speciﬁc character encoding. The DECODE function takes the binary data and converts it back to a readable string format using the speciﬁed character encoding. This function is useful when you need to work with binary data stored in a database and need to present it in a human-readable format, or when you need to convert data between diﬀerent character encodings. Syntax decode(expr, charset) Data type formatting functions 161 AWS Clean Rooms Arguments expr A BINARY expression encoded in charset. charset A STRING expression. SQL Reference Supported character set encodings (case-insensitive): 'US-ASCII', 'ISO-8859-1', 'UTF-8', 'UTF-16BE', 'UTF-16LE', and 'UTF-16'. Return type The DECODE function returns a STRING. Example The following example has a table called messages with a column called message_text that stores message data in a binary format using the UTF-8 character encoding. The DECODE function converts the binary data back to a readable string format. The output of this query is the readable text of the message stored in the messages table, with the ID 123, converted from the binary format to a string using the 'utf-8' encoding. SELECT decode(message_text, 'utf-8') AS message FROM messages WHERE message_id = 123; ENCODE function The ENCODE function is used to convert a string to its binary representation using a speciﬁed character encoding. This function is useful when you need to work with binary data or when you need to convert between diﬀerent character encodings. For example, you might use the ENCODE function when storing data in a database that requires binary storage, or when you need to transfer data between systems that use diﬀerent character encodings. Syntax encode(str, charset) Data type formatting functions 162 AWS Clean Rooms Arguments str A STRING expression to be encoded. charset A STRING expression specifying the encoding. SQL
sql-reference-049	sql-reference.pdf	49	function The ENCODE function is used to convert a string to its binary representation using a speciﬁed character encoding. This function is useful when you need to work with binary data or when you need to convert between diﬀerent character encodings. For example, you might use the ENCODE function when storing data in a database that requires binary storage, or when you need to transfer data between systems that use diﬀerent character encodings. Syntax encode(str, charset) Data type formatting functions 162 AWS Clean Rooms Arguments str A STRING expression to be encoded. charset A STRING expression specifying the encoding. SQL Reference Supported character set encodings (case-insensitive): 'US-ASCII', 'ISO-8859-1', 'UTF-8', 'UTF-16BE', 'UTF-16LE', and 'UTF-16'. Return type The ENCODE function returns a BINARY. Example The following example converts the string 'abc' to its binary representation using the 'utf-8' encoding, which in this case results in the original string being returned. This is because the 'utf-8' encoding is a variable-width character encoding that can represent the entire ASCII character set (which includes the letters 'a', 'b', and 'c') using a single byte per character. Therefore, the binary representation of 'abc' using 'utf-8' is the same as the original string. SELECT encode('abc', 'utf-8'); abc HEX function The HEX function converts a numeric value (either an integer or a ﬂoating-point number) to its corresponding hexadecimal string representation. Hexadecimal is a numeral system that uses 16 distinct symbols (0-9 and A-F) to represent numeric values. It is commonly used in computer science and programming to represent binary data in a more compact and human-readable format. Syntax hex(expr) Data type formatting functions 163 AWS Clean Rooms Arguments expr A BIGINT, BINARY, or STRING expression. Return type SQL Reference HEX returns a STRING. The function returns the hexadecimal representation of the argument. Example The following example takes the integer value 17 as input and applies the HEX() function to it. The output is 11, which is the hexadecimal representation of the input value 17. SELECT hex(17); 11 The following example converts the string 'Spark_SQL' to its hexadecimal representation. The output is 537061726B2053514C, which is the hexadecimal representation of the input string 'Spark_SQL'. SELECT hex('Spark_SQL'); 537061726B2053514C In this example, the string 'Spark_SQL' is converted as follows: • 'S' -> 53 • 'p' -> 70 • 'a' -> 61 • 'r' -> 72 ' • k' -> 6B • '_' -> 20 • 'S' -> 53 • 'Q' -> 51 • 'L' -> 4C Data type formatting functions 164 AWS Clean Rooms SQL Reference The concatenation of these hexadecimal values results in the ﬁnal output "537061726B2053514C". STR_TO_MAP function The STR_TO_MAP function is a string-to-map conversion function. It converts a string representation of a map (or dictionary) into an actual map data structure. This function is useful when you need to work with map data structures in SQL, but the data is initially stored as a string. By converting the string representation to an actual map, you can then perform operations and manipulations on the map data. Syntax str_to_map(text[, pairDelim[, keyValueDelim]]) Arguments text A STRING expression that represents the map. pairDelim An optional STRING literal that speciﬁes how to separate entries. It defaults to a comma (','). keyValueDelim An optional STRING literal that speciﬁes how to separate each key-value pair. It defaults to a colon (':'). Return type The STR_TO_MAP function returns a MAP of STRING for both keys and values. Both pairDelim and keyValueDelim are treated as regular expressions. Example The following example takes the input string and the two delimiter arguments, and converts the string representation into an actual map data structure. In this speciﬁc example, the input string 'a:1,b:2,c:3' represents a map with the following key-value pairs: 'a' is the key, and '1' is the value. 'b' is the key, and '2' is the value. 'c' is the key, and '3' is the value. The ',' Data type formatting functions 165 AWS Clean Rooms SQL Reference delimiter is used to separate the key-value pairs, and the ':' delimiter is used to separate the key and value within each pair. The output of this query is: {"a":"1","b":"2","c":"3"}. This is the resulting map data structure, where the keys are 'a', 'b', and 'c', and the corresponding values are '1', '2', and '3'. SELECT str_to_map('a:1,b:2,c:3', ',', ':'); {"a":"1","b":"2","c":"3"} The following example demonstrates that the STR_TO_MAP function expects the input string to be in a speciﬁc format, with the key-value pairs delimited correctly. If the input string doesn't match the expected format, the function will still attempt to create a map, but the resulting values may not be as expected. SELECT str_to_map('a'); {"a":null} TO_CHAR TO_CHAR converts a timestamp or numeric expression to a character-string data format. Syntax TO_CHAR (timestamp_expression \| numeric_expression , 'format') Arguments timestamp_expression An expression that results in a TIMESTAMP or TIMESTAMPTZ type value or a value that can implicitly
sql-reference-050	sql-reference.pdf	50	and '3'. SELECT str_to_map('a:1,b:2,c:3', ',', ':'); {"a":"1","b":"2","c":"3"} The following example demonstrates that the STR_TO_MAP function expects the input string to be in a speciﬁc format, with the key-value pairs delimited correctly. If the input string doesn't match the expected format, the function will still attempt to create a map, but the resulting values may not be as expected. SELECT str_to_map('a'); {"a":null} TO_CHAR TO_CHAR converts a timestamp or numeric expression to a character-string data format. Syntax TO_CHAR (timestamp_expression \| numeric_expression , 'format') Arguments timestamp_expression An expression that results in a TIMESTAMP or TIMESTAMPTZ type value or a value that can implicitly be coerced to a timestamp. numeric_expression An expression that results in a numeric data type value or a value that can implicitly be coerced to a numeric type. For more information, see Numeric types. TO_CHAR inserts a space to the left of the numeral string. Note TO_CHAR doesn't support 128-bit DECIMAL values. Data type formatting functions 166 AWS Clean Rooms format SQL Reference The format for the new value. For valid formats, see Datetime format strings and Numeric format strings. Return type VARCHAR Examples The following example converts a timestamp to a value with the date and time in a format with the name of the month padded to nine characters, the name of the day of the week, and the day number of the month. select to_char(timestamp '2009-12-31 23:15:59', 'MONTH-DY-DD-YYYY HH12:MIPM'); to_char ------------------------- DECEMBER -THU-31-2009 11:15PM The following example converts a timestamp to a value with day number of the year. select to_char(timestamp '2009-12-31 23:15:59', 'DDD'); to_char ------------------------- 365 The following example converts a timestamp to an ISO day number of the week. select to_char(timestamp '2022-05-16 23:15:59', 'ID'); to_char ------------------------- 1 The following example extracts the month name from a date. select to_char(date '2009-12-31', 'MONTH'); to_char Data type formatting functions 167 AWS Clean Rooms ------------------------- DECEMBER SQL Reference The following example converts each STARTTIME value in the EVENT table to a string that consists of hours, minutes, and seconds. select to_char(starttime, 'HH12:MI:SS') from event where eventid between 1 and 5 order by eventid; to_char ---------- 02:30:00 08:00:00 02:30:00 02:30:00 07:00:00 (5 rows) The following example converts an entire timestamp value into a diﬀerent format. select starttime, to_char(starttime, 'MON-DD-YYYY HH12:MIPM') from event where eventid=1; starttime \| to_char ---------------------+--------------------- 2008-01-25 14:30:00 \| JAN-25-2008 02:30PM (1 row) The following example converts a timestamp literal to a character string. select to_char(timestamp '2009-12-31 23:15:59','HH24:MI:SS'); to_char ---------- 23:15:59 (1 row) The following example converts a number to a character string with the negative sign at the end. select to_char(-125.8, '999D99S'); to_char Data type formatting functions 168 AWS Clean Rooms --------- 125.80- (1 row) SQL Reference The following example converts a number to a character string with the currency symbol. select to_char(-125.88, '$S999D99'); to_char --------- $-125.88 (1 row) The following example converts a number to a character string using angle brackets for negative numbers. select to_char(-125.88, '$999D99PR'); to_char --------- $<125.88> (1 row) The following example converts a number to a Roman numeral string. select to_char(125, 'RN'); to_char --------- CXXV (1 row) The following example displays the day of the week. SELECT to_char(current_timestamp, 'FMDay, FMDD HH12:MI:SS'); to_char ----------------------- Wednesday, 31 09:34:26 The following example displays the ordinal number suﬃx for a number. SELECT to_char(482, '999th'); to_char ----------------------- Data type formatting functions 169 AWS Clean Rooms 482nd SQL Reference The following example subtracts the commission from the price paid in the sales table. The diﬀerence is then rounded up and converted to a roman numeral, shown in the to_char column: select salesid, pricepaid, commission, (pricepaid - commission) as difference, to_char(pricepaid - commission, 'rn') from sales group by sales.pricepaid, sales.commission, salesid order by salesid limit 10; salesid \| pricepaid \| commission \| difference \| to_char ---------+-----------+------------+------------+----------------- 1 \| 728.00 \| 109.20 \| 618.80 \| dcxix 2 \| 76.00 \| 11.40 \| 64.60 \| lxv 3 \| 350.00 \| 52.50 \| 297.50 \| ccxcviii 4 \| 175.00 \| 26.25 \| 148.75 \| cxlix 5 \| 154.00 \| 23.10 \| 130.90 \| cxxxi 6 \| 394.00 \| 59.10 \| 334.90 \| cccxxxv 7 \| 788.00 \| 118.20 \| 669.80 \| dclxx 8 \| 197.00 \| 29.55 \| 167.45 \| clxvii 9 \| 591.00 \| 88.65 \| 502.35 \| dii 10 \| 65.00 \| 9.75 \| 55.25 \| lv (10 rows) The following example adds the currency symbol to the diﬀerence values shown in the to_char column: select salesid, pricepaid, commission, (pricepaid - commission) as difference, to_char(pricepaid - commission, 'l99999D99') from sales group by sales.pricepaid, sales.commission, salesid order by salesid limit 10; salesid \| pricepaid \| commission \| difference \| to_char --------+-----------+------------+------------+------------ 1 \| 728.00 \| 109.20 \| 618.80 \| $ 618.80 2 \| 76.00 \| 11.40 \| 64.60 \| $ 64.60 3 \| 350.00 \| 52.50 \| 297.50 \| $ 297.50 4 \| 175.00 \| 26.25 \| 148.75 \| $ 148.75
sql-reference-051	sql-reference.pdf	51	65.00 \| 9.75 \| 55.25 \| lv (10 rows) The following example adds the currency symbol to the diﬀerence values shown in the to_char column: select salesid, pricepaid, commission, (pricepaid - commission) as difference, to_char(pricepaid - commission, 'l99999D99') from sales group by sales.pricepaid, sales.commission, salesid order by salesid limit 10; salesid \| pricepaid \| commission \| difference \| to_char --------+-----------+------------+------------+------------ 1 \| 728.00 \| 109.20 \| 618.80 \| $ 618.80 2 \| 76.00 \| 11.40 \| 64.60 \| $ 64.60 3 \| 350.00 \| 52.50 \| 297.50 \| $ 297.50 4 \| 175.00 \| 26.25 \| 148.75 \| $ 148.75 5 \| 154.00 \| 23.10 \| 130.90 \| $ 130.90 6 \| 394.00 \| 59.10 \| 334.90 \| $ 334.90 7 \| 788.00 \| 118.20 \| 669.80 \| $ 669.80 8 \| 197.00 \| 29.55 \| 167.45 \| $ 167.45 Data type formatting functions 170 AWS Clean Rooms SQL Reference 9 \| 591.00 \| 88.65 \| 502.35 \| $ 502.35 10 \| 65.00 \| 9.75 \| 55.25 \| $ 55.25 (10 rows) The following example lists the century in which each sale was made. select salesid, saletime, to_char(saletime, 'cc') from sales order by salesid limit 10; salesid \| saletime \| to_char ---------+---------------------+--------- 1 \| 2008-02-18 02:36:48 \| 21 2 \| 2008-06-06 05:00:16 \| 21 3 \| 2008-06-06 08:26:17 \| 21 4 \| 2008-06-09 08:38:52 \| 21 5 \| 2008-08-31 09:17:02 \| 21 6 \| 2008-07-16 11:59:24 \| 21 7 \| 2008-06-26 12:56:06 \| 21 8 \| 2008-07-10 02:12:36 \| 21 9 \| 2008-07-22 02:23:17 \| 21 10 \| 2008-08-06 02:51:55 \| 21 (10 rows) The following example converts each STARTTIME value in the EVENT table to a string that consists of hours, minutes, seconds, and time zone. select to_char(starttime, 'HH12:MI:SS TZ') from event where eventid between 1 and 5 order by eventid; to_char ---------- 02:30:00 UTC 08:00:00 UTC 02:30:00 UTC 02:30:00 UTC 07:00:00 UTC (5 rows) (10 rows) The following example shows formatting for seconds, milliseconds, and microseconds. Data type formatting functions 171 AWS Clean Rooms SQL Reference select sysdate, to_char(sysdate, 'HH24:MI:SS') as seconds, to_char(sysdate, 'HH24:MI:SS.MS') as milliseconds, to_char(sysdate, 'HH24:MI:SS:US') as microseconds; timestamp \| seconds \| milliseconds \| microseconds --------------------+----------+--------------+---------------- 2015-04-10 18:45:09 \| 18:45:09 \| 18:45:09.325 \| 18:45:09:325143 TO_DATE function TO_DATE converts a date represented by a character string to a DATE data type. Syntax TO_DATE(string, format) Arguments string A string to be converted. format A string literal that deﬁnes the format of the input string, in terms of its date parts. For a list of valid day, month, and year formats, see Datetime format strings. is_strict An optional Boolean value that speciﬁes whether an error is returned if an input date value is out of range. When is_strict is set to TRUE, an error is returned if there is an out of range value. When is_strict is set to FALSE, which is the default, then overﬂow values are accepted. Return type TO_DATE returns a DATE, depending on the format value. If the conversion to format fails, then an error is returned. Examples The following SQL statement converts the date 02 Oct 2001 into a date data type. Data type formatting functions 172 AWS Clean Rooms SQL Reference select to_date('02 Oct 2001', 'DD Mon YYYY'); to_date ------------ 2001-10-02 (1 row) The following SQL statement converts the string 20010631 to a date. select to_date('20010631', 'YYYYMMDD', FALSE); The result is July 1, 2001, because there are only 30 days in June. to_date ------------ 2001-07-01 The following SQL statement converts the string 20010631 to a date: to_date('20010631', 'YYYYMMDD', TRUE); The result is an error because there are only 30 days in June. ERROR: date/time field date value out of range: 2001-6-31 TO_NUMBER TO_NUMBER converts a string to a numeric (decimal) value. Syntax to_number(string, format) Arguments string String to be converted. The format must be a literal value. Data type formatting functions 173 AWS Clean Rooms format SQL Reference The second argument is a format string that indicates how the character string should be parsed to create the numeric value. For example, the format '99D999' speciﬁes that the string to be converted consists of ﬁve digits with the decimal point in the third position. For example, to_number('12.345','99D999') returns 12.345 as a numeric value. For a list of valid formats, see Numeric format strings. Return type TO_NUMBER returns a DECIMAL number. If the conversion to format fails, then an error is returned. Examples The following example converts the string 12,454.8- to a number: select to_number('12,454.8-', '99G999D9S'); to_number ----------- -12454.8 The following example converts the string $ 12,454.88 to a number: select to_number('$ 12,454.88', 'L 99G999D99'); to_number ----------- 12454.88 The following example converts the string $ 2,012,454.88 to a number: select to_number('$ 2,012,454.88', 'L 9,999,999.99'); to_number ----------- 2012454.88 Data type formatting functions 174 AWS Clean Rooms UNBASE64 function SQL Reference The UNBASE64 function converts an argument
sql-reference-052	sql-reference.pdf	52	a list of valid formats, see Numeric format strings. Return type TO_NUMBER returns a DECIMAL number. If the conversion to format fails, then an error is returned. Examples The following example converts the string 12,454.8- to a number: select to_number('12,454.8-', '99G999D9S'); to_number ----------- -12454.8 The following example converts the string $ 12,454.88 to a number: select to_number('$ 12,454.88', 'L 99G999D99'); to_number ----------- 12454.88 The following example converts the string $ 2,012,454.88 to a number: select to_number('$ 2,012,454.88', 'L 9,999,999.99'); to_number ----------- 2012454.88 Data type formatting functions 174 AWS Clean Rooms UNBASE64 function SQL Reference The UNBASE64 function converts an argument from a base 64 string to a binary. Base64 encoding is commonly used to represent binary data (such as images, ﬁles, or encrypted information) in a textual format that is safe for transmission over various communication channels (such as email, URL parameters, or database storage). The UNBASE64 function allows you to reverse this process and recover the original binary data. This type of functionality can be useful in scenarios where you need to work with data that has been encoded in Base64 format, such as when integrating with external systems or APIs that use Base64 as a data transfer mechanism. Syntax unbase64(expr) Arguments expr A STRING expression in a base64 format. Return type BINARY Example In the following example, the Base64-encoded string 'U3BhcmsgU1FM' is converted back to the original string 'Spark SQL'. SELECT unbase64('U3BhcmsgU1FM'); Spark SQL UNHEX function The UNHEX function converts a hexadecimal string back to its original string representation. Data type formatting functions 175 AWS Clean Rooms SQL Reference This function can be useful in scenarios where you need to work with data that has been stored or transmitted in a hexadecimal format, and you need to restore the original string representation for further processing or display. The UNHEX function is the counterpart to the HEX function. Syntax unhex(expr) Arguments expr A STRING expression of hexadecimal characters. Return type UNHEX returns a BINARY. If the length of expr is odd, the ﬁrst character is discarded and the result is padded with a null byte. If expr contains non hex characters the result is NULL. Example The following example converts a hexadecimal string back to its original string representation by using the UNHEX() and DECODE() functions together. The ﬁrst part of the query, uses the UNHEX() function to convert the hexadecimal string '537061726B2053514C' to its binary representation. The second part of the query, uses the DECODE() function to convert the binary data obtained from the UNHEX() function back to a string, using the 'UTF-8' character encoding. The output of the query, is he original string 'Spark_SQL' that was converted to hexadecimal and then back to a string. SELECT decode(unhex('537061726B2053514C'), 'UTF-8'); Spark SQL Datetime format strings The following datetime format strings apply to functions such as TO_CHAR. These strings can contain datetime separators (such as '-', '/', or ':') and the following "dateparts" and "timeparts". Data type formatting functions 176 AWS Clean Rooms SQL Reference For examples of formatting dates as strings, see TO_CHAR. Datepart or timepart Meaning BC or B.C., AD or A.D., b.c. or bc, ad or a.d. Upper and lowercase era indicators CC Two-digit century number YYYY, YYY, YY, Y 4-digit, 3-digit, 2-digit, 1-digit year number Y,YYY IYYY, IYY, IY, I 4-digit year number with comma 4-digit, 3-digit, 2-digit, 1-digit International Organization for Standardization (ISO) year number Q Quarter number (1 to 4) MONTH, Month, month MON, Mon, mon MM RM, rm W WW IW DAY, Day, day Month name (uppercase, mixed-case, lowercase, blank-padded to 9 characters) Abbreviated month name (uppercase, mixed- case, lowercase, blank-padded to 3 characters) Month number (01-12) Month number in Roman numerals (I–XII, with I being January, uppercase or lowercase) Week of month (1–5; the ﬁrst week starts on the ﬁrst day of the month.) Week number of year (1–53; the ﬁrst week starts on the ﬁrst day of the year.) ISO week number of year (the ﬁrst Thursday of the new year is in week 1.) Day name (uppercase, mixed-case, lowercase, blank-padded to 9 characters) Data type formatting functions 177 AWS Clean Rooms SQL Reference Datepart or timepart Meaning DY, Dy, dy DDD IDDD DD D ID J HH24 HH or HH12 MI SS MS Abbreviated day name (uppercase, mixed-case, lowercase, blank-padded to 3 characters) Day of year (001–366) Day of ISO 8601 week-numbering year (001-371; day 1 of the year is Monday of the ﬁrst ISO week) Day of month as a number (01–31) Day of week (1–7; Sunday is 1) Note The D datepart behaves diﬀerently from the day of week (DOW) datepart used for the datetime functions DATE_PART and EXTRACT. DOW is based on integers 0–6, where Sunday is 0. For more information, see Date parts for date or timestamp functions. ISO 8601 day of the week, Monday (1)
sql-reference-053	sql-reference.pdf	53	MI SS MS Abbreviated day name (uppercase, mixed-case, lowercase, blank-padded to 3 characters) Day of year (001–366) Day of ISO 8601 week-numbering year (001-371; day 1 of the year is Monday of the ﬁrst ISO week) Day of month as a number (01–31) Day of week (1–7; Sunday is 1) Note The D datepart behaves diﬀerently from the day of week (DOW) datepart used for the datetime functions DATE_PART and EXTRACT. DOW is based on integers 0–6, where Sunday is 0. For more information, see Date parts for date or timestamp functions. ISO 8601 day of the week, Monday (1) to Sunday (7) Julian day (days since January 1, 4712 BC) Hour (24-hour clock, 00–23) Hour (12-hour clock, 01–12) Minutes (00–59) Seconds (00–59) Milliseconds (.000) Data type formatting functions 178 AWS Clean Rooms SQL Reference Datepart or timepart Meaning US Microseconds (.000000) AM or PM, A.M. or P.M., a.m. or p.m., am or pm Upper and lowercase meridian indicators (for 12-hour clock) Upper and lowercase time zone abbreviation; valid for TIMESTAMPTZ only Oﬀset from UTC; valid for TIMESTAMPTZ only TZ, tz OF Note You must surround datetime separators (such as '-', '/' or ':') with single quotation marks, but you must surround the "dateparts" and "timeparts" listed in the preceding table with double quotation marks. Numeric format strings The following numeric format strings apply to functions such as TO_NUMBER and TO_CHAR. • For examples of formatting strings as numbers, see TO_NUMBER. • For examples of formatting numbers as strings, see TO_CHAR. Format Description 9 0 . (period), D , (comma) Numeric value with the speciﬁed number of digits. Numeric value with leading zeros. Decimal point. Thousands separator. Data type formatting functions 179 AWS Clean Rooms Format CC FM PR S L G MI PL SG RN TH or th SQL Reference Description Century code. For example, the 21st century started on 2001-01-01 (supported for TO_CHAR only). Fill mode. Suppress padding blanks and zeroes. Negative value in angle brackets. Sign anchored to a number. Currency symbol in the speciﬁed position. Group separator. Minus sign in the speciﬁed position for numbers that are less than 0. Plus sign in the speciﬁed position for numbers that are greater than 0. Plus or minus sign in the speciﬁed position. Roman numeral between 1 and 3999 (supported for TO_CHAR only). Ordinal number suﬃx. Does not convert fractional numbers or values that are less than zero. Teradata-style formatting characters for numeric data This topic shows you how the TEXT_TO_INT_ALT and TEXT_TO_NUMERIC_ALT functions interpret the characters in the input expression string. In the following table, you can also ﬁnd a list of the characters that you can specify in the format phrase. In addition, you can ﬁnd a description of the diﬀerences between Teradata-style formatting and AWS Clean Rooms for the format option. Data type formatting functions 180 AWS Clean Rooms Format G D / , : % . B SQL Reference Description Not supported as a group separator in the input expression string. You can't specify this character in the format phrase. Radix symbol. You can specify this character in the format phrase. This character is equivalent to the . (period). The radix symbol can't appear in a format phrase that contains any of the following characters: • . (period) • S (uppercase 's') • V (uppercase 'v') Insertion characters / (forward slash), comma (,), : (colon), and % (percent sign). You can't include these characters in the format phrase. AWS Clean Rooms ignores these characters in the input expression string. Period as a radix character, that is, a decimal point. This character can't appear in a format phrase that contains any of the following characters: • D (uppercase 'd') • S (uppercase 's') • V (uppercase 'v') You can't include the blank space character (B) in the format phrase. In the input expression Data type formatting functions 181 AWS Clean Rooms Format + - V Z 9 SQL Reference Description string, leading and trailing spaces are ignored and spaces between digits aren't allowed. You can't include the plus sign (+) or minus sign (-) in the format phrase. However, the plus sign (+) and minus sign (-) are parsed implicitly as part of numeric value if they appear in the input expression string. Decimal point position indicator. This character can't appear in a format phrase that contains any of the following characters: • D (uppercase 'd') • . (period) Zero-suppressed decimal digit. AWS Clean Rooms trims leading zeros. The Z character can't follow a 9 character. The Z character must be to the left of the radix character if the fraction part contains the 9 character. Decimal digit. CHAR(n) For this format, you can specify the following: • CHAR consists of Z or 9 characters. AWS Clean Rooms doesn't support a + (plus) or -
sql-reference-054	sql-reference.pdf	54	if they appear in the input expression string. Decimal point position indicator. This character can't appear in a format phrase that contains any of the following characters: • D (uppercase 'd') • . (period) Zero-suppressed decimal digit. AWS Clean Rooms trims leading zeros. The Z character can't follow a 9 character. The Z character must be to the left of the radix character if the fraction part contains the 9 character. Decimal digit. CHAR(n) For this format, you can specify the following: • CHAR consists of Z or 9 characters. AWS Clean Rooms doesn't support a + (plus) or - (minus) in the CHAR value. • n is an integer constant, I, or F. For I, this is the number of characters necessary to display the integer portion of numeric or integer data. For F, this is the number of characters necessary to display the fractiona l portion of numeric data. Data type formatting functions 182 AWS Clean Rooms Format - S E FN9 FNE SQL Reference Description Hyphen (-) character. You can't include this character in the format phrase. AWS Clean Rooms ignores this character in the input expression string. Signed zoned decimal. The S character must follow the last decimal digit in the format phrase. The last character of the input expression string and the correspon ding numeric conversion are listed in Data formatting characters for signed zoned decimal, Teradata–style numeric data formatting . The S character can't appear in a format phrase that contains any of the following characters: • + (plus sign) • . (period) • D (uppercase 'd') • Z (uppercase 'z') • F (uppercase 'f') • E (uppercase 'e') Exponential notation. The input expression string can include the exponent character. You can't specify E as an exponent character in format phrase. Not supported in AWS Clean Rooms. Not supported in AWS Clean Rooms. Data type formatting functions 183 AWS Clean Rooms Format $, USD, US Dollars L C N O U A SQL Reference Description Dollar sign ($), ISO currency symbol (USD), and the currency name US Dollars. The ISO currency symbol USD and the currency name US Dollars are case-sensitive. AWS Clean Rooms supports only the USD currency. The input expression string can include spaces between the USD currency symbol and the numeric value, for example ‘$ 123E2’ or ‘123E2 $’. Currency symbol. This currency symbol character can only appear once in the format phrase. You can't specify repeated currency symbol characters. ISO currency symbol. This currency symbol character can only appear once in the format phrase. You can't specify repeated currency symbol characters. Full currency name. This currency symbol character can only appear once in the format phrase. You can't specify repeated currency symbol characters. Dual currency symbol. You can't specify this character in the format phrase. Dual ISO currency symbol. You can't specify this character in the format phrase. Full dual currency name. You can't specify this character in the format phrase. Data type formatting functions 184 AWS Clean Rooms SQL Reference Data formatting characters for signed zoned decimal, Teradata–style numeric data formatting You can use the following characters in the format phrase of the TEXT_TO_INT_ALT and TEXT_TO_NUMERIC_ALT functions for a signed zoned decimal value. Last character of the input string Numeric conversion { or 0 A or 1 B or 2 C or 3 D or 4 E or 5 F or 6 G or 7 H or 8 I or 9 } J K L M N O P n … 0 n … 1 n … 2 n … 3 n … 4 n … 5 n … 6 n … 7 n … 8 n … 9 -n … 0 -n … 1 -n … 2 -n … 3 -n … 4 -n … 5 -n … 6 -n … 7 Data type formatting functions 185 AWS Clean Rooms SQL Reference Last character of the input string Numeric conversion Q R -n … 8 -n … 9 Date and time functions Date and time functions allow you to perform a wide range of operations on date and time data, such as extracting parts of a date, performing date calculations, formatting dates and times, and working with the current date and time. These functions are essential for tasks such as data analysis, reporting, and data manipulation involving temporal data. AWS Clean Rooms supports the following date and time functions: Topics • ADD_MONTHS function • CONVERT_TIMEZONE function • CURRENT_DATE function • CURRENT_TIMESTAMP function • DATEADD function • DATEDIFF function • DATE_PART function • DATE_TRUNC function • DAY function • DAYOFMONTH function • DAYOFWEEK function • DAYOFYEAR function • EXTRACT function • FROM_UTC_TIMESTAMP function • GETDATE function • HOUR function Date and time functions 186 SQL Reference AWS Clean Rooms • MINUTE function • MONTH function • SECOND function •
sql-reference-055	sql-reference.pdf	55	current date and time. These functions are essential for tasks such as data analysis, reporting, and data manipulation involving temporal data. AWS Clean Rooms supports the following date and time functions: Topics • ADD_MONTHS function • CONVERT_TIMEZONE function • CURRENT_DATE function • CURRENT_TIMESTAMP function • DATEADD function • DATEDIFF function • DATE_PART function • DATE_TRUNC function • DAY function • DAYOFMONTH function • DAYOFWEEK function • DAYOFYEAR function • EXTRACT function • FROM_UTC_TIMESTAMP function • GETDATE function • HOUR function Date and time functions 186 SQL Reference AWS Clean Rooms • MINUTE function • MONTH function • SECOND function • SYSDATE function • TIMEOFDAY function • TIMESTAMP function • TO_TIMESTAMP function • WEEKDAY function • YEAR function • Date parts for date or timestamp functions ADD_MONTHS function ADD_MONTHS adds the speciﬁed number of months to a date or timestamp value or expression. The DATEADD function provides similar functionality. Syntax ADD_MONTHS( {date \| timestamp}, integer) Arguments date \| timestamp A date or timestamp column or an expression that implicitly converts to a date or timestamp. If the date is the last day of the month, or if the resulting month is shorter, the function returns the last day of the month in the result. For other dates, the result contains the same day number as the date expression. integer A positive or negative integer. Use a negative number to subtract months from dates. Return type TIMESTAMP Date and time functions 187 AWS Clean Rooms Example SQL Reference The following query uses the ADD_MONTHS function inside a TRUNC function. The TRUNC function removes the time of day from the result of ADD_MONTHS. The ADD_MONTHS function adds 12 months to each value from the CALDATE column. select distinct trunc(add_months(caldate, 12)) as calplus12, trunc(caldate) as cal from date order by 1 asc; calplus12 \| cal ------------+------------ 2009-01-01 \| 2008-01-01 2009-01-02 \| 2008-01-02 2009-01-03 \| 2008-01-03 ... (365 rows) The following examples demonstrate the behavior when the ADD_MONTHS function operates on dates with months that have diﬀerent numbers of days. select add_months('2008-03-31',1); add_months --------------------- 2008-04-30 00:00:00 (1 row) select add_months('2008-04-30',1); add_months --------------------- 2008-05-31 00:00:00 (1 row) CONVERT_TIMEZONE function CONVERT_TIMEZONE converts a timestamp from one time zone to another. The function automatically adjusts for daylight saving time. Date and time functions 188 AWS Clean Rooms Syntax SQL Reference CONVERT_TIMEZONE ( ['source_timezone',] 'target_timezone', 'timestamp') Arguments source_timezone (Optional) The time zone of the current timestamp. The default is UTC. target_timezone The time zone for the new timestamp. timestamp A timestamp column or an expression that implicitly converts to a timestamp. Return type TIMESTAMP Examples The following example converts the timestamp value from the default UTC time zone to PST. select convert_timezone('PST', '2008-08-21 07:23:54'); convert_timezone ------------------------ 2008-08-20 23:23:54 The following example converts the timestamp value in the LISTTIME column from the default UTC time zone to PST. Though the timestamp is within the daylight time period, it's converted to standard time because the target time zone is speciﬁed as an abbreviation (PST). select listtime, convert_timezone('PST', listtime) from listing where listid = 16; listtime \| convert_timezone --------------------+------------------- Date and time functions 189 AWS Clean Rooms SQL Reference 2008-08-24 09:36:12 2008-08-24 01:36:12 The following example converts a timestamp LISTTIME column from the default UTC time zone to US/Paciﬁc time zone. The target time zone uses a time zone name, and the timestamp is within the daylight time period, so the function returns the daylight time. select listtime, convert_timezone('US/Pacific', listtime) from listing where listid = 16; listtime \| convert_timezone --------------------+--------------------- 2008-08-24 09:36:12 \| 2008-08-24 02:36:12 The following example converts a timestamp string from EST to PST: select convert_timezone('EST', 'PST', '20080305 12:25:29'); convert_timezone ------------------- 2008-03-05 09:25:29 The following example converts a timestamp to US Eastern Standard Time because the target time zone uses a time zone name (America/New_York) and the timestamp is within the standard time period. select convert_timezone('America/New_York', '2013-02-01 08:00:00'); convert_timezone --------------------- 2013-02-01 03:00:00 (1 row) The following example converts the timestamp to US Eastern Daylight Time because the target time zone uses a time zone name (America/New_York) and the timestamp is within the daylight time period. select convert_timezone('America/New_York', '2013-06-01 08:00:00'); convert_timezone --------------------- Date and time functions 190 AWS Clean Rooms 2013-06-01 04:00:00 (1 row) The following example demonstrates the use of oﬀsets. SQL Reference SELECT CONVERT_TIMEZONE('GMT','NEWZONE +2','2014-05-17 12:00:00') as newzone_plus_2, CONVERT_TIMEZONE('GMT','NEWZONE-2:15','2014-05-17 12:00:00') as newzone_minus_2_15, CONVERT_TIMEZONE('GMT','America/Los_Angeles+2','2014-05-17 12:00:00') as la_plus_2, CONVERT_TIMEZONE('GMT','GMT+2','2014-05-17 12:00:00') as gmt_plus_2; newzone_plus_2 \| newzone_minus_2_15 \| la_plus_2 \| gmt_plus_2 ---------------------+---------------------+---------------------+--------------------- 2014-05-17 10:00:00 \| 2014-05-17 14:15:00 \| 2014-05-17 10:00:00 \| 2014-05-17 10:00:00 (1 row) CURRENT_DATE function CURRENT_DATE returns a date in the current session time zone (UTC by default) in the default format: YYYY-MM-DD. Note CURRENT_DATE returns the start date for the current transaction, not for the start of the current statement. Consider the scenario where you start a transaction containing multiple statements on 10/01/08 23:59, and the statement containing CURRENT_DATE runs
sql-reference-056	sql-reference.pdf	56	oﬀsets. SQL Reference SELECT CONVERT_TIMEZONE('GMT','NEWZONE +2','2014-05-17 12:00:00') as newzone_plus_2, CONVERT_TIMEZONE('GMT','NEWZONE-2:15','2014-05-17 12:00:00') as newzone_minus_2_15, CONVERT_TIMEZONE('GMT','America/Los_Angeles+2','2014-05-17 12:00:00') as la_plus_2, CONVERT_TIMEZONE('GMT','GMT+2','2014-05-17 12:00:00') as gmt_plus_2; newzone_plus_2 \| newzone_minus_2_15 \| la_plus_2 \| gmt_plus_2 ---------------------+---------------------+---------------------+--------------------- 2014-05-17 10:00:00 \| 2014-05-17 14:15:00 \| 2014-05-17 10:00:00 \| 2014-05-17 10:00:00 (1 row) CURRENT_DATE function CURRENT_DATE returns a date in the current session time zone (UTC by default) in the default format: YYYY-MM-DD. Note CURRENT_DATE returns the start date for the current transaction, not for the start of the current statement. Consider the scenario where you start a transaction containing multiple statements on 10/01/08 23:59, and the statement containing CURRENT_DATE runs at 10/02/08 00:00. CURRENT_DATE returns 10/01/08, not 10/02/08. Syntax CURRENT_DATE Return type DATE Example The following example returns the current date (in the AWS Region where the function runs). select current_date; Date and time functions 191 AWS Clean Rooms date ------------ 2008-10-01 SQL Reference CURRENT_TIMESTAMP function CURRENT_TIMESTAMP returns the current date and time, including the date, time, and (optionally) the milliseconds or microseconds. This function is useful when you need to get the current date and time, for example, to record the timestamp of an event, to perform time-based calculations, or to populate date/time columns. Syntax current_timestamp() Return type The CURRENT_TIMESTAMP function returns a DATE. Example The following example returns current date and time at the moment the query is executed, which is April 25, 2020, at 15:49:11.914 (3:49:11.914 PM). SELECT current_timestamp(); 2020-04-25 15:49:11.914 The following example retrieves the current date and time for each row in the squirrels table. SELECT current_timestamp() FROM squirrels DATEADD function Increments a DATE, TIME, TIMETZ, or TIMESTAMP value by a speciﬁed interval. Syntax DATEADD( datepart, interval ) Date and time functions 192 AWS Clean Rooms Arguments datepart SQL Reference The date part (year, month, day, or hour, for example) that the function operates on. For more information, see Date parts for date or timestamp functions. interval An integer that speciﬁed the interval (number of days, for example) to add to the target expression. A negative integer subtracts the interval. Return type TIMESTAMP or TIME or TIMETZ depending on the input data type. Examples with a DATE column The following example adds 30 days to each date in November that exists in the DATE table. select dateadd(day,30,caldate) as novplus30 from date where month='NOV' order by dateid; novplus30 --------------------- 2008-12-01 00:00:00 2008-12-02 00:00:00 2008-12-03 00:00:00 ... (30 rows) The following example adds 18 months to a literal date value. select dateadd(month,18,'2008-02-28'); date_add --------------------- 2009-08-28 00:00:00 (1 row) Date and time functions 193 AWS Clean Rooms SQL Reference The default column name for a DATEADD function is DATE_ADD. The default timestamp for a date value is 00:00:00. The following example adds 30 minutes to a date value that doesn't specify a timestamp. select dateadd(m,30,'2008-02-28'); date_add --------------------- 2008-02-28 00:30:00 (1 row) You can name date parts in full or abbreviate them. In this case, m stands for minutes, not months. Examples with a TIME column The following example table TIME_TEST has a column TIME_VAL (type TIME) with three values inserted. select time_val from time_test; time_val --------------------- 20:00:00 00:00:00.5550 00:58:00 The following example adds 5 minutes to each TIME_VAL in the TIME_TEST table. select dateadd(minute,5,time_val) as minplus5 from time_test; minplus5 --------------- 20:05:00 00:05:00.5550 01:03:00 The following example adds 8 hours to a literal time value. select dateadd(hour, 8, time '13:24:55'); Date and time functions 194 AWS Clean Rooms date_add --------------- 21:24:55 SQL Reference The following example shows when a time goes over 24:00:00 or under 00:00:00. select dateadd(hour, 12, time '13:24:55'); date_add --------------- 01:24:55 Examples with a TIMESTAMP column The output values in these examples are in UTC which is the default timezone. The following example table TIMESTAMP_TEST has a column TIMESTAMP_VAL (type TIMESTAMP) with three values inserted. SELECT timestamp_val FROM timestamp_test; timestamp_val ------------------ 1988-05-15 10:23:31 2021-03-18 17:20:41 2023-06-02 18:11:12 The following example adds 20 years only to the TIMESTAMP_VAL values in TIMESTAMP_TEST from before the year 2000. SELECT dateadd(year,20,timestamp_val) FROM timestamp_test WHERE timestamp_val < to_timestamp('2000-01-01 00:00:00', 'YYYY-MM-DD HH:MI:SS'); date_add --------------- 2008-05-15 10:23:31 The following example adds 5 seconds to a literal timestamp value written without a seconds indicator. Date and time functions 195 AWS Clean Rooms SQL Reference SELECT dateadd(second, 5, timestamp '2001-06-06'); date_add --------------- 2001-06-06 00:00:05 Usage notes The DATEADD(month, ...) and ADD_MONTHS functions handle dates that fall at the ends of months diﬀerently: • ADD_MONTHS: If the date you are adding to is the last day of the month, the result is always the last day of the result month, regardless of the length of the month. For example, April 30 + 1 month is May 31. select add_months('2008-04-30',1); add_months --------------------- 2008-05-31 00:00:00 (1 row) • DATEADD: If there are fewer days in the date you are adding to than in the result month, the result is the corresponding day of the result month, not the
sql-reference-057	sql-reference.pdf	57	notes The DATEADD(month, ...) and ADD_MONTHS functions handle dates that fall at the ends of months diﬀerently: • ADD_MONTHS: If the date you are adding to is the last day of the month, the result is always the last day of the result month, regardless of the length of the month. For example, April 30 + 1 month is May 31. select add_months('2008-04-30',1); add_months --------------------- 2008-05-31 00:00:00 (1 row) • DATEADD: If there are fewer days in the date you are adding to than in the result month, the result is the corresponding day of the result month, not the last day of that month. For example, April 30 + 1 month is May 30. select dateadd(month,1,'2008-04-30'); date_add --------------------- 2008-05-30 00:00:00 (1 row) The DATEADD function handles the leap year date 02-29 diﬀerently when using dateadd(month, 12,…) or dateadd(year, 1, …). select dateadd(month,12,'2016-02-29'); date_add --------------------- Date and time functions 196 SQL Reference AWS Clean Rooms 2017-02-28 00:00:00 select dateadd(year, 1, '2016-02-29'); date_add --------------------- 2017-03-01 00:00:00 DATEDIFF function DATEDIFF returns the diﬀerence between the date parts of two date or time expressions. Syntax date_diff(endDate, startDate) Arguments endDate A DATE expression. startDate A DATE expression. Return type BIGINT Examples with a DATE column The following example ﬁnds the diﬀerence, in number of weeks, between two literal date values. select datediff(week,'2009-01-01','2009-12-31') as numweeks; numweeks ---------- 52 (1 row) The following example ﬁnds the diﬀerence, in hours, between two literal date values. When you don't provide the time value for a date, it defaults to 00:00:00. Date and time functions 197 AWS Clean Rooms SQL Reference select datediff(hour, '2023-01-01', '2023-01-03 05:04:03'); date_diff ---------- 53 (1 row) The following example ﬁnds the diﬀerence, in days, between two literal TIMESTAMETZ values. Select datediff(days, 'Jun 1,2008 09:59:59 EST', 'Jul 4,2008 09:59:59 EST') date_diff ---------- 33 The following example ﬁnds the diﬀerence, in days, between two dates in the same row of a table. select * from date_table; start_date \| end_date -----------+----------- 2009-01-01 \| 2009-03-23 2023-01-04 \| 2024-05-04 (2 rows) select datediff(day, start_date, end_date) as duration from date_table; duration --------- 81 486 (2 rows) The following example ﬁnds the diﬀerence, in number of quarters, between a literal value in the past and today's date. This example assumes that the current date is June 5, 2008. You can name date parts in full or abbreviate them. The default column name for the DATEDIFF function is DATE_DIFF. select datediff(qtr, '1998-07-01', current_date); date_diff Date and time functions 198 AWS Clean Rooms ----------- 40 (1 row) SQL Reference The following example joins the SALES and LISTING tables to calculate how many days after they were listed any tickets were sold for listings 1000 through 1005. The longest wait for sales of these listings was 15 days, and the shortest was less than one day (0 days). select priceperticket, datediff(day, listtime, saletime) as wait from sales, listing where sales.listid = listing.listid and sales.listid between 1000 and 1005 order by wait desc, priceperticket desc; priceperticket \| wait ---------------+------ 96.00 \| 15 123.00 \| 11 131.00 \| 9 123.00 \| 6 129.00 \| 4 96.00 \| 4 96.00 \| 0 (7 rows) This example calculates the average number of hours sellers waited for all ticket sales. select avg(datediff(hours, listtime, saletime)) as avgwait from sales, listing where sales.listid = listing.listid; avgwait --------- 465 (1 row) Examples with a TIME column The following example table TIME_TEST has a column TIME_VAL (type TIME) with three values inserted. select time_val from time_test; Date and time functions 199 AWS Clean Rooms SQL Reference time_val --------------------- 20:00:00 00:00:00.5550 00:58:00 The following example ﬁnds the diﬀerence in number of hours between the TIME_VAL column and a time literal. select datediff(hour, time_val, time '15:24:45') from time_test; date_diff ----------- -5 15 15 The following example ﬁnds the diﬀerence in number of minutes between two literal time values. select datediff(minute, time '20:00:00', time '21:00:00') as nummins; nummins ---------- 60 Examples with a TIMETZ column The following example table TIMETZ_TEST has a column TIMETZ_VAL (type TIMETZ) with three values inserted. select timetz_val from timetz_test; timetz_val ------------------ 04:00:00+00 00:00:00.5550+00 05:58:00+00 The following example ﬁnds the diﬀerences in number of hours, between a TIMETZ literal and timetz_val. Date and time functions 200 AWS Clean Rooms SQL Reference select datediff(hours, timetz '20:00:00 PST', timetz_val) as numhours from timetz_test; numhours ---------- 0 -4 1 The following example ﬁnds the diﬀerence in number of hours, between two literal TIMETZ values. select datediff(hours, timetz '20:00:00 PST', timetz '00:58:00 EST') as numhours; numhours ---------- 1 DATE_PART function DATE_PART extracts date part values from an expression. DATE_PART is a synonym of the PGDATE_PART function. Syntax datepart(field, source) Arguments ﬁeld Which part of the source should be extracted, and supported string values are the same as the ﬁelds of the equivalent function EXTRACT. source A DATE or INTERVAL column from where ﬁeld should be extracted.
sql-reference-058	sql-reference.pdf	58	datediff(hours, timetz '20:00:00 PST', timetz_val) as numhours from timetz_test; numhours ---------- 0 -4 1 The following example ﬁnds the diﬀerence in number of hours, between two literal TIMETZ values. select datediff(hours, timetz '20:00:00 PST', timetz '00:58:00 EST') as numhours; numhours ---------- 1 DATE_PART function DATE_PART extracts date part values from an expression. DATE_PART is a synonym of the PGDATE_PART function. Syntax datepart(field, source) Arguments ﬁeld Which part of the source should be extracted, and supported string values are the same as the ﬁelds of the equivalent function EXTRACT. source A DATE or INTERVAL column from where ﬁeld should be extracted. Return type If ﬁeld is 'SECOND', a DECIMAL(8, 6). In all other cases, an INTEGER. Date and time functions 201 AWS Clean Rooms Example SQL Reference The following example extracts the day of the year (DOY) from a date value. The output shows that the day of the year for the date "2019-08-12" is 224. This means that August 12, 2019 is the 224th day of the year 2019. SELECT datepart('doy', DATE'2019-08-12'); 224 DATE_TRUNC function The DATE_TRUNC function truncates a timestamp expression or literal based on the date part that you specify, such as hour, day, or month. Syntax date_trunc(format, datetime) Arguments format The format representing the unit to be truncated to. Valid formats are as follows: • "YEAR", "YYYY", "YY" - truncate to the ﬁrst date of the year that the ts falls in, the time part will be zero out • "QUARTER" - truncate to the ﬁrst date of the quarter that the ts falls in, the time part will be zero out • "MONTH", "MM", "MON" - truncate to the ﬁrst date of the month that the ts falls in, the time part will be zero out • "WEEK" - truncate to the Monday of the week that the ts falls in, the time part will be zero out • "DAY", "DD" - zero out the time part • "HOUR" - zero out the minute and second with fraction part • "MINUTE"- zero out the second with fraction part • "SECOND" - zero out the second fraction part • "MILLISECOND" - zero out the microseconds • "MICROSECOND" - everything remains Date and time functions 202 AWS Clean Rooms ts A datetime value Return type SQL Reference Returns timestamp ts truncated to the unit speciﬁed by the format model Examples The following example truncates a date value to the beginning of the year. The output shows that the date "2015-03-05" has been truncated to "2015-01-01", which is the beginning of the year 2015. SELECT date_trunc('YEAR', '2015-03-05'); date_trunc ----------- 2015-01-01 DAY function The DAY function returns the day of month of the date/timestamp. Date extraction functions are useful when you need to work with speciﬁc components of a date or timestamp, such as when performing date-based calculations, ﬁltering data, or formatting date values. Syntax day(date) Arguments date A DATE or TIMESTAMP expression. Returns The DAY function returns an INTEGER. Date and time functions 203 AWS Clean Rooms Examples SQL Reference The following example extracts the day of the month (30) from the input date '2009-07-30'. SELECT day('2009-07-30'); 30 The following example extracts the day of the month from the birthday column of the squirrels table and returns the results as the output of the SELECT statement. The output of this query will be a list of day values, one for each row in the squirrels table, representing the day of the month for each squirrel's birthday. SELECT day(birthday) FROM squirrels DAYOFMONTH function The DAYOFMONTH function returns the day of month of the date/timestamp (a value between 1 and 31, depending on the month and year). The DAYOFMONTH function is similar to the DAY function, but they have slightly diﬀerent names and slightly diﬀerent behavior. The DAY function is more commonly used, but the DAYOFMONTH function can be used as an alternative. This type of query can be useful when you need to perform date-based analysis or ﬁltering on a table that contains date or timestamp data, such as extracting speciﬁc components of a date for further processing or reporting. Syntax dayofmonth(date) Arguments date A DATE or TIMESTAMP expression. Returns The DAYOFMONTH function returns an INTEGER. Date and time functions 204 AWS Clean Rooms Example SQL Reference The following example extracts the day of the month (30) from the input date '2009-07-30'. SELECT dayofmonth('2009-07-30'); 30 The following example applies the DAYOFMONTH function to the birthday column of the squirrels table. For each row in the squirrels table, the day of the month from the birthday column will be extracted and returned as the output of the SELECT statement. The output of this query will be a list of day values, one for each row in the squirrels table, representing the day of the month for each squirrel's birthday. SELECT dayofmonth(birthday)
sql-reference-059	sql-reference.pdf	59	204 AWS Clean Rooms Example SQL Reference The following example extracts the day of the month (30) from the input date '2009-07-30'. SELECT dayofmonth('2009-07-30'); 30 The following example applies the DAYOFMONTH function to the birthday column of the squirrels table. For each row in the squirrels table, the day of the month from the birthday column will be extracted and returned as the output of the SELECT statement. The output of this query will be a list of day values, one for each row in the squirrels table, representing the day of the month for each squirrel's birthday. SELECT dayofmonth(birthday) FROM squirrels DAYOFWEEK function The DAYOFWEEK function takes a date or timestamp as input and returns the day of the week as a number (1 for Sunday, 2 for Monday, ..., 7 for Saturday). This date extraction function is useful when you need to work with speciﬁc components of a date or timestamp, such as when performing date-based calculations, ﬁltering data, or formatting date values. Syntax dayofweek(date) Arguments date A DATE or TIMESTAMP expression. Returns The DAYOFWEEK function returns an INTEGER where 1 = Sunday Date and time functions 205 SQL Reference AWS Clean Rooms 2 = Monday 3 = Tuesday 4 = Wednesday 5 = Thursday 6 = Friday 7 = Saturday Examples The following example extracts the day of the week from this date, which is 5 (representing Thursday). SELECT dayofweek('2009-07-30'); 5 The following example extracts the day of the week from the birthday column of the squirrels table and returns the results as the output of the SELECT statement. The output of this query will be a list of day of the week values, one for each row in the squirrels table, representing the day of the week for each squirrel's birthday. SELECT dayofweek(birthday) FROM squirrels DAYOFYEAR function The DAYOFYEAR function is a date extraction function that takes a date or timestamp as input and returns the day of the year (a value between 1 and 366, depending on the year and whether it's a leap year). This function is useful when you need to work with speciﬁc components of a date or timestamp, such as when performing date-based calculations, ﬁltering data, or formatting date values. Syntax dayofyear(date) Date and time functions 206 AWS Clean Rooms Arguments date A DATE or TIMESTAMP expression. Returns SQL Reference The DAYOFYEAR function returns an INTEGER (between 1 and 366, depending on the year and whether it's a leap year). Examples The following example extracts the day of the year (100) from the input date '2016-04-09'. SELECT dayofyear('2016-04-09'); 100 The following example extracts the day of the year from the birthday column of the squirrels table and returns the results as the output of the SELECT statement. SELECT dayofyear(birthday) FROM squirrels EXTRACT function The EXTRACT function returns a date or time part from a TIMESTAMP, TIMESTAMPTZ, TIME, or TIMETZ value. Examples include a day, month, year, hour, minute, second, millisecond, or microsecond from a timestamp. Syntax EXTRACT(datepart FROM source) Arguments datepart The subﬁeld of a date or time to extract, such as a day, month, year, hour, minute, second, millisecond, or microsecond. For possible values, see Date parts for date or timestamp functions. Date and time functions 207 AWS Clean Rooms source SQL Reference A column or expression that evaluates to a data type of TIMESTAMP, TIMESTAMPTZ, TIME, or TIMETZ. Return type INTEGER if the source value evaluates to data type TIMESTAMP, TIME, or TIMETZ. DOUBLE PRECISION if the source value evaluates to data type TIMESTAMPTZ. Examples with TIME The following example table TIME_TEST has a column TIME_VAL (type TIME) with three values inserted. select time_val from time_test; time_val --------------------- 20:00:00 00:00:00.5550 00:58:00 The following example extracts the minutes from each time_val. select extract(minute from time_val) as minutes from time_test; minutes ----------- 0 0 58 The following example extracts the hours from each time_val. select extract(hour from time_val) as hours from time_test; hours ----------- 20 Date and time functions 208 AWS Clean Rooms 0 0 FROM_UTC_TIMESTAMP function SQL Reference The FROM_UTC_TIMESTAMP function converts the input date from UTC (Coordinated Universal Time) to the speciﬁed time zone. This function is useful when you need to convert date and time values from UTC to a speciﬁc time zone. This can be important when working with data that originates from diﬀerent parts of the world and needs to be presented in the appropriate local time. Syntax from_utc_timestamp(timestamp, timezone Arguments timestamp A TIMESTAMP expression with a UTC timestamp. timezone A STRING expression that is a valid timezone to which the input date or timestamp should be converted. Returns The FROM_UTC_TIMESTAMP function returns a TIMESTAMP. Example The following example converts the input date from UTC to the speciﬁed time zone ('Asia/ Seoul'), which in this case is 9 hours ahead of UTC. The resulting output is
sql-reference-060	sql-reference.pdf	60	to a speciﬁc time zone. This can be important when working with data that originates from diﬀerent parts of the world and needs to be presented in the appropriate local time. Syntax from_utc_timestamp(timestamp, timezone Arguments timestamp A TIMESTAMP expression with a UTC timestamp. timezone A STRING expression that is a valid timezone to which the input date or timestamp should be converted. Returns The FROM_UTC_TIMESTAMP function returns a TIMESTAMP. Example The following example converts the input date from UTC to the speciﬁed time zone ('Asia/ Seoul'), which in this case is 9 hours ahead of UTC. The resulting output is the date and time in the Seoul time zone, which is 2016-08-31 09:00:00. SELECT from_utc_timestamp('2016-08-31', 'Asia/Seoul'); 2016-08-31 09:00:00 Date and time functions 209 AWS Clean Rooms GETDATE function SQL Reference The GETDATE function returns the current date and time in the current session time zone (UTC by default). It returns the start date or time of the current statement, even when it is within a transaction block. Syntax GETDATE() The parentheses are required. Return type TIMESTAMP Example The following example uses the GETDATE function to return the full timestamp for the current date. select getdate(); HOUR function The HOUR function is a time extraction function that takes a time or timestamp as input and returns the hour component (a value between 0 and 23). This time extraction function is useful when you need to work with speciﬁc components of a time or timestamp, such as when performing time-based calculations, ﬁltering data, or formatting time values. Syntax hour(timestamp) Date and time functions 210 SQL Reference AWS Clean Rooms Arguments timestamp A TIMESTAMP expression. Returns The HOUR function returns an INTEGER. Example The following example extracts the hour component (12) from the input timestamp '2009-07-30 12:58:59'. SELECT hour('2009-07-30 12:58:59'); 12 MINUTE function The MINUTE function is a time extraction function that takes a time or timestamp as input and returns the minute component (a value between 0 and 60). Syntax minute(timestamp) Arguments timestamp A TIMESTAMP expression or a STRING of a valid timestamp format. Returns The MINUTE function returns an INTEGER. Example The following example extracts the minute component (58) from the input timestamp '2009-07-30 12:58:59'. Date and time functions 211 AWS Clean Rooms SQL Reference SELECT minute('2009-07-30 12:58:59'); 58 MONTH function The MONTH function is a time extraction function that takes a time or timestamp as input and returns the month component (a value between 0 and 12). Syntax month(date) Arguments date A TIMESTAMP expression or a STRING of a valid timestamp format. Returns The MONTH function returns an INTEGER. Example The following example extracts the month component (7) from the input timestamp '2016-07-30'. SELECT month('2016-07-30'); 7 SECOND function The SECOND function is a time extraction function that takes a time or timestamp as input and returns the second component (a value between 0 and 60). Syntax second(timestamp) Date and time functions 212 SQL Reference AWS Clean Rooms Arguments timestamp A TIMESTAMP expression. Returns The SECOND function returns an INTEGER. Example The following example extracts the second component (59) from the input timestamp '2009-07-30 12:58:59'. SELECT second('2009-07-30 12:58:59'); 59 SYSDATE function SYSDATE returns the current date and time in the current session time zone (UTC by default). Note SYSDATE returns the start date and time for the current transaction, not for the start of the current statement. Syntax SYSDATE This function requires no arguments. Return type TIMESTAMP Examples The following example uses the SYSDATE function to return the full timestamp for the current date. Date and time functions 213 AWS Clean Rooms select sysdate; timestamp ---------------------------- 2008-12-04 16:10:43.976353 (1 row) SQL Reference The following example uses the SYSDATE function inside the TRUNC function to return the current date without the time. select trunc(sysdate); trunc ------------ 2008-12-04 (1 row) The following query returns sales information for dates that fall between the date when the query is issued and whatever date is 120 days earlier. select salesid, pricepaid, trunc(saletime) as saletime, trunc(sysdate) as now from sales where saletime between trunc(sysdate)-120 and trunc(sysdate) order by saletime asc; salesid \| pricepaid \| saletime \| now ---------+-----------+------------+------------ 91535 \| 670.00 \| 2008-08-07 \| 2008-12-05 91635 \| 365.00 \| 2008-08-07 \| 2008-12-05 91901 \| 1002.00 \| 2008-08-07 \| 2008-12-05 ... TIMEOFDAY function TIMEOFDAY is a special alias used to return the weekday, date, and time as a string value. It returns the time of day string for the current statement, even when it is within a transaction block. Syntax TIMEOFDAY() Date and time functions 214 AWS Clean Rooms Return type VARCHAR Examples SQL Reference The following example returns the current date and time by using the TIMEOFDAY function. select timeofday(); timeofday ------------ Thu Sep 19 22:53:50.333525 2013 UTC (1 row) TIMESTAMP function The TIMESTAMP function takes a value (typically a number) and converts it to a timestamp
sql-reference-061	sql-reference.pdf	61	\| 2008-12-05 ... TIMEOFDAY function TIMEOFDAY is a special alias used to return the weekday, date, and time as a string value. It returns the time of day string for the current statement, even when it is within a transaction block. Syntax TIMEOFDAY() Date and time functions 214 AWS Clean Rooms Return type VARCHAR Examples SQL Reference The following example returns the current date and time by using the TIMEOFDAY function. select timeofday(); timeofday ------------ Thu Sep 19 22:53:50.333525 2013 UTC (1 row) TIMESTAMP function The TIMESTAMP function takes a value (typically a number) and converts it to a timestamp data type. This function is useful when you need to convert a numeric value representing a time or date to a timestamp data type. This can be helpful when you are working with data that is stored in a numeric format, such as Unix timestamps or epoch time. Syntax timestamp(expr) Arguments expr Any expression that can be cast to TIMESTAMP. Returns The TIMESTAMP function returns a TIMESTAMP. Example The following example converts a numeric Unix timestamp (1632416400) to its corresponding timestamp data type: September 22, 2021 at 12:00:00 PM UTC. Date and time functions 215 AWS Clean Rooms SQL Reference SELECT timestamp(1632416400); 2021-09-22 12:00:00 UTC TO_TIMESTAMP function TO_TIMESTAMP converts a TIMESTAMP string to TIMESTAMPTZ. Syntax to_timestamp (timestamp, format) to_timestamp (timestamp, format, is_strict) Arguments timestamp A string that represents a timestamp value in the format speciﬁed by format. If this argument is left as empty, the timestamp value defaults to 0001-01-01 00:00:00. format A string literal that deﬁnes the format of the timestamp value. Formats that include a time zone (TZ, tz, or OF) are not supported as input. For valid timestamp formats, see Datetime format strings. is_strict An optional Boolean value that speciﬁes whether an error is returned if an input timestamp value is out of range. When is_strict is set to TRUE, an error is returned if there is an out of range value. When is_strict is set to FALSE, which is the default, then overﬂow values are accepted. Return type TIMESTAMPTZ Examples The following example demonstrates using the TO_TIMESTAMP function to convert a TIMESTAMP string to a TIMESTAMPTZ. Date and time functions 216 AWS Clean Rooms SQL Reference select sysdate, to_timestamp(sysdate, 'YYYY-MM-DD HH24:MI:SS') as second; timestamp \| second -------------------------- ---------------------- 2021-04-05 19:27:53.281812 \| 2021-04-05 19:27:53+00 It's possible to pass TO_TIMESTAMP part of a date. The remaining date parts are set to default values. The time is included in the output: SELECT TO_TIMESTAMP('2017','YYYY'); to_timestamp -------------------------- 2017-01-01 00:00:00+00 The following SQL statement converts the string '2011-12-18 24:38:15' to a TIMESTAMPTZ. The result is a TIMESTAMPTZ that falls on the next day because the number of hours is more than 24 hours: SELECT TO_TIMESTAMP('2011-12-18 24:38:15', 'YYYY-MM-DD HH24:MI:SS'); to_timestamp ---------------------- 2011-12-19 00:38:15+00 The following SQL statement converts the string '2011-12-18 24:38:15' to a TIMESTAMPTZ. The result is an error because the time value in the timestamp is more than 24 hours: SELECT TO_TIMESTAMP('2011-12-18 24:38:15', 'YYYY-MM-DD HH24:MI:SS', TRUE); ERROR: date/time field time value out of range: 24:38:15.0 WEEKDAY function The WEEKDAY function is a date extraction function that takes a date or timestamp as input and returns the day of the week as a number (0 for Monday, 1 for Tuesday, ..., 6 for Sunday). The WEEKDAY function is similar to the DAYOFTHEWEEK function, but they have slightly diﬀerent behavior. The DAYOFTHEWEEK function returns the day of the week as a number (1 for Sunday, 2 Date and time functions 217 AWS Clean Rooms SQL Reference for Monday, ..., 7 for Saturday), whereas the WEEKDAY function returns the day of the week as a number (0 for Monday, 1 for Tuesday, ..., 6 for Sunday). This function is useful when you need to work with speciﬁc components of a date or timestamp, such as when performing date-based calculations, ﬁltering data, or formatting date values. Syntax weekday(date) Arguments date A DATE or TIMESTAMP expression. Returns The WEEKDAY function returns an INTEGER where 0 = Monday 1 = Tuesday 2 = Wednesday 3 = Thursday 4 = Friday 5 = Saturday 6 = Sunday Example The following example extracts the weekday (Thursday, which is represented by the number 3) from the input date '2009-07-30'. SELECT weekday('2009-07-30'); 3 Date and time functions 218 AWS Clean Rooms SQL Reference The following example extracts the weekday from the birthday column of the squirrels table and returns the results as the output of the SELECT statement. The output of this query will be a list of weekday values, one for each row in the squirrels table, representing the day of the week for each squirrel's birthday. SELECT weekday(birthday) FROM squirrels YEAR function The YEAR function is a date extraction function that takes a date or timestamp as input and returns the year component (a four-digit number). Syntax year(date) Arguments
sql-reference-062	sql-reference.pdf	62	weekday('2009-07-30'); 3 Date and time functions 218 AWS Clean Rooms SQL Reference The following example extracts the weekday from the birthday column of the squirrels table and returns the results as the output of the SELECT statement. The output of this query will be a list of weekday values, one for each row in the squirrels table, representing the day of the week for each squirrel's birthday. SELECT weekday(birthday) FROM squirrels YEAR function The YEAR function is a date extraction function that takes a date or timestamp as input and returns the year component (a four-digit number). Syntax year(date) Arguments date A DATE or TIMESTAMP expression. Returns The YEAR function returns an INTEGER. Example The following example extracts the year component (2016) from the input date '2016-07-30'. SELECT year('2016-07-30'); 2016 The following example extracts the year component from the birthday column of the squirrels table and returns the results as the output of the SELECT statement. The output of this query will be a list of year values, one for each row in the squirrels table, representing the year of each squirrel's birthday. SELECT year(birthday) FROM squirrels Date and time functions 219 AWS Clean Rooms SQL Reference Date parts for date or timestamp functions The following table identiﬁes the date part and time part names and abbreviations that are accepted as arguments to the following functions: • DATEADD • DATEDIFF • DATE_PART • EXTRACT Date part or time part Abbreviations millennium, millennia mil, mils century, centuries c, cent, cents decade, decades dec, decs epoch year, years quarter, quarters epoch (supported by the EXTRACT) y, yr, yrs qtr, qtrs month, months mon, mons week, weeks w day of week dayofweek, dow, dw, weekday (supported by the DATE_PART and the EXTRACT function) Returns an integer from 0–6, starting with Sunday. Note The DOW date part behaves diﬀerently from the day of week (D) date part used for datetime format strings. D Date and time functions 220 AWS Clean Rooms SQL Reference Date part or time part Abbreviations is based on integers 1–7, where Sunday is 1. For more information, see Datetime format strings. dayofyear, doy, dy, yearday (supported by the EXTRACT) d h, hr, hrs day of year day, days hour, hours minute, minutes m, min, mins second, seconds s, sec, secs millisecond, milliseconds ms, msec, msecs, msecond, mseconds, millisec, millisecs, millisecon microsecond, microseconds microsec, microsecs, microsecond, usecond, useconds, us, usec, usecs timezone, timezone_hour, timezone_minute Supported by the EXTRACT for timestamp with time zone (TIMESTAMPTZ) only. Variations in results with seconds, milliseconds, and microseconds Minor diﬀerences in query results occur when diﬀerent date functions specify seconds, milliseconds, or microseconds as date parts: • The EXTRACT function return integers for the speciﬁed date part only, ignoring higher- and lower-level date parts. If the speciﬁed date part is seconds, milliseconds and microseconds are not included in the result. If the speciﬁed date part is milliseconds, seconds and microseconds are not included. If the speciﬁed date part is microseconds, seconds and milliseconds are not included. • The DATE_PART function returns the complete seconds portion of the timestamp, regardless of the speciﬁed date part, returning either a decimal value or an integer as required. Date and time functions 221 AWS Clean Rooms SQL Reference CENTURY, EPOCH, DECADE, and MIL notes CENTURY or CENTURIES AWS Clean Rooms interprets a CENTURY to start with year ###1 and end with year ###0: select extract (century from timestamp '2000-12-16 12:21:13'); date_part ----------- 20 (1 row) select extract (century from timestamp '2001-12-16 12:21:13'); date_part ----------- 21 (1 row) EPOCH The AWS Clean Rooms implementation of EPOCH is relative to 1970-01-01 00:00:00.000000 independent of the time zone where the cluster resides. You might need to oﬀset the results by the diﬀerence in hours depending on the time zone where the cluster is located. DECADE or DECADES AWS Clean Rooms interprets the DECADE or DECADES DATEPART based on the common calendar. For example, because the common calendar starts from the year 1, the ﬁrst decade (decade 1) is 0001-01-01 through 0009-12-31, and the second decade (decade 2) is 0010-01-01 through 0019-12-31. For example, decade 201 spans from 2000-01-01 to 2009-12-31: select extract(decade from timestamp '1999-02-16 20:38:40'); date_part ----------- 200 (1 row) select extract(decade from timestamp '2000-02-16 20:38:40'); date_part ----------- 201 (1 row) Date and time functions 222 AWS Clean Rooms SQL Reference select extract(decade from timestamp '2010-02-16 20:38:40'); date_part ----------- 202 (1 row) MIL or MILS AWS Clean Rooms interprets a MIL to start with the ﬁrst day of year #001 and end with the last day of year #000: select extract (mil from timestamp '2000-12-16 12:21:13'); date_part ----------- 2 (1 row) select extract (mil from timestamp '2001-12-16 12:21:13'); date_part ----------- 3 (1 row) Encryption and decryption functions Encryption and decryption functions help SQL developers protect sensitive data from unauthorized access or misuse
sql-reference-063	sql-reference.pdf	63	from timestamp '2000-02-16 20:38:40'); date_part ----------- 201 (1 row) Date and time functions 222 AWS Clean Rooms SQL Reference select extract(decade from timestamp '2010-02-16 20:38:40'); date_part ----------- 202 (1 row) MIL or MILS AWS Clean Rooms interprets a MIL to start with the ﬁrst day of year #001 and end with the last day of year #000: select extract (mil from timestamp '2000-12-16 12:21:13'); date_part ----------- 2 (1 row) select extract (mil from timestamp '2001-12-16 12:21:13'); date_part ----------- 3 (1 row) Encryption and decryption functions Encryption and decryption functions help SQL developers protect sensitive data from unauthorized access or misuse by converting it between a readable, plaintext form and an unreadable, ciphertext form. AWS Clean Rooms Spark SQL supports the following encryption and decryption functions: Topics • AES_ENCRYPT function • AES_DECRYPT function AES_ENCRYPT function The AES_ENCRYPT function is used for encrypting data using the Advanced Encryption Standard (AES) algorithm. Encryption and decryption functions 223 AWS Clean Rooms Syntax SQL Reference aes_encrypt(expr, key[, mode[, padding[, iv[, aad]]]]) Arguments expr The binary value to encrypt. key The passphrase to use to encrypt the data. Key lengths of 16, 24 and 32 bits are supported. mode Speciﬁes which block cipher mode should be used to encrypt messages. Valid modes: ECB (Electronic CodeBook), GCM (Galois/Counter Mode), CBC (Cipher-Block Chaining). padding Speciﬁes how to pad messages whose length isn't a multiple of the block size. Valid values: PKCS, NONE, DEFAULT. The DEFAULT padding means PKCS (Public Key Cryptography Standards) for ECB, NONE for GCM and PKCS for CBC. Supported combinations of (mode, padding) are ('ECB', 'PKCS'), ('GCM', 'NONE') and ('CBC', 'PKCS'). iv Optional initialization vector (IV). Only supported for CBC and GCM modes. Valid values: 12-bytes long for GCM and 16 bytes for CBC. aad Optional additional authenticated data (AAD). Only supported for GCM mode. This can be any free-form input and must be provided for both encryption and decryption. Encryption and decryption functions 224 AWS Clean Rooms Return type SQL Reference The AES_ENCRYPT function returns an encrypted value of expr using AES in given mode with the speciﬁed padding. Examples The following example demonstrates how to use the Spark SQL AES_ENCRYPT function to securely encrypt a string of data (in this case, the word "Spark") using a speciﬁed encryption key. The resulting ciphertext is then Base64-encoded to make it easier to store or transmit. SELECT base64(aes_encrypt('Spark', 'abcdefghijklmnop')); 4A5jOAh9FNGwoMeuJukfllrLdHEZxA2DyuSQAWz77dfn The following example demonstrates how to use the Spark SQL AES_ENCRYPT function to securely encrypt a string of data (in this case, the word "Spark") using a speciﬁed encryption key. The resulting ciphertext is then represented in hexadecimal format, which can be useful for tasks such as data storage, transmission, or debugging. SELECT hex(aes_encrypt('Spark', '0000111122223333')); 83F16B2AA704794132802D248E6BFD4E380078182D1544813898AC97E709B28A94 The following example demonstrates how to use the Spark SQL AES_ENCRYPT function to securely encrypt a string of data (in this case, "Spark SQL") using a speciﬁed encryption key, encryption mode, and padding mode. The resulting ciphertext is then Base64-encoded to make it easier to store or transmit. SELECT base64(aes_encrypt('Spark SQL', '1234567890abcdef', 'ECB', 'PKCS')); 3lmwu+Mw0H3fi5NDvcu9lg== AES_DECRYPT function The AES_DECRYPT function is used for decrypting data using the Advanced Encryption Standard (AES) algorithm. Syntax aes_decrypt(expr, key[, mode[, padding[, aad]]]) Encryption and decryption functions 225 AWS Clean Rooms Arguments expr The binary value to decrypt. key The passphrase to use to decrypt the data. SQL Reference The passphrase must match the key originally used to produce the encrypted value and be 16, 24, or 32 bytes long. mode Speciﬁes which block cipher mode should be used to decrypt messages. Valid modes: ECB, GCM, CBC. padding Speciﬁes how to pad messages whose length isn't a multiple of the block size. Valid values: PKCS, NONE, DEFAULT. The DEFAULT padding means PKCS for ECB, NONE for GCM and PKCS for CBC. aad Optional additional authenticated data (AAD). Only supported for GCM mode. This can be any free-form input and must be provided for both encryption and decryption. Return type Returns a decrypted value of expr using AES in mode with padding. Examples The following example demonstrates how to use the Spark SQL AES_ENCRYPT function to securely encrypt a string of data (in this case, the word "Spark") using a speciﬁed encryption key. The resulting ciphertext is then Base64-encoded to make it easier to store or transmit. SELECT base64(aes_encrypt('Spark', 'abcdefghijklmnop')); 4A5jOAh9FNGwoMeuJukfllrLdHEZxA2DyuSQAWz77dfn Encryption and decryption functions 226 AWS Clean Rooms SQL Reference The following example demonstrates how to use the Spark SQL AES_DECRYPT function to decrypt data that has been previously encrypted and Base64-encoded. The decryption process requires the correct encryption key and parameters (encryption mode and padding mode) to successfully recover the original plaintext data. SELECT aes_decrypt(unbase64('3lmwu+Mw0H3fi5NDvcu9lg=='), '1234567890abcdef', 'ECB', 'PKCS'); Spark SQL Hash functions A hash function is a mathematical function that converts a numerical input value into another value. AWS Clean Rooms Spark SQL supports
sql-reference-064	sql-reference.pdf	64	is then Base64-encoded to make it easier to store or transmit. SELECT base64(aes_encrypt('Spark', 'abcdefghijklmnop')); 4A5jOAh9FNGwoMeuJukfllrLdHEZxA2DyuSQAWz77dfn Encryption and decryption functions 226 AWS Clean Rooms SQL Reference The following example demonstrates how to use the Spark SQL AES_DECRYPT function to decrypt data that has been previously encrypted and Base64-encoded. The decryption process requires the correct encryption key and parameters (encryption mode and padding mode) to successfully recover the original plaintext data. SELECT aes_decrypt(unbase64('3lmwu+Mw0H3fi5NDvcu9lg=='), '1234567890abcdef', 'ECB', 'PKCS'); Spark SQL Hash functions A hash function is a mathematical function that converts a numerical input value into another value. AWS Clean Rooms Spark SQL supports the following hash functions: Topics • MD5 function • SHA function • SHA1 function • SHA2 function • MURMUR3_32_HASH function • xxHASH64 function MD5 function Uses the MD5 cryptographic hash function to convert a variable-length string into a 32-character string that is a text representation of the hexadecimal value of a 128-bit checksum. Syntax MD5(string) Arguments string A variable-length string. Hash functions 227 AWS Clean Rooms Return type SQL Reference The MD5 function returns a 32-character string that is a text representation of the hexadecimal value of a 128-bit checksum. Examples The following example shows the 128-bit value for the string 'AWS Clean Rooms': select md5('AWS Clean Rooms'); md5 ---------------------------------- f7415e33f972c03abd4f3fed36748f7a (1 row) SHA function Synonym of SHA1 function. See SHA1 function. SHA1 function The SHA1 function uses the SHA1 cryptographic hash function to convert a variable-length string into a 40-character string that is a text representation of the hexadecimal value of a 160-bit checksum. Syntax SHA1 is a synonym of SHA function. SHA1(string) Arguments string A variable-length string. Hash functions 228 AWS Clean Rooms Return type SQL Reference The SHA1 function returns a 40-character string that is a text representation of the hexadecimal value of a 160-bit checksum. Example The following example returns the 160-bit value for the word 'AWS Clean Rooms': select sha1('AWS Clean Rooms'); SHA2 function The SHA2 function uses the SHA2 cryptographic hash function to convert a variable-length string into a character string. The character string is a text representation of the hexadecimal value of the checksum with the speciﬁed number of bits. Syntax SHA2(string, bits) Arguments string A variable-length string. integer The number of bits in the hash functions. Valid values are 0 (same as 256), 224, 256, 384, and 512. Return type The SHA2 function returns a character string that is a text representation of the hexadecimal value of the checksum or an empty string if the number of bits is invalid. Example The following example returns the 256-bit value for the word 'AWS Clean Rooms': select sha2('AWS Clean Rooms', 256); Hash functions 229 AWS Clean Rooms SQL Reference MURMUR3_32_HASH function The MURMUR3_32_HASH function computes the 32-bit Murmur3A non-cryptographic hash for all common data types including numeric and string types. Syntax MURMUR3_32_HASH(value [, seed]) Arguments value The input value to hash. AWS Clean Rooms hashes the binary representation of the input value. This behavior is similar to FNV_HASH, but the value is converted to the binary representation speciﬁed by the Apache Iceberg 32-bit Murmur3 hash speciﬁcation. seed The INT seed of the hash function. This argument is optional. If not given, AWS Clean Rooms uses the default seed of 0. This enables combining the hash of multiple columns without any conversions or concatenations. Return type The function returns an INT. Example The following examples return the Murmur3 hash of a number, the string 'AWS Clean Rooms', and the concatenation of the two. select MURMUR3_32_HASH(1); MURMUR3_32_HASH ---------------------- -5968735742475085980 (1 row) select MURMUR3_32_HASH('AWS Clean Rooms'); Hash functions 230 AWS Clean Rooms MURMUR3_32_HASH ---------------------- 7783490368944507294 (1 row) SQL Reference select MURMUR3_32_HASH('AWS Clean Rooms', MURMUR3_32_HASH(1)); MURMUR3_32_HASH ---------------------- -2202602717770968555 (1 row) Usage notes To compute the hash of a table with multiple columns, you can compute the Murmur3 hash of the ﬁrst column and pass it as a seed to the hash of the second column. Then, it passes the Murmur3 hash of the second column as a seed to the hash of the third column. The following example creates seeds to hash a table with multiple columns. select MURMUR3_32_HASH(column_3, MURMUR3_32_HASH(column_2, MURMUR3_32_HASH(column_1))) from sample_table; The same property can be used to compute the hash of a concatenation of strings. select MURMUR3_32_HASH('abcd'); MURMUR3_32_HASH --------------------- -281581062704388899 (1 row) select MURMUR3_32_HASH('cd', MURMUR3_32_HASH('ab')); MURMUR3_32_HASH --------------------- -281581062704388899 (1 row) The hash function uses the type of the input to determine the number of bytes to hash. Use casting to enforce a speciﬁc type, if necessary. Hash functions 231 AWS Clean Rooms SQL Reference The following examples use diﬀerent input types to produce diﬀerent results. select MURMUR3_32_HASH(1::smallint); MURMUR3_32_HASH -------------------- 589727492704079044 (1 row) select MURMUR3_32_HASH(1); MURMUR3_32_HASH ---------------------- -5968735742475085980 (1 row) select MURMUR3_32_HASH(1::bigint); MURMUR3_32_HASH ---------------------- -8517097267634966620 (1 row) xxHASH64 function The xxhash64 function returns a 64-bit hash value of the arguments. The xxhash64() function is a non-cryptographic
sql-reference-065	sql-reference.pdf	65	select MURMUR3_32_HASH('abcd'); MURMUR3_32_HASH --------------------- -281581062704388899 (1 row) select MURMUR3_32_HASH('cd', MURMUR3_32_HASH('ab')); MURMUR3_32_HASH --------------------- -281581062704388899 (1 row) The hash function uses the type of the input to determine the number of bytes to hash. Use casting to enforce a speciﬁc type, if necessary. Hash functions 231 AWS Clean Rooms SQL Reference The following examples use diﬀerent input types to produce diﬀerent results. select MURMUR3_32_HASH(1::smallint); MURMUR3_32_HASH -------------------- 589727492704079044 (1 row) select MURMUR3_32_HASH(1); MURMUR3_32_HASH ---------------------- -5968735742475085980 (1 row) select MURMUR3_32_HASH(1::bigint); MURMUR3_32_HASH ---------------------- -8517097267634966620 (1 row) xxHASH64 function The xxhash64 function returns a 64-bit hash value of the arguments. The xxhash64() function is a non-cryptographic hash function designed to be fast and eﬃcient. It's often used in data processing and storage applications, where a unique identiﬁer for a piece of data is needed, but the exact contents of the data don't need to be kept secret. In the context of a SQL query, the xxhash64() function could be used for various purposes, such as: • Generating a unique identiﬁer for a row in a table • Partitioning data based on a hash value • Implementing custom indexing or data distribution strategies The speciﬁc use case would depend on the requirements of the application and the data being processed. Hash functions 232 SQL Reference AWS Clean Rooms Syntax xxhash64(expr1, expr2, ...) Arguments expr1 An expression of any type. expr2 An expression of any type. Returns Returns a 64-bit hash value of the arguments (BIGINT). Hash seed is 42. Example The following example generates a 64-bit hash value (5602566077635097486) based on the provided input. The ﬁrst argument is a string value, in this case, the word "Spark". The second argument is an array containing the single integer value 123. The third argument is an integer value representing the seed for the hash function. SELECT xxhash64('Spark', array(123), 2); 5602566077635097486 Hyperloglog functions The HyperLogLog (HLL) functions in SQL provide a way to eﬃciently estimate the number of unique elements (cardinality) in a large dataset, even when the actual set of unique elements isn't stored. The main beneﬁts of using HLL functions are: • Memory eﬃciency: HLL sketches require much less memory than storing the full set of unique elements, making them suitable for large datasets. • Distributed computing: HLL sketches can be combined across multiple data sources or processing nodes, allowing for eﬃcient distributed unique count estimation. Hyperloglog functions 233 AWS Clean Rooms SQL Reference • Approximate results: HLL provides an approximate unique count estimation, with a tunable trade-oﬀ between accuracy and memory usage (via the precision parameter). These functions are particularly useful in scenarios where you need to estimate the number of unique items, such as in analytics, data warehousing, and real-time stream processing applications. AWS Clean Rooms supports the following HLL functions. Topics • HLL_SKETCH_AGG function • HLL_SKETCH_ESTIMATE function • HLL_UNION function • HLL_UNION_AGG function HLL_SKETCH_AGG function The HLL_SKETCH_AGG aggregate function creates an HLL sketch from the values in the speciﬁed column. It returns an HLLSKETCH data type that encapsulates the input expression values. The HLL_SKETCH_AGG aggregate function works with any data type and ignores NULL values. When there are no rows in a table or all rows are NULL, the resulting sketch has no index-value pairs such as {"version":1,"logm":15,"sparse":{"indices":[],"values":[]}}. Syntax HLL_SKETCH_AGG (aggregate_expression[, lgConfigK ] ) Argument aggregate_expression Any expression of type INT, BIGINT, STRING, or BINARY against which unique counting will occur. Any NULL values are ignored. lgConﬁgK An optional INT constant between 4 and 21 inclusive with default 12. The log-base-2 of K, where K is the number of buckets or slots for the sketch. Hyperloglog functions 234 AWS Clean Rooms Return type SQL Reference The HLL_SKETCH_AGG function returns a non-NULL BINARY buﬀer containing the HyperLogLog sketch computed because of consuming and aggregating all input values in the aggregation group. Examples The following examples use the HyperLogLog (HLL) algorithm to estimate the distinct count of values in the col column. The hll_sketch_agg(col, 12) function aggregates the values in the col column, creating an HLL sketch using a precision of 12. The hll_sketch_estimate() function is then used to estimate the distinct count of values based on the generated HLL sketch. The ﬁnal result of the query is 3, which represents the estimated distinct count of values in the col column. In this case, the distinct values are 1, 2, and 3. SELECT hll_sketch_estimate(hll_sketch_agg(col, 12)) FROM VALUES (1), (1), (2), (2), (3) tab(col); 3 The following example also uses the HLL algorithm to estimate the distinct count of values in the col column, but it doesn't specify a precision value for the HLL sketch. In this case, it uses the default precision of 14. The hll_sketch_agg(col) function takes the values in the col column and creates an HyperLogLog (HLL) sketch, which is a compact data structure that can be used to estimate the distinct count of elements.
sql-reference-066	sql-reference.pdf	66	in the col column. In this case, the distinct values are 1, 2, and 3. SELECT hll_sketch_estimate(hll_sketch_agg(col, 12)) FROM VALUES (1), (1), (2), (2), (3) tab(col); 3 The following example also uses the HLL algorithm to estimate the distinct count of values in the col column, but it doesn't specify a precision value for the HLL sketch. In this case, it uses the default precision of 14. The hll_sketch_agg(col) function takes the values in the col column and creates an HyperLogLog (HLL) sketch, which is a compact data structure that can be used to estimate the distinct count of elements. The hll_sketch_estimate(hll_sketch_agg(col)) function takes the HLL sketch created in the previous step and calculates an estimate of the distinct count of values in the col column. The ﬁnal result of the query is 3, which represents the estimated distinct count of values in the col column. In this case, the distinct values are 1, 2, and 3. SELECT hll_sketch_estimate(hll_sketch_agg(col)) FROM VALUES (1), (1), (2), (2), (3) tab(col); 3 HLL_SKETCH_ESTIMATE function The HLL_SKETCH_ESTIMATE function takes an HLL sketch and estimates the number of unique elements represented by the sketch. It uses the HyperLogLog (HLL) algorithm to count a probabilistic approximation of the number of unique values in a given column, consuming a binary representation known as a sketch buﬀer previously generated by the HLL_SKETCH_AGG function and returning the result as a big integer. Hyperloglog functions 235 AWS Clean Rooms SQL Reference The HLL sketching algorithm provides an eﬃcient way to estimate the number of unique elements, even for large datasets, without having to store the full set of unique values. The hll_union and hll_union_agg functions can also combine sketches together by consuming and merging these buﬀers as inputs. Syntax HLL_SKETCH_ESTIMATE (hllsketch_expression) Argument hllsketch_expression A BINARY expression holding a sketch generated by HLL_SKETCH_AGG Return type The HLL_SKETCH_ESTIMATE function returns a BIGINT value that is the approximate distinct count represented by the input sketch. Examples The following examples use the HyperLogLog (HLL) sketching algorithm to estimate the cardinality (unique count) of values in the col column. The hll_sketch_agg(col, 12) function takes the col column and creates an HLL sketch using a precision of 12 bits. The HLL sketch is an approximate data structure that can eﬃciently estimate the number of unique elements in a set. The hll_sketch_estimate() function takes the HLL sketch created by hll_sketch_agg and estimates the cardinality (unique count) of the values represented by the sketch. The FROM VALUES (1), (1), (2), (2), (3) tab(col); generates a test dataset with 5 rows, where the col column contains the values 1, 1, 2, 2, and 3. The result of this query is the estimated unique count of the values in the col column, which is 3. SELECT hll_sketch_estimate(hll_sketch_agg(col, 12)) FROM VALUES (1), (1), (2), (2), (3) tab(col); 3 The diﬀerence between the following example and the previous one is that the precision parameter (12 bits) isn't speciﬁed in the hll_sketch_agg function call. In this case, the default Hyperloglog functions 236 AWS Clean Rooms SQL Reference precision of 14 bits is used, which may provide a more accurate estimate for the unique count compared to the previous example that used 12 bits of precision. SELECT hll_sketch_estimate(hll_sketch_agg(col)) FROM VALUES (1), (1), (2), (2), (3) tab(col); 3 HLL_UNION function The HLL_UNION function combines two HLL sketches into a single, uniﬁed sketch. It uses the HyperLogLog (HLL) algorithm to combine two sketches into a single sketch. Queries can use the resulting buﬀers to compute approximate unique counts as long integers with the hll_sketch_estimate function. Syntax HLL_UNION (( expr1, expr2 [, allowDifferentLgConfigK ] )) Argument exprN A BINARY expression holding a sketch generated by HLL_SKETCH_AGG. allowDiﬀerentLgConﬁgK A optional BOOLEAN expression controlling whether to allow merging two sketches with diﬀerent lgConﬁgK values. The default value is false. Return type The HLL_UNION function returns a BINARY buﬀer containing the HyperLogLog sketch computed as a result of combining the input expressions. When the allowDifferentLgConfigK parameter is true, the result sketch uses the smaller of the two provided lgConfigK values. Examples The following examples use the HyperLogLog (HLL) sketching algorithm to estimate the unique count of values across two columns, col1 and col2, in a dataset. Hyperloglog functions 237 AWS Clean Rooms SQL Reference The hll_sketch_agg(col1) function creates an HLL sketch for the unique values in the col1 column. The hll_sketch_agg(col2) function creates an HLL sketch for the unique values in the col2 column. The hll_union(...) function combines the two HLL sketches created in steps 1 and 2 into a single, uniﬁed HLL sketch. The hll_sketch_estimate(...) function takes the combined HLL sketch and estimates the unique count of values across both col1 and col2. The FROM VALUES clause generates a test dataset with 5 rows, where col1 contains the values 1, 1, 2, 2, and 3, and col2 contains the values
sql-reference-067	sql-reference.pdf	67	Rooms SQL Reference The hll_sketch_agg(col1) function creates an HLL sketch for the unique values in the col1 column. The hll_sketch_agg(col2) function creates an HLL sketch for the unique values in the col2 column. The hll_union(...) function combines the two HLL sketches created in steps 1 and 2 into a single, uniﬁed HLL sketch. The hll_sketch_estimate(...) function takes the combined HLL sketch and estimates the unique count of values across both col1 and col2. The FROM VALUES clause generates a test dataset with 5 rows, where col1 contains the values 1, 1, 2, 2, and 3, and col2 contains the values 4, 4, 5, 5, and 6. The result of this query is the estimated unique count of values across both col1 and col2, which is 6. The HLL sketching algorithm provides an eﬃcient way to estimate the number of unique elements, even for large datasets, without having to store the full set of unique values. In this example, the hll_union function is used to combine the HLL sketches from the two columns, which allows the unique count to be estimated across the entire dataset, rather than just for each column individually. SELECT hll_sketch_estimate( hll_union( hll_sketch_agg(col1), hll_sketch_agg(col2))) FROM VALUES (1, 4), (1, 4), (2, 5), (2, 5), (3, 6) AS tab(col1, col2); 6 The diﬀerence between the following example and the previous one is that the precision parameter (12 bits) isn't speciﬁed in the hll_sketch_agg function call. In this case, the default precision of 14 bits is used, which may provide a more accurate estimate for the unique count compared to the previous example that used 12 bits of precision. SELECT hll_sketch_estimate( hll_union( Hyperloglog functions 238 AWS Clean Rooms SQL Reference hll_sketch_agg(col1, 14), hll_sketch_agg(col2, 14))) FROM VALUES (1, 4), (1, 4), (2, 5), (2, 5), (3, 6) AS tab(col1, col2); HLL_UNION_AGG function The HLL_UNION_AGG function combines multiple HLL sketches into a single, uniﬁed sketch. It uses the HyperLogLog (HLL) algorithm to combine a group of sketches into a single one. Queries can use the resulting buﬀers to compute approximate unique counts with the hll_sketch_estimate function. Syntax HLL_UNION_AGG ( expr [, allowDifferentLgConfigK ] ) Argument expr A BINARY expression holding a sketch generated by HLL_SKETCH_AGG. allowDiﬀerentLgConﬁgK A optional BOOLEAN expression controlling whether to allow merging two sketches with diﬀerent lgConﬁgK values. The default value is false. Return type The HLL_UNION_AGG function returns a BINARY buﬀer containing the HyperLogLog sketch computed as a result of combining the input expressions of the same group. When the allowDifferentLgConfigK parameter is true, the result sketch uses the smaller of the two provided lgConfigK values. Examples The following examples use the HyperLogLog (HLL) sketching algorithm to estimate the unique count of values across multiple HLL sketches. Hyperloglog functions 239 AWS Clean Rooms SQL Reference The ﬁrst example estimates the unique count of values in a dataset. SELECT hll_sketch_estimate(hll_union_agg(sketch, true)) FROM (SELECT hll_sketch_agg(col) as sketch FROM VALUES (1) AS tab(col) UNION ALL SELECT hll_sketch_agg(col, 20) as sketch FROM VALUES (1) AS tab(col)); 1 The inner query creates two HLL sketches: • The ﬁrst SELECT statement creates a sketch from a single value of 1. • The second SELECT statement creates a sketch from another single value of 1, but with a precision of 20. The outer query uses the HLL_UNION_AGG function to combine the two sketches into a single sketch. Then it applies the HLL_SKETCH_ESTIMATE function to this combined sketch to estimate the unique count of values. The result of this query is the estimated unique count of the values in the col column, which is 1. This means that the two input values of 1 are considered to be unique, even though they have the same value. The second example includes a diﬀerent precision parameter for the HLL_UNION_AGG function. In this case, both HLL sketches are created with a precision of 14 bits, which allows them to be successfully combined using hll_union_agg with the true parameter. SELECT hll_sketch_estimate(hll_union_agg(sketch, true)) FROM (SELECT hll_sketch_agg(col, 14) as sketch FROM VALUES (1) AS tab(col) UNION ALL SELECT hll_sketch_agg(col, 14) as sketch FROM VALUES (1) AS tab(col)); 1 The ﬁnal result of the query is the estimated unique count, which in this case is also 1. This means that the two input values of 1 are considered to be unique, even though they have the same value. Hyperloglog functions 240 AWS Clean Rooms JSON functions SQL Reference When you need to store a relatively small set of key-value pairs, you might save space by storing the data in JSON format. Because JSON strings can be stored in a single column, using JSON might be more eﬃcient than storing your data in tabular format. Example For example, suppose you have a sparse table, where you need to have many columns to fully represent all possible attributes. However, most of the column values are
sql-reference-068	sql-reference.pdf	68	values of 1 are considered to be unique, even though they have the same value. Hyperloglog functions 240 AWS Clean Rooms JSON functions SQL Reference When you need to store a relatively small set of key-value pairs, you might save space by storing the data in JSON format. Because JSON strings can be stored in a single column, using JSON might be more eﬃcient than storing your data in tabular format. Example For example, suppose you have a sparse table, where you need to have many columns to fully represent all possible attributes. However, most of the column values are NULL for any given row or any given column. By using JSON for storage, you might be able to store the data for a row in key-value pairs in a single JSON string and eliminate the sparsely-populated table columns. In addition, you can easily modify JSON strings to store additional key:value pairs without needing to add columns to a table. We recommend using JSON sparingly. JSON isn't a good choice for storing larger datasets because, by storing disparate data in a single column, JSON doesn't use the AWS Clean Rooms column store architecture. JSON uses UTF-8 encoded text strings, so JSON strings can be stored as CHAR or VARCHAR data types. Use VARCHAR if the strings include multi-byte characters. JSON strings must be properly formatted JSON, according to the following rules: • The root level JSON can either be a JSON object or a JSON array. A JSON object is an unordered set of comma-separated key:value pairs enclosed by curly braces. For example, {"one":1, "two":2} • A JSON array is an ordered set of comma-separated values enclosed by brackets. An example is the following: ["first", {"one":1}, "second", 3, null] • JSON arrays use a zero-based index; the ﬁrst element in an array is at position 0. In a JSON key:value pair, the key is a string in double quotation marks. • A JSON value can be any of the following: • JSON object • JSON array • String in double quotation marks JSON functions 241 AWS Clean Rooms • Number (integer and ﬂoat) • Boolean • Null • Empty objects and empty arrays are valid JSON values. • JSON ﬁelds are case-sensitive. • White space between JSON structural elements (such as { }, [ ]) is ignored. SQL Reference Topics • GET_JSON_OBJECT function • TO_JSON function GET_JSON_OBJECT function The GET_JSON_OBJECT function extracts a json object from path. Syntax get_json_object(json_txt, path) Arguments json_txt A STRING expression containing well formed JSON. path A STRING literal with a well formed JSON path expression. Returns Returns a STRING. A NULL is returned if the object can't be found. Example The following example extracts a value from a JSON object.. The ﬁrst argument is a JSON string that represents a simple object with a single key-value pair. The second argument is a JSON path JSON functions 242 AWS Clean Rooms SQL Reference expression. The $ symbol represents the root of the JSON object, and the .a part speciﬁes that we want to extract the value associated with the "a" key. The output of the function is 'b', which is the value associated with the "a" key in the input JSON object. SELECT get_json_object('{"a":"b"}', '$.a'); b TO_JSON function The TO_JSON function converts an input expression into a JSON string representation. The function handles the conversion of diﬀerent data types (such as numbers, strings, and booleans) into their corresponding JSON representations. The TO_JSON function is useful when you need to convert structured data (such as database rows or JSON objects) into a more portable, self-describing format like JSON. This can be particularly helpful when you need to interact with other systems or services that expect JSON-formatted data. Syntax to_json(expr[, options]) Arguments expr The input expression that you want to convert to a JSON string. It can be a value, a column, or any other valid SQL expression. options An optional set of conﬁguration options that can be used to customize the JSON conversion process. These options may include things like the handling of null values, the representation of numeric values, and the treatment of special characters.. Returns Returns a JSON string with a given struct value JSON functions 243 AWS Clean Rooms Examples SQL Reference The following example converts a named struct (a type of structured data) into a JSON string. The ﬁrst argument (named_struct('a', 1, 'b', 2)) is the input expression that is passed to the to_json() function. It creates a named struct with two ﬁelds: "a" with a value of 1, and "b" with a value of 2. The to_json() function takes the named struct as its argument and converts it into a JSON string representation. The output is {"a":1,"b":2}, which is a valid JSON string that represented the named struct. SELECT to_json(named_struct('a', 1, 'b',
sql-reference-069	sql-reference.pdf	69	functions 243 AWS Clean Rooms Examples SQL Reference The following example converts a named struct (a type of structured data) into a JSON string. The ﬁrst argument (named_struct('a', 1, 'b', 2)) is the input expression that is passed to the to_json() function. It creates a named struct with two ﬁelds: "a" with a value of 1, and "b" with a value of 2. The to_json() function takes the named struct as its argument and converts it into a JSON string representation. The output is {"a":1,"b":2}, which is a valid JSON string that represented the named struct. SELECT to_json(named_struct('a', 1, 'b', 2)); {"a":1,"b":2} The following example converts a named struct that contains a timestamp value into a JSON string, with a customized timestamp format. The ﬁrst argument (named_struct('time', to_timestamp('2015-08-26', 'yyyy-MM-dd'))) creates a named struct with a single ﬁeld 'time' that contains the timestamp value. The second argument (map('timestampFormat', 'dd/MM/yyyy')) creates a map (key-value dictionary) with a single key-value pair, where the key is 'timestampFormat' and the value is 'dd/MM/yyyy'. This map is used to specify the desired format for the timestamp value when converting it to JSON. The to_json() function converts the named struct into a JSON string. The second argument, the map, is used to customize the timestamp format to 'dd/MM/yyyy'. The output is {"time":"26/08/2015"}, which is a JSON string with a single ﬁeld 'time' that contains the timestamp value in the desired 'dd/MM/yyyy' format. SELECT to_json(named_struct('time', to_timestamp('2015-08-26', 'yyyy-MM-dd')), map('timestampFormat', 'dd/MM/yyyy')); {"time":"26/08/2015"} Math functions This section describes the mathematical operators and functions supported in AWS Clean Rooms Spark SQL. Topics • Mathematical operator symbols • ABS function • ACOS function • ACOSH function Math functions 244 SQL Reference AWS Clean Rooms • ASIN function • ATAN function • ATAN2 function • CBRT function • CEILING (or CEIL) function • COS function • COT function • DEGREES function • DIV function • DLOG1 function • DLOG10 function • EXP function • FLOOR function • LN function • LOG function • MOD function • PI function • POW function • POWER function • RADIANS function • RAND function • RANDOM function • ROUND function • SIGN function • SIN function • SQRT function • TRUNC function Mathematical operator symbols The following table lists the supported mathematical operators. Math functions 245 SQL Reference AWS Clean Rooms Supported operators Operator Description Example Result + - * / % ^ \|/ \|\|/ @ addition 2 + 3 subtraction 2 - 3 multiplic ation 2 * 3 division 4 / 2 modulo 5 % 4 exponenti ation 2.0 ^ 3.0 square root \| / 25.0 cube root \|\| / 27.0 absolute value @ -5.0 5 -1 6 2 1 8 5 3 5 Examples Calculate the commission paid plus a $2.00 handling fee for a given transaction: select commission, (commission + 2.00) as comm from sales where salesid=10000; commission \| comm -----------+------- 28.05 \| 30.05 (1 row) Calculate 20 percent of the sales price for a given transaction: select pricepaid, (pricepaid * .20) as twentypct Math functions 246 AWS Clean Rooms SQL Reference from sales where salesid=10000; pricepaid \| twentypct ----------+----------- 187.00 \| 37.400 (1 row) Forecast ticket sales based on a continuous growth pattern. In this example, the subquery returns the number of tickets sold in 2008. That result is multiplied exponentially by a continuous growth rate of 5 percent over 10 years. select (select sum(qtysold) from sales, date where sales.dateid=date.dateid and year=2008) ^ ((5::float/100)*10) as qty10years; qty10years ------------------ 587.664019657491 (1 row) Find the total price paid and commission for sales with a date ID that is greater than or equal to 2,000. Then subtract the total commission from the total price paid. select sum (pricepaid) as sum_price, dateid, sum (commission) as sum_comm, (sum (pricepaid) - sum (commission)) as value from sales where dateid >= 2000 group by dateid order by dateid limit 10; sum_price \| dateid \| sum_comm \| value -----------+--------+----------+----------- 364445.00 \| 2044 \| 54666.75 \| 309778.25 349344.00 \| 2112 \| 52401.60 \| 296942.40 343756.00 \| 2124 \| 51563.40 \| 292192.60 378595.00 \| 2116 \| 56789.25 \| 321805.75 328725.00 \| 2080 \| 49308.75 \| 279416.25 349554.00 \| 2028 \| 52433.10 \| 297120.90 249207.00 \| 2164 \| 37381.05 \| 211825.95 285202.00 \| 2064 \| 42780.30 \| 242421.70 320945.00 \| 2012 \| 48141.75 \| 272803.25 321096.00 \| 2016 \| 48164.40 \| 272931.60 (10 rows) Math functions 247 AWS Clean Rooms ABS function SQL Reference ABS calculates the absolute value of a number, where that number can be a literal or an expression that evaluates to a number. Syntax ABS (number) Arguments number Number or expression that evaluates to a number. It can be the SMALLINT, INTEGER, BIGINT, DECIMAL, FLOAT4, or FLOAT8 type. Return type ABS returns the same data type as its argument. Examples Calculate the absolute value of -38: select abs (-38); abs -------
sql-reference-070	sql-reference.pdf	70	42780.30 \| 242421.70 320945.00 \| 2012 \| 48141.75 \| 272803.25 321096.00 \| 2016 \| 48164.40 \| 272931.60 (10 rows) Math functions 247 AWS Clean Rooms ABS function SQL Reference ABS calculates the absolute value of a number, where that number can be a literal or an expression that evaluates to a number. Syntax ABS (number) Arguments number Number or expression that evaluates to a number. It can be the SMALLINT, INTEGER, BIGINT, DECIMAL, FLOAT4, or FLOAT8 type. Return type ABS returns the same data type as its argument. Examples Calculate the absolute value of -38: select abs (-38); abs ------- 38 (1 row) Calculate the absolute value of (14-76): select abs (14-76); abs ------- 62 (1 row) ACOS function ACOS is a trigonometric function that returns the arc cosine of a number. The return value is in radians and is between 0 and PI. Math functions 248 SQL Reference AWS Clean Rooms Syntax ACOS(number) Arguments number The input parameter is a DOUBLE PRECISION number. Return type DOUBLE PRECISION Examples To return the arc cosine of -1, use the following example. SELECT ACOS(-1); +-------------------+ \| acos \| +-------------------+ \| 3.141592653589793 \| +-------------------+ ACOSH function The ACOSH function returns the inverse hyperbolic cosine of a numeric expression. Syntax ACOSH(expression) Arguments expression A numeric expression. Math functions 249 AWS Clean Rooms Return type DOUBLE SQL Reference If the argument is out of bounds, NaN (or "Not a Number") is returned. Example The following example selects two columns from the squirrels table: the id column, which contains the unique identiﬁer for each squirrel, and a calculated column, acosh(latitude), which represents the inverse hyperbolic cosine (acosh) of the latitude column value. SELECT id, acosh(latitude) FROM squirrels The acosh() function is a mathematical function that calculates the inverse hyperbolic cosine of a given value. It's used to transform the latitude values in a way that may be useful for further data analysis or calculations. The resulting output would provide the id and the transformed latitude value (using the acosh() function) for each squirrel in the squirrels table. This type of query might be used in scenarios where the original latitude values need to be transformed or scaled in a speciﬁc way for the purposes of data analysis, modeling, or other computational processes. ASIN function ASIN is a trigonometric function that returns the arc sine of a number. The return value is in radians and is between PI/2 and -PI/2. Syntax ASIN(number) Arguments number The input parameter is a DOUBLE PRECISION number. Return type DOUBLE PRECISION Math functions 250 AWS Clean Rooms Examples To return the arc sine of 1, use the following example. SQL Reference SELECT ASIN(1) AS halfpi; +--------------------+ \| halfpi \| +--------------------+ \| 1.5707963267948966 \| +--------------------+ ATAN function ATAN is a trigonometric function that returns the arc tangent of a number. The return value is in radians and is between -PI and PI. Syntax ATAN(number) Arguments number The input parameter is a DOUBLE PRECISION number. Return type DOUBLE PRECISION Examples To return the arc tangent of 1 and multiply it by 4, use the following example. SELECT ATAN(1) * 4 AS pi; +-------------------+ \| pi \| +-------------------+ Math functions 251 AWS Clean Rooms \| 3.141592653589793 \| +-------------------+ ATAN2 function SQL Reference ATAN2 is a trigonometric function that returns the arc tangent of one number divided by another number. The return value is in radians and is between PI/2 and -PI/2. Syntax ATAN2(number1, number2) Arguments number1 A DOUBLE PRECISION number. number2 A DOUBLE PRECISION number. Return type DOUBLE PRECISION Examples To return the arc tangent of 2/2 and multiply it by 4, use the following example. SELECT ATAN2(2,2) * 4 AS PI; +-------------------+ \| pi \| +-------------------+ \| 3.141592653589793 \| +-------------------+ CBRT function The CBRT function is a mathematical function that calculates the cube root of a number. Math functions 252 SQL Reference AWS Clean Rooms Syntax CBRT (number) Argument CBRT takes a DOUBLE PRECISION number as an argument. Return type CBRT returns a DOUBLE PRECISION number. Examples Calculate the cube root of the commission paid for a given transaction: select cbrt(commission) from sales where salesid=10000; cbrt ------------------ 3.03839539048843 (1 row) CEILING (or CEIL) function The CEILING or CEIL function is used to round a number up to the next whole number. (The FLOOR function rounds a number down to the next whole number.) Syntax CEIL \| CEILING(number) Arguments number The number or expression that evaluates to a number. It can be the SMALLINT, INTEGER, BIGINT, DECIMAL, FLOAT4, or FLOAT8 type. Return type CEILING and CEIL return the same data type as its argument. Math functions 253 AWS Clean Rooms Example SQL Reference Calculate the ceiling of the commission paid for a given sales transaction: select ceiling(commission) from sales where salesid=10000; ceiling --------- 29 (1 row) COS function COS is a trigonometric function that returns
sql-reference-071	sql-reference.pdf	71	number up to the next whole number. (The FLOOR function rounds a number down to the next whole number.) Syntax CEIL \| CEILING(number) Arguments number The number or expression that evaluates to a number. It can be the SMALLINT, INTEGER, BIGINT, DECIMAL, FLOAT4, or FLOAT8 type. Return type CEILING and CEIL return the same data type as its argument. Math functions 253 AWS Clean Rooms Example SQL Reference Calculate the ceiling of the commission paid for a given sales transaction: select ceiling(commission) from sales where salesid=10000; ceiling --------- 29 (1 row) COS function COS is a trigonometric function that returns the cosine of a number. The return value is in radians and is between -1 and 1, inclusive. Syntax COS(double_precision) Argument number The input parameter is a double precision number. Return type The COS function returns a double precision number. Examples The following example returns cosine of 0: select cos(0); cos ----- 1 (1 row) The following example returns the cosine of PI: Math functions 254 AWS Clean Rooms select cos(pi()); cos ----- -1 (1 row) COT function SQL Reference COT is a trigonometric function that returns the cotangent of a number. The input parameter must be nonzero. Syntax COT(number) Argument number The input parameter is a DOUBLE PRECISION number. Return type DOUBLE PRECISION Examples To return the cotangent of 1, use the following example. SELECT COT(1); +--------------------+ \| cot \| +--------------------+ \| 0.6420926159343306 \| +--------------------+ DEGREES function Converts an angle in radians to its equivalent in degrees. Math functions 255 SQL Reference AWS Clean Rooms Syntax DEGREES(number) Argument number The input parameter is a DOUBLE PRECISION number. Return type DOUBLE PRECISION Example To return the degree equivalent of .5 radians, use the following example. SELECT DEGREES(.5); +-------------------+ \| degrees \| +-------------------+ \| 28.64788975654116 \| +-------------------+ To convert PI radians to degrees, use the following example. SELECT DEGREES(pi()); +---------+ \| degrees \| +---------+ \| 180 \| +---------+ DIV function The DIV operator returns the integral part of the division of dividend by divisor. Math functions 256 AWS Clean Rooms Syntax dividend div divisor Arguments dividend SQL Reference An expression that evaluates to a numeric or interval. divisor A matching interval type if dividend is an interval, a numeric otherwise. Return type BIGINT Examples The following example selects two columns from the squirrels table: the id column, which contains the unique identiﬁer for each squirrel, and a calculated column, age div 2, which represents the integer division of the age column by 2. The age div 2 calculation performs integer division on the age column, eﬀectively rounding down the age to the nearest even integer. For example, if the age column contains values like 3, 5, 7, and 10, the age div 2 column would contain the values 1, 2, 3, and 5, respectively. SELECT id, age div 2 FROM squirrels This query can be useful in scenarios where you need to group or analyze data based on age ranges, and you want to simplify the age values by rounding them down to the nearest even integer. The resulting output would provide the id and the age divided by 2 for each squirrel in the squirrels table. DLOG1 function The DLOG1 function returns the natural logarithm of the input parameter. The DLOG1 function is a synonym of the LN function. Math functions 257 AWS Clean Rooms DLOG10 function SQL Reference The DLOG10 returns the base 10 logarithm of the input parameter. The DLOG10 function is a synonym of the LOG function. Syntax DLOG10(number) Argument number The input parameter is a double precision number. Return type The DLOG10 function returns a double precision number. Example The following example returns the base 10 logarithm of the number 100: select dlog10(100); dlog10 -------- 2 (1 row) EXP function The EXP function implements the exponential function for a numeric expression, or the base of the natural logarithm, e, raised to the power of expression. The EXP function is the inverse of LN function. Syntax EXP (expression) Math functions 258 AWS Clean Rooms Argument expression SQL Reference The expression must be an INTEGER, DECIMAL, or DOUBLE PRECISION data type. Return type EXP returns a DOUBLE PRECISION number. Example Use the EXP function to forecast ticket sales based on a continuous growth pattern. In this example, the subquery returns the number of tickets sold in 2008. That result is multiplied by the result of the EXP function, which speciﬁes a continuous growth rate of 7% over 10 years. select (select sum(qtysold) from sales, date where sales.dateid=date.dateid and year=2008) * exp((7::float/100)*10) qty2018; qty2018 ------------------ 695447.483772222 (1 row) FLOOR function The FLOOR function rounds a number down to the next whole number. Syntax FLOOR (number) Argument number The number or expression that evaluates to a number. It can be the SMALLINT, INTEGER, BIGINT, DECIMAL, FLOAT4, or FLOAT8 type. Math
sql-reference-072	sql-reference.pdf	72	based on a continuous growth pattern. In this example, the subquery returns the number of tickets sold in 2008. That result is multiplied by the result of the EXP function, which speciﬁes a continuous growth rate of 7% over 10 years. select (select sum(qtysold) from sales, date where sales.dateid=date.dateid and year=2008) * exp((7::float/100)*10) qty2018; qty2018 ------------------ 695447.483772222 (1 row) FLOOR function The FLOOR function rounds a number down to the next whole number. Syntax FLOOR (number) Argument number The number or expression that evaluates to a number. It can be the SMALLINT, INTEGER, BIGINT, DECIMAL, FLOAT4, or FLOAT8 type. Math functions 259 AWS Clean Rooms Return type FLOOR returns the same data type as its argument. Example SQL Reference The example shows the value of the commission paid for a given sales transaction before and after using the FLOOR function. select commission from sales where salesid=10000; floor ------- 28.05 (1 row) select floor(commission) from sales where salesid=10000; floor ------- 28 (1 row) LN function The LN function returns the natural logarithm of the input parameter. The LN function is a synonym of the DLOG1 function. Syntax LN(expression) Argument expression The target column or expression that the function operates on. Math functions 260 AWS Clean Rooms Note SQL Reference This function returns an error for some data types if the expression references an AWS Clean Rooms user-created table or an AWS Clean Rooms STL or STV system table. Expressions with the following data types produce an error if they reference a user-created or system table. • BOOLEAN • CHAR • DATE • DECIMAL or NUMERIC • TIMESTAMP • VARCHAR Expressions with the following data types run successfully on user-created tables and STL or STV system tables: • BIGINT • DOUBLE PRECISION • INTEGER • REAL • SMALLINT Return type The LN function returns the same type as the expression. Example The following example returns the natural logarithm, or base e logarithm, of the number 2.718281828: select ln(2.718281828); ln -------------------- 0.9999999998311267 Math functions 261 AWS Clean Rooms (1 row) Note that the answer is nearly equal to 1. SQL Reference This example returns the natural logarithm of the values in the USERID column in the USERS table: select username, ln(userid) from users order by userid limit 10; username \| ln ----------+------------------- JSG99FHE \| 0 PGL08LJI \| 0.693147180559945 IFT66TXU \| 1.09861228866811 XDZ38RDD \| 1.38629436111989 AEB55QTM \| 1.6094379124341 NDQ15VBM \| 1.79175946922805 OWY35QYB \| 1.94591014905531 AZG78YIP \| 2.07944154167984 MSD36KVR \| 2.19722457733622 WKW41AIW \| 2.30258509299405 (10 rows) LOG function Returns the base 10 logarithm of a number. Synonym of DLOG10 function. Syntax LOG(number) Argument number The input parameter is a double precision number. Return type The LOG function returns a double precision number. Math functions 262 AWS Clean Rooms Example SQL Reference The following example returns the base 10 logarithm of the number 100: select log(100); dlog10 -------- 2 (1 row) MOD function Returns the remainder of two numbers, otherwise known as a modulo operation. To calculate the result, the ﬁrst parameter is divided by the second. Syntax MOD(number1, number2) Arguments number1 The ﬁrst input parameter is an INTEGER, SMALLINT, BIGINT, or DECIMAL number. If either parameter is a DECIMAL type, the other parameter must also be a DECIMAL type. If either parameter is an INTEGER, the other parameter can be an INTEGER, SMALLINT, or BIGINT. Both parameters can also be SMALLINT or BIGINT, but one parameter cannot be a SMALLINT if the other is a BIGINT. number2 The second parameter is an INTEGER, SMALLINT, BIGINT, or DECIMAL number. The same data type rules apply to number2 as to number1. Return type Valid return types are DECIMAL, INT, SMALLINT, and BIGINT. The return type of the MOD function is the same numeric type as the input parameters, if both input parameters are the same type. If either input parameter is an INTEGER, however, the return type will also be an INTEGER. Math functions 263 AWS Clean Rooms Usage notes You can use % as a modulo operator. Examples The following example return the remainder when a number is divided by another: SQL Reference SELECT MOD(10, 4); mod ------ 2 The following example returns a decimal result: SELECT MOD(10.5, 4); mod ------ 2.5 You can cast parameter values: SELECT MOD(CAST(16.4 as integer), 5); mod ------ 1 Check if the ﬁrst parameter is even by dividing it by 2: SELECT mod(5,2) = 0 as is_even; is_even -------- false You can use the % as a modulo operator: SELECT 11 % 4 as remainder; Math functions 264 AWS Clean Rooms remainder ----------- 3 SQL Reference The following example returns information for odd-numbered categories in the CATEGORY table: select catid, catname from category where mod(catid,2)=1 order by 1,2; catid \| catname -------+----------- 1 \| MLB 3 \| NFL 5 \| MLS 7 \| Plays 9 \| Pop 11 \| Classical
sql-reference-073	sql-reference.pdf	73	values: SELECT MOD(CAST(16.4 as integer), 5); mod ------ 1 Check if the ﬁrst parameter is even by dividing it by 2: SELECT mod(5,2) = 0 as is_even; is_even -------- false You can use the % as a modulo operator: SELECT 11 % 4 as remainder; Math functions 264 AWS Clean Rooms remainder ----------- 3 SQL Reference The following example returns information for odd-numbered categories in the CATEGORY table: select catid, catname from category where mod(catid,2)=1 order by 1,2; catid \| catname -------+----------- 1 \| MLB 3 \| NFL 5 \| MLS 7 \| Plays 9 \| Pop 11 \| Classical (6 rows) PI function The PI function returns the value of pi to 14 decimal places. Syntax PI() Return type DOUBLE PRECISION Examples To return the value of pi, use the following example. SELECT PI(); +-------------------+ \| pi \| Math functions 265 AWS Clean Rooms +-------------------+ \| 3.141592653589793 \| +-------------------+ POW function SQL Reference The POW function is an exponential function that raises a numeric expression to the power of a second numeric expression. For example, 2 to the third power is calculated as POWER(2,3), with a result of 8. Syntax The following syntax comes from the Apache Spark SQL Reference. POW(expression1, expression2) Arguments expression1 Numeric expression to be raised. Must be an INTEGER, DECIMAL, or FLOAT data type. expression2 Power to raise expression1. Must be an INTEGER, DECIMAL, or FLOAT data type. Return type DOUBLE PRECISION Example The following example calculates the square of the age value for each row in the squirrels table and returns those squared age values. The resulting squared age values are selected and returned as the output of the query. SELECT pow(age, 2) FROM squirrels The output of the SQL query SELECT pow(age, 2) FROM squirrels would be a single column containing the squared age values for each row in the squirrels table. For example, if the squirrels table had the following data: Math functions 266 AWS Clean Rooms SQL Reference age 3 5 7 10 Then the output of the query would be: pow(age, 2) 9 25 49 100 This query is useful for generating a new column or dataset that contains the squared age values for the squirrels, which could be useful for further analysis or calculations involving the squirrel age data. POWER function The POWER function is an exponential function that raises a numeric expression to the power of a second numeric expression. For example, 2 to the third power is calculated as POWER(2,3), with a result of 8. Syntax {POWER(expression1, expression2) Arguments expression1 Numeric expression to be raised. Must be an INTEGER, DECIMAL, or FLOAT data type. expression2 Power to raise expression1. Must be an INTEGER, DECIMAL, or FLOAT data type. Math functions 267 SQL Reference AWS Clean Rooms Return type DOUBLE PRECISION Example SELECT (SELECT SUM(qtysold) FROM sales, date WHERE sales.dateid=date.dateid AND year=2008) * POW((1+7::FLOAT/100),10) qty2010; +-------------------+ \| qty2010 \| +-------------------+ \| 679353.7540885945 \| +-------------------+ RADIANS function The RADIANS function converts an angle in degrees to its equivalent in radians. Syntax RADIANS(number) Argument number The input parameter is a DOUBLE PRECISION number. Return type DOUBLE PRECISION Example To return the radian equivalent of 180 degrees, use the following example. SELECT RADIANS(180); +-------------------+ \| radians \| Math functions 268 AWS Clean Rooms +-------------------+ \| 3.141592653589793 \| +-------------------+ RAND function SQL Reference The RAND function generates a random ﬂoating-point number between 0 and 1. The RAND function generates a new random number each time it's called. Syntax RAND() Return type RANDOM returns a DOUBLE. Example The following example generates a column of random ﬂoating-point numbers between 0 and 1 for each row in the squirrels table. The resulting output would be a single column containing a list of random decimal values, with one value for each row in the squirrels table. SELECT rand() FROM squirrels This type of query is useful when you need to generate random numbers, for example, to simulate random events or to introduce randomness into your data analysis. In the context of the squirrels table, it might be used to assign random values to each squirrel, which could then be used for further processing or analysis. RANDOM function The RANDOM function generates a random value between 0.0 (inclusive) and 1.0 (exclusive). Syntax RANDOM() Return type RANDOM returns a DOUBLE PRECISION number. Math functions 269 AWS Clean Rooms Examples SQL Reference 1. Compute a random value between 0 and 99. If the random number is 0 to 1, this query produces a random number from 0 to 100: select cast (random() * 100 as int); INTEGER ------ 24 (1 row) 2. Retrieve a uniform random sample of 10 items: select * from sales order by random() limit 10; Now retrieve a random sample of 10 items, but choose the items in proportion to their prices. For example, an
sql-reference-074	sql-reference.pdf	74	(inclusive) and 1.0 (exclusive). Syntax RANDOM() Return type RANDOM returns a DOUBLE PRECISION number. Math functions 269 AWS Clean Rooms Examples SQL Reference 1. Compute a random value between 0 and 99. If the random number is 0 to 1, this query produces a random number from 0 to 100: select cast (random() * 100 as int); INTEGER ------ 24 (1 row) 2. Retrieve a uniform random sample of 10 items: select * from sales order by random() limit 10; Now retrieve a random sample of 10 items, but choose the items in proportion to their prices. For example, an item that is twice the price of another would be twice as likely to appear in the query results: select * from sales order by log(1 - random()) / pricepaid limit 10; 3. This example uses the SET command to set a SEED value so that RANDOM generates a predictable sequence of numbers. First, return three RANDOM integers without setting the SEED value ﬁrst: select cast (random() * 100 as int); INTEGER ------ 6 (1 row) select cast (random() * 100 as int); INTEGER Math functions 270 SQL Reference AWS Clean Rooms ------ 68 (1 row) select cast (random() * 100 as int); INTEGER ------ 56 (1 row) Now, set the SEED value to .25, and return three more RANDOM numbers: set seed to .25; select cast (random() * 100 as int); INTEGER ------ 21 (1 row) select cast (random() * 100 as int); INTEGER ------ 79 (1 row) select cast (random() * 100 as int); INTEGER ------ 12 (1 row) Finally, reset the SEED value to .25, and verify that RANDOM returns the same results as the previous three calls: set seed to .25; select cast (random() * 100 as int); INTEGER ------ 21 (1 row) select cast (random() * 100 as int); Math functions 271 SQL Reference AWS Clean Rooms INTEGER ------ 79 (1 row) select cast (random() * 100 as int); INTEGER ------ 12 (1 row) ROUND function The ROUND function rounds numbers to the nearest integer or decimal. The ROUND function can optionally include a second argument as an integer to indicate the number of decimal places for rounding, in either direction. When you don't provide the second argument, the function rounds to the nearest whole number. When the second argument >n is speciﬁed, the function rounds to the nearest number with n decimal places of precision. Syntax ROUND (number [ , integer ] ) Argument number A number or expression that evaluates to a number. It can be the DECIMAL or FLOAT8 type. AWS Clean Rooms can convert other data types per the implicit conversion rules. integer (optional) An integer that indicates the number of decimal places for rounding in either directions. Return type ROUND returns the same numeric data type as the input argument(s). Examples Round the commission paid for a given transaction to the nearest whole number. Math functions 272 AWS Clean Rooms SQL Reference select commission, round(commission) from sales where salesid=10000; commission \| round -----------+------- 28.05 \| 28 (1 row) Round the commission paid for a given transaction to the ﬁrst decimal place. select commission, round(commission, 1) from sales where salesid=10000; commission \| round -----------+------- 28.05 \| 28.1 (1 row) For the same query, extend the precision in the opposite direction. select commission, round(commission, -1) from sales where salesid=10000; commission \| round -----------+------- 28.05 \| 30 (1 row) SIGN function The SIGN function returns the sign (positive or negative) of a number. The result of the SIGN function is 1, -1, or 0 indicating the sign of the argument. Syntax SIGN (number) Math functions 273 AWS Clean Rooms Argument number SQL Reference Number or expression that evaluates to a number. It can be the DECIMALor FLOAT8 type. AWS Clean Rooms can convert other data types per the implicit conversion rules. Return type SIGN returns the same numeric data type as the input argument(s). If the input is DECIMAL, the output is DECIMAL(1,0). Examples To determine the sign of the commission paid for a given transaction from the SALES table, use the following example. SELECT commission, SIGN(commission) FROM sales WHERE salesid=10000; +------------+------+ \| commission \| sign \| +------------+------+ \| 28.05 \| 1 \| +------------+------+ SIN function SIN is a trigonometric function that returns the sine of a number. The return value is between -1 and 1. Syntax SIN(number) Argument number A DOUBLE PRECISION number in radians. Math functions 274 SQL Reference AWS Clean Rooms Return type DOUBLE PRECISION Example To return the sine of -PI, use the following example. SELECT SIN(-PI()); +-------------------------+ \| sin \| +-------------------------+ \| -0.00000000000000012246 \| +-------------------------+ SQRT function The SQRT function returns the square root of a numeric value. The square root is a number multiplied by itself to get the given value. Syntax SQRT (expression) Argument expression The
sql-reference-075	sql-reference.pdf	75	+------------+------+ SIN function SIN is a trigonometric function that returns the sine of a number. The return value is between -1 and 1. Syntax SIN(number) Argument number A DOUBLE PRECISION number in radians. Math functions 274 SQL Reference AWS Clean Rooms Return type DOUBLE PRECISION Example To return the sine of -PI, use the following example. SELECT SIN(-PI()); +-------------------------+ \| sin \| +-------------------------+ \| -0.00000000000000012246 \| +-------------------------+ SQRT function The SQRT function returns the square root of a numeric value. The square root is a number multiplied by itself to get the given value. Syntax SQRT (expression) Argument expression The expression must have an integer, decimal, or ﬂoating-point data type. The expression can include functions. The system might perform implicit type conversions. Return type SQRT returns a DOUBLE PRECISION number. Examples The following example returns the square root of a number. select sqrt(16); sqrt Math functions 275 AWS Clean Rooms --------------- 4 The following example performs an implicit type conversion. SQL Reference select sqrt('16'); sqrt --------------- 4 The following example nests functions to perform a more complex task. select sqrt(round(16.4)); sqrt --------------- 4 The following example results in the length of the radius when given the area of a circle. It calculates the radius in inches, for instance, when given the area in square inches. The area in the sample is 20. select sqrt(20/pi()); This returns the value 5.046265044040321. The following example returns the square root for COMMISSION values from the SALES table. The COMMISSION column is a DECIMAL column. This example shows how you can use the function in a query with more complex conditional logic. select sqrt(commission) from sales where salesid < 10 order by salesid; sqrt ------------------ 10.4498803820905 3.37638860322683 7.24568837309472 5.1234753829798 ... Math functions 276 AWS Clean Rooms SQL Reference The following query returns the rounded square root for the same set of COMMISSION values. select salesid, commission, round(sqrt(commission)) from sales where salesid < 10 order by salesid; salesid \| commission \| round --------+------------+------- 1 \| 109.20 \| 10 2 \| 11.40 \| 3 3 \| 52.50 \| 7 4 \| 26.25 \| 5 ... For more information about sample data in AWS Clean Rooms, see Sample database. TRUNC function The TRUNC function truncates numbers to the previous integer or decimal. The TRUNC function can optionally include a second argument as an integer to indicate the number of decimal places for rounding, in either direction. When you don't provide the second argument, the function rounds to the nearest whole number. When the second argument >nis speciﬁed, the function rounds to the nearest number with >n decimal places of precision. This function also truncates a timestamp and returns a date. Syntax TRUNC (number [ , integer ] \| timestamp ) Arguments number A number or expression that evaluates to a number. It can be the DECIMAL or FLOAT8 type. AWS Clean Rooms can convert other data types per the implicit conversion rules. integer (optional) An integer that indicates the number of decimal places of precision, in either direction. If no integer is provided, the number is truncated as a whole number; if an integer is speciﬁed, the number is truncated to the speciﬁed decimal place. Math functions 277 AWS Clean Rooms timestamp SQL Reference The function can also return the date from a timestamp. (To return a timestamp value with 00:00:00 as the time, cast the function result to a timestamp.) Return type TRUNC returns the same data type as the ﬁrst input argument. For timestamps, TRUNC returns a date. Examples Truncate the commission paid for a given sales transaction. select commission, trunc(commission) from sales where salesid=784; commission \| trunc -----------+------- 111.15 \| 111 (1 row) Truncate the same commission value to the ﬁrst decimal place. select commission, trunc(commission,1) from sales where salesid=784; commission \| trunc -----------+------- 111.15 \| 111.1 (1 row) Truncate the commission with a negative value for the second argument; 111.15 is rounded down to 110. select commission, trunc(commission,-1) from sales where salesid=784; Math functions 278 AWS Clean Rooms commission \| trunc -----------+------- 111.15 \| 110 (1 row) SQL Reference Return the date portion from the result of the SYSDATE function (which returns a timestamp): select sysdate; timestamp ---------------------------- 2011-07-21 10:32:38.248109 (1 row) select trunc(sysdate); trunc ------------ 2011-07-21 (1 row) Apply the TRUNC function to a TIMESTAMP column. The return type is a date. select trunc(starttime) from event order by eventid limit 1; trunc ------------ 2008-01-25 (1 row) Scalar functions This section describes the scalar functions supported in AWS Clean Rooms Spark SQL. A scalar function is a function that takes one or more values as input and returns a single value as output. Scalar functions operate on individual rows or elements and produce a single result for each input. Scalar functions, such as SIZE, are diﬀerent from other types of SQL functions, such as
sql-reference-076	sql-reference.pdf	76	trunc ------------ 2011-07-21 (1 row) Apply the TRUNC function to a TIMESTAMP column. The return type is a date. select trunc(starttime) from event order by eventid limit 1; trunc ------------ 2008-01-25 (1 row) Scalar functions This section describes the scalar functions supported in AWS Clean Rooms Spark SQL. A scalar function is a function that takes one or more values as input and returns a single value as output. Scalar functions operate on individual rows or elements and produce a single result for each input. Scalar functions, such as SIZE, are diﬀerent from other types of SQL functions, such as aggregate functions (count, sum, avg) and table-generating functions (explode, ﬂatten). These other function types operate on multiple rows or generate multiple rows, whereas scalar functions work on individual rows or elements. Scalar functions 279 AWS Clean Rooms Topics • SIZE function SIZE function SQL Reference The SIZE function takes an existing array, map, or string as an argument and returns a single value representing the size or length of that data structure. It doesn't create a new data structure. It's used for querying and analyzing the properties of existing data structures, rather than for creating new ones. This function is a useful for determining the number of elements in an array or the length of a string. It can be particularly helpful when working with arrays and other data structures in SQL, because it allows you to get information about the size or cardinality of the data. Syntax size(expr) Arguments expr An ARRAY, MAP, or STRING expression. Return type The SIZE function returns an INTEGER. Example In this example, the SIZE function is applied to the array ['b', 'd', 'c', 'a'], and it returns the value 4, which is the number of elements in the array. SELECT size(array('b', 'd', 'c', 'a')); 4 In this example, the SIZE function is applied to the map {'a': 1, 'b': 2}, and it returns the value 2, which is the number of key-value pairs in the map. Scalar functions 280 AWS Clean Rooms SQL Reference SELECT size(map('a', 1, 'b', 2)); 2 In this example, the SIZE function is applied to the string 'hello world', and it returns the value 11, which is the number of characters in the string. SELECT size('hello world'); 11 String functions String functions process and manipulate character strings or expressions that evaluate to character strings. When the string argument in these functions is a literal value, it must be enclosed in single quotation marks. Supported data types include CHAR and VARCHAR. The following section provides the function names, syntax, and descriptions for supported functions. All oﬀsets into strings are one-based. Topics • \|\| (Concatenation) operator • BTRIM function • CHAR_LENGTH function • CHARACTER_LENGTH function • CHARINDEX function • CONCAT function • FORMAT_STRING function • LEFT and RIGHT functions • LEN function • LENGTH function • LOWER function • LPAD and RPAD functions • LTRIM function • POSITION function String functions 281 SQL Reference AWS Clean Rooms • REGEXP_COUNT function • REGEXP_INSTR function • REGEXP_REPLACE function • REGEXP_SUBSTR function • REPEAT function • REPLACE function • REPLICATE function • REVERSE function • RTRIM function • SOUNDEX function • SPLIT function • SPLIT_PART function • STRPOS function • SUBSTR function • SUBSTRING function • TEXTLEN function • TRANSLATE function • TRIM function • UPPER function • UUID function \|\| (Concatenation) operator Concatenates two expressions on either side of the \|\| symbol and returns the concatenated expression. The concatentation operator is similar to CONCAT function. Note For both the CONCAT function and the concatenation operator, if one or both expressions is null, the result of the concatenation is null. String functions 282 AWS Clean Rooms Syntax expression1 \|\| expression2 Arguments expression1, expression2 SQL Reference Both arguments can be ﬁxed-length or variable-length character strings or expressions. Return type The \|\| operator returns a string. The type of string is the same as the input arguments. Example The following example concatenates the FIRSTNAME and LASTNAME ﬁelds from the USERS table: select firstname \|\| ' ' \|\| lastname from users order by 1 limit 10; concat ----------------- Aaron Banks Aaron Booth Aaron Browning Aaron Burnett Aaron Casey Aaron Cash Aaron Castro Aaron Dickerson Aaron Dixon Aaron Dotson (10 rows) To concatenate columns that might contain nulls, use the NVL and COALESCE functions expression. The following example uses NVL to return a 0 whenever NULL is encountered. select venuename \|\| ' seats ' \|\| nvl(venueseats, 0) String functions 283 AWS Clean Rooms SQL Reference from venue where venuestate = 'NV' or venuestate = 'NC' order by 1 limit 10; seating ----------------------------------- Ballys Hotel seats 0 Bank of America Stadium seats 73298 Bellagio Hotel seats 0 Caesars Palace seats 0 Harrahs Hotel seats 0 Hilton Hotel seats 0 Luxor Hotel seats 0 Mandalay Bay Hotel seats 0 Mirage
sql-reference-077	sql-reference.pdf	77	Dotson (10 rows) To concatenate columns that might contain nulls, use the NVL and COALESCE functions expression. The following example uses NVL to return a 0 whenever NULL is encountered. select venuename \|\| ' seats ' \|\| nvl(venueseats, 0) String functions 283 AWS Clean Rooms SQL Reference from venue where venuestate = 'NV' or venuestate = 'NC' order by 1 limit 10; seating ----------------------------------- Ballys Hotel seats 0 Bank of America Stadium seats 73298 Bellagio Hotel seats 0 Caesars Palace seats 0 Harrahs Hotel seats 0 Hilton Hotel seats 0 Luxor Hotel seats 0 Mandalay Bay Hotel seats 0 Mirage Hotel seats 0 New York New York seats 0 BTRIM function The BTRIM function trims a string by removing leading and trailing blanks or by removing leading and trailing characters that match an optional speciﬁed string. Syntax BTRIM(string [, trim_chars ] ) Arguments string The input VARCHAR string to be trimmed. trim_chars The VARCHAR string containing the characters to be matched. Return type The BTRIM function returns a VARCHAR string. Examples The following example trims leading and trailing blanks from the string ' abc ': String functions 284 AWS Clean Rooms SQL Reference select ' abc ' as untrim, btrim(' abc ') as trim; untrim \| trim ----------+------ abc \| abc The following example removes the leading and trailing 'xyz' strings from the string 'xyzaxyzbxyzcxyz'. The leading and trailing occurrences of 'xyz' are removed, but occurrences that are internal within the string are not removed. select 'xyzaxyzbxyzcxyz' as untrim, btrim('xyzaxyzbxyzcxyz', 'xyz') as trim; untrim \| trim -----------------+----------- xyzaxyzbxyzcxyz \| axyzbxyzc The following example removes the leading and trailing parts from the string 'setuphistorycassettes' that match any of the characters in the trim_chars list 'tes'. Any t, e, or s that occur before another character that is not in the trim_chars list at the beginning or ending of the input string are removed. SELECT btrim('setuphistorycassettes', 'tes'); btrim ----------------- uphistoryca CHAR_LENGTH function Synonym of the LEN function. See LEN function. CHARACTER_LENGTH function Synonym of the LEN function. See LEN function. String functions 285 AWS Clean Rooms CHARINDEX function SQL Reference Returns the location of the speciﬁed substring within a string. See POSITION function and STRPOS function for similar functions. Syntax CHARINDEX( substring, string ) Arguments substring The substring to search for within the string. string The string or column to be searched. Return type The CHARINDEX function returns an integer corresponding to the position of the substring (one- based, not zero-based). The position is based on the number of characters, not bytes, so that multi- byte characters are counted as single characters. Usage notes CHARINDEX returns 0 if the substring is not found within the string: select charindex('dog', 'fish'); charindex ---------- 0 (1 row) Examples The following example shows the position of the string fish within the word dogfish: select charindex('fish', 'dogfish'); String functions 286 AWS Clean Rooms charindex ---------- 4 (1 row) SQL Reference The following example returns the number of sales transactions with a COMMISSION over 999.00 from the SALES table: select distinct charindex('.', commission), count (charindex('.', commission)) from sales where charindex('.', commission) > 4 group by charindex('.', commission) order by 1,2; charindex \| count ----------+------- 5 \| 629 (1 row) CONCAT function The CONCAT function concatenates two expressions and returns the resulting expression. To concatenate more than two expressions, use nested CONCAT functions. The concatenation operator (\|\|) between two expressions produces the same results as the CONCAT function. Note For both the CONCAT function and the concatenation operator, if one or both expressions is null, the result of the concatenation is null. Syntax CONCAT ( expression1, expression2 ) Arguments expression1, expression2 Both arguments can be a ﬁxed-length character string, a variable-length character string, a binary expression, or an expression that evaluates to one of these inputs. String functions 287 AWS Clean Rooms Return type SQL Reference CONCAT returns an expression. The data type of the expression is the same type as the input arguments. If the input expressions are of diﬀerent types, AWS Clean Rooms tries to implicitly type casts one of the expressions. If values can't be cast, an error is returned. Examples The following example concatenates two character literals: select concat('December 25, ', '2008'); concat ------------------- December 25, 2008 (1 row) The following query, using the \|\| operator instead of CONCAT, produces the same result: select 'December 25, '\|\|'2008'; concat ------------------- December 25, 2008 (1 row) The following example uses two CONCAT functions to concatenate three character strings: select concat('Thursday, ', concat('December 25, ', '2008')); concat ----------------------------- Thursday, December 25, 2008 (1 row) To concatenate columns that might contain nulls, use the NVL and COALESCE functions. The following example uses NVL to return a 0 whenever NULL is encountered. select concat(venuename, concat(' seats ', nvl(venueseats, 0))) as seating from venue where venuestate = 'NV' or venuestate =
sql-reference-078	sql-reference.pdf	78	25, 2008 (1 row) The following query, using the \|\| operator instead of CONCAT, produces the same result: select 'December 25, '\|\|'2008'; concat ------------------- December 25, 2008 (1 row) The following example uses two CONCAT functions to concatenate three character strings: select concat('Thursday, ', concat('December 25, ', '2008')); concat ----------------------------- Thursday, December 25, 2008 (1 row) To concatenate columns that might contain nulls, use the NVL and COALESCE functions. The following example uses NVL to return a 0 whenever NULL is encountered. select concat(venuename, concat(' seats ', nvl(venueseats, 0))) as seating from venue where venuestate = 'NV' or venuestate = 'NC' String functions 288 SQL Reference AWS Clean Rooms order by 1 limit 5; seating ----------------------------------- Ballys Hotel seats 0 Bank of America Stadium seats 73298 Bellagio Hotel seats 0 Caesars Palace seats 0 Harrahs Hotel seats 0 (5 rows) The following query concatenates CITY and STATE values from the VENUE table: select concat(venuecity, venuestate) from venue where venueseats > 75000 order by venueseats; concat ------------------- DenverCO Kansas CityMO East RutherfordNJ LandoverMD (4 rows) The following query uses nested CONCAT functions. The query concatenates CITY and STATE values from the VENUE table but delimits the resulting string with a comma and a space: select concat(concat(venuecity,', '),venuestate) from venue where venueseats > 75000 order by venueseats; concat --------------------- Denver, CO Kansas City, MO East Rutherford, NJ Landover, MD (4 rows) String functions 289 AWS Clean Rooms FORMAT_STRING function SQL Reference The FORMAT_STRING function creates a formatted string by substituting placeholders in a template string with the provided arguments. It returns a formatted string from printf-style format strings. The FORMAT_STRING function works by replacing the placeholders in the template string with the corresponding values passed as arguments. This type of string formatting can be useful when you need to dynamically construct strings that include a mix of static text and dynamic data, such as when generating output messages, reports, or other types of informative text. The FORMAT_STRING function provides a concise and readable way to create these types of formatted strings, making it easier to maintain and update the code that generates the output. Syntax format_string(strfmt, obj, ...) Arguments strfmt A STRING expression. obj A STRING or numeric expression. Return type FORMAT_STRING returns a STRING. Example The following example contains a template string that contains two placeholders: %d for a decimal (integer) value, and %s for a string value. The %d placeholder is replaced with the decimal (integer) value (100), and the %s placeholder is replaced with the string value ("days"). The output is a template string with the placeholders replaced by the provided arguments: "Hello World 100 days". SELECT format_string("Hello World %d %s", 100, "days"); Hello World 100 days String functions 290 AWS Clean Rooms LEFT and RIGHT functions SQL Reference These functions return the speciﬁed number of leftmost or rightmost characters from a character string. The number is based on the number of characters, not bytes, so that multibyte characters are counted as single characters. Syntax LEFT ( string, integer ) RIGHT ( string, integer ) Arguments string Any character string or any expression that evaluates to a character string. integer A positive integer. Return type LEFT and RIGHT return a VARCHAR string. Example The following example returns the leftmost 5 and rightmost 5 characters from event names that have IDs between 1000 and 1005: select eventid, eventname, left(eventname,5) as left_5, right(eventname,5) as right_5 from event where eventid between 1000 and 1005 order by 1; eventid \| eventname \| left_5 \| right_5 --------+----------------+--------+--------- String functions 291 AWS Clean Rooms SQL Reference 1000 \| Gypsy \| Gypsy \| Gypsy 1001 \| Chicago \| Chica \| icago 1002 \| The King and I \| The K \| and I 1003 \| Pal Joey \| Pal J \| Joey 1004 \| Grease \| Greas \| rease 1005 \| Chicago \| Chica \| icago (6 rows) LEN function Returns the length of the speciﬁed string as the number of characters. Syntax LEN is a synonym of LENGTH function, CHAR_LENGTH function, CHARACTER_LENGTH function, and TEXTLEN function. LEN(expression) Argument expression The input parameter is a CHAR or VARCHAR or an alias of one of the valid input types. Return type The LEN function returns an integer indicating the number of characters in the input string. If the input string is a character string, the LEN function returns the actual number of characters in multi-byte strings, not the number of bytes. For example, a VARCHAR(12) column is required to store three four-byte Chinese characters. The LEN function will return 3 for that same string. Usage notes Length calculations do not count trailing spaces for ﬁxed-length character strings but do count them for variable-length strings. Example The following example returns the number of bytes and the number of characters in the string français. String functions 292 AWS Clean
sql-reference-079	sql-reference.pdf	79	indicating the number of characters in the input string. If the input string is a character string, the LEN function returns the actual number of characters in multi-byte strings, not the number of bytes. For example, a VARCHAR(12) column is required to store three four-byte Chinese characters. The LEN function will return 3 for that same string. Usage notes Length calculations do not count trailing spaces for ﬁxed-length character strings but do count them for variable-length strings. Example The following example returns the number of bytes and the number of characters in the string français. String functions 292 AWS Clean Rooms SQL Reference select octet_length('français'), len('français'); octet_length \| len --------------+----- 9 \| 8 The following example returns the number of characters in the strings cat with no trailing spaces and cat with three trailing spaces: select len('cat'), len('cat '); len \| len -----+----- 3 \| 6 The following example returns the ten longest VENUENAME entries in the VENUE table: select venuename, len(venuename) from venue order by 2 desc, 1 limit 10; venuename \| len ----------------------------------------+----- Saratoga Springs Performing Arts Center \| 39 Lincoln Center for the Performing Arts \| 38 Nassau Veterans Memorial Coliseum \| 33 Jacksonville Municipal Stadium \| 30 Rangers BallPark in Arlington \| 29 University of Phoenix Stadium \| 29 Circle in the Square Theatre \| 28 Hubert H. Humphrey Metrodome \| 28 Oriole Park at Camden Yards \| 27 Dick's Sporting Goods Park \| 26 LENGTH function Synonym of the LEN function. See LEN function. String functions 293 AWS Clean Rooms LOWER function SQL Reference Converts a string to lowercase. LOWER supports UTF-8 multibyte characters, up to a maximum of four bytes per character. Syntax LOWER(string) Argument string The input parameter is a VARCHAR string (or any other data type, such as CHAR, that can be implicitly converted to VARCHAR). Return type The LOWER function returns a character string that is the same data type as the input string. Examples The following example converts the CATNAME ﬁeld to lowercase: select catname, lower(catname) from category order by 1,2; catname \| lower ----------+----------- Classical \| classical Jazz \| jazz MLB \| mlb MLS \| mls Musicals \| musicals NBA \| nba NFL \| nfl NHL \| nhl Opera \| opera Plays \| plays Pop \| pop (11 rows) String functions 294 AWS Clean Rooms LPAD and RPAD functions SQL Reference These functions prepend or append characters to a string, based on a speciﬁed length. Syntax LPAD (string1, length, [ string2 ]) RPAD (string1, length, [ string2 ]) Arguments string1 A character string or an expression that evaluates to a character string, such as the name of a character column. length An integer that deﬁnes the length of the result of the function. The length of a string is based on the number of characters, not bytes, so that multi-byte characters are counted as single characters. If string1 is longer than the speciﬁed length, it is truncated (on the right). If length is a negative number, the result of the function is an empty string. string2 One or more characters that are prepended or appended to string1. This argument is optional; if it is not speciﬁed, spaces are used. Return type These functions return a VARCHAR data type. Examples Truncate a speciﬁed set of event names to 20 characters and prepend the shorter names with spaces: select lpad(eventname,20) from event where eventid between 1 and 5 order by 1; lpad String functions 295 AWS Clean Rooms -------------------- Salome Il Trovatore Boris Godunov Gotterdammerung La Cenerentola (Cind (5 rows) SQL Reference Truncate the same set of event names to 20 characters but append the shorter names with 0123456789. select rpad(eventname,20,'0123456789') from event where eventid between 1 and 5 order by 1; rpad -------------------- Boris Godunov0123456 Gotterdammerung01234 Il Trovatore01234567 La Cenerentola (Cind Salome01234567890123 (5 rows) LTRIM function Trims characters from the beginning of a string. Removes the longest string containing only characters in the trim characters list. Trimming is complete when a trim character doesn't appear in the input string. Syntax LTRIM( string [, trim_chars] ) Arguments string A string column, expression, or string literal to be trimmed. trim_chars A string column, expression, or string literal that represents the characters to be trimmed from the beginning of string. If not speciﬁed, a space is used as the trim character. String functions 296 AWS Clean Rooms Return type SQL Reference The LTRIM function returns a character string that is the same data type as the input string (CHAR or VARCHAR). Examples The following example trims the year from the listime column. The trim characters in string literal '2008-' indicate the characters to be trimmed from the left. If you use the trim characters '028-', you achieve the same result. select listid, listtime, ltrim(listtime, '2008-') from listing order by 1, 2, 3 limit
sql-reference-080	sql-reference.pdf	80	be trimmed from the beginning of string. If not speciﬁed, a space is used as the trim character. String functions 296 AWS Clean Rooms Return type SQL Reference The LTRIM function returns a character string that is the same data type as the input string (CHAR or VARCHAR). Examples The following example trims the year from the listime column. The trim characters in string literal '2008-' indicate the characters to be trimmed from the left. If you use the trim characters '028-', you achieve the same result. select listid, listtime, ltrim(listtime, '2008-') from listing order by 1, 2, 3 limit 10; listid \| listtime \| ltrim -------+---------------------+---------------- 1 \| 2008-01-24 06:43:29 \| 1-24 06:43:29 2 \| 2008-03-05 12:25:29 \| 3-05 12:25:29 3 \| 2008-11-01 07:35:33 \| 11-01 07:35:33 4 \| 2008-05-24 01:18:37 \| 5-24 01:18:37 5 \| 2008-05-17 02:29:11 \| 5-17 02:29:11 6 \| 2008-08-15 02:08:13 \| 15 02:08:13 7 \| 2008-11-15 09:38:15 \| 11-15 09:38:15 8 \| 2008-11-09 05:07:30 \| 11-09 05:07:30 9 \| 2008-09-09 08:03:36 \| 9-09 08:03:36 10 \| 2008-06-17 09:44:54 \| 6-17 09:44:54 LTRIM removes any of the characters in trim_chars when they appear at the beginning of string. The following example trims the characters 'C', 'D', and 'G' when they appear at the beginning of VENUENAME, which is a VARCHAR column. select venueid, venuename, ltrim(venuename, 'CDG') from venue where venuename like '%Park' order by 2 limit 7; venueid \| venuename \| btrim --------+----------------------------+-------------------------- 121 \| ATT Park \| ATT Park String functions 297 AWS Clean Rooms SQL Reference 109 \| Citizens Bank Park \| itizens Bank Park 102 \| Comerica Park \| omerica Park 9 \| Dick's Sporting Goods Park \| ick's Sporting Goods Park 97 \| Fenway Park \| Fenway Park 112 \| Great American Ball Park \| reat American Ball Park 114 \| Miller Park \| Miller Park The following example uses the trim character 2 which is retrieved from the venueid column. select ltrim('2008-01-24 06:43:29', venueid) from venue where venueid=2; ltrim ------------------ 008-01-24 06:43:29 The following example does not trim any characters because a 2 is found before the '0' trim character. select ltrim('2008-01-24 06:43:29', '0'); ltrim ------------------- 2008-01-24 06:43:29 The following example uses the default space trim character and trims the two spaces from the beginning of the string. select ltrim(' 2008-01-24 06:43:29'); ltrim ------------------- 2008-01-24 06:43:29 POSITION function Returns the location of the speciﬁed substring within a string. See CHARINDEX function and STRPOS function for similar functions. String functions 298 SQL Reference AWS Clean Rooms Syntax POSITION(substring IN string ) Arguments substring The substring to search for within the string. string The string or column to be searched. Return type The POSITION function returns an integer corresponding to the position of the substring (one- based, not zero-based). The position is based on the number of characters, not bytes, so that multi- byte characters are counted as single characters. Usage notes POSITION returns 0 if the substring is not found within the string: select position('dog' in 'fish'); position ---------- 0 (1 row) Examples The following example shows the position of the string fish within the word dogfish: select position('fish' in 'dogfish'); position ---------- 4 (1 row) String functions 299 AWS Clean Rooms SQL Reference The following example returns the number of sales transactions with a COMMISSION over 999.00 from the SALES table: select distinct position('.' in commission), count (position('.' in commission)) from sales where position('.' in commission) > 4 group by position('.' in commission) order by 1,2; position \| count ---------+------- 5 \| 629 (1 row) REGEXP_COUNT function Searches a string for a regular expression pattern and returns an integer that indicates the number of times the pattern occurs in the string. If no match is found, then the function returns 0. Syntax REGEXP_COUNT ( source_string, pattern [, position [, parameters ] ] ) Arguments source_string A string expression, such as a column name, to be searched. pattern A string literal that represents a regular expression pattern. position A positive integer that indicates the position within source_string to begin searching. The position is based on the number of characters, not bytes, so that multibyte characters are counted as single characters. The default is 1. If position is less than 1, the search begins at the ﬁrst character of source_string. If position is greater than the number of characters in source_string, the result is 0. parameters One or more string literals that indicate how the function matches the pattern. The possible values are the following: String functions 300 AWS Clean Rooms SQL Reference • c – Perform case-sensitive matching. The default is to use case-sensitive matching. • i – Perform case-insensitive matching. • p – Interpret the pattern with Perl Compatible Regular Expression (PCRE) dialect. Return type Integer Example The following example counts the number of times a three-letter sequence occurs. SELECT
sql-reference-081	sql-reference.pdf	81	search begins at the ﬁrst character of source_string. If position is greater than the number of characters in source_string, the result is 0. parameters One or more string literals that indicate how the function matches the pattern. The possible values are the following: String functions 300 AWS Clean Rooms SQL Reference • c – Perform case-sensitive matching. The default is to use case-sensitive matching. • i – Perform case-insensitive matching. • p – Interpret the pattern with Perl Compatible Regular Expression (PCRE) dialect. Return type Integer Example The following example counts the number of times a three-letter sequence occurs. SELECT regexp_count('abcdefghijklmnopqrstuvwxyz', '[a-z]{3}'); regexp_count -------------- 8 The following example counts the number of times the top-level domain name is either org or edu. SELECT email, regexp_count(email,'@[^.]\\.(org\|edu)')FROM users ORDER BY userid LIMIT 4; email \| regexp_count -----------------------------------------------+-------------- Etiam.laoreet.libero@sodalesMaurisblandit.edu \| 1 Suspendisse.tristique@nonnisiAenean.edu \| 1 amet.faucibus.ut@condimentumegetvolutpat.ca \| 0 sed@lacusUtnec.ca \| 0 The following example counts the occurrences of the string FOX, using case-insensitive matching. SELECT regexp_count('the fox', 'FOX', 1, 'i'); regexp_count -------------- 1 The following example uses a pattern written in the PCRE dialect to locate words containing at least one number and one lowercase letter. It uses the ?= operator, which has a speciﬁc look-ahead String functions 301 AWS Clean Rooms SQL Reference connotation in PCRE. This example counts the number of occurrences of such words, with case- sensitive matching. SELECT regexp_count('passwd7 plain A1234 a1234', '(?=[^ ][a-z])(?=[^ ][0-9])[^ ]+', 1, 'p'); regexp_count -------------- 2 The following example uses a pattern written in the PCRE dialect to locate words containing at least one number and one lowercase letter. It uses the ?= operator, which has a speciﬁc connotation in PCRE. This example counts the number of occurrences of such words, but diﬀers from the previous example in that it uses case-insensitive matching. SELECT regexp_count('passwd7 plain A1234 a1234', '(?=[^ ][a-z])(?=[^ ][0-9])[^ ]+', 1, 'ip'); regexp_count -------------- 3 REGEXP_INSTR function Searches a string for a regular expression pattern and returns an integer that indicates the beginning position or ending position of the matched substring. If no match is found, then the function returns 0. REGEXP_INSTR is similar to the POSITION function, but lets you search a string for a regular expression pattern. Syntax REGEXP_INSTR ( source_string, pattern [, position [, occurrence] [, option [, parameters ] ] ] ] ) Arguments source_string A string expression, such as a column name, to be searched. String functions 302 AWS Clean Rooms pattern A string literal that represents a regular expression pattern. position SQL Reference A positive integer that indicates the position within source_string to begin searching. The position is based on the number of characters, not bytes, so that multibyte characters are counted as single characters. The default is 1. If position is less than 1, the search begins at the ﬁrst character of source_string. If position is greater than the number of characters in source_string, the result is 0. occurrence A positive integer that indicates which occurrence of the pattern to use. REGEXP_INSTR skips the ﬁrst occurrence -1 matches. The default is 1. If occurrence is less than 1 or greater than the number of characters in source_string, the search is ignored and the result is 0. option A value that indicates whether to return the position of the ﬁrst character of the match (0) or the position of the ﬁrst character following the end of the match (1). A nonzero value is the same as 1. The default value is 0. parameters One or more string literals that indicate how the function matches the pattern. The possible values are the following: • c – Perform case-sensitive matching. The default is to use case-sensitive matching. • i – Perform case-insensitive matching. • e – Extract a substring using a subexpression. If pattern includes a subexpression, REGEXP_INSTR matches a substring using the ﬁrst subexpression in pattern. REGEXP_INSTR considers only the ﬁrst subexpression; additional subexpressions are ignored. If the pattern doesn't have a subexpression, REGEXP_INSTR ignores the 'e' parameter. • p – Interpret the pattern with Perl Compatible Regular Expression (PCRE) dialect. Return type Integer String functions 303 AWS Clean Rooms Example SQL Reference The following example searches for the @ character that begins a domain name and returns the starting position of the ﬁrst match. SELECT email, regexp_instr(email, '@[^.]') FROM users ORDER BY userid LIMIT 4; email \| regexp_instr -----------------------------------------------+-------------- Etiam.laoreet.libero@example.com \| 21 Suspendisse.tristique@nonnisiAenean.edu \| 22 amet.faucibus.ut@condimentumegetvolutpat.ca \| 17 sed@lacusUtnec.ca \| 4 The following example searches for variants of the word Center and returns the starting position of the ﬁrst match. SELECT venuename, regexp_instr(venuename,'[cC]ent(er\|re)$') FROM venue WHERE regexp_instr(venuename,'[cC]ent(er\|re)$') > 0 ORDER BY venueid LIMIT 4; venuename \| regexp_instr -----------------------+-------------- The Home Depot Center \| 16 Izod Center \| 6 Wachovia Center \| 10 Air Canada Centre \| 12 The following example ﬁnds the starting position of the ﬁrst
sql-reference-082	sql-reference.pdf	82	the starting position of the ﬁrst match. SELECT email, regexp_instr(email, '@[^.]') FROM users ORDER BY userid LIMIT 4; email \| regexp_instr -----------------------------------------------+-------------- Etiam.laoreet.libero@example.com \| 21 Suspendisse.tristique@nonnisiAenean.edu \| 22 amet.faucibus.ut@condimentumegetvolutpat.ca \| 17 sed@lacusUtnec.ca \| 4 The following example searches for variants of the word Center and returns the starting position of the ﬁrst match. SELECT venuename, regexp_instr(venuename,'[cC]ent(er\|re)$') FROM venue WHERE regexp_instr(venuename,'[cC]ent(er\|re)$') > 0 ORDER BY venueid LIMIT 4; venuename \| regexp_instr -----------------------+-------------- The Home Depot Center \| 16 Izod Center \| 6 Wachovia Center \| 10 Air Canada Centre \| 12 The following example ﬁnds the starting position of the ﬁrst occurrence of the string FOX, using case-insensitive matching logic. SELECT regexp_instr('the fox', 'FOX', 1, 1, 0, 'i'); regexp_instr -------------- 5 String functions 304 AWS Clean Rooms SQL Reference The following example uses a pattern written in PCRE dialect to locate words containing at least one number and one lowercase letter. It uses the ?= operator, which has a speciﬁc look-ahead connotation in PCRE. This example ﬁnds the starting position of the second such word. SELECT regexp_instr('passwd7 plain A1234 a1234', '(?=[^ ][a-z])(?=[^ ][0-9])[^ ]+', 1, 2, 0, 'p'); regexp_instr -------------- 21 The following example uses a pattern written in PCRE dialect to locate words containing at least one number and one lowercase letter. It uses the ?= operator, which has a speciﬁc look-ahead connotation in PCRE. This example ﬁnds the starting position of the second such word, but diﬀers from the previous example in that it uses case-insensitive matching. SELECT regexp_instr('passwd7 plain A1234 a1234', '(?=[^ ][a-z])(?=[^ ][0-9])[^ ]+', 1, 2, 0, 'ip'); regexp_instr -------------- 15 REGEXP_REPLACE function Searches a string for a regular expression pattern and replaces every occurrence of the pattern with the speciﬁed string. REGEXP_REPLACE is similar to the REPLACE function, but lets you search a string for a regular expression pattern. REGEXP_REPLACE is similar to the TRANSLATE function and the REPLACE function, except that TRANSLATE makes multiple single-character substitutions and REPLACE substitutes one entire string with another string, while REGEXP_REPLACE lets you search a string for a regular expression pattern. Syntax REGEXP_REPLACE ( source_string, pattern [, replace_string [ , position [, parameters ] ] ] ) String functions 305 AWS Clean Rooms Arguments source_string A string expression, such as a column name, to be searched. pattern A string literal that represents a regular expression pattern. replace_string SQL Reference A string expression, such as a column name, that will replace each occurrence of pattern. The default is an empty string ( "" ). position A positive integer that indicates the position within source_string to begin searching. The position is based on the number of characters, not bytes, so that multibyte characters are counted as single characters. The default is 1. If position is less than 1, the search begins at the ﬁrst character of source_string. If position is greater than the number of characters in source_string, the result is source_string. parameters One or more string literals that indicate how the function matches the pattern. The possible values are the following: • c – Perform case-sensitive matching. The default is to use case-sensitive matching. • i – Perform case-insensitive matching. • p – Interpret the pattern with Perl Compatible Regular Expression (PCRE) dialect. Return type VARCHAR If either pattern or replace_string is NULL, the return is NULL. Example The following example deletes the @ and domain name from email addresses. SELECT email, regexp_replace(email, '@.\\.(org\|gov\|com\|edu\|ca)$') FROM users ORDER BY userid LIMIT 4; String functions 306 AWS Clean Rooms SQL Reference email \| regexp_replace -----------------------------------------------+---------------- Etiam.laoreet.libero@sodalesMaurisblandit.edu \| Etiam.laoreet.libero Suspendisse.tristique@nonnisiAenean.edu \| Suspendisse.tristique amet.faucibus.ut@condimentumegetvolutpat.ca \| amet.faucibus.ut sed@lacusUtnec.ca \| sed The following example replaces the domain names of email addresses with this value: internal.company.com. SELECT email, regexp_replace(email, '@.\\.[[:alpha:]]{2,3}', '@internal.company.com') FROM users ORDER BY userid LIMIT 4; email \| regexp_replace ----------------------------------------------- +-------------------------------------------- Etiam.laoreet.libero@sodalesMaurisblandit.edu \| Etiam.laoreet.libero@internal.company.com Suspendisse.tristique@nonnisiAenean.edu \| Suspendisse.tristique@internal.company.com amet.faucibus.ut@condimentumegetvolutpat.ca \| amet.faucibus.ut@internal.company.com sed@lacusUtnec.ca \| sed@internal.company.com The following example replaces all occurrences of the string FOX within the value quick brown fox, using case-insensitive matching. SELECT regexp_replace('the fox', 'FOX', 'quick brown fox', 1, 'i'); regexp_replace --------------------- the quick brown fox The following example uses a pattern written in the PCRE dialect to locate words containing at least one number and one lowercase letter. It uses the ?= operator, which has a speciﬁc look- ahead connotation in PCRE. This example replaces each occurrence of such a word with the value [hidden]. SELECT regexp_replace('passwd7 plain A1234 a1234', '(?=[^ ][a-z])(?=[^ ]*[0-9])[^ ]+', '[hidden]', 1, 'p'); String functions 307 AWS Clean Rooms SQL Reference regexp_replace ------------------------------- [hidden] plain A1234 [hidden] The following example uses a pattern written in the PCRE dialect to locate words containing at least one number and one lowercase letter. It uses the ?= operator, which has a speciﬁc look- ahead connotation in PCRE. This example replaces each occurrence of such a word with the value [hidden], but diﬀers
sql-reference-083	sql-reference.pdf	83	which has a speciﬁc look- ahead connotation in PCRE. This example replaces each occurrence of such a word with the value [hidden]. SELECT regexp_replace('passwd7 plain A1234 a1234', '(?=[^ ][a-z])(?=[^ ][0-9])[^ ]+', '[hidden]', 1, 'p'); String functions 307 AWS Clean Rooms SQL Reference regexp_replace ------------------------------- [hidden] plain A1234 [hidden] The following example uses a pattern written in the PCRE dialect to locate words containing at least one number and one lowercase letter. It uses the ?= operator, which has a speciﬁc look- ahead connotation in PCRE. This example replaces each occurrence of such a word with the value [hidden], but diﬀers from the previous example in that it uses case-insensitive matching. SELECT regexp_replace('passwd7 plain A1234 a1234', '(?=[^ ][a-z])(?=[^ ][0-9])[^ ]+', '[hidden]', 1, 'ip'); regexp_replace ---------------------------------- [hidden] plain [hidden] [hidden] REGEXP_SUBSTR function Returns characters from a string by searching it for a regular expression pattern. REGEXP_SUBSTR is similar to the SUBSTRING function function, but lets you search a string for a regular expression pattern. If the function can't match the regular expression to any characters in the string, it returns an empty string. Syntax REGEXP_SUBSTR ( source_string, pattern [, position [, occurrence [, parameters ] ] ] ) Arguments source_string A string expression to be searched. pattern A string literal that represents a regular expression pattern. position A positive integer that indicates the position within source_string to begin searching. The position is based on the number of characters, not bytes, so that multi-byte characters are String functions 308 AWS Clean Rooms SQL Reference counted as single characters. The default is 1. If position is less than 1, the search begins at the ﬁrst character of source_string. If position is greater than the number of characters in source_string, the result is an empty string (""). occurrence A positive integer that indicates which occurrence of the pattern to use. REGEXP_SUBSTR skips the ﬁrst occurrence -1 matches. The default is 1. If occurrence is less than 1 or greater than the number of characters in source_string, the search is ignored and the result is NULL. parameters One or more string literals that indicate how the function matches the pattern. The possible values are the following: • c – Perform case-sensitive matching. The default is to use case-sensitive matching. • i – Perform case-insensitive matching. • e – Extract a substring using a subexpression. If pattern includes a subexpression, REGEXP_SUBSTR matches a substring using the ﬁrst subexpression in pattern. A subexpression is an expression within the pattern that is bracketed with parentheses. For example, for the pattern 'This is a (\\w+)' matches the ﬁrst expression with the string 'This is a ' followed by a word. Instead of returning pattern, REGEXP_SUBSTR with the e parameter returns only the string inside the subexpression. REGEXP_SUBSTR considers only the ﬁrst subexpression; additional subexpressions are ignored. If the pattern doesn't have a subexpression, REGEXP_SUBSTR ignores the 'e' parameter. • p – Interpret the pattern with Perl Compatible Regular Expression (PCRE) dialect. Return type VARCHAR Example The following example returns the portion of an email address between the @ character and the domain extension. SELECT email, regexp_substr(email,'@[^.]') String functions 309 AWS Clean Rooms FROM users ORDER BY userid LIMIT 4; SQL Reference email \| regexp_substr -----------------------------------------------+-------------------------- Etiam.laoreet.libero@sodalesMaurisblandit.edu \| @sodalesMaurisblandit Suspendisse.tristique@nonnisiAenean.edu \| @nonnisiAenean amet.faucibus.ut@condimentumegetvolutpat.ca \| @condimentumegetvolutpat sed@lacusUtnec.ca \| @lacusUtnec The following example returns the portion of the input corresponding to the ﬁrst occurrence of the string FOX, using case-insensitive matching. SELECT regexp_substr('the fox', 'FOX', 1, 1, 'i'); regexp_substr --------------- fox The following example returns the ﬁrst portion of the input that begins with lowercase letters. This is functionally identical to the same SELECT statement without the c parameter. SELECT regexp_substr('THE SECRET CODE IS THE LOWERCASE PART OF 1931abc0EZ.', '[a-z]+', 1, 1, 'c'); regexp_substr --------------- abc The following example uses a pattern written in the PCRE dialect to locate words containing at least one number and one lowercase letter. It uses the ?= operator, which has a speciﬁc look-ahead connotation in PCRE. This example returns the portion of the input corresponding to the second such word. SELECT regexp_substr('passwd7 plain A1234 a1234', '(?=[^ ][a-z])(?=[^ ][0-9])[^ ]+', 1, 2, 'p'); regexp_substr --------------- a1234 String functions 310 AWS Clean Rooms SQL Reference The following example uses a pattern written in the PCRE dialect to locate words containing at least one number and one lowercase letter. It uses the ?= operator, which has a speciﬁc look-ahead connotation in PCRE. This example returns the portion of the input corresponding to the second such word, but diﬀers from the previous example in that it uses case-insensitive matching. SELECT regexp_substr('passwd7 plain A1234 a1234', '(?=[^ ][a-z])(?=[^ ]*[0-9])[^ ]+', 1, 2, 'ip'); regexp_substr --------------- A1234 The following example uses a subexpression to ﬁnd the second string matching the pattern 'this is a (\\w+)' using case-insensitive matching. It returns the subexpression inside the
sql-reference-084	sql-reference.pdf	84	a pattern written in the PCRE dialect to locate words containing at least one number and one lowercase letter. It uses the ?= operator, which has a speciﬁc look-ahead connotation in PCRE. This example returns the portion of the input corresponding to the second such word, but diﬀers from the previous example in that it uses case-insensitive matching. SELECT regexp_substr('passwd7 plain A1234 a1234', '(?=[^ ][a-z])(?=[^ ][0-9])[^ ]+', 1, 2, 'ip'); regexp_substr --------------- A1234 The following example uses a subexpression to ﬁnd the second string matching the pattern 'this is a (\\w+)' using case-insensitive matching. It returns the subexpression inside the parentheses. select regexp_substr( 'This is a cat, this is a dog. This is a mouse.', 'this is a (\\w+)', 1, 2, 'ie'); regexp_substr --------------- dog REPEAT function Repeats a string the speciﬁed number of times. If the input parameter is numeric, REPEAT treats it as a string. Synonym for REPLICATE function. Syntax REPEAT(string, integer) Arguments string The ﬁrst input parameter is the string to be repeated. String functions 311 AWS Clean Rooms integer SQL Reference The second parameter is an integer indicating the number of times to repeat the string. Return type The REPEAT function returns a string. Examples The following example repeats the value of the CATID column in the CATEGORY table three times: select catid, repeat(catid,3) from category order by 1,2; catid \| repeat -------+-------- 1 \| 111 2 \| 222 3 \| 333 4 \| 444 5 \| 555 6 \| 666 7 \| 777 8 \| 888 9 \| 999 10 \| 101010 11 \| 111111 (11 rows) REPLACE function Replaces all occurrences of a set of characters within an existing string with other speciﬁed characters. REPLACE is similar to the TRANSLATE function and the REGEXP_REPLACE function, except that TRANSLATE makes multiple single-character substitutions and REGEXP_REPLACE lets you search a string for a regular expression pattern, while REPLACE substitutes one entire string with another string. String functions 312 SQL Reference AWS Clean Rooms Syntax REPLACE(string1, old_chars, new_chars) Arguments string CHAR or VARCHAR string to be searched search old_chars CHAR or VARCHAR string to replace. new_chars New CHAR or VARCHAR string replacing the old_string. Return type VARCHAR If either old_chars or new_chars is NULL, the return is NULL. Examples The following example converts the string Shows to Theatre in the CATGROUP ﬁeld: select catid, catgroup, replace(catgroup, 'Shows', 'Theatre') from category order by 1,2,3; catid \| catgroup \| replace -------+----------+---------- 1 \| Sports \| Sports 2 \| Sports \| Sports 3 \| Sports \| Sports 4 \| Sports \| Sports 5 \| Sports \| Sports 6 \| Shows \| Theatre 7 \| Shows \| Theatre 8 \| Shows \| Theatre String functions 313 AWS Clean Rooms SQL Reference 9 \| Concerts \| Concerts 10 \| Concerts \| Concerts 11 \| Concerts \| Concerts (11 rows) REPLICATE function Synonym for the REPEAT function. See REPEAT function. REVERSE function The REVERSE function operates on a string and returns the characters in reverse order. For example, reverse('abcde') returns edcba. This function works on numeric and date data types as well as character data types; however, in most cases it has practical value for character strings. Syntax REVERSE ( expression ) Argument expression An expression with a character, date, timestamp, or numeric data type that represents the target of the character reversal. All expressions are implicitly converted to variable-length character strings. Trailing blanks in ﬁxed-width character strings are ignored. Return type REVERSE returns a VARCHAR. Examples Select ﬁve distinct city names and their corresponding reversed names from the USERS table: select distinct city as cityname, reverse(cityname) from users order by city limit 5; cityname \| reverse String functions 314 AWS Clean Rooms ---------+---------- Aberdeen \| needrebA Abilene \| enelibA Ada \| adA Agat \| tagA Agawam \| mawagA (5 rows) SQL Reference Select ﬁve sales IDs and their corresponding reversed IDs cast as character strings: select salesid, reverse(salesid)::varchar from sales order by salesid desc limit 5; salesid \| reverse --------+--------- 172456 \| 654271 172455 \| 554271 172454 \| 454271 172453 \| 354271 172452 \| 254271 (5 rows) RTRIM function The RTRIM function trims a speciﬁed set of characters from the end of a string. Removes the longest string containing only characters in the trim characters list. Trimming is complete when a trim character doesn't appear in the input string. Syntax RTRIM( string, trim_chars ) Arguments string A string column, expression, or string literal to be trimmed. trim_chars A string column, expression, or string literal that represents the characters to be trimmed from the end of string. If not speciﬁed, a space is used as the trim character. String functions 315 AWS Clean Rooms Return type SQL Reference A string that is the same data type as the string argument. Example The following example trims leading and trailing blanks from the string '
sql-reference-085	sql-reference.pdf	85	the trim characters list. Trimming is complete when a trim character doesn't appear in the input string. Syntax RTRIM( string, trim_chars ) Arguments string A string column, expression, or string literal to be trimmed. trim_chars A string column, expression, or string literal that represents the characters to be trimmed from the end of string. If not speciﬁed, a space is used as the trim character. String functions 315 AWS Clean Rooms Return type SQL Reference A string that is the same data type as the string argument. Example The following example trims leading and trailing blanks from the string ' abc ': select ' abc ' as untrim, rtrim(' abc ') as trim; untrim \| trim ----------+------ abc \| abc The following example removes the trailing 'xyz' strings from the string 'xyzaxyzbxyzcxyz'. The trailing occurrences of 'xyz' are removed, but occurrences that are internal within the string are not removed. select 'xyzaxyzbxyzcxyz' as untrim, rtrim('xyzaxyzbxyzcxyz', 'xyz') as trim; untrim \| trim -----------------+----------- xyzaxyzbxyzcxyz \| xyzaxyzbxyzc The following example removes the trailing parts from the string 'setuphistorycassettes' that match any of the characters in the trim_chars list 'tes'. Any t, e, or s that occur before another character that is not in the trim_chars list at the ending of the input string are removed. SELECT rtrim('setuphistorycassettes', 'tes'); rtrim ----------------- setuphistoryca The following example trims the characters 'Park' from the end of VENUENAME where present: select venueid, venuename, rtrim(venuename, 'Park') from venue order by 1, 2, 3 String functions 316 AWS Clean Rooms limit 10; SQL Reference venueid \| venuename \| rtrim --------+----------------------------+------------------------- 1 \| Toyota Park \| Toyota 2 \| Columbus Crew Stadium \| Columbus Crew Stadium 3 \| RFK Stadium \| RFK Stadium 4 \| CommunityAmerica Ballpark \| CommunityAmerica Ballp 5 \| Gillette Stadium \| Gillette Stadium 6 \| New York Giants Stadium \| New York Giants Stadium 7 \| BMO Field \| BMO Field 8 \| The Home Depot Center \| The Home Depot Cente 9 \| Dick's Sporting Goods Park \| Dick's Sporting Goods 10 \| Pizza Hut Park \| Pizza Hut Note that RTRIM removes any of the characters P, a, r, or k when they appear at the end of a VENUENAME. SOUNDEX function The SOUNDEX function returns the American Soundex value consisting of the ﬁrst letter followed by a 3–digit encoding of the sounds that represent the English pronunciation of the string that you specify. Syntax SOUNDEX(string) Arguments string You specify a CHAR or VARCHAR string that you want to convert to an American Soundex code value. Return type The SOUNDEX function returns a VARCHAR(4) string consisting of a capital letter followed by a three–digit encoding of the sounds that represent the English pronunciation. String functions 317 AWS Clean Rooms Usage notes SQL Reference The SOUNDEX function converts only English alphabetical lowercase and uppercase ASCII characters, including a–z and A–Z. SOUNDEX ignores other characters. SOUNDEX returns a single Soundex value for a string of multiple words separated by spaces. select soundex('AWS Amazon'); soundex --------- A252 SOUNDEX returns an empty string if the input string doesn't contain any English letters. select soundex('+-*/%'); soundex --------- Example The following example returns the Soundex A525 for the word Amazon. select soundex('Amazon'); soundex --------- A525 SPLIT function The SPLIT function allows you to extract substrings from a larger string and work with them as an array. The SPLIT function is useful when you need to break down a string into individual components based on a speciﬁc delimiter or pattern. Syntax split(str, regex, limit) String functions 318 AWS Clean Rooms Arguments str A string expression to split. regex SQL Reference A string representing a regular expression. The regex string should be a Java regular expression. limit An integer expression which controls the number of times the regex is applied. • limit > 0: The resulting array's length will not be more than limit, and the resulting array's last entry will contain all input beyond the last matched regex. • limit <= 0: regex will be applied as many times as possible, and the resulting array can be of any size. Return type The SPLIT function returns an ARRAY<STRING>. If limit > 0: The resulting array’s length will not be more than limit, and the resulting array’s last entry will contain all input beyond the last matched regex. If limit <= 0: regex will be applied as many times as possible, and the resulting array can be of any size. Example In this example, the SPLIT function splits the input string 'oneAtwoBthreeC' wherever it encounters the characters 'A', 'B', or 'C' (as speciﬁed by the regular expression pattern '[ABC]'). The resulting output is an array of four elements: "one", "two", "three", and an empty string "". SELECT split('oneAtwoBthreeC', '[ABC]'); ["one","two","three",""] SPLIT_PART function Splits a string on the speciﬁed delimiter and returns the
sql-reference-086	sql-reference.pdf	86	limit, and the resulting array’s last entry will contain all input beyond the last matched regex. If limit <= 0: regex will be applied as many times as possible, and the resulting array can be of any size. Example In this example, the SPLIT function splits the input string 'oneAtwoBthreeC' wherever it encounters the characters 'A', 'B', or 'C' (as speciﬁed by the regular expression pattern '[ABC]'). The resulting output is an array of four elements: "one", "two", "three", and an empty string "". SELECT split('oneAtwoBthreeC', '[ABC]'); ["one","two","three",""] SPLIT_PART function Splits a string on the speciﬁed delimiter and returns the part at the speciﬁed position. String functions 319 AWS Clean Rooms Syntax SPLIT_PART(string, delimiter, position) Arguments string SQL Reference A string column, expression, or string literal to be split. The string can be CHAR or VARCHAR. delimiter The delimiter string indicating sections of the input string. If delimiter is a literal, enclose it in single quotation marks. position Position of the portion of string to return (counting from 1). Must be an integer greater than 0. If position is larger than the number of string portions, SPLIT_PART returns an empty string. If delimiter is not found in string, then the returned value contains the contents of the speciﬁed part, which might be the entire string or an empty value. Return type A CHAR or VARCHAR string, the same as the string parameter. Examples The following example splits a string literal into parts using the $ delimiter and returns the second part. select split_part('abc$def$ghi','$',2) split_part ---------- def The following example splits a string literal into parts using the $ delimiter. It returns an empty string because part 4 is not found. select split_part('abc$def$ghi','$',4) String functions 320 AWS Clean Rooms split_part ---------- SQL Reference The following example splits a string literal into parts using the # delimiter. It returns the entire string, which is the ﬁrst part, because the delimiter is not found. select split_part('abc$def$ghi','#',1) split_part ------------ abc$def$ghi The following example splits the timestamp ﬁeld LISTTIME into year, month, and day components. select listtime, split_part(listtime,'-',1) as year, split_part(listtime,'-',2) as month, split_part(split_part(listtime,'-',3),' ',1) as day from listing limit 5; listtime \| year \| month \| day ---------------------+------+-------+------ 2008-03-05 12:25:29 \| 2008 \| 03 \| 05 2008-09-09 08:03:36 \| 2008 \| 09 \| 09 2008-09-26 05:43:12 \| 2008 \| 09 \| 26 2008-10-04 02:00:30 \| 2008 \| 10 \| 04 2008-01-06 08:33:11 \| 2008 \| 01 \| 06 The following example selects the LISTTIME timestamp ﬁeld and splits it on the '-' character to get the month (the second part of the LISTTIME string), then counts the number of entries for each month: select split_part(listtime,'-',2) as month, count(*) from listing group by split_part(listtime,'-',2) order by 1, 2; month \| count -------+------- 01 \| 18543 String functions 321 SQL Reference AWS Clean Rooms 02 \| 16620 03 \| 17594 04 \| 16822 05 \| 17618 06 \| 17158 07 \| 17626 08 \| 17881 09 \| 17378 10 \| 17756 11 \| 12912 12 \| 4589 STRPOS function Returns the position of a substring within a speciﬁed string. See CHARINDEX function and POSITION function for similar functions. Syntax STRPOS(string, substring ) Arguments string The ﬁrst input parameter is the string to be searched. substring The second parameter is the substring to search for within the string. Return type The STRPOS function returns an integer corresponding to the position of the substring (one-based, not zero-based). The position is based on the number of characters, not bytes, so that multi-byte characters are counted as single characters. Usage notes STRPOS returns 0 if the substring is not found within the string: String functions 322 AWS Clean Rooms SQL Reference select strpos('dogfish', 'fist'); strpos -------- 0 (1 row) Examples The following example shows the position of the string fish within the word dogfish: select strpos('dogfish', 'fish'); strpos -------- 4 (1 row) The following example returns the number of sales transactions with a COMMISSION over 999.00 from the SALES table: select distinct strpos(commission, '.'), count (strpos(commission, '.')) from sales where strpos(commission, '.') > 4 group by strpos(commission, '.') order by 1, 2; strpos \| count -------+------- 5 \| 629 (1 row) SUBSTR function Synonym of the SUBSTRING function. See SUBSTRING function. SUBSTRING function Returns the subset of a string based on the speciﬁed start position. String functions 323 AWS Clean Rooms SQL Reference If the input is a character string, the start position and number of characters extracted are based on characters, not bytes, so that multi-byte characters are counted as single characters. If the input is a binary expression, the start position and extracted substring are based on bytes. You can't specify a negative length, but you can specify a negative starting position. Syntax SUBSTRING(charactestring FROM start_position [ FOR numbecharacters ] ) SUBSTRING(charactestring, start_position, numbecharacters ) SUBSTRING(binary_expression, start_byte, numbebytes
sql-reference-087	sql-reference.pdf	87	function. SUBSTRING function Returns the subset of a string based on the speciﬁed start position. String functions 323 AWS Clean Rooms SQL Reference If the input is a character string, the start position and number of characters extracted are based on characters, not bytes, so that multi-byte characters are counted as single characters. If the input is a binary expression, the start position and extracted substring are based on bytes. You can't specify a negative length, but you can specify a negative starting position. Syntax SUBSTRING(charactestring FROM start_position [ FOR numbecharacters ] ) SUBSTRING(charactestring, start_position, numbecharacters ) SUBSTRING(binary_expression, start_byte, numbebytes ) SUBSTRING(binary_expression, start_byte ) Arguments charactestring The string to be searched. Non-character data types are treated like a string. start_position The position within the string to begin the extraction, starting at 1. The start_position is based on the number of characters, not bytes, so that multi-byte characters are counted as single characters. This number can be negative. numbecharacters The number of characters to extract (the length of the substring). The numbecharacters is based on the number of characters, not bytes, so that multi-byte characters are counted as single characters. This number cannot be negative. start_byte The position within the binary expression to begin the extraction, starting at 1. This number can be negative. numbebytes The number of bytes to extract, that is, the length of the substring. This number can't be negative. String functions 324 AWS Clean Rooms Return type VARCHAR Usage notes for character strings SQL Reference The following example returns a four-character string beginning with the sixth character. select substring('caterpillar',6,4); substring ----------- pill (1 row) If the start_position + numbecharacters exceeds the length of the string, SUBSTRING returns a substring starting from the start_position until the end of the string. For example: select substring('caterpillar',6,8); substring ----------- pillar (1 row) If the start_position is negative or 0, the SUBSTRING function returns a substring beginning at the ﬁrst character of string with a length of start_position + numbecharacters -1. For example: select substring('caterpillar',-2,6); substring ----------- cat (1 row) If start_position + numbecharacters -1 is less than or equal to zero, SUBSTRING returns an empty string. For example: select substring('caterpillar',-5,4); substring ----------- String functions 325 AWS Clean Rooms (1 row) Examples SQL Reference The following example returns the month from the LISTTIME string in the LISTING table: select listid, listtime, substring(listtime, 6, 2) as month from listing order by 1, 2, 3 limit 10; listid \| listtime \| month --------+---------------------+------- 1 \| 2008-01-24 06:43:29 \| 01 2 \| 2008-03-05 12:25:29 \| 03 3 \| 2008-11-01 07:35:33 \| 11 4 \| 2008-05-24 01:18:37 \| 05 5 \| 2008-05-17 02:29:11 \| 05 6 \| 2008-08-15 02:08:13 \| 08 7 \| 2008-11-15 09:38:15 \| 11 8 \| 2008-11-09 05:07:30 \| 11 9 \| 2008-09-09 08:03:36 \| 09 10 \| 2008-06-17 09:44:54 \| 06 (10 rows) The following example is the same as above, but uses the FROM...FOR option: select listid, listtime, substring(listtime from 6 for 2) as month from listing order by 1, 2, 3 limit 10; listid \| listtime \| month --------+---------------------+------- 1 \| 2008-01-24 06:43:29 \| 01 2 \| 2008-03-05 12:25:29 \| 03 3 \| 2008-11-01 07:35:33 \| 11 4 \| 2008-05-24 01:18:37 \| 05 5 \| 2008-05-17 02:29:11 \| 05 6 \| 2008-08-15 02:08:13 \| 08 7 \| 2008-11-15 09:38:15 \| 11 String functions 326 AWS Clean Rooms SQL Reference 8 \| 2008-11-09 05:07:30 \| 11 9 \| 2008-09-09 08:03:36 \| 09 10 \| 2008-06-17 09:44:54 \| 06 (10 rows) You can't use SUBSTRING to predictably extract the preﬁx of a string that might contain multi-byte characters because you need to specify the length of a multi-byte string based on the number of bytes, not the number of characters. To extract the beginning segment of a string based on the length in bytes, you can CAST the string as VARCHAR(byte_length) to truncate the string, where byte_length is the required length. The following example extracts the ﬁrst 5 bytes from the string 'Fourscore and seven'. select cast('Fourscore and seven' as varchar(5)); varchar ------- Fours The following example returns the ﬁrst name Ana which appears after the last space in the input string Silva, Ana. select reverse(substring(reverse('Silva, Ana'), 1, position(' ' IN reverse('Silva, Ana')))) reverse ----------- Ana TEXTLEN function Synonym of LEN function. See LEN function. TRANSLATE function For a given expression, replaces all occurrences of speciﬁed characters with speciﬁed substitutes. Existing characters are mapped to replacement characters by their positions in the characters_to_replace and characters_to_substitute arguments. If more characters are speciﬁed String functions 327 AWS Clean Rooms SQL Reference in the characters_to_replace argument than in the characters_to_substitute argument, the extra characters from the characters_to_replace argument are omitted in the return value. TRANSLATE is similar to the REPLACE function and the REGEXP_REPLACE function, except that REPLACE substitutes one entire string with another string
sql-reference-088	sql-reference.pdf	88	Ana')))) reverse ----------- Ana TEXTLEN function Synonym of LEN function. See LEN function. TRANSLATE function For a given expression, replaces all occurrences of speciﬁed characters with speciﬁed substitutes. Existing characters are mapped to replacement characters by their positions in the characters_to_replace and characters_to_substitute arguments. If more characters are speciﬁed String functions 327 AWS Clean Rooms SQL Reference in the characters_to_replace argument than in the characters_to_substitute argument, the extra characters from the characters_to_replace argument are omitted in the return value. TRANSLATE is similar to the REPLACE function and the REGEXP_REPLACE function, except that REPLACE substitutes one entire string with another string and REGEXP_REPLACE lets you search a string for a regular expression pattern, while TRANSLATE makes multiple single-character substitutions. If any argument is null, the return is NULL. Syntax TRANSLATE ( expression, characters_to_replace, characters_to_substitute ) Arguments expression The expression to be translated. characters_to_replace A string containing the characters to be replaced. characters_to_substitute A string containing the characters to substitute. Return type VARCHAR Examples The following example replaces several characters in a string: select translate('mint tea', 'inea', 'osin'); translate ----------- most tin String functions 328 AWS Clean Rooms SQL Reference The following example replaces the at sign (@) with a period for all values in a column: select email, translate(email, '@', '.') as obfuscated_email from users limit 10; email obfuscated_email ------------------------------------------------------------------------------------------- Etiam.laoreet.libero@sodalesMaurisblandit.edu Etiam.laoreet.libero.sodalesMaurisblandit.edu amet.faucibus.ut@condimentumegetvolutpat.ca amet.faucibus.ut.condimentumegetvolutpat.ca turpis@accumsanlaoreet.org turpis.accumsanlaoreet.org ullamcorper.nisl@Cras.edu ullamcorper.nisl.Cras.edu arcu.Curabitur@senectusetnetus.com arcu.Curabitur.senectusetnetus.com ac@velit.ca ac.velit.ca Aliquam.vulputate.ullamcorper@amalesuada.org Aliquam.vulputate.ullamcorper.amalesuada.org vel.est@velitegestas.edu vel.est.velitegestas.edu dolor.nonummy@ipsumdolorsit.ca dolor.nonummy.ipsumdolorsit.ca et@Nunclaoreet.ca et.Nunclaoreet.ca The following example replaces spaces with underscores and strips out periods for all values in a column: select city, translate(city, ' .', '_') from users where city like 'Sain%' or city like 'St%' group by city order by city; city translate --------------+------------------ Saint Albans Saint_Albans Saint Cloud Saint_Cloud Saint Joseph Saint_Joseph Saint Louis Saint_Louis Saint Paul Saint_Paul St. George St_George St. Marys St_Marys St. Petersburg St_Petersburg Stafford Stafford Stamford Stamford Stanton Stanton Starkville Starkville String functions 329 AWS Clean Rooms SQL Reference Statesboro Statesboro Staunton Staunton Steubenville Steubenville Stevens Point Stevens_Point Stillwater Stillwater Stockton Stockton Sturgis Sturgis TRIM function Trims a string by removing leading and trailing blanks or by removing leading and trailing characters that match an optional speciﬁed string. Syntax TRIM( [ BOTH ] [ trim_chars FROM ] string Arguments trim_chars (Optional) The characters to be trimmed from the string. If this parameter is omitted, blanks are trimmed. string The string to be trimmed. Return type The TRIM function returns a VARCHAR or CHAR string. If you use the TRIM function with a SQL command, AWS Clean Rooms implicitly converts the results to VARCHAR. If you use the TRIM function in the SELECT list for a SQL function, AWS Clean Rooms does not implicitly convert the results, and you might need to perform an explicit conversion to avoid a data type mismatch error. See the CAST function and CONVERT function functions for information about explicit conversions. Example The following example trims leading and trailing blanks from the string ' abc ': select ' abc ' as untrim, trim(' abc ') as trim; String functions 330 AWS Clean Rooms untrim \| trim ----------+------ abc \| abc SQL Reference The following example removes the double quotation marks that surround the string "dog": select trim('"' FROM '"dog"'); btrim ------- dog TRIM removes any of the characters in trim_chars when they appear at the beginning of string. The following example trims the characters 'C', 'D', and 'G' when they appear at the beginning of VENUENAME, which is a VARCHAR column. select venueid, venuename, trim(venuename, 'CDG') from venue where venuename like '%Park' order by 2 limit 7; venueid \| venuename \| btrim --------+----------------------------+-------------------------- 121 \| ATT Park \| ATT Park 109 \| Citizens Bank Park \| itizens Bank Park 102 \| Comerica Park \| omerica Park 9 \| Dick's Sporting Goods Park \| ick's Sporting Goods Park 97 \| Fenway Park \| Fenway Park 112 \| Great American Ball Park \| reat American Ball Park 114 \| Miller Park \| Miller Park UPPER function Converts a string to uppercase. UPPER supports UTF-8 multibyte characters, up to a maximum of four bytes per character. Syntax UPPER(string) String functions 331 AWS Clean Rooms Arguments string SQL Reference The input parameter is a VARCHAR string (or any other data type, such as CHAR, that can be implicitly converted to VARCHAR). Return type The UPPER function returns a character string that is the same data type as the input string. Examples The following example converts the CATNAME ﬁeld to uppercase: select catname, upper(catname) from category order by 1,2; catname \| upper ----------+----------- Classical \| CLASSICAL Jazz \| JAZZ MLB \| MLB MLS \| MLS Musicals \| MUSICALS NBA \| NBA NFL \| NFL NHL \| NHL Opera \| OPERA Plays \| PLAYS Pop \| POP (11 rows) UUID function The UUID function generates a Universally Unique Identiﬁer (UUID).
sql-reference-089	sql-reference.pdf	89	other data type, such as CHAR, that can be implicitly converted to VARCHAR). Return type The UPPER function returns a character string that is the same data type as the input string. Examples The following example converts the CATNAME ﬁeld to uppercase: select catname, upper(catname) from category order by 1,2; catname \| upper ----------+----------- Classical \| CLASSICAL Jazz \| JAZZ MLB \| MLB MLS \| MLS Musicals \| MUSICALS NBA \| NBA NFL \| NFL NHL \| NHL Opera \| OPERA Plays \| PLAYS Pop \| POP (11 rows) UUID function The UUID function generates a Universally Unique Identiﬁer (UUID). UUIDs are globally unique identiﬁers that are commonly used to provide unique identiﬁers for various purposes, such as: • Identifying database records or other data entities. • Generating unique names or keys for ﬁles, directories, or other resources. String functions 332 AWS Clean Rooms SQL Reference • Tracking and correlating data across distributed systems. • Providing unique identiﬁers for network packets, software components, or other digital assets. The UUID function generates a UUID value that is unique with a very high probability, even across distributed systems and over long periods of time. UUIDs are typically generated using a combination of the current timestamp, the computer's network address, and other random or pseudo-random data, ensuring that each generated UUID is highly unlikely to conﬂict with any other UUID. In the context of a SQL query, the UUID function can be used to generate unique identiﬁers for new records being inserted into a database, or to provide unique keys for data partitioning, indexing, or other purposes where a unique identiﬁer is required. Note The UUID function is non-deterministic. Syntax uuid() Arguments The UUID function takes no argument. Return type UUID returns a universally unique identiﬁer (UUID) string. The value is returned as a canonical UUID 36-character string. Example The following example generates a Universally Unique Identiﬁer (UUID). The output is a 36- character string representing a Universally Unique Identiﬁer. SELECT uuid(); 46707d92-02f4-4817-8116-a4c3b23e6266 String functions 333 AWS Clean Rooms SQL Reference Privacy-related functions AWS Clean Rooms provides functions to help you comply with privacy-related compliance for the following speciﬁcations. • Global Privacy Platform (GPP) – A speciﬁcation from the Interactive Advertising Bureau (IAB) that establishes a global, standardized framework for online privacy and use of data. For more information about the technical speciﬁcations of the GPP, see the Global Privacy Platform documentation on GitHub. • Transparency and Consent Framework (TCF) – A key component of the GPP, launched in 2020, which provides a standardized technical framework to help companies comply with privacy regulations such as the EU General Data Protection Regulation (GDPR). The TCF enables customers to grant or withhold consent to data collection and processing. For more information about the technical speciﬁcations of TCF, see the TCF documentation on GitHub. Topics • consent_gpp_v1_decode function • consent_tcf_v2_decode function consent_gpp_v1_decode function The consent_gpp_v1_decode function is used to decode Global Privacy Platform (GPP) v1 consent data. It takes the encoded consent string as input and returns the decoded consent data, which includes information about the user's privacy preferences and consent choices. This function is useful when working with data that includes GPP v1 consent information, as it allows you to access and analyze the consent data in a structured format. Syntax consent_gpp_v1_decode(gpp_string) Arguments gpp_string The encoded GPP v1 consent string. Privacy-related functions 334 AWS Clean Rooms Returns SQL Reference The returned dictionary includes the following key-value pairs: • version: The version of the GPP speciﬁcation used (currently 1). • cmpId: The ID of the Consent Management Platform (CMP) that encoded the consent string. • cmpVersion: The version of the CMP that encoded the consent string. • consentScreen: The ID of the screen in the CMP UI where the user provided consent. • consentLanguage: The language code of the consent information. • vendorListVersion: The version of the vendor list used. • publisherCountryCode: The country code of the publisher. • purposeConsent: A list of integers representing the purposes for which the user has consented to. • purposeLegitimateInterest: A list of purpose IDs for which the user's legitimate interest has been transparently communicated. • specialFeatureOptIns: A list of integers representing the special features that the user has opted into. • vendorConsent: A list of vendor IDs that the user has consented to. • vendorLegitimateInterest: A list of vendor IDs for which the user's legitimate interest has been transparently communicated. Example The following example takes a single argument, which is the encoded consent string. It returns a dictionary containing the decoded consent data, including information about the user's privacy preferences, consent choices, and other metadata. SELECT * FROM consent_gpp_v1_decode('ABCDEFGHIJK'); The basic structure of the returned consent data includes information about the consent string version, the CMP (Consent Management Platform) details, the user's consent and legitimate interest choices for diﬀerent
sql-reference-090	sql-reference.pdf	90	into. • vendorConsent: A list of vendor IDs that the user has consented to. • vendorLegitimateInterest: A list of vendor IDs for which the user's legitimate interest has been transparently communicated. Example The following example takes a single argument, which is the encoded consent string. It returns a dictionary containing the decoded consent data, including information about the user's privacy preferences, consent choices, and other metadata. SELECT * FROM consent_gpp_v1_decode('ABCDEFGHIJK'); The basic structure of the returned consent data includes information about the consent string version, the CMP (Consent Management Platform) details, the user's consent and legitimate interest choices for diﬀerent purposes and vendors, and other metadata. { Privacy-related functions 335 AWS Clean Rooms SQL Reference "version": 1, "cmpId": 12, "cmpVersion": 34, "consentScreen": 5, "consentLanguage": "en", "vendorListVersion": 89, "publisherCountryCode": "US", "purposeConsent": [1], "purposeLegitimateInterests": [1], "specialFeatureOptins": [1], "vendorConsent": [1], "vendorLegitimateInterests": [1]} } consent_tcf_v2_decode function The consent_tcf_v2_decode function is used to decode Transparency and Consent Framework (TCF) v2 consent data. It takes the encoded consent string as input and returns the decoded consent data, which includes information about the user's privacy preferences and consent choices. This function is useful when working with data that includes TCF v2 consent information, as it allows you to access and analyze the consent data in a structured format. Syntax consent_tcf_v2_decode(tcf_string) Arguments tcf_string The encoded TCF v2 consent string. Returns The consent_tcf_v2_decode function returns a dictionary containing the decoded consent data from a Transparency and Consent Framework (TCF) v2 consent string. The returned dictionary includes the following key-value pairs: Core segment Privacy-related functions 336 AWS Clean Rooms SQL Reference • version: The version of the TCF speciﬁcation used (currently 2). • created: The date and time when the consent string was created. • lastUpdated: The date and time when the consent string was last updated. • cmpId: The ID of the Consent Management Platform (CMP) that encoded the consent string. • cmpVersion: The version of the CMP that encoded the consent string. • consentScreen: The ID of the screen in the CMP UI where the user provided consent. • consentLanguage: The language code of the consent information. • vendorListVersion: The version of the vendor list used. • tcfPolicyVersion: The version of the TCF policy that the consent string is based on. • isServiceSpecific: A Boolean value indicating whether the consent is speciﬁc to a particular service or applies to all services. • useNonStandardStacks: A Boolean value indicating whether non-standard stacks are used. • specialFeatureOptIns: A list of integers representing the special features that the user has opted into. • purposeConsent: A list of integers representing the purposes for which the user has consented to. • purposesLITransparency: A list of integers representing the purposes for which the user has given legitimate interest transparency. • purposeOneTreatment: A Boolean value indicating whether the user has requested the "purpose one treatment" (that is, all purposes are treated equally). • publisherCountryCode: The country code of the publisher. • vendorConsent: A list of vendor IDs that the user has consented to. • vendorLegitimateInterest: A list of vendor IDs for which the user's legitimate interest has been transparently communicated. • pubRestrictionEntry: A list of publisher restrictions. This ﬁeld contains the Purpose ID, Restriction Type, and List of Vendor IDs under that Purpose restriction. Disclosed vendor segment • disclosedVendors: A list of integers representing the vendors that have been disclosed to the user. Privacy-related functions 337 AWS Clean Rooms Publisher purposes segment SQL Reference • pubPurposesConsent: A list of integers representing the publisher-speciﬁc purposes for which the user has given consent. • pubPurposesLITransparency: A list of integers representing the publisher-speciﬁc purposes for which the user has given legitimate interest transparency. • customPurposesConsent: A list of integers representing the custom purposes for which the user has given consent. • customPurposesLITransparency: A list of integers representing the custom purposes for which the user has given legitimate interest transparency. This detailed consent data can be used to understand and respect the user's privacy preferences when working with personal data. Example The following example takes a single argument, which is the encoded consent string. It returns a dictionary containing the decoded consent data, including information about the user's privacy preferences, consent choices, and other metadata. from aws_clean_rooms.functions import consent_tcf_v2_decode consent_string = "CO1234567890abcdef" consent_data = consent_tcf_v2_decode(consent_string) print(consent_data) The basic structure of the returned consent data includes information about the consent string version, the CMP (Consent Management Platform) details, the user's consent and legitimate interest choices for diﬀerent purposes and vendors, and other metadata. / core segment / version: 2, created: "2023-10-01T12:00:00Z", lastUpdated: "2023-10-01T12:00:00Z", cmpId: 1234, cmpVersion: 5, Privacy-related functions 338 AWS Clean Rooms SQL Reference consentScreen: 1, consentLanguage: "en", vendorListVersion: 2, tcfPolicyVersion: 2, isServiceSpecific: false, useNonStandardStacks: false, specialFeatureOptIns: [1, 2, 3], purposeConsent: [1, 2, 3], purposesLITransparency: [1, 2, 3], purposeOneTreatment: true, publisherCountryCode: "US", vendorConsent: [1, 2, 3], vendorLegitimateInterest: [1, 2, 3],
sql-reference-091	sql-reference.pdf	91	= "CO1234567890abcdef" consent_data = consent_tcf_v2_decode(consent_string) print(consent_data) The basic structure of the returned consent data includes information about the consent string version, the CMP (Consent Management Platform) details, the user's consent and legitimate interest choices for diﬀerent purposes and vendors, and other metadata. / core segment / version: 2, created: "2023-10-01T12:00:00Z", lastUpdated: "2023-10-01T12:00:00Z", cmpId: 1234, cmpVersion: 5, Privacy-related functions 338 AWS Clean Rooms SQL Reference consentScreen: 1, consentLanguage: "en", vendorListVersion: 2, tcfPolicyVersion: 2, isServiceSpecific: false, useNonStandardStacks: false, specialFeatureOptIns: [1, 2, 3], purposeConsent: [1, 2, 3], purposesLITransparency: [1, 2, 3], purposeOneTreatment: true, publisherCountryCode: "US", vendorConsent: [1, 2, 3], vendorLegitimateInterest: [1, 2, 3], pubRestrictionEntry: [ { purpose: 1, restrictionType: 2, restrictionDescription: "Example restriction" }, ], / disclosed vendor segment / disclosedVendors: [1, 2, 3], / publisher purposes segment / pubPurposesConsent: [1, 2, 3], pubPurposesLITransparency: [1, 2, 3], customPurposesConsent: [1, 2, 3], customPurposesLITransparency: [1, 2, 3], }; Window functions By using window functions, you can create analytic business queries more eﬃciently. Window functions operate on a partition or "window" of a result set, and return a value for every row in that window. In contrast, non-windowed functions perform their calculations with respect to every row in the result set. Unlike group functions that aggregate result rows, window functions retain all rows in the table expression. The values returned are calculated by using values from the sets of rows in that window. For each row in the table, the window deﬁnes a set of rows that is used to compute additional attributes. A window is deﬁned using a window speciﬁcation (the OVER clause), and is based on three main concepts: • Window partitioning, which forms groups of rows (PARTITION clause) Window functions 339 AWS Clean Rooms SQL Reference • Window ordering, which deﬁnes an order or sequence of rows within each partition (ORDER BY clause) • Window frames, which are deﬁned relative to each row to further restrict the set of rows (ROWS speciﬁcation) Window functions are the last set of operations performed in a query except for the ﬁnal ORDER BY clause. All joins and all WHERE, GROUP BY, and HAVING clauses are completed before the window functions are processed. Therefore, window functions can appear only in the select list or ORDER BY clause. You can use multiple window functions within a single query with diﬀerent frame clauses. You can also use window functions in other scalar expressions, such as CASE. Window function syntax summary Window functions follow a standard syntax, which is as follows. function (expression) OVER ( [ PARTITION BY expr_list ] [ ORDER BY order_list [ frame_clause ] ] ) Here, function is one of the functions described in this section. The expr_list is as follows. expression \| column_name [, expr_list ] The order_list is as follows. expression \| column_name [ ASC \| DESC ] [ NULLS FIRST \| NULLS LAST ] [, order_list ] The frame_clause is as follows. ROWS { UNBOUNDED PRECEDING \| unsigned_value PRECEDING \| CURRENT ROW } \| { BETWEEN { UNBOUNDED PRECEDING \| unsigned_value { PRECEDING \| FOLLOWING } \| CURRENT ROW} AND Window functions 340 AWS Clean Rooms SQL Reference { UNBOUNDED FOLLOWING \| unsigned_value { PRECEDING \| FOLLOWING } \| CURRENT ROW }} Arguments function The window function. For details, see the individual function descriptions. OVER The clause that deﬁnes the window speciﬁcation. The OVER clause is mandatory for window functions, and diﬀerentiates window functions from other SQL functions. PARTITION BY expr_list (Optional) The PARTITION BY clause subdivides the result set into partitions, much like the GROUP BY clause. If a partition clause is present, the function is calculated for the rows in each partition. If no partition clause is speciﬁed, a single partition contains the entire table, and the function is computed for that complete table. The ranking functions DENSE_RANK, NTILE, RANK, and ROW_NUMBER require a global comparison of all the rows in the result set. When a PARTITION BY clause is used, the query optimizer can run each aggregation in parallel by spreading the workload across multiple slices according to the partitions. If the PARTITION BY clause is not present, the aggregation step must be run serially on a single slice, which can have a signiﬁcant negative impact on performance, especially for large clusters. AWS Clean Rooms doesn't support string literals in PARTITION BY clauses. ORDER BY order_list (Optional) The window function is applied to the rows within each partition sorted according to the order speciﬁcation in ORDER BY. This ORDER BY clause is distinct from and completely unrelated to ORDER BY clauses in the frame_clause. The ORDER BY clause can be used without the PARTITION BY clause. For ranking functions, the ORDER BY clause identiﬁes the measures for the ranking values. For aggregation functions, the partitioned rows must be ordered before the aggregate function is computed for each frame. For more
sql-reference-092	sql-reference.pdf	92	large clusters. AWS Clean Rooms doesn't support string literals in PARTITION BY clauses. ORDER BY order_list (Optional) The window function is applied to the rows within each partition sorted according to the order speciﬁcation in ORDER BY. This ORDER BY clause is distinct from and completely unrelated to ORDER BY clauses in the frame_clause. The ORDER BY clause can be used without the PARTITION BY clause. For ranking functions, the ORDER BY clause identiﬁes the measures for the ranking values. For aggregation functions, the partitioned rows must be ordered before the aggregate function is computed for each frame. For more about window function types, see Window functions. Column identiﬁers or expressions that evaluate to column identiﬁers are required in the order list. Neither constants nor constant expressions can be used as substitutes for column names. Window functions 341 AWS Clean Rooms SQL Reference NULLS values are treated as their own group, sorted and ranked according to the NULLS FIRST or NULLS LAST option. By default, NULL values are sorted and ranked last in ASC ordering, and sorted and ranked ﬁrst in DESC ordering. AWS Clean Rooms doesn't support string literals in ORDER BY clauses. If the ORDER BY clause is omitted, the order of the rows is nondeterministic. Note In any parallel system such as AWS Clean Rooms, when an ORDER BY clause doesn't produce a unique and total ordering of the data, the order of the rows is nondeterministic. That is, if the ORDER BY expression produces duplicate values (a partial ordering), the return order of those rows might vary from one run of AWS Clean Rooms to the next. In turn, window functions might return unexpected or inconsistent results. For more information, see Unique ordering of data for window functions. column_name Name of a column to be partitioned by or ordered by. ASC \| DESC Option that deﬁnes the sort order for the expression, as follows: • ASC: ascending (for example, low to high for numeric values and 'A' to 'Z' for character strings). If no option is speciﬁed, data is sorted in ascending order by default. • DESC: descending (high to low for numeric values; 'Z' to 'A' for strings). NULLS FIRST \| NULLS LAST Option that speciﬁes whether NULLS should be ordered ﬁrst, before non-null values, or last, after non-null values. By default, NULLS are sorted and ranked last in ASC ordering, and sorted and ranked ﬁrst in DESC ordering. frame_clause For aggregate functions, the frame clause further reﬁnes the set of rows in a function's window when using ORDER BY. It enables you to include or exclude sets of rows within the ordered result. The frame clause consists of the ROWS keyword and associated speciﬁers. Window functions 342 AWS Clean Rooms SQL Reference The frame clause doesn't apply to ranking functions. Also, the frame clause isn't required when no ORDER BY clause is used in the OVER clause for an aggregate function. If an ORDER BY clause is used for an aggregate function, an explicit frame clause is required. When no ORDER BY clause is speciﬁed, the implied frame is unbounded, equivalent to ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING. ROWS This clause deﬁnes the window frame by specifying a physical oﬀset from the current row. This clause speciﬁes the rows in the current window or partition that the value in the current row is to be combined with. It uses arguments that specify row position, which can be before or after the current row. The reference point for all window frames is the current row. Each row becomes the current row in turn as the window frame slides forward in the partition. The frame can be a simple set of rows up to and including the current row. {UNBOUNDED PRECEDING \| offset PRECEDING \| CURRENT ROW} Or it can be a set of rows between two boundaries. BETWEEN { UNBOUNDED PRECEDING \| offset { PRECEDING \| FOLLOWING } \| CURRENT ROW } AND { UNBOUNDED FOLLOWING \| offset { PRECEDING \| FOLLOWING } \| CURRENT ROW } UNBOUNDED PRECEDING indicates that the window starts at the ﬁrst row of the partition; oﬀset PRECEDING indicates that the window starts a number of rows equivalent to the value of oﬀset before the current row. UNBOUNDED PRECEDING is the default. CURRENT ROW indicates the window begins or ends at the current row. UNBOUNDED FOLLOWING indicates that the window ends at the last row of the partition; oﬀset FOLLOWING indicates that the window ends a number of rows equivalent to the value of oﬀset after the current row. oﬀset identiﬁes a physical number of rows before or after the current row. In this case, oﬀset must be a constant that evaluates to a positive numeric value. For example, 5 FOLLOWING ends the frame ﬁve rows after the current
sql-reference-093	sql-reference.pdf	93	the value of oﬀset before the current row. UNBOUNDED PRECEDING is the default. CURRENT ROW indicates the window begins or ends at the current row. UNBOUNDED FOLLOWING indicates that the window ends at the last row of the partition; oﬀset FOLLOWING indicates that the window ends a number of rows equivalent to the value of oﬀset after the current row. oﬀset identiﬁes a physical number of rows before or after the current row. In this case, oﬀset must be a constant that evaluates to a positive numeric value. For example, 5 FOLLOWING ends the frame ﬁve rows after the current row. Window functions 343 AWS Clean Rooms SQL Reference Where BETWEEN is not speciﬁed, the frame is implicitly bounded by the current row. For example, ROWS 5 PRECEDING is equal to ROWS BETWEEN 5 PRECEDING AND CURRENT ROW. Also, ROWS UNBOUNDED FOLLOWING is equal to ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING. Note You can't specify a frame in which the starting boundary is greater than the ending boundary. For example, you can't specify any of the following frames. between 5 following and 5 preceding between current row and 2 preceding between 3 following and current row Unique ordering of data for window functions If an ORDER BY clause for a window function doesn't produce a unique and total ordering of the data, the order of the rows is nondeterministic. If the ORDER BY expression produces duplicate values (a partial ordering), the return order of those rows can vary in multiple runs. In this case, window functions can also return unexpected or inconsistent results. For example, the following query returns diﬀerent results over multiple runs. These diﬀerent results occur because order by dateid doesn't produce a unique ordering of the data for the SUM window function. select dateid, pricepaid, sum(pricepaid) over(order by dateid rows unbounded preceding) as sumpaid from sales group by dateid, pricepaid; dateid \| pricepaid \| sumpaid --------+-----------+------------- 1827 \| 1730.00 \| 1730.00 1827 \| 708.00 \| 2438.00 1827 \| 234.00 \| 2672.00 ... select dateid, pricepaid, sum(pricepaid) over(order by dateid rows unbounded preceding) as sumpaid Window functions 344 AWS Clean Rooms from sales group by dateid, pricepaid; dateid \| pricepaid \| sumpaid --------+-----------+------------- 1827 \| 234.00 \| 234.00 1827 \| 472.00 \| 706.00 1827 \| 347.00 \| 1053.00 ... SQL Reference In this case, adding a second ORDER BY column to the window function can solve the problem. select dateid, pricepaid, sum(pricepaid) over(order by dateid, pricepaid rows unbounded preceding) as sumpaid from sales group by dateid, pricepaid; dateid \| pricepaid \| sumpaid --------+-----------+--------- 1827 \| 234.00 \| 234.00 1827 \| 337.00 \| 571.00 1827 \| 347.00 \| 918.00 ... Supported functions AWS Clean Rooms Spark SQL supports two types of window functions: aggregate and ranking. Following are the supported aggregate functions: • CUME_DIST window function • DENSE_RANK window function • FIRST_VALUE window function • LAG window function • LAST_VALUE window function • LEAD window function Following are the supported ranking functions: • DENSE_RANK window function Window functions 345 AWS Clean Rooms SQL Reference • PERCENT_RANK window function • RANK window function • ROW_NUMBER window function Sample table for window function examples You can ﬁnd speciﬁc window function examples with each function description. Some of the examples use a table named WINSALES, which contains 11 rows, as shown in the following table. SALESID DATEID SELLERID BUYERID QTY QTY_SHIPP ED 30001 10001 10005 40001 10006 20001 40005 20002 30003 30004 30007 8/2/2003 12/24/2003 12/24/2003 1/9/2004 1/18/2004 2/12/2004 2/12/2004 2/16/2004 4/18/2004 4/18/2004 9/7/2004 3 1 1 4 1 2 4 2 3 3 3 B C A A C B A C B B C 10 10 20 10 20 10 10 30 40 10 20 10 20 15 20 30 CUME_DIST window function Calculates the cumulative distribution of a value within a window or partition. Assuming ascending ordering, the cumulative distribution is determined using this formula: Window functions 346 AWS Clean Rooms SQL Reference count of rows with values <= x / count of rows in the window or partition where x equals the value in the current row of the column speciﬁed in the ORDER BY clause. The following dataset illustrates use of this formula: Row# Value Calculation CUME_DIST 1 2500 (1)/(5) 0.2 2 2600 (2)/(5) 0.4 3 2800 (3)/(5) 0.6 4 2900 (4)/(5) 0.8 5 3100 (5)/(5) 1.0 The return value range is >0 to 1, inclusive. Syntax CUME_DIST () OVER ( [ PARTITION BY partition_expression ] [ ORDER BY order_list ] ) Arguments OVER A clause that speciﬁes the window partitioning. The OVER clause cannot contain a window frame speciﬁcation. PARTITION BY partition_expression Optional. An expression that sets the range of records for each group in the OVER clause. ORDER BY order_list The expression on which to calculate cumulative distribution. The expression must have either a
sql-reference-094	sql-reference.pdf	94	CUME_DIST 1 2500 (1)/(5) 0.2 2 2600 (2)/(5) 0.4 3 2800 (3)/(5) 0.6 4 2900 (4)/(5) 0.8 5 3100 (5)/(5) 1.0 The return value range is >0 to 1, inclusive. Syntax CUME_DIST () OVER ( [ PARTITION BY partition_expression ] [ ORDER BY order_list ] ) Arguments OVER A clause that speciﬁes the window partitioning. The OVER clause cannot contain a window frame speciﬁcation. PARTITION BY partition_expression Optional. An expression that sets the range of records for each group in the OVER clause. ORDER BY order_list The expression on which to calculate cumulative distribution. The expression must have either a numeric data type or be implicitly convertible to one. If ORDER BY is omitted, the return value is 1 for all rows. If ORDER BY doesn't produce a unique ordering, the order of the rows is nondeterministic. For more information, see Unique ordering of data for window functions. Window functions 347 AWS Clean Rooms Return type FLOAT8 Examples SQL Reference The following example calculates the cumulative distribution of the quantity for each seller: select sellerid, qty, cume_dist() over (partition by sellerid order by qty) from winsales; sellerid qty cume_dist -------------------------------------------------- 1 10.00 0.33 1 10.64 0.67 1 30.37 1 3 10.04 0.25 3 15.15 0.5 3 20.75 0.75 3 30.55 1 2 20.09 0.5 2 20.12 1 4 10.12 0.5 4 40.23 1 For a description of the WINSALES table, see Sample table for window function examples. DENSE_RANK window function The DENSE_RANK window function determines the rank of a value in a group of values, based on the ORDER BY expression in the OVER clause. If the optional PARTITION BY clause is present, the rankings are reset for each group of rows. Rows with equal values for the ranking criteria receive the same rank. The DENSE_RANK function diﬀers from RANK in one respect: If two or more rows tie, there is no gap in the sequence of ranked values. For example, if two rows are ranked 1, the next rank is 2. You can have ranking functions with diﬀerent PARTITION BY and ORDER BY clauses in the same query. Syntax DENSE_RANK () OVER Window functions 348 AWS Clean Rooms ( [ PARTITION BY expr_list ] [ ORDER BY order_list ] ) Arguments ( ) SQL Reference The function takes no arguments, but the empty parentheses are required. OVER The window clauses for the DENSE_RANK function. PARTITION BY expr_list Optional. One or more expressions that deﬁne the window. ORDER BY order_list Optional. The expression on which the ranking values are based. If no PARTITION BY is speciﬁed, ORDER BY uses the entire table. If ORDER BY is omitted, the return value is 1 for all rows. If ORDER BY doesn't produce a unique ordering, the order of the rows is nondeterministic. For more information, see Unique ordering of data for window functions. Return type INTEGER Examples The following example orders the table by the quantity sold (in descending order), and assign both a dense rank and a regular rank to each row. The results are sorted after the window function results are applied. select salesid, qty, dense_rank() over(order by qty desc) as d_rnk, rank() over(order by qty desc) as rnk from winsales order by 2,1; Window functions 349 AWS Clean Rooms SQL Reference salesid \| qty \| d_rnk \| rnk ---------+-----+-------+----- 10001 \| 10 \| 5 \| 8 10006 \| 10 \| 5 \| 8 30001 \| 10 \| 5 \| 8 40005 \| 10 \| 5 \| 8 30003 \| 15 \| 4 \| 7 20001 \| 20 \| 3 \| 4 20002 \| 20 \| 3 \| 4 30004 \| 20 \| 3 \| 4 10005 \| 30 \| 2 \| 2 30007 \| 30 \| 2 \| 2 40001 \| 40 \| 1 \| 1 (11 rows) Note the diﬀerence in rankings assigned to the same set of rows when the DENSE_RANK and RANK functions are used side by side in the same query. For a description of the WINSALES table, see Sample table for window function examples. The following example partitions the table by SELLERID and orders each partition by the quantity (in descending order) and assign a dense rank to each row. The results are sorted after the window function results are applied. select salesid, sellerid, qty, dense_rank() over(partition by sellerid order by qty desc) as d_rnk from winsales order by 2,3,1; salesid \| sellerid \| qty \| d_rnk ---------+----------+-----+------- 10001 \| 1 \| 10 \| 2 10006 \| 1 \| 10 \| 2 10005 \| 1 \| 30 \| 1 20001 \| 2 \| 20 \| 1 20002 \| 2 \| 20 \| 1 30001 \| 3 \| 10 \| 4 30003 \| 3 \| 15 \| 3 30004 \| 3 \| 20 \| 2 30007 \| 3 \| 30 \| 1 40005 \| 4
sql-reference-095	sql-reference.pdf	95	after the window function results are applied. select salesid, sellerid, qty, dense_rank() over(partition by sellerid order by qty desc) as d_rnk from winsales order by 2,3,1; salesid \| sellerid \| qty \| d_rnk ---------+----------+-----+------- 10001 \| 1 \| 10 \| 2 10006 \| 1 \| 10 \| 2 10005 \| 1 \| 30 \| 1 20001 \| 2 \| 20 \| 1 20002 \| 2 \| 20 \| 1 30001 \| 3 \| 10 \| 4 30003 \| 3 \| 15 \| 3 30004 \| 3 \| 20 \| 2 30007 \| 3 \| 30 \| 1 40005 \| 4 \| 10 \| 2 40001 \| 4 \| 40 \| 1 (11 rows) Window functions 350 AWS Clean Rooms SQL Reference For a description of the WINSALES table, see Sample table for window function examples. FIRST_VALUE window function Given an ordered set of rows, FIRST_VALUE returns the value of the speciﬁed expression with respect to the ﬁrst row in the window frame. For information about selecting the last row in the frame, see LAST_VALUE window function . Syntax FIRST_VALUE( expression )[ IGNORE NULLS \| RESPECT NULLS ] OVER ( [ PARTITION BY expr_list ] [ ORDER BY order_list frame_clause ] ) Arguments expression The target column or expression that the function operates on. IGNORE NULLS When this option is used with FIRST_VALUE, the function returns the ﬁrst value in the frame that is not NULL (or NULL if all values are NULL). RESPECT NULLS Indicates that AWS Clean Rooms should include null values in the determination of which row to use. RESPECT NULLS is supported by default if you do not specify IGNORE NULLS. OVER Introduces the window clauses for the function. PARTITION BY expr_list Deﬁnes the window for the function in terms of one or more expressions. ORDER BY order_list Sorts the rows within each partition. If no PARTITION BY clause is speciﬁed, ORDER BY sorts the entire table. If you specify an ORDER BY clause, you must also specify a frame_clause. Window functions 351 AWS Clean Rooms SQL Reference The results of the FIRST_VALUE function depends on the ordering of the data. The results are nondeterministic in the following cases: • When no ORDER BY clause is speciﬁed and a partition contains two diﬀerent values for an expression • When the expression evaluates to diﬀerent values that correspond to the same value in the ORDER BY list. frame_clause If an ORDER BY clause is used for an aggregate function, an explicit frame clause is required. The frame clause reﬁnes the set of rows in a function's window, including or excluding sets of rows in the ordered result. The frame clause consists of the ROWS keyword and associated speciﬁers. See Window function syntax summary. Return type These functions support expressions that use primitive AWS Clean Rooms data types. The return type is the same as the data type of the expression. Examples The following example returns the seating capacity for each venue in the VENUE table, with the results ordered by capacity (high to low). The FIRST_VALUE function is used to select the name of the venue that corresponds to the ﬁrst row in the frame: in this case, the row with the highest number of seats. The results are partitioned by state, so when the VENUESTATE value changes, a new ﬁrst value is selected. The window frame is unbounded so the same ﬁrst value is selected for each row in each partition. For California, Qualcomm Stadium has the highest number of seats (70561), so this name is the ﬁrst value for all of the rows in the CA partition. select venuestate, venueseats, venuename, first_value(venuename) over(partition by venuestate order by venueseats desc rows between unbounded preceding and unbounded following) from (select * from venue where venueseats >0) order by venuestate; venuestate \| venueseats \| venuename \| first_value Window functions 352 AWS Clean Rooms SQL Reference -----------+------------+-------------------------------- +------------------------------ CA \| 70561 \| Qualcomm Stadium \| Qualcomm Stadium CA \| 69843 \| Monster Park \| Qualcomm Stadium CA \| 63026 \| McAfee Coliseum \| Qualcomm Stadium CA \| 56000 \| Dodger Stadium \| Qualcomm Stadium CA \| 45050 \| Angel Stadium of Anaheim \| Qualcomm Stadium CA \| 42445 \| PETCO Park \| Qualcomm Stadium CA \| 41503 \| AT&T Park \| Qualcomm Stadium CA \| 22000 \| Shoreline Amphitheatre \| Qualcomm Stadium CO \| 76125 \| INVESCO Field \| INVESCO Field CO \| 50445 \| Coors Field \| INVESCO Field DC \| 41888 \| Nationals Park \| Nationals Park FL \| 74916 \| Dolphin Stadium \| Dolphin Stadium FL \| 73800 \| Jacksonville Municipal Stadium \| Dolphin Stadium FL \| 65647 \| Raymond James Stadium \| Dolphin Stadium FL \| 36048 \| Tropicana Field \| Dolphin Stadium ... LAG window function The LAG window function returns the values for a
sql-reference-096	sql-reference.pdf	96	Park \| Qualcomm Stadium CA \| 41503 \| AT&T Park \| Qualcomm Stadium CA \| 22000 \| Shoreline Amphitheatre \| Qualcomm Stadium CO \| 76125 \| INVESCO Field \| INVESCO Field CO \| 50445 \| Coors Field \| INVESCO Field DC \| 41888 \| Nationals Park \| Nationals Park FL \| 74916 \| Dolphin Stadium \| Dolphin Stadium FL \| 73800 \| Jacksonville Municipal Stadium \| Dolphin Stadium FL \| 65647 \| Raymond James Stadium \| Dolphin Stadium FL \| 36048 \| Tropicana Field \| Dolphin Stadium ... LAG window function The LAG window function returns the values for a row at a given oﬀset above (before) the current row in the partition. Syntax LAG (value_expr [, offset ]) [ IGNORE NULLS \| RESPECT NULLS ] OVER ( [ PARTITION BY window_partition ] ORDER BY window_ordering ) Arguments value_expr The target column or expression that the function operates on. oﬀset An optional parameter that speciﬁes the number of rows before the current row to return values for. The oﬀset can be a constant integer or an expression that evaluates to an integer. If you do not specify an oﬀset, AWS Clean Rooms uses 1 as the default value. An oﬀset of 0 indicates the current row. Window functions 353 AWS Clean Rooms IGNORE NULLS SQL Reference An optional speciﬁcation that indicates that AWS Clean Rooms should skip null values in the determination of which row to use. Null values are included if IGNORE NULLS is not listed. Note You can use an NVL or COALESCE expression to replace the null values with another value. RESPECT NULLS Indicates that AWS Clean Rooms should include null values in the determination of which row to use. RESPECT NULLS is supported by default if you do not specify IGNORE NULLS. OVER Speciﬁes the window partitioning and ordering. The OVER clause cannot contain a window frame speciﬁcation. PARTITION BY window_partition An optional argument that sets the range of records for each group in the OVER clause. ORDER BY window_ordering Sorts the rows within each partition. The LAG window function supports expressions that use any of the AWS Clean Rooms data types. The return type is the same as the type of the value_expr. Examples The following example shows the quantity of tickets sold to the buyer with a buyer ID of 3 and the time that buyer 3 bought the tickets. To compare each sale with the previous sale for buyer 3, the query returns the previous quantity sold for each sale. Since there is no purchase before 1/16/2008, the ﬁrst previous quantity sold value is null: select buyerid, saletime, qtysold, lag(qtysold,1) over (order by buyerid, saletime) as prev_qtysold Window functions 354 AWS Clean Rooms SQL Reference from sales where buyerid = 3 order by buyerid, saletime; buyerid \| saletime \| qtysold \| prev_qtysold ---------+---------------------+---------+-------------- 3 \| 2008-01-16 01:06:09 \| 1 \| 3 \| 2008-01-28 02:10:01 \| 1 \| 1 3 \| 2008-03-12 10:39:53 \| 1 \| 1 3 \| 2008-03-13 02:56:07 \| 1 \| 1 3 \| 2008-03-29 08:21:39 \| 2 \| 1 3 \| 2008-04-27 02:39:01 \| 1 \| 2 3 \| 2008-08-16 07:04:37 \| 2 \| 1 3 \| 2008-08-22 11:45:26 \| 2 \| 2 3 \| 2008-09-12 09:11:25 \| 1 \| 2 3 \| 2008-10-01 06:22:37 \| 1 \| 1 3 \| 2008-10-20 01:55:51 \| 2 \| 1 3 \| 2008-10-28 01:30:40 \| 1 \| 2 (12 rows) LAST_VALUE window function Given an ordered set of rows, The LAST_VALUE function returns the value of the expression with respect to the last row in the frame. For information about selecting the ﬁrst row in the frame, see FIRST_VALUE window function . Syntax LAST_VALUE( expression )[ IGNORE NULLS \| RESPECT NULLS ] OVER ( [ PARTITION BY expr_list ] [ ORDER BY order_list frame_clause ] ) Arguments expression The target column or expression that the function operates on. IGNORE NULLS The function returns the last value in the frame that is not NULL (or NULL if all values are NULL). Window functions 355 AWS Clean Rooms RESPECT NULLS SQL Reference Indicates that AWS Clean Rooms should include null values in the determination of which row to use. RESPECT NULLS is supported by default if you do not specify IGNORE NULLS. OVER Introduces the window clauses for the function. PARTITION BY expr_list Deﬁnes the window for the function in terms of one or more expressions. ORDER BY order_list Sorts the rows within each partition. If no PARTITION BY clause is speciﬁed, ORDER BY sorts the entire table. If you specify an ORDER BY clause, you must also specify a frame_clause. The results depend on the ordering of the data. The results are nondeterministic in the following cases: • When no ORDER BY clause is speciﬁed and a partition contains two diﬀerent values for an
sql-reference-097	sql-reference.pdf	97	if you do not specify IGNORE NULLS. OVER Introduces the window clauses for the function. PARTITION BY expr_list Deﬁnes the window for the function in terms of one or more expressions. ORDER BY order_list Sorts the rows within each partition. If no PARTITION BY clause is speciﬁed, ORDER BY sorts the entire table. If you specify an ORDER BY clause, you must also specify a frame_clause. The results depend on the ordering of the data. The results are nondeterministic in the following cases: • When no ORDER BY clause is speciﬁed and a partition contains two diﬀerent values for an expression • When the expression evaluates to diﬀerent values that correspond to the same value in the ORDER BY list. frame_clause If an ORDER BY clause is used for an aggregate function, an explicit frame clause is required. The frame clause reﬁnes the set of rows in a function's window, including or excluding sets of rows in the ordered result. The frame clause consists of the ROWS keyword and associated speciﬁers. See Window function syntax summary. Return type These functions support expressions that use primitive AWS Clean Rooms data types. The return type is the same as the data type of the expression. Examples The following example returns the seating capacity for each venue in the VENUE table, with the results ordered by capacity (high to low). The LAST_VALUE function is used to select the name of the venue that corresponds to the last row in the frame: in this case, the row with the least number of seats. The results are partitioned by state, so when the VENUESTATE value changes, a new last Window functions 356 AWS Clean Rooms SQL Reference value is selected. The window frame is unbounded so the same last value is selected for each row in each partition. For California, Shoreline Amphitheatre is returned for every row in the partition because it has the lowest number of seats (22000). select venuestate, venueseats, venuename, last_value(venuename) over(partition by venuestate order by venueseats desc rows between unbounded preceding and unbounded following) from (select * from venue where venueseats >0) order by venuestate; venuestate \| venueseats \| venuename \| last_value -----------+------------+-------------------------------- +------------------------------ CA \| 70561 \| Qualcomm Stadium \| Shoreline Amphitheatre CA \| 69843 \| Monster Park \| Shoreline Amphitheatre CA \| 63026 \| McAfee Coliseum \| Shoreline Amphitheatre CA \| 56000 \| Dodger Stadium \| Shoreline Amphitheatre CA \| 45050 \| Angel Stadium of Anaheim \| Shoreline Amphitheatre CA \| 42445 \| PETCO Park \| Shoreline Amphitheatre CA \| 41503 \| AT&T Park \| Shoreline Amphitheatre CA \| 22000 \| Shoreline Amphitheatre \| Shoreline Amphitheatre CO \| 76125 \| INVESCO Field \| Coors Field CO \| 50445 \| Coors Field \| Coors Field DC \| 41888 \| Nationals Park \| Nationals Park FL \| 74916 \| Dolphin Stadium \| Tropicana Field FL \| 73800 \| Jacksonville Municipal Stadium \| Tropicana Field FL \| 65647 \| Raymond James Stadium \| Tropicana Field FL \| 36048 \| Tropicana Field \| Tropicana Field ... LEAD window function The LEAD window function returns the values for a row at a given oﬀset below (after) the current row in the partition. Syntax LEAD (value_expr [, offset ]) Window functions 357 AWS Clean Rooms SQL Reference [ IGNORE NULLS \| RESPECT NULLS ] OVER ( [ PARTITION BY window_partition ] ORDER BY window_ordering ) Arguments value_expr The target column or expression that the function operates on. oﬀset An optional parameter that speciﬁes the number of rows below the current row to return values for. The oﬀset can be a constant integer or an expression that evaluates to an integer. If you do not specify an oﬀset, AWS Clean Rooms uses 1 as the default value. An oﬀset of 0 indicates the current row. IGNORE NULLS An optional speciﬁcation that indicates that AWS Clean Rooms should skip null values in the determination of which row to use. Null values are included if IGNORE NULLS is not listed. Note You can use an NVL or COALESCE expression to replace the null values with another value. RESPECT NULLS Indicates that AWS Clean Rooms should include null values in the determination of which row to use. RESPECT NULLS is supported by default if you do not specify IGNORE NULLS. OVER Speciﬁes the window partitioning and ordering. The OVER clause cannot contain a window frame speciﬁcation. PARTITION BY window_partition An optional argument that sets the range of records for each group in the OVER clause. ORDER BY window_ordering Sorts the rows within each partition. Window functions 358 AWS Clean Rooms SQL Reference The LEAD window function supports expressions that use any of the AWS Clean Rooms data types. The return type is the same as the type of the value_expr. Examples The following example provides the commission for events
sql-reference-098	sql-reference.pdf	98	is supported by default if you do not specify IGNORE NULLS. OVER Speciﬁes the window partitioning and ordering. The OVER clause cannot contain a window frame speciﬁcation. PARTITION BY window_partition An optional argument that sets the range of records for each group in the OVER clause. ORDER BY window_ordering Sorts the rows within each partition. Window functions 358 AWS Clean Rooms SQL Reference The LEAD window function supports expressions that use any of the AWS Clean Rooms data types. The return type is the same as the type of the value_expr. Examples The following example provides the commission for events in the SALES table for which tickets were sold on January 1, 2008 and January 2, 2008 and the commission paid for ticket sales for the subsequent sale. select eventid, commission, saletime, lead(commission, 1) over (order by saletime) as next_comm from sales where saletime between '2008-01-01 00:00:00' and '2008-01-02 12:59:59' order by saletime; eventid \| commission \| saletime \| next_comm ---------+------------+---------------------+----------- 6213 \| 52.05 \| 2008-01-01 01:00:19 \| 106.20 7003 \| 106.20 \| 2008-01-01 02:30:52 \| 103.20 8762 \| 103.20 \| 2008-01-01 03:50:02 \| 70.80 1150 \| 70.80 \| 2008-01-01 06:06:57 \| 50.55 1749 \| 50.55 \| 2008-01-01 07:05:02 \| 125.40 8649 \| 125.40 \| 2008-01-01 07:26:20 \| 35.10 2903 \| 35.10 \| 2008-01-01 09:41:06 \| 259.50 6605 \| 259.50 \| 2008-01-01 12:50:55 \| 628.80 6870 \| 628.80 \| 2008-01-01 12:59:34 \| 74.10 6977 \| 74.10 \| 2008-01-02 01:11:16 \| 13.50 4650 \| 13.50 \| 2008-01-02 01:40:59 \| 26.55 4515 \| 26.55 \| 2008-01-02 01:52:35 \| 22.80 5465 \| 22.80 \| 2008-01-02 02:28:01 \| 45.60 5465 \| 45.60 \| 2008-01-02 02:28:02 \| 53.10 7003 \| 53.10 \| 2008-01-02 02:31:12 \| 70.35 4124 \| 70.35 \| 2008-01-02 03:12:50 \| 36.15 1673 \| 36.15 \| 2008-01-02 03:15:00 \| 1300.80 ... (39 rows) PERCENT_RANK window function Calculates the percent rank of a given row. The percent rank is determined using this formula: (x - 1) / (the number of rows in the window or partition - 1) where x is the rank of the current row. The following dataset illustrates use of this formula: Window functions 359 AWS Clean Rooms SQL Reference Row# Value Rank Calculation PERCENT_RANK 1 15 1 (1-1)/(7-1) 0.0000 2 20 2 (2-1)/(7-1) 0.1666 3 20 2 (2-1)/(7-1) 0.1666 4 20 2 (2-1)/(7-1) 0.1666 5 30 5 (5-1)/(7-1) 0.6666 6 30 5 (5-1)/(7-1) 0.6666 7 40 7 (7-1)/(7-1) 1.0000 The return value range is 0 to 1, inclusive. The ﬁrst row in any set has a PERCENT_RANK of 0. Syntax PERCENT_RANK () OVER ( [ PARTITION BY partition_expression ] [ ORDER BY order_list ] ) Arguments ( ) The function takes no arguments, but the empty parentheses are required. OVER A clause that speciﬁes the window partitioning. The OVER clause cannot contain a window frame speciﬁcation. PARTITION BY partition_expression Optional. An expression that sets the range of records for each group in the OVER clause. ORDER BY order_list Optional. The expression on which to calculate percent rank. The expression must have either a numeric data type or be implicitly convertible to one. If ORDER BY is omitted, the return value is 0 for all rows. If ORDER BY does not produce a unique ordering, the order of the rows is nondeterministic. For more information, see Unique ordering of data for window functions. Window functions 360 AWS Clean Rooms Return type FLOAT8 Examples SQL Reference The following example calculates the percent rank of the sales quantities for each seller: select sellerid, qty, percent_rank() over (partition by sellerid order by qty) from winsales; sellerid qty percent_rank ---------------------------------------- 1 10.00 0.0 1 10.64 0.5 1 30.37 1.0 3 10.04 0.0 3 15.15 0.33 3 20.75 0.67 3 30.55 1.0 2 20.09 0.0 2 20.12 1.0 4 10.12 0.0 4 40.23 1.0 For a description of the WINSALES table, see Sample table for window function examples. RANK window function The RANK window function determines the rank of a value in a group of values, based on the ORDER BY expression in the OVER clause. If the optional PARTITION BY clause is present, the rankings are reset for each group of rows. Rows with equal values for the ranking criteria receive the same rank. AWS Clean Rooms adds the number of tied rows to the tied rank to calculate the next rank and thus the ranks might not be consecutive numbers. For example, if two rows are ranked 1, the next rank is 3. RANK diﬀers from the DENSE_RANK window function in one respect: For DENSE_RANK, if two or more rows tie, there is no gap in the sequence of ranked values. For example, if two rows are ranked 1, the next rank is 2. You can have ranking functions with diﬀerent PARTITION BY and ORDER BY clauses in the same query. Window functions 361
sql-reference-099	sql-reference.pdf	99	Clean Rooms adds the number of tied rows to the tied rank to calculate the next rank and thus the ranks might not be consecutive numbers. For example, if two rows are ranked 1, the next rank is 3. RANK diﬀers from the DENSE_RANK window function in one respect: For DENSE_RANK, if two or more rows tie, there is no gap in the sequence of ranked values. For example, if two rows are ranked 1, the next rank is 2. You can have ranking functions with diﬀerent PARTITION BY and ORDER BY clauses in the same query. Window functions 361 SQL Reference AWS Clean Rooms Syntax RANK () OVER ( [ PARTITION BY expr_list ] [ ORDER BY order_list ] ) Arguments ( ) The function takes no arguments, but the empty parentheses are required. OVER The window clauses for the RANK function. PARTITION BY expr_list Optional. One or more expressions that deﬁne the window. ORDER BY order_list Optional. Deﬁnes the columns on which the ranking values are based. If no PARTITION BY is speciﬁed, ORDER BY uses the entire table. If ORDER BY is omitted, the return value is 1 for all rows. If ORDER BY does not produce a unique ordering, the order of the rows is nondeterministic. For more information, see Unique ordering of data for window functions. Return type INTEGER Examples The following example orders the table by the quantity sold (default ascending), and assign a rank to each row. A rank value of 1 is the highest ranked value. The results are sorted after the window function results are applied: select salesid, qty, rank() over (order by qty) as rnk from winsales Window functions 362 SQL Reference AWS Clean Rooms order by 2,1; salesid \| qty \| rnk --------+-----+----- 10001 \| 10 \| 1 10006 \| 10 \| 1 30001 \| 10 \| 1 40005 \| 10 \| 1 30003 \| 15 \| 5 20001 \| 20 \| 6 20002 \| 20 \| 6 30004 \| 20 \| 6 10005 \| 30 \| 9 30007 \| 30 \| 9 40001 \| 40 \| 11 (11 rows) Note that the outer ORDER BY clause in this example includes columns 2 and 1 to make sure that AWS Clean Rooms returns consistently sorted results each time this query is run. For example, rows with sales IDs 10001 and 10006 have identical QTY and RNK values. Ordering the ﬁnal result set by column 1 ensures that row 10001 always falls before 10006. For a description of the WINSALES table, see Sample table for window function examples. In the following example, the ordering is reversed for the window function (order by qty desc). Now the highest rank value applies to the largest QTY value. select salesid, qty, rank() over (order by qty desc) as rank from winsales order by 2,1; salesid \| qty \| rank ---------+-----+----- 10001 \| 10 \| 8 10006 \| 10 \| 8 30001 \| 10 \| 8 40005 \| 10 \| 8 30003 \| 15 \| 7 20001 \| 20 \| 4 20002 \| 20 \| 4 30004 \| 20 \| 4 10005 \| 30 \| 2 30007 \| 30 \| 2 Window functions 363 AWS Clean Rooms 40001 \| 40 \| 1 (11 rows) SQL Reference For a description of the WINSALES table, see Sample table for window function examples. The following example partitions the table by SELLERID and order each partition by the quantity (in descending order) and assign a rank to each row. The results are sorted after the window function results are applied. select salesid, sellerid, qty, rank() over (partition by sellerid order by qty desc) as rank from winsales order by 2,3,1; salesid \| sellerid \| qty \| rank --------+----------+-----+----- 10001 \| 1 \| 10 \| 2 10006 \| 1 \| 10 \| 2 10005 \| 1 \| 30 \| 1 20001 \| 2 \| 20 \| 1 20002 \| 2 \| 20 \| 1 30001 \| 3 \| 10 \| 4 30003 \| 3 \| 15 \| 3 30004 \| 3 \| 20 \| 2 30007 \| 3 \| 30 \| 1 40005 \| 4 \| 10 \| 2 40001 \| 4 \| 40 \| 1 (11 rows) ROW_NUMBER window function Determines the ordinal number of the current row within a group of rows, counting from 1, based on the ORDER BY expression in the OVER clause. If the optional PARTITION BY clause is present, the ordinal numbers are reset for each group of rows. Rows with equal values for the ORDER BY expressions receive the diﬀerent row numbers nondeterministically. Syntax ROW_NUMBER () OVER ( [ PARTITION BY expr_list ] Window functions 364 AWS Clean Rooms [ ORDER BY order_list ] ) Arguments ( ) SQL Reference The function takes no arguments, but the empty parentheses are required. OVER
sql-reference-100	sql-reference.pdf	100	window function Determines the ordinal number of the current row within a group of rows, counting from 1, based on the ORDER BY expression in the OVER clause. If the optional PARTITION BY clause is present, the ordinal numbers are reset for each group of rows. Rows with equal values for the ORDER BY expressions receive the diﬀerent row numbers nondeterministically. Syntax ROW_NUMBER () OVER ( [ PARTITION BY expr_list ] Window functions 364 AWS Clean Rooms [ ORDER BY order_list ] ) Arguments ( ) SQL Reference The function takes no arguments, but the empty parentheses are required. OVER The window clauses for the ROW_NUMBER function. PARTITION BY expr_list Optional. One or more expressions that deﬁne the ROW_NUMBER function. ORDER BY order_list Optional. The expression that deﬁnes the columns on which the row numbers are based. If no PARTITION BY is speciﬁed, ORDER BY uses the entire table. If ORDER BY does not produce a unique ordering or is omitted, the order of the rows is nondeterministic. For more information, see Unique ordering of data for window functions. Return type BIGINT Examples The following example partitions the table by SELLERID and orders each partition by QTY (in ascending order), then assigns a row number to each row. The results are sorted after the window function results are applied. select salesid, sellerid, qty, row_number() over (partition by sellerid order by qty asc) as row from winsales order by 2,4; salesid \| sellerid \| qty \| row ---------+----------+-----+----- 10006 \| 1 \| 10 \| 1 Window functions 365 AWS Clean Rooms SQL Reference 10001 \| 1 \| 10 \| 2 10005 \| 1 \| 30 \| 3 20001 \| 2 \| 20 \| 1 20002 \| 2 \| 20 \| 2 30001 \| 3 \| 10 \| 1 30003 \| 3 \| 15 \| 2 30004 \| 3 \| 20 \| 3 30007 \| 3 \| 30 \| 4 40005 \| 4 \| 10 \| 1 40001 \| 4 \| 40 \| 2 (11 rows) For a description of the WINSALES table, see Sample table for window function examples. AWS Clean Rooms Spark SQL conditions Conditions are statements of one or more expressions and logical operators that evaluate to true, false, or unknown. Conditions are also sometimes referred to as predicates. Syntax comparison_condition \| logical_condition \| range_condition \| pattern_matching_condition \| null_condition \| EXISTS_condition \| IN_condition Note All string comparisons and LIKE pattern matches are case-sensitive. For example, 'A' and 'a' do not match. However, you can do a case-insensitive pattern match by using the ILIKE predicate. The following SQL conditions are supported in AWS Clean Rooms Spark SQL. Topics • Comparison operators SQL conditions 366 SQL Reference AWS Clean Rooms • Logical conditions • Pattern-matching conditions • BETWEEN range condition • Null condition • EXISTS condition • IN condition Comparison operators Comparison conditions state logical relationships between two values. All comparison conditions are binary operators with a Boolean return type. AWS Clean Rooms Spark SQL supports the comparison operators described in the following table. Operator Syntax Description ! < > !expression a < b a > b The logical NOT operator. Used to negate a boolean expression, meaning it returns the opposite of the expressio n's value. The ! operator can also be combined with other logical operators, such as AND and OR, to create more complex boolean expressions. The less than comparison operator. Used to compare two values and determine if the value on the left is less than the value on the right. The greater than compariso n operator. Used to compare two values and determine Comparison operators 367 AWS Clean Rooms Operator Syntax Description SQL Reference <= >= = a <= b a >= b a = b <> or != a <> b or a != b if the value on the left is greater than the value on the right. The less than or equal to comparison operator. Used to compare two values and returns true if the value on the left is less than or equal to the value on the right, and false otherwise. The greater than or equal to comparison operator. Used to compare two values and determine if the value on the left is greater than or equal to the value on the right. The equality comparison operator, which compares two values and returns true if they're equal, and false otherwise. The not equal to comparison operator, which compares two values and returns true if they're not equal, and false otherwise. <=> expression1 <=> A null-safe equality operator expression2 that returns true if the two expressions are equal, including when both are NULL, and false otherwise. Comparison operators 368 AWS Clean Rooms Operator == Syntax a == b SQL Reference Description The standard equality comparison operator, which compares two values and returns
sql-reference-101	sql-reference.pdf	101	left is greater than or equal to the value on the right. The equality comparison operator, which compares two values and returns true if they're equal, and false otherwise. The not equal to comparison operator, which compares two values and returns true if they're not equal, and false otherwise. <=> expression1 <=> A null-safe equality operator expression2 that returns true if the two expressions are equal, including when both are NULL, and false otherwise. Comparison operators 368 AWS Clean Rooms Operator == Syntax a == b SQL Reference Description The standard equality comparison operator, which compares two values and returns true if they're equal, and false otherwise. Note The == operator is case-sensitive when comparing string values. If you need to perform a case-inse nsitive comparison, you can use functions like UPPER() or LOWER() to convert the values to the same case before the comparison. Examples Here are some simple examples of comparison conditions: a = 5 a < b min(x) >= 5 qtysold = any (select qtysold from sales where dateid = 1882 The following query returns the id values for all the squirrels that are not currently foraging. SELECT id FROM squirrels WHERE !is_foraging Comparison operators 369 AWS Clean Rooms SQL Reference The following query returns venues with more than 10,000 seats from the VENUE table: select venueid, venuename, venueseats from venue where venueseats > 10000 order by venueseats desc; venueid \| venuename \| venueseats ---------+--------------------------------+------------ 83 \| FedExField \| 91704 6 \| New York Giants Stadium \| 80242 79 \| Arrowhead Stadium \| 79451 78 \| INVESCO Field \| 76125 69 \| Dolphin Stadium \| 74916 67 \| Ralph Wilson Stadium \| 73967 76 \| Jacksonville Municipal Stadium \| 73800 89 \| Bank of America Stadium \| 73298 72 \| Cleveland Browns Stadium \| 73200 86 \| Lambeau Field \| 72922 ... (57 rows) This example selects the users (USERID) from the USERS table who like rock music: select userid from users where likerock = 't' order by 1 limit 5; userid -------- 3 5 6 13 16 (5 rows) This example selects the users (USERID) from the USERS table where it is unknown whether they like rock music: select firstname, lastname, likerock from users where likerock is unknown order by userid limit 10; Comparison operators 370 AWS Clean Rooms SQL Reference firstname \| lastname \| likerock ----------+----------+---------- Rafael \| Taylor \| Vladimir \| Humphrey \| Barry \| Roy \| Tamekah \| Juarez \| Mufutau \| Watkins \| Naida \| Calderon \| Anika \| Huff \| Bruce \| Beck \| Mallory \| Farrell \| Scarlett \| Mayer \| (10 rows Examples with a TIME column The following example table TIME_TEST has a column TIME_VAL (type TIME) with three values inserted. select time_val from time_test; time_val --------------------- 20:00:00 00:00:00.5550 00:58:00 The following example extracts the hours from each timetz_val. select time_val from time_test where time_val < '3:00'; time_val --------------- 00:00:00.5550 00:58:00 The following example compares two time literals. select time '18:25:33.123456' = time '18:25:33.123456'; ?column? ---------- t Comparison operators 371 AWS Clean Rooms SQL Reference Examples with a TIMETZ column The following example table TIMETZ_TEST has a column TIMETZ_VAL (type TIMETZ) with three values inserted. select timetz_val from timetz_test; timetz_val ------------------ 04:00:00+00 00:00:00.5550+00 05:58:00+00 The following example selects only the TIMETZ values less than 3:00:00 UTC. The comparison is made after converting the value to UTC. select timetz_val from timetz_test where timetz_val < '3:00:00 UTC'; timetz_val --------------- 00:00:00.5550+00 The following example compares two TIMETZ literals. The time zone is ignored for the comparison. select time '18:25:33.123456 PST' < time '19:25:33.123456 EST'; ?column? ---------- t Logical conditions Logical conditions combine the result of two conditions to produce a single result. All logical conditions are binary operators with a Boolean return type. Syntax expression { AND \| OR } Logical conditions 372 AWS Clean Rooms expression NOT expression SQL Reference Logical conditions use a three-valued Boolean logic where the null value represents an unknown relationship. The following table describes the results for logical conditions, where E1 and E2 represent expressions: E1 TRUE TRUE TRUE FALSE FALSE FALSE E1 AND E2 E1 OR E2 NOT E2 E2 TRUE FALSE TRUE FALSE UNKNOWN UNKNOWN TRUE FALSE FALSE FALSE FALSE TRUE UNKNOWN TRUE TRUE TRUE TRUE FALSE UNKNOWN FALSE UNKNOWN UNKNOWN TRUE UNKNOWN TRUE UNKNOWN FALSE FALSE UNKNOWN UNKNOWN UNKNOWN UNKNOWN UNKNOWN The NOT operator is evaluated before AND, and the AND operator is evaluated before the OR operator. Any parentheses used may override this default order of evaluation. Examples The following example returns USERID and USERNAME from the USERS table where the user likes both Las Vegas and sports: select userid, username from users where likevegas = 1 and likesports = 1 order by userid; userid \| username Logical conditions 373 SQL Reference AWS Clean Rooms --------+---------- 1 \| JSG99FHE 67 \|
sql-reference-102	sql-reference.pdf	102	TRUE TRUE TRUE FALSE UNKNOWN FALSE UNKNOWN UNKNOWN TRUE UNKNOWN TRUE UNKNOWN FALSE FALSE UNKNOWN UNKNOWN UNKNOWN UNKNOWN UNKNOWN The NOT operator is evaluated before AND, and the AND operator is evaluated before the OR operator. Any parentheses used may override this default order of evaluation. Examples The following example returns USERID and USERNAME from the USERS table where the user likes both Las Vegas and sports: select userid, username from users where likevegas = 1 and likesports = 1 order by userid; userid \| username Logical conditions 373 SQL Reference AWS Clean Rooms --------+---------- 1 \| JSG99FHE 67 \| TWU10MZT 87 \| DUF19VXU 92 \| HYP36WEQ 109 \| FPL38HZK 120 \| DMJ24GUZ 123 \| QZR22XGQ 130 \| ZQC82ALK 133 \| LBN45WCH 144 \| UCX04JKN 165 \| TEY68OEB 169 \| AYQ83HGO 184 \| TVX65AZX ... (2128 rows) The next example returns the USERID and USERNAME from the USERS table where the user likes Las Vegas, or sports, or both. This query returns all of the output from the previous example plus the users who like only Las Vegas or sports. select userid, username from users where likevegas = 1 or likesports = 1 order by userid; userid \| username --------+---------- 1 \| JSG99FHE 2 \| PGL08LJI 3 \| IFT66TXU 5 \| AEB55QTM 6 \| NDQ15VBM 9 \| MSD36KVR 10 \| WKW41AIW 13 \| QTF33MCG 15 \| OWU78MTR 16 \| ZMG93CDD 22 \| RHT62AGI 27 \| KOY02CVE 29 \| HUH27PKK ... (18968 rows) Logical conditions 374 AWS Clean Rooms SQL Reference The following query uses parentheses around the OR condition to ﬁnd venues in New York or California where Macbeth was performed: select distinct venuename, venuecity from venue join event on venue.venueid=event.venueid where (venuestate = 'NY' or venuestate = 'CA') and eventname='Macbeth' order by 2,1; venuename \| venuecity ----------------------------------------+--------------- Geffen Playhouse \| Los Angeles Greek Theatre \| Los Angeles Royce Hall \| Los Angeles American Airlines Theatre \| New York City August Wilson Theatre \| New York City Belasco Theatre \| New York City Bernard B. Jacobs Theatre \| New York City ... Removing the parentheses in this example changes the logic and results of the query. The following example uses the NOT operator: select * from category where not catid=1 order by 1; catid \| catgroup \| catname \| catdesc -------+----------+-----------+-------------------------------------------- 2 \| Sports \| NHL \| National Hockey League 3 \| Sports \| NFL \| National Football League 4 \| Sports \| NBA \| National Basketball Association 5 \| Sports \| MLS \| Major League Soccer ... The following example uses a NOT condition followed by an AND condition: select * from category where (not catid=1) and catgroup='Sports' order by catid; catid \| catgroup \| catname \| catdesc Logical conditions 375 AWS Clean Rooms SQL Reference -------+----------+---------+--------------------------------- 2 \| Sports \| NHL \| National Hockey League 3 \| Sports \| NFL \| National Football League 4 \| Sports \| NBA \| National Basketball Association 5 \| Sports \| MLS \| Major League Soccer (4 rows) Pattern-matching conditions A pattern-matching operator searches a string for a pattern speciﬁed in the conditional expression and returns true or false depending on whether it ﬁnds a match. AWS Clean Rooms Spark SQL uses the following methods for pattern matching: • LIKE expressions The LIKE operator compares a string expression, such as a column name, with a pattern that uses the wildcard characters % (percent) and _ (underscore). LIKE pattern matching always covers the entire string. LIKE performs a case-sensitive match and ILIKE performs a case-insensitive match. • SIMILAR TO regular expressions The SIMILAR TO operator matches a string expression with a SQL standard regular expression pattern, which can include a set of pattern-matching metacharacters that includes the two supported by the LIKE operator. SIMILAR TO matches the entire string and performs a case- sensitive match. Topics • LIKE • RLIKE • SIMILAR TO LIKE The LIKE operator compares a string expression, such as a column name, with a pattern that uses the wildcard characters % (percent) and _ (underscore). LIKE pattern matching always covers the entire string. To match a sequence anywhere within a string, the pattern must start and end with a percent sign. LIKE is case-sensitive; ILIKE is case-insensitive. Pattern-matching conditions 376 AWS Clean Rooms Syntax SQL Reference expression [ NOT ] LIKE \| ILIKE pattern [ ESCAPE 'escape_char' ] Arguments expression A valid UTF-8 character expression, such as a column name. LIKE \| ILIKE LIKE performs a case-sensitive pattern match. ILIKE performs a case-insensitive pattern match for single-byte UTF-8 (ASCII) characters. To perform a case-insensitive pattern match for multibyte characters, use the LOWER function on expression and pattern with a LIKE condition. In contrast to comparison predicates, such as = and <>, LIKE and ILIKE predicates do not implicitly ignore trailing spaces. To ignore trailing spaces, use RTRIM or explicitly
sql-reference-103	sql-reference.pdf	103	376 AWS Clean Rooms Syntax SQL Reference expression [ NOT ] LIKE \| ILIKE pattern [ ESCAPE 'escape_char' ] Arguments expression A valid UTF-8 character expression, such as a column name. LIKE \| ILIKE LIKE performs a case-sensitive pattern match. ILIKE performs a case-insensitive pattern match for single-byte UTF-8 (ASCII) characters. To perform a case-insensitive pattern match for multibyte characters, use the LOWER function on expression and pattern with a LIKE condition. In contrast to comparison predicates, such as = and <>, LIKE and ILIKE predicates do not implicitly ignore trailing spaces. To ignore trailing spaces, use RTRIM or explicitly cast a CHAR column to VARCHAR. The ~~ operator is equivalent to LIKE, and ~~* is equivalent to ILIKE. Also the !~~ and !~~* operators are equivalent to NOT LIKE and NOT ILIKE. pattern A valid UTF-8 character expression with the pattern to be matched. escape_char A character expression that will escape metacharacters characters in the pattern. The default is two backslashes ('\\'). If pattern does not contain metacharacters, then the pattern only represents the string itself; in that case LIKE acts the same as the equals operator. Either of the character expressions can be CHAR or VARCHAR data types. If they diﬀer, AWS Clean Rooms converts pattern to the data type of expression. LIKE supports the following pattern-matching metacharacters: Operator Description % Matches any sequence of zero or more characters. Pattern-matching conditions 377 AWS Clean Rooms SQL Reference Operator Description _ Matches any single character. Examples The following table shows examples of pattern matching using LIKE: Expression Returns 'abc' LIKE 'abc' 'abc' LIKE 'a%' 'abc' LIKE '_B_' 'abc' ILIKE '_B_' 'abc' LIKE 'c%' True True False True False The following example ﬁnds all cities whose names start with "E": select distinct city from users where city like 'E%' order by city; city --------------- East Hartford East Lansing East Rutherford East St. Louis Easthampton Easton Eatontown Eau Claire ... The following example ﬁnds users whose last name contains "ten" : select distinct lastname from users Pattern-matching conditions 378 AWS Clean Rooms SQL Reference where lastname like '%ten%' order by lastname; lastname ------------- Christensen Wooten ... The following example ﬁnds cities whose third and fourth characters are "ea". The command uses ILIKE to demonstrate case insensitivity: select distinct city from users where city ilike '__EA%' order by city; city ------------- Brea Clearwater Great Falls Ocean City Olean Wheaton (6 rows) The following example uses the default escape string (\\) to search for strings that include "start_" (the text start followed by an underscore _): select tablename, "column" from my_table_def where "column" like '%start\\_%' limit 5; tablename \| column -------------------+--------------- my_s3client \| start_time my_tr_conflict \| xact_start_ts my_undone \| undo_start_ts my_unload_log \| start_time my_vacuum_detail \| start_row (5 rows) The following example speciﬁes '^' as the escape character, then uses the escape character to search for strings that include "start_" (the text start followed by an underscore _): Pattern-matching conditions 379 AWS Clean Rooms SQL Reference select tablename, "column" from my_table_def where "column" like '%start^_%' escape '^' limit 5; tablename \| column -------------------+--------------- my_s3client \| start_time my_tr_conflict \| xact_start_ts my_undone \| undo_start_ts my_unload_log \| start_time my_vacuum_detail \| start_row (5 rows) The following example uses the ~~* operator to do a case-insensitive (ILIKE) search for cities that start with "Ag". select distinct city from users where city ~~* 'Ag%' order by city; city ------------ Agat Agawam Agoura Hills Aguadilla RLIKE The RLIKE operator allows you to check if a string matches a speciﬁed regular expression pattern. Returns true if str matches regexp, or false otherwise. Syntax rlike(str, regexp) Arguments str A string expression Pattern-matching conditions 380 AWS Clean Rooms regexp SQL Reference A string expression. The regex string should be a Java regular expression. String literals (including regex patterns) are unescaped in our SQL parser. For example, to match "\abc", a regular expression for regexp can be "^\abc$". Examples The following example sets the value of the spark.sql.parser.escapedStringLiterals conﬁguration parameter to true. This parameter is speciﬁc to the Spark SQL engine. The spark.sql.parser.escapedStringLiterals parameter in Spark SQL controls how the SQL parser handles escaped string literals. When set to true, the parser will interpret backslash characters (\) within string literals as escape characters, allowing you to include special characters like newlines, tabs, and quotation marks within your string values. SET spark.sql.parser.escapedStringLiterals=true; spark.sql.parser.escapedStringLiterals true For example, with spark.sql.parser.escapedStringLiterals=true, you could use the following string literal in your SQL query: SELECT 'Hello, world!\n' The newline character \n would be interpreted as a literal newline character in the output. The following example performs a regular expression pattern match. The ﬁrst argument is passed to the RLIKE operator. It's a string that represents a ﬁle path, where the actual username is replaced with the pattern '****'. The second argument is the regular expression pattern used for the matching. The
sql-reference-104	sql-reference.pdf	104	to include special characters like newlines, tabs, and quotation marks within your string values. SET spark.sql.parser.escapedStringLiterals=true; spark.sql.parser.escapedStringLiterals true For example, with spark.sql.parser.escapedStringLiterals=true, you could use the following string literal in your SQL query: SELECT 'Hello, world!\n' The newline character \n would be interpreted as a literal newline character in the output. The following example performs a regular expression pattern match. The ﬁrst argument is passed to the RLIKE operator. It's a string that represents a ﬁle path, where the actual username is replaced with the pattern '**'. The second argument is the regular expression pattern used for the matching. The output (true) indicates that the ﬁrst string ('%SystemDrive%\Users\*') matches the regular expression pattern ('%SystemDrive%\\Users.'). SELECT rlike('%SystemDrive%\Users\John', '%SystemDrive%\Users.'); true Pattern-matching conditions 381 AWS Clean Rooms SIMILAR TO SQL Reference The SIMILAR TO operator matches a string expression, such as a column name, with a SQL standard regular expression pattern. A SQL regular expression pattern can include a set of pattern-matching metacharacters, including the two supported by the LIKE operator. The SIMILAR TO operator returns true only if its pattern matches the entire string, unlike POSIX regular expression behavior, where the pattern can match any portion of the string. SIMILAR TO performs a case-sensitive match. Note Regular expression matching using SIMILAR TO is computationally expensive. We recommend using LIKE whenever possible, especially when processing a very large number of rows. For example, the following queries are functionally identical, but the query that uses LIKE runs several times faster than the query that uses a regular expression: select count() from event where eventname SIMILAR TO '%(Ring\|Die)%'; select count() from event where eventname LIKE '%Ring%' OR eventname LIKE '%Die %'; Syntax expression [ NOT ] SIMILAR TO pattern [ ESCAPE 'escape_char' ] Arguments expression A valid UTF-8 character expression, such as a column name. SIMILAR TO SIMILAR TO performs a case-sensitive pattern match for the entire string in expression. pattern A valid UTF-8 character expression representing a SQL standard regular expression pattern. Pattern-matching conditions 382 AWS Clean Rooms escape_char SQL Reference A character expression that will escape metacharacters in the pattern. The default is two backslashes ('\\'). If pattern does not contain metacharacters, then the pattern only represents the string itself. Either of the character expressions can be CHAR or VARCHAR data types. If they diﬀer, AWS Clean Rooms converts pattern to the data type of expression. SIMILAR TO supports the following pattern-matching metacharacters: Operator Description Matches any sequence of zero or more characters. Matches any single character. Denotes alternation (either of two alternatives). Repeat the previous item zero or more times. Repeat the previous item one or more times. Repeat the previous item zero or one time. Repeat the previous item exactly m times. Repeat the previous item m or more times. Repeat the previous item at least m and not more than n times. Parentheses group items into a single logical item. A bracket expression speciﬁes a character class, just as in POSIX regular expressions. % _ \| + ? {m} {m,} {m,n} () [...] Examples The following table shows examples of pattern matching using SIMILAR TO: Pattern-matching conditions 383 SQL Reference AWS Clean Rooms Expression 'abc' SIMILAR TO 'abc' 'abc' SIMILAR TO '_b_' 'abc' SIMILAR TO '_A_' 'abc' SIMILAR TO '%(b\|d)%' 'abc' SIMILAR TO '(b\|c)%' Returns True True False True False 'AbcAbcdefgefg12efgefg12' SIMILAR True TO '((Ab)?c)+d((efg)+(12))+' 'aaaaaab11111xy' SIMILAR TO 'a{6}_ True [0-9]{5}(x\|y){2}' '$0.87' SIMILAR TO '$[0-9]+(.[0-9] True [0-9])?' The following example ﬁnds cities whose names contain "E" or "H": SELECT DISTINCT city FROM users WHERE city SIMILAR TO '%E%\|%H%' ORDER BY city LIMIT 5; city ----------------- Agoura Hills Auburn Hills Benton Harbor Beverly Hills Chicago Heights The following example uses the default escape string ('\\') to search for strings that include "_": SELECT tablename, "column" FROM my_table_def WHERE "column" SIMILAR TO '%start\\_%' ORDER BY tablename, "column" LIMIT 5; Pattern-matching conditions 384 AWS Clean Rooms SQL Reference tablename \| column --------------------------+--------------------- my_abort_idle \| idle_start_time my_abort_idle \| txn_start_time my_analyze_compression \| start_time my_auto_worker_levels \| start_level my_auto_worker_levels \| start_wlm_occupancy The following example speciﬁes '^' as the escape string, then uses the escape string to search for strings that include "_": SELECT tablename, "column" FROM my_table_def WHERE "column" SIMILAR TO '%start^_%' ESCAPE '^' ORDER BY tablename, "column" LIMIT 5; tablename \| column --------------------------+--------------------- stcs_abort_idle \| idle_start_time stcs_abort_idle \| txn_start_time stcs_analyze_compression \| start_time stcs_auto_worker_levels \| start_level stcs_auto_worker_levels \| start_wlm_occupancy BETWEEN range condition A BETWEEN condition tests expressions for inclusion in a range of values, using the keywords BETWEEN and AND. Syntax expression [ NOT ] BETWEEN expression AND expression Expressions can be numeric, character, or datetime data types, but they must be compatible. The range is inclusive. Examples The ﬁrst example counts how many transactions registered sales of either 2, 3, or 4 tickets: BETWEEN range condition 385 AWS Clean Rooms SQL Reference select count(*) from
sql-reference-105	sql-reference.pdf	105	tablename, "column" LIMIT 5; tablename \| column --------------------------+--------------------- stcs_abort_idle \| idle_start_time stcs_abort_idle \| txn_start_time stcs_analyze_compression \| start_time stcs_auto_worker_levels \| start_level stcs_auto_worker_levels \| start_wlm_occupancy BETWEEN range condition A BETWEEN condition tests expressions for inclusion in a range of values, using the keywords BETWEEN and AND. Syntax expression [ NOT ] BETWEEN expression AND expression Expressions can be numeric, character, or datetime data types, but they must be compatible. The range is inclusive. Examples The ﬁrst example counts how many transactions registered sales of either 2, 3, or 4 tickets: BETWEEN range condition 385 AWS Clean Rooms SQL Reference select count() from sales where qtysold between 2 and 4; count -------- 104021 (1 row) The range condition includes the begin and end values. select min(dateid), max(dateid) from sales where dateid between 1900 and 1910; min \| max -----+----- 1900 \| 1910 The ﬁrst expression in a range condition must be the lesser value and the second expression the greater value. The following example will always return zero rows due to the values of the expressions: select count() from sales where qtysold between 4 and 2; count ------- 0 (1 row) However, applying the NOT modiﬁer will invert the logic and produce a count of all rows: select count() from sales where qtysold not between 4 and 2; count -------- 172456 (1 row) The following query returns a list of venues with 20000 to 50000 seats: BETWEEN range condition 386 AWS Clean Rooms SQL Reference select venueid, venuename, venueseats from venue where venueseats between 20000 and 50000 order by venueseats desc; venueid \| venuename \| venueseats ---------+-------------------------------+------------ 116 \| Busch Stadium \| 49660 106 \| Rangers BallPark in Arlington \| 49115 96 \| Oriole Park at Camden Yards \| 48876 ... (22 rows) The following example demonstrates using BETWEEN for date values: select salesid, qtysold, pricepaid, commission, saletime from sales where eventid between 1000 and 2000 and saletime between '2008-01-01' and '2008-01-03' order by saletime asc; salesid \| qtysold \| pricepaid \| commission \| saletime --------+---------+-----------+------------+--------------- 65082 \| 4 \| 472 \| 70.8 \| 1/1/2008 06:06 110917 \| 1 \| 337 \| 50.55 \| 1/1/2008 07:05 112103 \| 1 \| 241 \| 36.15 \| 1/2/2008 03:15 137882 \| 3 \| 1473 \| 220.95 \| 1/2/2008 05:18 40331 \| 2 \| 58 \| 8.7 \| 1/2/2008 05:57 110918 \| 3 \| 1011 \| 151.65 \| 1/2/2008 07:17 96274 \| 1 \| 104 \| 15.6 \| 1/2/2008 07:18 150499 \| 3 \| 135 \| 20.25 \| 1/2/2008 07:20 68413 \| 2 \| 158 \| 23.7 \| 1/2/2008 08:12 Note that although BETWEEN's range is inclusive, dates default to having a time value of 00:00:00. The only valid January 3 row for the sample query would be a row with a saletime of 1/3/2008 00:00:00. Null condition The NULL condition tests for nulls, when a value is missing or unknown. Null condition 387 SQL Reference AWS Clean Rooms Syntax expression IS [ NOT ] NULL Arguments expression Any expression such as a column. IS NULL Is true when the expression's value is null and false when it has a value. IS NOT NULL Is false when the expression's value is null and true when it has a value. Example This example indicates how many times the SALES table contains null in the QTYSOLD ﬁeld: select count() from sales where qtysold is null; count ------- 0 (1 row) EXISTS condition EXISTS conditions test for the existence of rows in a subquery, and return true if a subquery returns at least one row. If NOT is speciﬁed, the condition returns true if a subquery returns no rows. Syntax [ NOT ] EXISTS (table_subquery) EXISTS condition 388 SQL Reference AWS Clean Rooms Arguments EXISTS Is true when the table_subquery returns at least one row. NOT EXISTS Is true when the table_subquery returns no rows. table_subquery A subquery that evaluates to a table with one or more columns and one or more rows. Example This example returns all date identiﬁers, one time each, for each date that had a sale of any kind: select dateid from date where exists ( select 1 from sales where date.dateid = sales.dateid ) order by dateid; dateid -------- 1827 1828 1829 ... IN condition An IN condition tests a value for membership in a set of values or in a subquery. Syntax expression [ NOT ] IN (expr_list \| table_subquery) IN condition 389 AWS Clean Rooms Arguments expression SQL Reference A numeric, character, or datetime expression that is evaluated against the expr_list or table_subquery and must be compatible with the data type of that list or subquery. expr_list One or more comma-delimited expressions, or one or more sets of comma-delimited expressions bounded by parentheses. table_subquery A subquery that evaluates to a table with one or more rows, but is limited to only one column
sql-reference-106	sql-reference.pdf	106	An IN condition tests a value for membership in a set of values or in a subquery. Syntax expression [ NOT ] IN (expr_list \| table_subquery) IN condition 389 AWS Clean Rooms Arguments expression SQL Reference A numeric, character, or datetime expression that is evaluated against the expr_list or table_subquery and must be compatible with the data type of that list or subquery. expr_list One or more comma-delimited expressions, or one or more sets of comma-delimited expressions bounded by parentheses. table_subquery A subquery that evaluates to a table with one or more rows, but is limited to only one column in its select list. IN \| NOT IN IN returns true if the expression is a member of the expression list or query. NOT IN returns true if the expression is not a member. IN and NOT IN return NULL and no rows are returned in the following cases: If expression yields null; or if there are no matching expr_list or table_subquery values and at least one of these comparison rows yields null. Examples The following conditions are true only for those values listed: qtysold in (2, 4, 5) date.day in ('Mon', 'Tues') date.month not in ('Oct', 'Nov', 'Dec') Optimization for Large IN Lists To optimize query performance, an IN list that includes more than 10 values is internally evaluated as a scalar array. IN lists with fewer than 10 values are evaluated as a series of OR predicates. This optimization is supported for SMALLINT, INTEGER, BIGINT, REAL, DOUBLE PRECISION, BOOLEAN, CHAR, VARCHAR, DATE, TIMESTAMP, and TIMESTAMPTZ data types. Look at the EXPLAIN output for the query to see the eﬀect of this optimization. For example: explain select * from sales IN condition 390 AWS Clean Rooms QUERY PLAN -------------------------------------------------------------------- XN Seq Scan on sales (cost=0.00..6035.96 rows=86228 width=53) Filter: (salesid = ANY ('{1,2,3,4,5,6,7,8,9,10,11}'::integer[])) (2 rows) SQL Reference IN condition 391 AWS Clean Rooms SQL Reference AWS Clean Rooms SQL AWS Clean Rooms SQL enforces rules regarding the use of data types, expressions, and literals. For more information about AWS Clean Rooms, see the AWS Clean Rooms User Guide and the AWS Clean Rooms API Reference. The following topics provide general information about the literals, data types, commands, functions, and conditions supported in AWS Clean Rooms SQL: Topics • Literals • Data types • AWS Clean Rooms SQL commands • AWS Clean Rooms SQL functions • AWS Clean Rooms SQL conditions Literals A literal or constant is a ﬁxed data value, composed of a sequence of characters or a numeric constant. AWS Clean Rooms supports several types of literals, including: • Numeric literals for integer, decimal, and ﬂoating-point numbers. • Date, time, and timestamp literals, used with datetime data types. For more information, see Date, time, and timestamp literals. • Interval literals. For more information, see Interval literals. • Null literals, used to specify a null value. • Only TAB, CARRIAGE RETURN (CR), and LINE FEED (LF) Unicode control characters from the Unicode general category (Cc) are supported. AWS Clean Rooms doesn't support direct references to string literals in the SELECT clause, but they can be used within functions such as CAST. Literals 392 AWS Clean Rooms SQL Reference + (Concatenation) operator Concatenates numeric literals, string literals, and/or datetime and interval literals. They are on either side of the + symbol and return diﬀerent types based on the inputs on either side of the + symbol. Syntax numeric + string date + time date + timetz The order of the arguments can be reversed. Arguments numeric literals Literals or constants that represent numbers can be integer or ﬂoating-point. string literals Strings, character strings, or character constants date A DATE column or an expression that implicitly converts to a DATE. time A TIME column or an expression that implicitly converts to a TIME. timetz A TIMETZ column or an expression that implicitly converts to a TIMETZ. Example The following example table TIME_TEST has a column TIME_VAL (type TIME) with three values inserted. + (Concatenation) operator 393 AWS Clean Rooms SQL Reference select date '2000-01-02' + time_val as ts from time_test; Data types Each value that AWS Clean Rooms stores or retrieves has a data type with a ﬁxed set of associated properties. Data types are declared when tables are created. A data type constrains the set of values that a column or argument can contain. The following table lists the data types that you can use in AWS Clean Rooms SQL. Data type name Data type Aliases Description ARRAY BIGINT BOOLEAN CHAR DATE DECIMAL DOUBLE PRECISION INTEGER MAP Data types the section called “Nested type” the section called “Numeric types” Not applicable Not applicable the section called “Boolean type” BOOL the section called “Character types” the section called “Datetime types” the section called “Numeric types” the section called “Numeric types” CHARACTER Not
sql-reference-107	sql-reference.pdf	107	Data types are declared when tables are created. A data type constrains the set of values that a column or argument can contain. The following table lists the data types that you can use in AWS Clean Rooms SQL. Data type name Data type Aliases Description ARRAY BIGINT BOOLEAN CHAR DATE DECIMAL DOUBLE PRECISION INTEGER MAP Data types the section called “Nested type” the section called “Numeric types” Not applicable Not applicable the section called “Boolean type” BOOL the section called “Character types” the section called “Datetime types” the section called “Numeric types” the section called “Numeric types” CHARACTER Not applicable NUMERIC FLOAT8, FLOAT the section called “Numeric types” INT Array nested data type Signed eight-byte integer Logical Boolean (true/false) Fixed-length character string Calendar date (year, month, day) Exact numeric of selectable precision Double precision ﬂoating-point number Signed four-byte integer the section called “Nested type” Not applicable Map nested data type 394 AWS Clean Rooms SQL Reference Data type name Data type REAL SMALLINT STRUCT SUPER TIME TIMESTAMP TIMESTAMPTZ TIMETZ VARBYTE VARCHAR the section called “Numeric types” the section called “Numeric types” the section called “Nested type” the section called “SUPER type” Aliases FLOAT4 Not applicable Not applicable Not applicable Description Single precision ﬂoating-point number Signed two-byte integer Struct nested data type Superset data type that encompasses all scalar types of AWS Clean Rooms including complex types such as ARRAY and STRUCTS. the section called “Datetime types” the section called “Datetime types” the section called “Datetime types” the section called “Datetime types” Not applicable Time of day Not applicable Not applicable Not applicable Date and the time of day Date, the time of day, and a time zone Time of day with time zone the section called “VARBYTE type” VARBINARY, BINARY VARYING Variable-length binary value the section called “Character types” CHARACTER VARYING Variable-length character string with a user-deﬁned limit Data types 395 AWS Clean Rooms Note SQL Reference The ARRAY, STRUCT, and MAP nested data types are currently only enabled for the custom analysis rule. For more information, see Nested type. Multibyte characters The VARCHAR data type supports UTF-8 multibyte characters up to a maximum of four bytes. Five-byte or longer characters are not supported. To calculate the size of a VARCHAR column that contains multibyte characters, multiply the number of characters by the number of bytes per character. For example, if a string has four Chinese characters, and each character is three bytes long, then you will need a VARCHAR(12) column to store the string. The VARCHAR data type doesn't support the following invalid UTF-8 codepoints: 0xD800 – 0xDFFF (Byte sequences: ED A0 80 – ED BF BF) The CHAR data type doesn't support multibyte characters. Numeric types Numeric data types include integers, decimals, and ﬂoating-point numbers. Topics • Integer types • DECIMAL or NUMERIC types • Floating-point types • Computations with numeric values Integer types Use the following data types to store whole numbers of various ranges. You can't store values outside of the allowed range for each type. Name SMALLINT Multibyte characters Storage 2 bytes Range -32768 to +32767 396 AWS Clean Rooms Name INTEGER or INT BIGINT Storage 4 bytes 8 bytes SQL Reference Range -2147483648 to +2147483647 -92233720 36854775808 to 922337203 6854775807 DECIMAL or NUMERIC types Use the DECIMAL or NUMERIC data type to store values with a user-deﬁned precision. The DECIMAL and NUMERIC keywords are interchangeable. In this document, decimal is the preferred term for this data type. The term numeric is used generically to refer to integer, decimal, and ﬂoating-point data types. Name Storage Range DECIMAL or NUMERIC Variable, up to 128 bits for uncompressed DECIMAL 128-bit signed integers with up to 38 digits of precision. types. Deﬁne a DECIMAL column in a table by specifying a precision and scale: decimal(precision, scale) precision The total number of signiﬁcant digits in the whole value: the number of digits on both sides of the decimal point. For example, the number 48.2891 has a precision of 6 and a scale of 4. The default precision, if not speciﬁed, is 18. The maximum precision is 38. If the number of digits to the left of the decimal point in an input value exceeds the precision of the column minus its scale, the value can't be copied into the column (or inserted or updated). This rule applies to any value that falls outside the range of the column deﬁnition. For example, the allowed range of values for a numeric(5,2) column is -999.99 to 999.99. Numeric types 397 AWS Clean Rooms scale SQL Reference The number of decimal digits in the fractional part of the value, to the right of the decimal point. Integers have a scale of zero. In a column speciﬁcation, the scale value must be less than or equal to the precision value. The default scale,
sql-reference-108	sql-reference.pdf	108	column minus its scale, the value can't be copied into the column (or inserted or updated). This rule applies to any value that falls outside the range of the column deﬁnition. For example, the allowed range of values for a numeric(5,2) column is -999.99 to 999.99. Numeric types 397 AWS Clean Rooms scale SQL Reference The number of decimal digits in the fractional part of the value, to the right of the decimal point. Integers have a scale of zero. In a column speciﬁcation, the scale value must be less than or equal to the precision value. The default scale, if not speciﬁed, is 0. The maximum scale is 37. If the scale of an input value that is loaded into a table is greater than the scale of the column, the value is rounded to the speciﬁed scale. For example, the PRICEPAID column in the SALES table is a DECIMAL(8,2) column. If a DECIMAL(8,4) value is inserted into the PRICEPAID column, the value is rounded to a scale of 2. insert into sales values (0, 8, 1, 1, 2000, 14, 5, 4323.8951, 11.00, null); select pricepaid, salesid from sales where salesid=0; pricepaid \| salesid -----------+--------- 4323.90 \| 0 (1 row) However, results of explicit casts of values selected from tables are not rounded. Note The maximum positive value that you can insert into a DECIMAL(19,0) column is 9223372036854775807 (263 -1). The maximum negative value is -9223372036854775807. For example, an attempt to insert the value 9999999999999999999 (19 nines) will cause an overﬂow error. Regardless of the placement of the decimal point, the largest string that AWS Clean Rooms can represent as a DECIMAL number is 9223372036854775807. For example, the largest value that you can load into a DECIMAL(19,18) column is 9.223372036854775807. These rules are because of the following: • DECIMAL values with 19 or fewer signiﬁcant digits of precision are stored internally as 8- byte integers. • DECIMAL values with 20 to 38 signiﬁcant digits of precision are stored as 16-byte integers. Numeric types 398 AWS Clean Rooms SQL Reference Notes about using 128-bit DECIMAL or NUMERIC columns Don't arbitrarily assign maximum precision to DECIMAL columns unless you are certain that your application requires that precision. 128-bit values use twice as much disk space as 64-bit values and can slow down query execution time. Floating-point types Use the REAL and DOUBLE PRECISION data types to store numeric values with variable precision. These types are inexact types, meaning that some values are stored as approximations, such that storing and returning a speciﬁc value may result in slight discrepancies. If you require exact storage and calculations (such as for monetary amounts), use the DECIMAL data type. REAL represents the single-precision ﬂoating point format, according to the IEEE Standard 754 for Floating-Point Arithmetic. It has a precision of about 6 digits, and a range of around 1E-37 to 1E +37. You can also specify this data type as FLOAT4. DOUBLE PRECISION represents the double-precision ﬂoating point format, according to the IEEE Standard 754 for Binary Floating-Point Arithmetic. It has a precision of about 15 digits, and a range of around 1E-307 to 1E+308. You can also specify this data type as FLOAT or FLOAT8. Computations with numeric values In AWS Clean Rooms, computation refers to binary mathematical operations: addition, subtraction, multiplication, and division. This section describes the expected return types for these operations, as well as the speciﬁc formula that is applied to determine precision and scale when DECIMAL data types are involved. When numeric values are computed during query processing, you might encounter cases where the computation is impossible and the query returns a numeric overﬂow error. You might also encounter cases where the scale of computed values varies or is unexpected. For some operations, you can use explicit casting (type promotion) or AWS Clean Rooms conﬁguration parameters to work around these problems. For information about the results of similar computations with SQL functions, see AWS Clean Rooms SQL functions. Return types for computations Given the set of numeric data types supported in AWS Clean Rooms, the following table shows the expected return types for addition, subtraction, multiplication, and division operations. The ﬁrst Numeric types 399 AWS Clean Rooms SQL Reference column on the left side of the table represents the ﬁrst operand in the calculation, and the top row represents the second operand. Operand 1 Operand 2 Return type SMALLINT or SHORT SMALLINT or SHORT SMALLINT or SHORT INTEGER BIGINT DECIMAL FLOAT8 FLOAT8 INTEGER SMALLINT or SHORT INTEGER SMALLINT or SHORT BIGINT SMALLINT or SHORT DECIMAL FLOAT4 FLOAT8 INTEGER SMALLINT or SHORT SMALLINT or SHORT INTEGER INTEGER INTEGER INTEGER INTEGER BIGINT or LONG BIGINT or LONG DECIMAL FLOAT4 FLOAT8 DECIMAL FLOAT8 FLOAT8 BIGINT or LONG BIGINT or LONG BIGINT or LONG BIGINT or LONG DECIMAL BIGINT or LONG BIGINT or LONG

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.