Spaces:
Building
on
T4
Building
on
T4
change container width param
Browse files
pages/Multimodal_Conversational_Search.py
CHANGED
@@ -1,566 +1,232 @@
|
|
|
|
|
|
1 |
import streamlit as st
|
2 |
import uuid
|
3 |
import os
|
4 |
-
import re
|
5 |
import sys
|
6 |
-
|
7 |
-
sys.path.insert(1, "/".join(os.path.realpath(__file__).split("/")[0:-2])+"/RAG")
|
8 |
-
sys.path.insert(1, "/".join(os.path.realpath(__file__).split("/")[0:-2])+"/utilities")
|
9 |
import boto3
|
10 |
-
import requests
|
11 |
-
from boto3 import Session
|
12 |
-
import botocore.session
|
13 |
import json
|
14 |
import random
|
15 |
import string
|
16 |
-
import rag_DocumentLoader
|
17 |
-
import rag_DocumentSearcher
|
18 |
import pandas as pd
|
19 |
-
from PIL import Image
|
20 |
-
import shutil
|
21 |
-
import base64
|
22 |
-
import time
|
23 |
-
import botocore
|
24 |
-
from requests_aws4auth import AWS4Auth
|
25 |
-
import colpali
|
26 |
from requests.auth import HTTPBasicAuth
|
27 |
-
import warnings
|
28 |
|
|
|
29 |
warnings.filterwarnings("ignore", category=DeprecationWarning)
|
30 |
|
|
|
|
|
|
|
|
|
|
|
31 |
|
|
|
|
|
|
|
|
|
32 |
|
33 |
-
|
34 |
-
|
35 |
-
|
36 |
-
|
37 |
-
|
|
|
|
|
|
|
|
|
38 |
)
|
39 |
-
|
|
|
|
|
|
|
|
|
|
|
40 |
USER_ICON = "images/user.png"
|
41 |
AI_ICON = "images/opensearch-twitter-card.png"
|
42 |
REGENERATE_ICON = "images/regenerate.png"
|
43 |
-
s3_bucket_ = "pdf-repo-uploads"
|
44 |
-
#"pdf-repo-uploads"
|
45 |
-
polly_client = boto3.client('polly',aws_access_key_id=st.secrets['user_access_key'],
|
46 |
-
aws_secret_access_key=st.secrets['user_secret_key'], region_name = 'us-east-1')
|
47 |
-
|
48 |
-
|
49 |
-
# Check if the user ID is already stored in the session state
|
50 |
-
if 'user_id' in st.session_state:
|
51 |
-
user_id = st.session_state['user_id']
|
52 |
-
#print(f"User ID: {user_id}")
|
53 |
-
|
54 |
-
# If the user ID is not yet stored in the session state, generate a random UUID
|
55 |
-
else:
|
56 |
-
user_id = str(uuid.uuid4())
|
57 |
-
st.session_state['user_id'] = user_id
|
58 |
-
|
59 |
-
|
60 |
-
if 'session_id' not in st.session_state:
|
61 |
-
st.session_state['session_id'] = ""
|
62 |
-
|
63 |
-
if "chats" not in st.session_state:
|
64 |
-
st.session_state.chats = [
|
65 |
-
{
|
66 |
-
'id': 0,
|
67 |
-
'question': '',
|
68 |
-
'answer': ''
|
69 |
-
}
|
70 |
-
]
|
71 |
-
|
72 |
-
if "questions_" not in st.session_state:
|
73 |
-
st.session_state.questions_ = []
|
74 |
-
|
75 |
-
if "show_columns" not in st.session_state:
|
76 |
-
st.session_state.show_columns = False
|
77 |
-
|
78 |
-
if "answers_" not in st.session_state:
|
79 |
-
st.session_state.answers_ = []
|
80 |
-
|
81 |
-
if "input_index" not in st.session_state:
|
82 |
-
st.session_state.input_index = "hpijan2024hometrack"#"globalwarmingnew"#"hpijan2024hometrack_no_img_no_table"
|
83 |
-
|
84 |
-
if "input_is_rerank" not in st.session_state:
|
85 |
-
st.session_state.input_is_rerank = True
|
86 |
-
|
87 |
-
if "input_is_colpali" not in st.session_state:
|
88 |
-
st.session_state.input_is_colpali = False
|
89 |
-
|
90 |
-
if "input_copali_rerank" not in st.session_state:
|
91 |
-
st.session_state.input_copali_rerank = False
|
92 |
-
|
93 |
-
if "input_table_with_sql" not in st.session_state:
|
94 |
-
st.session_state.input_table_with_sql = False
|
95 |
-
|
96 |
-
if "input_query" not in st.session_state:
|
97 |
-
st.session_state.input_query="which city has the highest average housing price in UK ?"#"What is the projected energy percentage from renewable sources in future?"#"Which city in United Kingdom has the highest average housing price ?"#"How many aged above 85 years died due to covid ?"# What is the projected energy from renewable sources ?"
|
98 |
-
|
99 |
-
if "input_rag_searchType" not in st.session_state:
|
100 |
-
st.session_state.input_rag_searchType = ["Vector Search"]
|
101 |
-
|
102 |
-
|
103 |
-
|
104 |
-
region = 'us-east-1'
|
105 |
-
bedrock_runtime_client = boto3.client('bedrock-runtime',region_name=region)
|
106 |
-
output = []
|
107 |
-
service = 'es'
|
108 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
109 |
st.markdown("""
|
110 |
<style>
|
111 |
-
[data-testid=column]:nth-of-type(
|
112 |
-
|
113 |
-
}
|
114 |
-
[data-testid=column]:nth-of-type(1) [data-testid=stVerticalBlock]{
|
115 |
gap: 0rem;
|
116 |
}
|
117 |
</style>
|
118 |
-
""",unsafe_allow_html=True)
|
119 |
-
|
120 |
-
|
121 |
-
|
122 |
-
|
123 |
-
credentials = boto3.Session().get_credentials()
|
124 |
-
awsauth = HTTPBasicAuth('master',st.secrets['ml_search_demo_api_access'])
|
125 |
-
service = 'es'
|
126 |
-
|
127 |
-
|
128 |
-
|
129 |
-
|
130 |
-
# if "input_searchType" not in st.session_state:
|
131 |
-
# st.session_state.input_searchType = "Conversational Search (RAG)"
|
132 |
-
|
133 |
-
# if "input_temperature" not in st.session_state:
|
134 |
-
# st.session_state.input_temperature = "0.001"
|
135 |
-
|
136 |
-
# if "input_topK" not in st.session_state:
|
137 |
-
# st.session_state.input_topK = 200
|
138 |
-
|
139 |
-
# if "input_topP" not in st.session_state:
|
140 |
-
# st.session_state.input_topP = 0.95
|
141 |
-
|
142 |
-
# if "input_maxTokens" not in st.session_state:
|
143 |
-
# st.session_state.input_maxTokens = 1024
|
144 |
-
|
145 |
-
|
146 |
-
def write_logo():
|
147 |
-
col1, col2, col3 = st.columns([5, 1, 5])
|
148 |
-
with col2:
|
149 |
-
st.image(AI_ICON, use_container_width='always')
|
150 |
|
|
|
151 |
def write_top_bar():
|
152 |
-
col1, col2 = st.columns([77,23])
|
153 |
with col1:
|
154 |
-
st.
|
155 |
-
st.header("Chat with your data",divider='rainbow')
|
156 |
-
|
157 |
-
#st.image(AI_ICON, use_container_width='always')
|
158 |
-
|
159 |
with col2:
|
160 |
-
st.write("")
|
161 |
-
st.write("")
|
162 |
clear = st.button("Clear")
|
163 |
-
st.write("")
|
164 |
-
st.write("")
|
165 |
return clear
|
166 |
|
167 |
-
|
168 |
-
|
169 |
-
if clear:
|
170 |
st.session_state.questions_ = []
|
171 |
st.session_state.answers_ = []
|
172 |
-
st.session_state.input_query=""
|
173 |
-
# st.session_state.input_searchType="Conversational Search (RAG)"
|
174 |
-
# st.session_state.input_temperature = "0.001"
|
175 |
-
# st.session_state.input_topK = 200
|
176 |
-
# st.session_state.input_topP = 0.95
|
177 |
-
# st.session_state.input_maxTokens = 1024
|
178 |
-
|
179 |
|
|
|
180 |
def handle_input():
|
181 |
-
|
182 |
-
|
183 |
-
|
184 |
-
|
185 |
-
return ""
|
186 |
-
inputs = {}
|
187 |
-
for key in st.session_state:
|
188 |
-
if key.startswith('input_'):
|
189 |
-
inputs[key.removeprefix('input_')] = st.session_state[key]
|
190 |
st.session_state.inputs_ = inputs
|
191 |
-
|
192 |
-
|
193 |
-
|
194 |
-
|
195 |
-
#st.write(inputs)
|
196 |
-
question_with_id = {
|
197 |
'question': inputs["query"],
|
198 |
'id': len(st.session_state.questions_)
|
199 |
-
}
|
200 |
-
|
201 |
-
if
|
202 |
out_ = colpali.colpali_search_rerank(st.session_state.input_query)
|
203 |
-
#print(out_)
|
204 |
else:
|
205 |
-
out_ = rag_DocumentSearcher.query_(
|
|
|
|
|
|
|
|
|
|
|
|
|
206 |
st.session_state.answers_.append({
|
207 |
'answer': out_['text'],
|
208 |
-
'source':out_['source'],
|
209 |
'id': len(st.session_state.questions_),
|
210 |
'image': out_['image'],
|
211 |
-
'table':out_['table']
|
212 |
})
|
213 |
-
st.session_state.input_query=""
|
214 |
-
|
215 |
-
|
216 |
-
|
217 |
-
# search_type = st.selectbox('Select the Search type',
|
218 |
-
# ('Conversational Search (RAG)',
|
219 |
-
# 'OpenSearch vector search',
|
220 |
-
# 'LLM Text Generation'
|
221 |
-
# ),
|
222 |
-
|
223 |
-
# key = 'input_searchType',
|
224 |
-
# help = "Select the type of retriever\n1. Conversational Search (Recommended) - This will include both the OpenSearch and LLM in the retrieval pipeline \n (note: This will put opensearch response as context to LLM to answer) \n2. OpenSearch vector search - This will put only OpenSearch's vector search in the pipeline, \n(Warning: this will lead to unformatted results )\n3. LLM Text Generation - This will include only LLM in the pipeline, \n(Warning: This will give hallucinated and out of context answers_)"
|
225 |
-
# )
|
226 |
-
|
227 |
-
# col1, col2, col3, col4 = st.columns(4)
|
228 |
-
|
229 |
-
# with col1:
|
230 |
-
# st.text_input('Temperature', value = "0.001", placeholder='LLM Temperature', key = 'input_temperature',help = "Set the temperature of the Large Language model. \n Note: 1. Set this to values lower to 1 in the order of 0.001, 0.0001, such low values reduces hallucination and creativity in the LLM response; 2. This applies only when LLM is a part of the retriever pipeline")
|
231 |
-
# with col2:
|
232 |
-
# st.number_input('Top K', value = 200, placeholder='Top K', key = 'input_topK', step = 50, help = "This limits the LLM's predictions to the top k most probable tokens at each step of generation, this applies only when LLM is a prt of the retriever pipeline")
|
233 |
-
# with col3:
|
234 |
-
# st.number_input('Top P', value = 0.95, placeholder='Top P', key = 'input_topP', step = 0.05, help = "This sets a threshold probability and selects the top tokens whose cumulative probability exceeds the threshold while the tokens are generated by the LLM")
|
235 |
-
# with col4:
|
236 |
-
# st.number_input('Max Output Tokens', value = 500, placeholder='Max Output Tokens', key = 'input_maxTokens', step = 100, help = "This decides the total number of tokens generated as the final response. Note: Values greater than 1000 takes longer response time")
|
237 |
-
|
238 |
-
# st.markdown('---')
|
239 |
-
|
240 |
|
241 |
-
|
242 |
-
|
243 |
-
|
244 |
with col1:
|
245 |
-
st.image(USER_ICON, use_container_width=
|
246 |
with col2:
|
247 |
-
|
248 |
-
|
249 |
-
|
250 |
-
|
251 |
|
|
|
|
|
|
|
252 |
|
253 |
-
def render_answer(question,answer,index,res_img):
|
254 |
-
|
255 |
-
|
256 |
-
col1, col2, col_3 = st.columns([4,74,22])
|
257 |
with col1:
|
258 |
-
st.image(AI_ICON, use_container_width=
|
|
|
259 |
with col2:
|
260 |
-
|
261 |
-
st.write(
|
262 |
-
|
263 |
-
|
264 |
-
|
265 |
-
|
266 |
-
|
267 |
-
|
268 |
-
|
269 |
-
|
270 |
-
|
271 |
-
|
272 |
-
|
273 |
-
|
274 |
-
|
275 |
-
|
276 |
-
|
277 |
-
|
278 |
-
|
279 |
-
|
280 |
-
|
281 |
-
|
282 |
-
|
283 |
-
|
284 |
-
|
285 |
-
|
286 |
-
|
287 |
-
|
288 |
-
|
289 |
-
|
290 |
-
# if(index == len(st.session_state.questions_)):
|
291 |
-
# st.write_stream(stream_)
|
292 |
-
# if(isinstance(st.session_state.answers_[index-1]['answer'],botocore.eventstream.EventStream)):
|
293 |
-
# st.session_state.answers_[index-1]['answer'] = "".join(output)
|
294 |
-
# else:
|
295 |
-
# st.write(ans_)
|
296 |
-
|
297 |
-
|
298 |
-
polly_response = polly_client.synthesize_speech(VoiceId='Joanna',
|
299 |
-
OutputFormat='ogg_vorbis',
|
300 |
-
Text = ans_,
|
301 |
-
Engine = 'neural')
|
302 |
-
|
303 |
-
audio_col1, audio_col2 = st.columns([50,50])
|
304 |
-
with audio_col1:
|
305 |
-
st.audio(polly_response['AudioStream'].read(), format="audio/ogg")
|
306 |
-
rdn_key_1 = ''.join([random.choice(string.ascii_letters)
|
307 |
-
for _ in range(10)])
|
308 |
-
def show_maxsim():
|
309 |
-
st.session_state.show_columns = True
|
310 |
-
st.session_state.maxSimImages = colpali.img_highlight(st.session_state.top_img, st.session_state.query_token_vectors, st.session_state.query_tokens)
|
311 |
-
handle_input()
|
312 |
-
with placeholder.container():
|
313 |
-
render_all()
|
314 |
-
if(st.session_state.input_is_colpali):
|
315 |
-
st.button("Show similarity map",key=rdn_key_1,on_click = show_maxsim)
|
316 |
-
|
317 |
-
|
318 |
-
|
319 |
-
#st.markdown("<div style='font-size:18px;padding:3px 7px 3px 7px;borderWidth: 0px;borderColor: red;borderStyle: solid;border-radius: 10px;'>"+ans_+"</div>", unsafe_allow_html = True)
|
320 |
-
#st.markdown("<div style='color:#e28743';padding:3px 7px 3px 7px;borderWidth: 0px;borderColor: red;borderStyle: solid;width: fit-content;height: fit-content;border-radius: 10px;'><b>Relevant images from the document :</b></div>", unsafe_allow_html = True)
|
321 |
-
#st.write("")
|
322 |
-
colu1,colu2,colu3 = st.columns([4,82,20])
|
323 |
-
with colu2:
|
324 |
-
with st.expander("Relevant Sources:"):
|
325 |
-
with st.container():
|
326 |
-
if(len(res_img)>0):
|
327 |
-
#with st.expander("Images:"):
|
328 |
-
|
329 |
-
idx = 0
|
330 |
-
print(res_img)
|
331 |
-
for i in range(0,len(res_img)):
|
332 |
-
|
333 |
-
if(st.session_state.input_is_colpali):
|
334 |
-
if(st.session_state.show_columns == True):
|
335 |
-
cols_per_row = 3
|
336 |
-
st.session_state.image_placeholder=st.empty()
|
337 |
-
with st.session_state.image_placeholder.container():
|
338 |
-
row = st.columns(cols_per_row)
|
339 |
-
for j, item in enumerate(res_img[i:i+cols_per_row]):
|
340 |
-
with row[j]:
|
341 |
-
st.image(item['file'])
|
342 |
-
|
343 |
-
else:
|
344 |
-
st.session_state.image_placeholder = st.empty()
|
345 |
-
with st.session_state.image_placeholder.container():
|
346 |
-
col3_,col4_,col5_ = st.columns([33,33,33])
|
347 |
-
with col3_:
|
348 |
-
st.image(res_img[i]['file'])
|
349 |
-
|
350 |
-
|
351 |
-
|
352 |
-
|
353 |
-
|
354 |
-
else:
|
355 |
-
if(res_img[i]['file'].lower()!='none' and idx < 1):
|
356 |
-
col3,col4,col5 = st.columns([33,33,33])
|
357 |
-
cols = [col3,col4]
|
358 |
-
img = res_img[i]['file'].split(".")[0]
|
359 |
-
caption = res_img[i]['caption']
|
360 |
-
|
361 |
-
with cols[idx]:
|
362 |
-
|
363 |
-
st.image(parent_dirname+"/figures/"+st.session_state.input_index+"/"+img+".jpg")
|
364 |
-
#st.write(caption)
|
365 |
-
idx = idx+1
|
366 |
-
if(st.session_state.show_columns == True):
|
367 |
-
st.session_state.show_columns = False
|
368 |
-
#st.markdown("<div style='color:#e28743';padding:3px 7px 3px 7px;borderWidth: 0px;borderColor: red;borderStyle: solid;width: fit-content;height: fit-content;border-radius: 10px;'><b>Sources from the document:</b></div>", unsafe_allow_html = True)
|
369 |
-
if(len(answer["table"] )>0):
|
370 |
-
#with st.expander("Table:"):
|
371 |
-
df = pd.read_csv(answer["table"][0]['name'],skipinitialspace = True, on_bad_lines='skip',delimiter='`')
|
372 |
df.fillna(method='pad', inplace=True)
|
373 |
st.table(df)
|
374 |
-
|
375 |
-
|
376 |
-
|
377 |
-
|
378 |
-
|
379 |
-
|
380 |
-
|
381 |
-
|
382 |
-
for _ in range(10)])
|
383 |
-
# rdn_key_1 = ''.join([random.choice(string.ascii_letters)
|
384 |
-
# for _ in range(10)])
|
385 |
-
currentValue = ''.join(st.session_state.input_rag_searchType)+str(st.session_state.input_is_rerank)+str(st.session_state.input_table_with_sql)+st.session_state.input_index
|
386 |
-
oldValue = ''.join(st.session_state.inputs_["rag_searchType"])+str(st.session_state.inputs_["is_rerank"])+str(st.session_state.inputs_["table_with_sql"])+str(st.session_state.inputs_["index"])
|
387 |
-
def on_button_click():
|
388 |
-
if(currentValue!=oldValue or 1==1):
|
389 |
-
st.session_state.input_query = st.session_state.questions_[-1]["question"]
|
390 |
-
st.session_state.answers_.pop()
|
391 |
-
st.session_state.questions_.pop()
|
392 |
-
|
393 |
-
handle_input()
|
394 |
-
with placeholder.container():
|
395 |
-
render_all()
|
396 |
-
# def show_maxsim():
|
397 |
-
# st.session_state.show_columns = True
|
398 |
-
# st.session_state.maxSimImages = colpali.img_highlight(st.session_state.top_img, st.session_state.query_token_vectors, st.session_state.query_tokens)
|
399 |
-
# handle_input()
|
400 |
-
# with placeholder.container():
|
401 |
-
# render_all()
|
402 |
-
if("currentValue" in st.session_state):
|
403 |
-
del st.session_state["currentValue"]
|
404 |
-
|
405 |
-
try:
|
406 |
-
del regenerate
|
407 |
-
except:
|
408 |
-
pass
|
409 |
-
placeholder__ = st.empty()
|
410 |
-
placeholder__.button("🔄",key=rdn_key,on_click=on_button_click)
|
411 |
-
|
412 |
-
|
413 |
-
#Each answer will have context of the question asked in order to associate the provided feedback with the respective question
|
414 |
-
def write_chat_message(md, q,index):
|
415 |
-
if(st.session_state.show_columns):
|
416 |
-
res_img = st.session_state.maxSimImages
|
417 |
-
else:
|
418 |
-
res_img = md['image']
|
419 |
-
chat = st.container()
|
420 |
-
with chat:
|
421 |
-
render_answer(q,md,index,res_img)
|
422 |
-
|
423 |
-
def render_all():
|
424 |
-
index = 0
|
425 |
-
for (q, a) in zip(st.session_state.questions_, st.session_state.answers_):
|
426 |
-
index = index +1
|
427 |
-
|
428 |
write_user_message(q)
|
429 |
-
write_chat_message(a, q,index)
|
430 |
|
|
|
431 |
placeholder = st.empty()
|
432 |
with placeholder.container():
|
433 |
-
|
434 |
|
435 |
-
|
436 |
-
col_2, col_3 = st.columns([75,20])
|
437 |
with col_2:
|
438 |
-
|
439 |
-
input = st.text_input( "Ask here",label_visibility = "collapsed",key="input_query")
|
440 |
with col_3:
|
441 |
-
|
442 |
-
|
|
|
443 |
with st.sidebar:
|
444 |
st.page_link("app.py", label=":orange[Home]", icon="🏠")
|
445 |
st.subheader(":blue[Sample Data]")
|
446 |
-
coln_1,coln_2 = st.columns([70,30])
|
447 |
with coln_1:
|
448 |
-
|
449 |
with coln_2:
|
450 |
-
st.markdown("<p style='font-size:15px'>Preview file</p>",unsafe_allow_html=True)
|
451 |
st.write("[:eyes:](https://github.com/aws-samples/AI-search-with-amazon-opensearch-service/blob/b559f82c07dfcca973f457c0a15d6444752553ab/rag/sample_pdfs/HPI-Jan-2024-Hometrack.pdf)")
|
452 |
st.write("[:eyes:](https://github.com/aws-samples/AI-search-with-amazon-opensearch-service/blob/b559f82c07dfcca973f457c0a15d6444752553ab/rag/sample_pdfs/global_warming.pdf)")
|
453 |
st.write("[:eyes:](https://github.com/aws-samples/AI-search-with-amazon-opensearch-service/blob/b559f82c07dfcca973f457c0a15d6444752553ab/rag/sample_pdfs/covid19_ie.pdf)")
|
454 |
-
st.markdown("""
|
455 |
-
<style>
|
456 |
-
[data-testid=column]:nth-of-type(2) [data-testid=stVerticalBlock]{
|
457 |
-
gap: 0rem;
|
458 |
-
}
|
459 |
-
[data-testid=column]:nth-of-type(1) [data-testid=stVerticalBlock]{
|
460 |
-
gap: 0rem;
|
461 |
-
}
|
462 |
-
</style>
|
463 |
-
""",unsafe_allow_html=True)
|
464 |
-
with st.expander("Sample questions:"):
|
465 |
-
st.markdown("<span style = 'color:#FF9900;'>UK Housing</span> - which city has the highest average housing price in UK ?",unsafe_allow_html=True)
|
466 |
-
st.markdown("<span style = 'color:#FF9900;'>Global Warming stats</span> - What is the projected energy percentage from renewable sources in future?",unsafe_allow_html=True)
|
467 |
-
st.markdown("<span style = 'color:#FF9900;'>Covid19 impacts</span> - How many aged above 85 years died due to covid ?",unsafe_allow_html=True)
|
468 |
-
|
469 |
-
|
470 |
-
#st.subheader(":blue[Your multi-modal documents]")
|
471 |
-
# pdf_doc_ = st.file_uploader(
|
472 |
-
# "Upload your PDFs here and click on 'Process'", accept_multiple_files=False)
|
473 |
-
|
474 |
-
|
475 |
-
# pdf_docs = [pdf_doc_]
|
476 |
-
# if st.button("Process"):
|
477 |
-
# with st.spinner("Processing"):
|
478 |
-
# if os.path.isdir(parent_dirname+"/pdfs") == False:
|
479 |
-
# os.mkdir(parent_dirname+"/pdfs")
|
480 |
-
|
481 |
-
# for pdf_doc in pdf_docs:
|
482 |
-
# print(type(pdf_doc))
|
483 |
-
# pdf_doc_name = (pdf_doc.name).replace(" ","_")
|
484 |
-
# with open(os.path.join(parent_dirname+"/pdfs",pdf_doc_name),"wb") as f:
|
485 |
-
# f.write(pdf_doc.getbuffer())
|
486 |
-
|
487 |
-
# request_ = { "bucket": s3_bucket_,"key": pdf_doc_name}
|
488 |
-
# # if(st.session_state.input_copali_rerank):
|
489 |
-
# # copali.process_doc(request_)
|
490 |
-
# # else:
|
491 |
-
# rag_DocumentLoader.load_docs(request_)
|
492 |
-
# print('lambda done')
|
493 |
-
# st.success('you can start searching on your PDF')
|
494 |
-
|
495 |
-
############## haystach demo temporary addition ############
|
496 |
-
# st.subheader(":blue[Multimodality]")
|
497 |
-
# colu1,colu2 = st.columns([50,50])
|
498 |
-
# with colu1:
|
499 |
-
# in_images = st.toggle('Images', key = 'in_images', disabled = False)
|
500 |
-
# with colu2:
|
501 |
-
# in_tables = st.toggle('Tables', key = 'in_tables', disabled = False)
|
502 |
-
# if(in_tables):
|
503 |
-
# st.session_state.input_table_with_sql = True
|
504 |
-
# else:
|
505 |
-
# st.session_state.input_table_with_sql = False
|
506 |
-
|
507 |
-
############## haystach demo temporary addition ############
|
508 |
-
#if(pdf_doc_ is None or pdf_doc_ == ""):
|
509 |
-
if(index_select == "Global Warming stats"):
|
510 |
-
st.session_state.input_index = "globalwarming"
|
511 |
-
if(index_select == "Covid19 impacts on Ireland"):
|
512 |
-
st.session_state.input_index = "covid19ie"#"choosetheknnalgorithmforyourbillionscaleusecasewithopensearchawsbigdatablog"
|
513 |
-
if(index_select == "BEIR"):
|
514 |
-
st.session_state.input_index = "2104"
|
515 |
-
if(index_select == "UK Housing"):
|
516 |
-
st.session_state.input_index = "hpijan2024hometrack"
|
517 |
-
|
518 |
-
# custom_index = st.text_input("If uploaded the file already, enter the original file name", value = "")
|
519 |
-
# if(custom_index!=""):
|
520 |
-
# st.session_state.input_index = re.sub('[^A-Za-z0-9]+', '', (custom_index.lower().replace(".pdf","").split("/")[-1].split(".")[0]).lower())
|
521 |
-
|
522 |
-
|
523 |
-
|
524 |
st.subheader(":blue[Retriever]")
|
525 |
-
|
526 |
-
|
527 |
-
'Vector Search',
|
528 |
-
'Sparse Search',
|
529 |
-
],
|
530 |
-
['Vector Search'],
|
531 |
-
|
532 |
-
key = 'input_rag_searchType',
|
533 |
-
help = "Select the type of Search, adding more than one search type will activate hybrid search"#\n1. Conversational Search (Recommended) - This will include both the OpenSearch and LLM in the retrieval pipeline \n (note: This will put opensearch response as context to LLM to answer) \n2. OpenSearch vector search - This will put only OpenSearch's vector search in the pipeline, \n(Warning: this will lead to unformatted results )\n3. LLM Text Generation - This will include only LLM in the pipeline, \n(Warning: This will give hallucinated and out of context answers)"
|
534 |
-
)
|
535 |
-
|
536 |
-
re_rank = st.checkbox('Re-rank results', key = 'input_re_rank', disabled = False, value = True, help = "Checking this box will re-rank the results using a cross-encoder model")
|
537 |
-
|
538 |
-
if(re_rank):
|
539 |
-
st.session_state.input_is_rerank = True
|
540 |
-
else:
|
541 |
-
st.session_state.input_is_rerank = False
|
542 |
-
|
543 |
st.subheader(":blue[Multi-vector retrieval]")
|
544 |
-
|
545 |
-
|
546 |
-
|
547 |
-
|
548 |
-
|
|
|
|
|
|
|
549 |
st.session_state.input_is_colpali = True
|
550 |
-
#st.session_state.input_query = ""
|
551 |
else:
|
552 |
st.session_state.input_is_colpali = False
|
553 |
-
|
554 |
with st.expander("Sample questions for Colpali retriever:"):
|
555 |
-
st.write("
|
556 |
-
|
557 |
-
|
558 |
-
|
559 |
-
|
560 |
-
|
561 |
-
|
562 |
-
# else:
|
563 |
-
# st.session_state.input_copali_rerank = False
|
564 |
-
|
565 |
|
566 |
-
|
|
|
1 |
+
# Streamlit app: Chat with PDFs using OpenSearch, RAG, and ColPali
|
2 |
+
|
3 |
import streamlit as st
|
4 |
import uuid
|
5 |
import os
|
|
|
6 |
import sys
|
7 |
+
import warnings
|
|
|
|
|
8 |
import boto3
|
|
|
|
|
|
|
9 |
import json
|
10 |
import random
|
11 |
import string
|
|
|
|
|
12 |
import pandas as pd
|
13 |
+
from PIL import Image
|
|
|
|
|
|
|
|
|
|
|
|
|
14 |
from requests.auth import HTTPBasicAuth
|
|
|
15 |
|
16 |
+
# Suppress Streamlit deprecation warnings
|
17 |
warnings.filterwarnings("ignore", category=DeprecationWarning)
|
18 |
|
19 |
+
# Add necessary module paths
|
20 |
+
base_path = "/".join(os.path.realpath(__file__).split("/")[:-2])
|
21 |
+
sys.path.insert(1, f"{base_path}/semantic_search")
|
22 |
+
sys.path.insert(1, f"{base_path}/RAG")
|
23 |
+
sys.path.insert(1, f"{base_path}/utilities")
|
24 |
|
25 |
+
# Local modules
|
26 |
+
import rag_DocumentLoader
|
27 |
+
import rag_DocumentSearcher
|
28 |
+
import colpali
|
29 |
|
30 |
+
# AWS & OpenSearch setup
|
31 |
+
region = 'us-east-1'
|
32 |
+
s3_bucket_ = "pdf-repo-uploads"
|
33 |
+
bedrock_runtime_client = boto3.client('bedrock-runtime', region_name=region)
|
34 |
+
polly_client = boto3.client(
|
35 |
+
'polly',
|
36 |
+
aws_access_key_id=st.secrets['user_access_key'],
|
37 |
+
aws_secret_access_key=st.secrets['user_secret_key'],
|
38 |
+
region_name=region
|
39 |
)
|
40 |
+
credentials = boto3.Session().get_credentials()
|
41 |
+
awsauth = HTTPBasicAuth('master', st.secrets['ml_search_demo_api_access'])
|
42 |
+
|
43 |
+
# App configuration
|
44 |
+
st.set_page_config(layout="wide", page_icon="images/opensearch_mark_default.png")
|
45 |
+
parent_dirname = "/".join((os.path.dirname(__file__)).split("/")[:-1])
|
46 |
USER_ICON = "images/user.png"
|
47 |
AI_ICON = "images/opensearch-twitter-card.png"
|
48 |
REGENERATE_ICON = "images/regenerate.png"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
49 |
|
50 |
+
# Session state setup
|
51 |
+
if 'user_id' not in st.session_state:
|
52 |
+
st.session_state['user_id'] = str(uuid.uuid4())
|
53 |
+
|
54 |
+
st.session_state.setdefault('session_id', "")
|
55 |
+
st.session_state.setdefault('chats', [{'id': 0, 'question': '', 'answer': ''}])
|
56 |
+
st.session_state.setdefault('questions_', [])
|
57 |
+
st.session_state.setdefault('answers_', [])
|
58 |
+
st.session_state.setdefault('show_columns', False)
|
59 |
+
st.session_state.setdefault('input_index', "hpijan2024hometrack")
|
60 |
+
st.session_state.setdefault('input_is_rerank', True)
|
61 |
+
st.session_state.setdefault('input_is_colpali', False)
|
62 |
+
st.session_state.setdefault('input_copali_rerank', False)
|
63 |
+
st.session_state.setdefault('input_table_with_sql', False)
|
64 |
+
st.session_state.setdefault('input_query', "which city has the highest average housing price in UK ?")
|
65 |
+
st.session_state.setdefault('input_rag_searchType', ["Vector Search"])
|
66 |
+
|
67 |
+
# Custom styling
|
68 |
st.markdown("""
|
69 |
<style>
|
70 |
+
[data-testid=column]:nth-of-type(1) [data-testid=stVerticalBlock],
|
71 |
+
[data-testid=column]:nth-of-type(2) [data-testid=stVerticalBlock] {
|
|
|
|
|
72 |
gap: 0rem;
|
73 |
}
|
74 |
</style>
|
75 |
+
""", unsafe_allow_html=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
76 |
|
77 |
+
# Top bar with app logo and clear button
|
78 |
def write_top_bar():
|
79 |
+
col1, col2 = st.columns([77, 23])
|
80 |
with col1:
|
81 |
+
st.header("Chat with your data", divider='rainbow')
|
|
|
|
|
|
|
|
|
82 |
with col2:
|
|
|
|
|
83 |
clear = st.button("Clear")
|
84 |
+
st.write("") # spacing
|
|
|
85 |
return clear
|
86 |
|
87 |
+
# Reset inputs when Clear is clicked
|
88 |
+
if write_top_bar():
|
|
|
89 |
st.session_state.questions_ = []
|
90 |
st.session_state.answers_ = []
|
91 |
+
st.session_state.input_query = ""
|
|
|
|
|
|
|
|
|
|
|
|
|
92 |
|
93 |
+
# Handle user query submission
|
94 |
def handle_input():
|
95 |
+
if st.session_state.input_query == '':
|
96 |
+
return
|
97 |
+
|
98 |
+
inputs = {key.removeprefix('input_'): st.session_state[key] for key in st.session_state if key.startswith('input_')}
|
|
|
|
|
|
|
|
|
|
|
99 |
st.session_state.inputs_ = inputs
|
100 |
+
|
101 |
+
st.session_state.questions_.append({
|
|
|
|
|
|
|
|
|
102 |
'question': inputs["query"],
|
103 |
'id': len(st.session_state.questions_)
|
104 |
+
})
|
105 |
+
|
106 |
+
if st.session_state.input_is_colpali:
|
107 |
out_ = colpali.colpali_search_rerank(st.session_state.input_query)
|
|
|
108 |
else:
|
109 |
+
out_ = rag_DocumentSearcher.query_(
|
110 |
+
awsauth,
|
111 |
+
inputs,
|
112 |
+
st.session_state['session_id'],
|
113 |
+
st.session_state.input_rag_searchType
|
114 |
+
)
|
115 |
+
|
116 |
st.session_state.answers_.append({
|
117 |
'answer': out_['text'],
|
118 |
+
'source': out_['source'],
|
119 |
'id': len(st.session_state.questions_),
|
120 |
'image': out_['image'],
|
121 |
+
'table': out_['table']
|
122 |
})
|
123 |
+
st.session_state.input_query = ""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
124 |
|
125 |
+
# Display user message block
|
126 |
+
def write_user_message(msg):
|
127 |
+
col1, col2 = st.columns([3, 97])
|
128 |
with col1:
|
129 |
+
st.image(USER_ICON, use_container_width=True)
|
130 |
with col2:
|
131 |
+
st.markdown(
|
132 |
+
f"<div style='color:#e28743;font-size:18px;padding:3px 7px;border-radius:10px;font-style:italic;'>{msg['question']}</div>",
|
133 |
+
unsafe_allow_html=True
|
134 |
+
)
|
135 |
|
136 |
+
# Render assistant answer block with optional images and tables
|
137 |
+
def write_chat_message(response, question, index):
|
138 |
+
col1, col2, col3 = st.columns([4, 74, 22])
|
139 |
|
|
|
|
|
|
|
|
|
140 |
with col1:
|
141 |
+
st.image(AI_ICON, use_container_width=True)
|
142 |
+
|
143 |
with col2:
|
144 |
+
answer_text = response['answer']
|
145 |
+
st.write(answer_text)
|
146 |
+
|
147 |
+
polly_response = polly_client.synthesize_speech(
|
148 |
+
VoiceId='Joanna', OutputFormat='ogg_vorbis', Text=answer_text, Engine='neural')
|
149 |
+
st.audio(polly_response['AudioStream'].read(), format="audio/ogg")
|
150 |
+
|
151 |
+
if st.session_state.input_is_colpali:
|
152 |
+
if st.button("Show similarity map", key=f"simmap_{index}"):
|
153 |
+
st.session_state.show_columns = True
|
154 |
+
st.session_state.maxSimImages = colpali.img_highlight(
|
155 |
+
st.session_state.top_img,
|
156 |
+
st.session_state.query_token_vectors,
|
157 |
+
st.session_state.query_tokens
|
158 |
+
)
|
159 |
+
handle_input()
|
160 |
+
with placeholder.container():
|
161 |
+
render_all()
|
162 |
+
|
163 |
+
with st.expander("Relevant Sources"):
|
164 |
+
for img in response.get('image', []):
|
165 |
+
if isinstance(img, dict) and 'file' in img:
|
166 |
+
st.image(img['file'])
|
167 |
+
|
168 |
+
for tbl in response.get('table', []):
|
169 |
+
try:
|
170 |
+
df = pd.read_csv(tbl['name'], skipinitialspace=True, on_bad_lines='skip', delimiter='`')
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
171 |
df.fillna(method='pad', inplace=True)
|
172 |
st.table(df)
|
173 |
+
except Exception as e:
|
174 |
+
st.warning(f"Failed to load table: {e}")
|
175 |
+
|
176 |
+
st.write(response.get("source", ""))
|
177 |
+
|
178 |
+
# Render all Q&A pairs
|
179 |
+
def render_all():
|
180 |
+
for index, (q, a) in enumerate(zip(st.session_state.questions_, st.session_state.answers_), start=1):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
181 |
write_user_message(q)
|
182 |
+
write_chat_message(a, q, index)
|
183 |
|
184 |
+
# Placeholder for dynamic rendering
|
185 |
placeholder = st.empty()
|
186 |
with placeholder.container():
|
187 |
+
render_all()
|
188 |
|
189 |
+
# Input field for user question
|
190 |
+
col_2, col_3 = st.columns([75, 20])
|
191 |
with col_2:
|
192 |
+
st.text_input("Ask here", label_visibility="collapsed", key="input_query")
|
|
|
193 |
with col_3:
|
194 |
+
st.button("GO", on_click=handle_input, key="play")
|
195 |
+
|
196 |
+
# Sidebar configuration
|
197 |
with st.sidebar:
|
198 |
st.page_link("app.py", label=":orange[Home]", icon="🏠")
|
199 |
st.subheader(":blue[Sample Data]")
|
200 |
+
coln_1, coln_2 = st.columns([70, 30])
|
201 |
with coln_1:
|
202 |
+
st.radio("Choose one index", ["UK Housing", "Global Warming stats", "Covid19 impacts on Ireland"], key="input_rad_index")
|
203 |
with coln_2:
|
204 |
+
st.markdown("<p style='font-size:15px'>Preview file</p>", unsafe_allow_html=True)
|
205 |
st.write("[:eyes:](https://github.com/aws-samples/AI-search-with-amazon-opensearch-service/blob/b559f82c07dfcca973f457c0a15d6444752553ab/rag/sample_pdfs/HPI-Jan-2024-Hometrack.pdf)")
|
206 |
st.write("[:eyes:](https://github.com/aws-samples/AI-search-with-amazon-opensearch-service/blob/b559f82c07dfcca973f457c0a15d6444752553ab/rag/sample_pdfs/global_warming.pdf)")
|
207 |
st.write("[:eyes:](https://github.com/aws-samples/AI-search-with-amazon-opensearch-service/blob/b559f82c07dfcca973f457c0a15d6444752553ab/rag/sample_pdfs/covid19_ie.pdf)")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
208 |
st.subheader(":blue[Retriever]")
|
209 |
+
st.multiselect("Select the Retriever(s)", ["Keyword Search", "Vector Search", "Sparse Search"], default=["Vector Search"], key="input_rag_searchType")
|
210 |
+
st.checkbox("Re-rank results", key="input_is_rerank", value=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
211 |
st.subheader(":blue[Multi-vector retrieval]")
|
212 |
+
|
213 |
+
colpali_search_rerank = st.checkbox('Try Colpali multi-vector retrieval on the [sample dataset](https://huggingface.co/datasets/vespa-engine/gpfg-QA)',
|
214 |
+
key='input_colpali',
|
215 |
+
disabled=False,
|
216 |
+
value=False,
|
217 |
+
help="Checking this box will use colpali as the embedding model and retrieval is performed using multi-vectors followed by re-ranking using MaxSim")
|
218 |
+
|
219 |
+
if colpali_search_rerank:
|
220 |
st.session_state.input_is_colpali = True
|
|
|
221 |
else:
|
222 |
st.session_state.input_is_colpali = False
|
223 |
+
|
224 |
with st.expander("Sample questions for Colpali retriever:"):
|
225 |
+
st.write("""
|
226 |
+
1. Proportion of female new hires 2021-2023?
|
227 |
+
2. First-half 2021 return on unlisted real estate investments?
|
228 |
+
3. Trend of the fund's expected absolute volatility between January 2014 and January 2016?
|
229 |
+
4. Fund return percentage in 2017?
|
230 |
+
5. Annualized gross return of the fund from 1997 to 2008?
|
231 |
+
""")
|
|
|
|
|
|
|
232 |
|
|