tests/tests_HTMLparser_htmlAutoCloseOnEnd.c

#include "unity/unity.h"
#include <libxml/HTMLparser.h>
#include <libxml/parser.h>
#include <string.h>
#include <stdlib.h>
#include <ctype.h>

/* Wrapper provided in the module under test for the static function */
void test_htmlAutoCloseOnEnd(htmlParserCtxtPtr ctxt);

/* Recorder for SAX endElement calls */
typedef struct {
    int call_count;
    char names[64][64]; /* store up to 64 element names, 63 chars each */
} EndRec;

static xmlSAXHandler g_sax; /* must be static/global so the pointer remains valid */

static void on_end_element(void *userData, const xmlChar *name) {
    EndRec *rec = (EndRec *)userData;
    if (rec == NULL || name == NULL)
        return;
    if (rec->call_count < 64) {
        strncpy(rec->names[rec->call_count], (const char *)name, sizeof(rec->names[0]) - 1);
        rec->names[rec->call_count][sizeof(rec->names[0]) - 1] = '\0';
        rec->call_count++;
    }
}

static int ci_equal(const char *a, const char *b) {
    if (a == NULL || b == NULL) return 0;
    while (*a && *b) {
        unsigned char ca = (unsigned char)*a;
        unsigned char cb = (unsigned char)*b;
        if (tolower(ca) != tolower(cb)) return 0;
        a++; b++;
    }
    return *a == *b;
}

/* Helper to create a push parser context with our SAX recorder */
static htmlParserCtxtPtr create_push_ctxt(EndRec *rec) {
    return htmlCreatePushParserCtxt(&g_sax, rec, NULL, 0, NULL, XML_CHAR_ENCODING_NONE);
}

/* Helper to feed a chunk to the push parser */
static void feed_chunk(htmlParserCtxtPtr ctxt, const char *data) {
    htmlParseChunk(ctxt, data, (int)strlen(data), 0);
}

void setUp(void) {
    /* initialize global SAX handler and libxml2 */
    memset(&g_sax, 0, sizeof(g_sax));
    g_sax.endElement = on_end_element;
    xmlInitParser();
}

void tearDown(void) {
    xmlCleanupParser();
}

/* Test: when there are no open elements (nameNr == 0), function does nothing */
void test_htmlAutoCloseOnEnd_no_open_elements_noop(void) {
    EndRec rec = {0};
    htmlParserCtxtPtr ctxt = create_push_ctxt(&rec);
    TEST_ASSERT_NOT_NULL(ctxt);

    /* Ensure no input was fed; nameNr should be 0 */
    TEST_ASSERT_EQUAL_INT(0, ctxt->nameNr);

    test_htmlAutoCloseOnEnd(ctxt);

    TEST_ASSERT_EQUAL_INT(0, ctxt->nameNr);
    TEST_ASSERT_EQUAL_INT(0, rec.call_count);

    htmlFreeParserCtxt(ctxt);
}

/* Test: closes all open elements and invokes endElement for each in reverse order */
void test_htmlAutoCloseOnEnd_closes_stack_and_calls_endElement_reverse_order(void) {
    EndRec rec = {0};
    htmlParserCtxtPtr ctxt = create_push_ctxt(&rec);
    TEST_ASSERT_NOT_NULL(ctxt);

    /* Feed partial HTML leaving elements unclosed */
    feed_chunk(ctxt, "<div><span>");

    /* The parser likely opened implicit elements like html/body; verify we have some open names */
    TEST_ASSERT_TRUE_MESSAGE(ctxt->nameNr > 0, "Expected at least one open element on the stack");

    /* Capture the top-of-stack element name before we call the function */
    char top_before[64] = {0};
    if (ctxt->name != NULL) {
        strncpy(top_before, (const char *)ctxt->name, sizeof(top_before) - 1);
    }

    /* Reset recorder to only capture calls from htmlAutoCloseOnEnd */
    rec.call_count = 0;

    int initial_nameNr = ctxt->nameNr;
    test_htmlAutoCloseOnEnd(ctxt);

    /* All names should be popped */
    TEST_ASSERT_EQUAL_INT(0, ctxt->nameNr);

    /* endElement should have been called once per previously open element */
    TEST_ASSERT_EQUAL_INT(initial_nameNr, rec.call_count);

    /* First callback should correspond to the previous top element (reverse order) */
    if (top_before[0] != '\0') {
        TEST_ASSERT_TRUE_MESSAGE(ci_equal(rec.names[0], top_before),
                                 "First endElement should be for the previous top-of-stack element");
    }

    htmlFreeParserCtxt(ctxt);
}

/* Test: returns early when HTML_PARSE_HTML5 option is set */
void test_htmlAutoCloseOnEnd_returns_early_with_HTML5_option(void) {
    EndRec rec = {0};
    htmlParserCtxtPtr ctxt = create_push_ctxt(&rec);
    TEST_ASSERT_NOT_NULL(ctxt);

    feed_chunk(ctxt, "<div><span>");

    TEST_ASSERT_TRUE_MESSAGE(ctxt->nameNr > 0, "Expected open elements before testing HTML5 early return");

    int before_nameNr = ctxt->nameNr;
    ctxt->options |= HTML_PARSE_HTML5;

    test_htmlAutoCloseOnEnd(ctxt);

    /* Should be unchanged and no callbacks should be invoked */
    TEST_ASSERT_EQUAL_INT(before_nameNr, ctxt->nameNr);
    TEST_ASSERT_EQUAL_INT(0, rec.call_count);

    htmlFreeParserCtxt(ctxt);
}

/* Test: with NULL SAX handler, the function still empties the stack but doesn't call endElement */
void test_htmlAutoCloseOnEnd_handles_null_sax_handler(void) {
    EndRec rec = {0};
    htmlParserCtxtPtr ctxt = create_push_ctxt(&rec);
    TEST_ASSERT_NOT_NULL(ctxt);

    feed_chunk(ctxt, "<div>");

    TEST_ASSERT_TRUE_MESSAGE(ctxt->nameNr > 0, "Expected open elements before null SAX test");

    int initial_nameNr = ctxt->nameNr;

    /* Remove SAX handler to simulate NULL sax/endElement */
    ctxt->sax = NULL;

    test_htmlAutoCloseOnEnd(ctxt);

    TEST_ASSERT_EQUAL_INT(0, ctxt->nameNr);
    TEST_ASSERT_EQUAL_INT(0, rec.call_count); /* no callbacks since sax is NULL */

    (void)initial_nameNr; /* suppress unused warning if not used in some configs */

    htmlFreeParserCtxt(ctxt);
}

int main(void) {
    UNITY_BEGIN();
    RUN_TEST(test_htmlAutoCloseOnEnd_no_open_elements_noop);
    RUN_TEST(test_htmlAutoCloseOnEnd_closes_stack_and_calls_endElement_reverse_order);
    RUN_TEST(test_htmlAutoCloseOnEnd_returns_early_with_HTML5_option);
    RUN_TEST(test_htmlAutoCloseOnEnd_handles_null_sax_handler);
    return UNITY_END();
}