tests/tests_HTMLparser_htmlAutoClose.c

#include "unity/unity.h"
#include <libxml/HTMLparser.h>

#include <stdlib.h>
#include <string.h>
#include <libxml/parser.h>
#include <libxml/parserInternals.h>

/* Wrapper provided in the module for testing the static function */
void test_htmlAutoClose(htmlParserCtxtPtr ctxt, const xmlChar * newtag);

typedef struct {
    int count;
    const xmlChar *last;
} TestSaxData;

static void onEndElement(void *userData, const xmlChar *name) {
    TestSaxData *d = (TestSaxData *)userData;
    if (d) {
        d->count++;
        d->last = name;
    }
}

/* Helper: create parser context with a prepared name stack (bottom..top). 
   tags[0] is bottom of stack, tags[count-1] is current top (ctxt->name). 
   If withSax is non-zero, install a SAX handler that counts endElement calls. */
static htmlParserCtxtPtr make_ctxt_with_stack(const char **tags, int count,
                                              int withSax,
                                              TestSaxData *saxDataOut) {
    htmlParserCtxtPtr ctxt = htmlNewParserCtxt();
    TEST_ASSERT_NOT_NULL(ctxt);

    /* Disable recording to avoid dereferencing uninitialized fields in htmlParserFinishElementParsing */
    ctxt->record_info = 0;

    /* Prepare name stack */
    if (count > 0) {
        ctxt->nameMax = count;
        ctxt->nameNr = count;
        ctxt->nameTab = (const xmlChar **)xmlMalloc(sizeof(xmlChar *) * (size_t)count);
        TEST_ASSERT_NOT_NULL(ctxt->nameTab);
        for (int i = 0; i < count; i++) {
            /* Duplicate to ensure safe lifetime if internals attempt to free */
            ctxt->nameTab[i] = xmlStrdup((const xmlChar *)tags[i]);
            TEST_ASSERT_NOT_NULL(ctxt->nameTab[i]);
        }
        ctxt->name = ctxt->nameTab[count - 1];
    } else {
        ctxt->nameMax = 0;
        ctxt->nameNr = 0;
        ctxt->nameTab = NULL;
        ctxt->name = NULL;
    }

    /* Options default to 0 unless overwritten by tests */
    ctxt->options = 0;

    if (withSax) {
        static xmlSAXHandler sax; /* static so its storage lives long enough */
        memset(&sax, 0, sizeof(sax));
        sax.endElement = onEndElement;

        if (saxDataOut) {
            saxDataOut->count = 0;
            saxDataOut->last = NULL;
        }

        ctxt->sax = &sax;
        ctxt->userData = saxDataOut;
    } else {
        ctxt->sax = NULL;
        ctxt->userData = NULL;
    }

    return ctxt;
}

void setUp(void) {
    /* Optional global setup */
}

void tearDown(void) {
    /* Optional global cleanup */
}

/* Test: returns immediately when HTML5 tokenizer option is set */
void test_htmlAutoClose_returns_immediately_with_HTML5_option(void) {
    const char *stack[] = { "ul", "li" };
    TestSaxData data;
    htmlParserCtxtPtr ctxt = make_ctxt_with_stack(stack, 2, 1, &data);

    /* Set HTML5 option to force early return */
    ctxt->options |= HTML_PARSE_HTML5;

    const xmlChar *prevName = ctxt->name;
    int prevNr = ctxt->nameNr;

    test_htmlAutoClose(ctxt, (const xmlChar *)"li");

    TEST_ASSERT_EQUAL_PTR(prevName, ctxt->name);
    TEST_ASSERT_EQUAL_INT(prevNr, ctxt->nameNr);
    TEST_ASSERT_EQUAL_INT(0, data.count);

    htmlFreeParserCtxt(ctxt);
}

/* Test: no action when newtag is NULL */
void test_htmlAutoClose_ignores_null_newtag(void) {
    const char *stack[] = { "ul", "li" };
    TestSaxData data;
    htmlParserCtxtPtr ctxt = make_ctxt_with_stack(stack, 2, 1, &data);

    const xmlChar *prevName = ctxt->name;
    int prevNr = ctxt->nameNr;

    test_htmlAutoClose(ctxt, NULL);

    TEST_ASSERT_EQUAL_PTR(prevName, ctxt->name);
    TEST_ASSERT_EQUAL_INT(prevNr, ctxt->nameNr);
    TEST_ASSERT_EQUAL_INT(0, data.count);

    htmlFreeParserCtxt(ctxt);
}

/* Test: single auto-close for matching pair (li/li), calls endElement and pops one name */
void test_htmlAutoClose_closes_single_matching_tag_and_calls_endElement(void) {
    const char *stack[] = { "ul", "li" };
    TestSaxData data;
    htmlParserCtxtPtr ctxt = make_ctxt_with_stack(stack, 2, 1, &data);

    test_htmlAutoClose(ctxt, (const xmlChar *)"li");

    TEST_ASSERT_EQUAL_INT(1, data.count);
    TEST_ASSERT_NOT_NULL_MESSAGE(data.last, "endElement should be called with closed tag name");
    TEST_ASSERT_EQUAL_INT(1, ctxt->nameNr); /* one pop */
    TEST_ASSERT_EQUAL_STRING("ul", (const char *)ctxt->name);
    TEST_ASSERT_EQUAL_INT(0, xmlStrcmp((const xmlChar *)"li", data.last));

    htmlFreeParserCtxt(ctxt);
}

/* Test: multiple auto-closes in a row (li closes li repeatedly), pops until non-li at bottom */
void test_htmlAutoClose_closes_multiple_in_a_row(void) {
    const char *stack[] = { "ul", "li", "li", "li" };
    TestSaxData data;
    htmlParserCtxtPtr ctxt = make_ctxt_with_stack(stack, 4, 1, &data);

    test_htmlAutoClose(ctxt, (const xmlChar *)"li");

    /* Expect three pops (all 'li') leaving 'ul' */
    TEST_ASSERT_EQUAL_INT(3, data.count);
    TEST_ASSERT_EQUAL_INT(1, ctxt->nameNr);
    TEST_ASSERT_EQUAL_STRING("ul", (const char *)ctxt->name);

    htmlFreeParserCtxt(ctxt);
}

/* Test: no action when top-of-stack doesn't match autoclose rule (li doesn't close ul) */
void test_htmlAutoClose_no_action_when_no_match(void) {
    const char *stack[] = { "ul" };
    TestSaxData data;
    htmlParserCtxtPtr ctxt = make_ctxt_with_stack(stack, 1, 1, &data);

    const xmlChar *prevName = ctxt->name;
    int prevNr = ctxt->nameNr;

    test_htmlAutoClose(ctxt, (const xmlChar *)"li");

    TEST_ASSERT_EQUAL_PTR(prevName, ctxt->name);
    TEST_ASSERT_EQUAL_INT(prevNr, ctxt->nameNr);
    TEST_ASSERT_EQUAL_INT(0, data.count);

    htmlFreeParserCtxt(ctxt);
}

/* Test: works safely with NULL SAX handler (no callback), still pops if match */
void test_htmlAutoClose_handles_null_sax(void) {
    const char *stack[] = { "ul", "li" };
    htmlParserCtxtPtr ctxt = make_ctxt_with_stack(stack, 2, 0, NULL);

    test_htmlAutoClose(ctxt, (const xmlChar *)"li");

    TEST_ASSERT_EQUAL_INT(1, ctxt->nameNr);
    TEST_ASSERT_EQUAL_STRING("ul", (const char *)ctxt->name);

    htmlFreeParserCtxt(ctxt);
}

/* Test: no action when name is NULL (empty stack) */
void test_htmlAutoClose_noop_when_name_is_null(void) {
    TestSaxData data;
    htmlParserCtxtPtr ctxt = make_ctxt_with_stack(NULL, 0, 1, &data);

    test_htmlAutoClose(ctxt, (const xmlChar *)"li");

    TEST_ASSERT_EQUAL_INT(0, data.count);
    TEST_ASSERT_EQUAL_INT(0, ctxt->nameNr);
    TEST_ASSERT_NULL(ctxt->name);

    htmlFreeParserCtxt(ctxt);
}

int main(void) {
    UNITY_BEGIN();
    RUN_TEST(test_htmlAutoClose_returns_immediately_with_HTML5_option);
    RUN_TEST(test_htmlAutoClose_ignores_null_newtag);
    RUN_TEST(test_htmlAutoClose_closes_single_matching_tag_and_calls_endElement);
    RUN_TEST(test_htmlAutoClose_closes_multiple_in_a_row);
    RUN_TEST(test_htmlAutoClose_no_action_when_no_match);
    RUN_TEST(test_htmlAutoClose_handles_null_sax);
    RUN_TEST(test_htmlAutoClose_noop_when_name_is_null);
    return UNITY_END();
}