tests/tests_HTMLparser_htmlParserFinishElementParsing.c

#include "unity/unity.h"
#include <libxml/HTMLparser.h>
#include <libxml/parser.h>
#include <libxml/tree.h>
#include <libxml/xmlmemory.h>

#include <stdlib.h>
#include <string.h>

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

static htmlParserCtxtPtr
create_ctxt_with_input(size_t base_len, size_t cur_offset, unsigned long consumed, unsigned long line,
                       xmlParserInputPtr *outInput, xmlChar **outBase) {
    htmlParserCtxtPtr ctxt = htmlNewParserCtxt();
    TEST_ASSERT_NOT_NULL_MESSAGE(ctxt, "Failed to create HTML parser context");

    xmlParserInputPtr in = xmlNewInputStream(ctxt);
    TEST_ASSERT_NOT_NULL_MESSAGE(in, "Failed to create new input stream");

    xmlChar *base = (xmlChar *)xmlMalloc(base_len);
    TEST_ASSERT_NOT_NULL_MESSAGE(base, "Failed to allocate base buffer");
    memset(base, 'A', base_len);

    /* Ensure cur_offset within base_len */
    TEST_ASSERT_MESSAGE(cur_offset <= base_len, "cur_offset exceeds base buffer length");

    in->base = base;
    in->cur = base + cur_offset;
    in->end = base + base_len;
    in->consumed = consumed;
    in->line = line;

    ctxt->input = in;

    if (outInput) *outInput = in;
    if (outBase) *outBase = base;

    return ctxt;
}

static void
destroy_ctxt_and_input(htmlParserCtxtPtr ctxt, xmlParserInputPtr in, xmlChar *base) {
    if (ctxt != NULL) {
        /* Detach input from context so we can free it explicitly to avoid double-free */
        ctxt->input = NULL;
    }
    if (in != NULL) {
        /* Avoid freeing base twice if libxml would attempt it; we own base */
        in->base = NULL;
        in->cur = NULL;
        in->end = NULL;
        xmlFreeInputStream(in);
    }
    if (base != NULL) {
        xmlFree(base);
    }
    if (ctxt != NULL) {
        htmlFreeParserCtxt(ctxt);
    }
}

void setUp(void) {
    xmlInitParser();
}

void tearDown(void) {
    xmlCleanupParser();
}

void test_htmlParserFinishElementParsing_records_end_info_basic(void) {
    xmlParserInputPtr in = NULL;
    xmlChar *base = NULL;
    /* base length 32, cur_offset 5, consumed 2, line 7 */
    htmlParserCtxtPtr ctxt = create_ctxt_with_input(32, 5, 2, 7, &in, &base);

    /* Prepare node and nodeInfo */
    xmlNodePtr node = xmlNewNode(NULL, (const xmlChar *)"test");
    TEST_ASSERT_NOT_NULL(node);
    ctxt->node = node;
    ctxt->record_info = 1;

    xmlParserNodeInfo *info = (xmlParserNodeInfo *)xmlMalloc(sizeof(xmlParserNodeInfo));
    TEST_ASSERT_NOT_NULL(info);
    memset(info, 0, sizeof(*info));
    ctxt->nodeInfo = info;

    /* Call the wrapper */
    test_htmlParserFinishElementParsing(ctxt);

    /* Verify info was added to the context's node info sequence */
    const xmlParserNodeInfo *found = xmlParserFindNodeInfo(ctxt, node);
    TEST_ASSERT_NOT_NULL_MESSAGE(found, "xmlParserNodeInfo not found for node");
    TEST_ASSERT_EQUAL_PTR(node, found->node);
    /* end_pos = consumed + (cur - base) = 2 + (5) = 7 */
    TEST_ASSERT_EQUAL_UINT(7u, (unsigned int)found->end_pos);
    TEST_ASSERT_EQUAL_UINT(7u, (unsigned int)found->end_line);

    /* Cleanup */
    xmlFreeNode(node);
    /* info was likely freed by htmlNodeInfoPop; freeing only if still referenced would be unsafe.
       We won't free 'info' here. */
    destroy_ctxt_and_input(ctxt, in, base);
}

void test_htmlParserFinishElementParsing_calculates_end_pos_with_consumed_offset(void) {
    xmlParserInputPtr in = NULL;
    xmlChar *base = NULL;
    /* base length 64, cur_offset 13, consumed 21, line 3 */
    htmlParserCtxtPtr ctxt = create_ctxt_with_input(64, 13, 21, 3, &in, &base);

    xmlNodePtr node = xmlNewNode(NULL, (const xmlChar *)"elem");
    TEST_ASSERT_NOT_NULL(node);
    ctxt->node = node;
    ctxt->record_info = 1;

    xmlParserNodeInfo *info = (xmlParserNodeInfo *)xmlMalloc(sizeof(xmlParserNodeInfo));
    TEST_ASSERT_NOT_NULL(info);
    memset(info, 0, sizeof(*info));
    ctxt->nodeInfo = info;

    test_htmlParserFinishElementParsing(ctxt);

    const xmlParserNodeInfo *found = xmlParserFindNodeInfo(ctxt, node);
    TEST_ASSERT_NOT_NULL_MESSAGE(found, "xmlParserNodeInfo not found for node");
    TEST_ASSERT_EQUAL_PTR(node, found->node);
    /* end_pos = 21 + 13 = 34; end_line = 3 */
    TEST_ASSERT_EQUAL_UINT(34u, (unsigned int)found->end_pos);
    TEST_ASSERT_EQUAL_UINT(3u, (unsigned int)found->end_line);

    xmlFreeNode(node);
    destroy_ctxt_and_input(ctxt, in, base);
}

void test_htmlParserFinishElementParsing_noop_when_node_null(void) {
    xmlParserInputPtr in = NULL;
    xmlChar *base = NULL;
    htmlParserCtxtPtr ctxt = create_ctxt_with_input(16, 4, 9, 11, &in, &base);

    /* node is NULL; enable record_info to test the node NULL guard */
    ctxt->node = NULL;
    ctxt->record_info = 1;

    /* Provide a nodeInfo struct; function must not use it since node is NULL */
    xmlParserNodeInfo *info = (xmlParserNodeInfo *)xmlMalloc(sizeof(xmlParserNodeInfo));
    TEST_ASSERT_NOT_NULL(info);
    memset(info, 0, sizeof(*info));
    ctxt->nodeInfo = info;

    test_htmlParserFinishElementParsing(ctxt);

    /* Create a node to search for; should not find any info since node was NULL */
    xmlNodePtr node = xmlNewNode(NULL, (const xmlChar *)"unused");
    TEST_ASSERT_NOT_NULL(node);
    const xmlParserNodeInfo *found = xmlParserFindNodeInfo(ctxt, node);
    TEST_ASSERT_NULL(found);

    xmlFreeNode(node);
    /* As above, assume htmlNodeInfoPop may have freed info; do not free directly. */
    destroy_ctxt_and_input(ctxt, in, base);
}

void test_htmlParserFinishElementParsing_noop_when_record_info_false_even_with_node(void) {
    /* Intentionally no input to ensure function doesn't touch input when record_info == 0 */
    htmlParserCtxtPtr ctxt = htmlNewParserCtxt();
    TEST_ASSERT_NOT_NULL(ctxt);

    xmlNodePtr node = xmlNewNode(NULL, (const xmlChar *)"n");
    TEST_ASSERT_NOT_NULL(node);
    ctxt->node = node;
    ctxt->record_info = 0; /* disabled */
    ctxt->input = NULL;    /* would crash if accessed */

    /* Provide a nodeInfo pointer; must not be used since record_info == 0 */
    xmlParserNodeInfo *info = (xmlParserNodeInfo *)xmlMalloc(sizeof(xmlParserNodeInfo));
    TEST_ASSERT_NOT_NULL(info);
    memset(info, 0, sizeof(*info));
    ctxt->nodeInfo = info;

    test_htmlParserFinishElementParsing(ctxt);

    /* Since recording is disabled, no node info should be present */
    const xmlParserNodeInfo *found = xmlParserFindNodeInfo(ctxt, node);
    TEST_ASSERT_NULL(found);

    xmlFreeNode(node);
    /* Do not free 'info' as it may be managed by internal pop logic in other cases. */
    htmlFreeParserCtxt(ctxt);
}

int main(void) {
    UNITY_BEGIN();
    RUN_TEST(test_htmlParserFinishElementParsing_records_end_info_basic);
    RUN_TEST(test_htmlParserFinishElementParsing_calculates_end_pos_with_consumed_offset);
    RUN_TEST(test_htmlParserFinishElementParsing_noop_when_node_null);
    RUN_TEST(test_htmlParserFinishElementParsing_noop_when_record_info_false_even_with_node);
    return UNITY_END();
}