tests/tests_HTMLparser_htmlParseElement.c

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

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

/* Simple event logger for SAX callbacks */
typedef struct {
    char buf[2048];
} EventLog;

static void log_append(EventLog *log, const char *s) {
    size_t cur = strlen(log->buf);
    size_t add = strlen(s);
    if (cur + add >= sizeof(log->buf)) {
        add = sizeof(log->buf) - cur - 1;
    }
    if (add > 0) {
        memcpy(log->buf + cur, s, add);
        log->buf[cur + add] = '\0';
    }
}

/* SAX callbacks we care about */
static void onStartElement(void *ctx, const xmlChar *name, const xmlChar **atts) {
    (void)atts;
    EventLog *log = (EventLog *)ctx;
    char tmp[256];
    snprintf(tmp, sizeof(tmp), "[S:%s]", (const char *)name);
    log_append(log, tmp);
}

static void onEndElement(void *ctx, const xmlChar *name) {
    EventLog *log = (EventLog *)ctx;
    char tmp[256];
    snprintf(tmp, sizeof(tmp), "[E:%s]", (const char *)name);
    log_append(log, tmp);
}

static void onCharacters(void *ctx, const xmlChar *ch, int len) {
    EventLog *log = (EventLog *)ctx;
    if (len <= 0)
        return;
    char tmp[512];
    int copy = len;
    if (copy > (int)sizeof(tmp) - 1)
        copy = (int)sizeof(tmp) - 1;
    memcpy(tmp, ch, (size_t)copy);
    tmp[copy] = '\0';

    char out[600];
    snprintf(out, sizeof(out), "[T:%s]", tmp);
    log_append(log, out);
}

/* Global SAX handler; configured in setUp */
static xmlSAXHandler gSAX;

void setUp(void) {
    memset(&gSAX, 0, sizeof(gSAX));
    gSAX.startElement = onStartElement;
    gSAX.endElement = onEndElement;
    gSAX.characters = onCharacters;
}

void tearDown(void) {
    /* nothing */
}

/* Helper to run one parsing case through htmlParseElement and check events */
static void run_case(const char *html, int options, const char *expectedLog) {
    EventLog log;
    log.buf[0] = '\0';

    htmlParserCtxtPtr ctxt = htmlCreateMemoryParserCtxt(html, (int)strlen(html));
    TEST_ASSERT_NOT_NULL_MESSAGE(ctxt, "htmlCreateMemoryParserCtxt returned NULL");

    ctxt->sax = &gSAX;
    ctxt->userData = &log;
    ctxt->options = options;

    htmlParseElement(ctxt);

    TEST_ASSERT_EQUAL_STRING_MESSAGE(expectedLog, log.buf, "Unexpected SAX event sequence");

    htmlFreeParserCtxt(ctxt);
}

/* Tests */

void test_htmlParseElement_null_context_safe(void) {
    /* Should be a no-op and must not crash */
    htmlParseElement(NULL);
    TEST_ASSERT_TRUE(1);
}

void test_htmlParseElement_empty_input_no_events(void) {
    run_case("", 0, "");
}

void test_htmlParseElement_parses_simple_element(void) {
    run_case("<p>hi</p>", 0, "[S:p][T:hi][E:p]");
}

void test_htmlParseElement_self_closing_xml_style_non_html5(void) {
    /* In non-HTML5 mode, '/>' triggers an explicit endElement */
    run_case("<br/>", 0, "[S:br][E:br]");
}

void test_htmlParseElement_self_closing_xml_style_html5(void) {
    /* In HTML5 mode, '/>' does NOT call endElement in htmlParseElementInternal */
    run_case("<br/>", HTML_PARSE_HTML5, "[S:br]");
}

void test_htmlParseElement_dtd_empty_element_non_html5(void) {
    /* DTD-empty element like <br> should auto-close in non-HTML5 mode */
    run_case("<br>", 0, "[S:br][E:br]");
}

void test_htmlParseElement_dtd_empty_element_html5(void) {
    /* In HTML5 mode, no endElement is emitted for DTD-empty element in this path */
    run_case("<br>", HTML_PARSE_HTML5, "[S:br]");
}

void test_htmlParseElement_auto_close_on_eof_non_html5(void) {
    /* Unclosed nested elements should be auto-closed at end of input in non-HTML5 mode */
    run_case("<div><span>t", 0, "[S:div][S:span][T:t][E:span][E:div]");
}

void test_htmlParseElement_auto_close_on_eof_html5(void) {
    /* In HTML5 mode, htmlAutoCloseOnEnd() is a no-op, so no endElement callbacks */
    run_case("<div><span>t", HTML_PARSE_HTML5, "[S:div][S:span][T:t]");
}

int main(void) {
    UNITY_BEGIN();
    RUN_TEST(test_htmlParseElement_null_context_safe);
    RUN_TEST(test_htmlParseElement_empty_input_no_events);
    RUN_TEST(test_htmlParseElement_parses_simple_element);
    RUN_TEST(test_htmlParseElement_self_closing_xml_style_non_html5);
    RUN_TEST(test_htmlParseElement_self_closing_xml_style_html5);
    RUN_TEST(test_htmlParseElement_dtd_empty_element_non_html5);
    RUN_TEST(test_htmlParseElement_dtd_empty_element_html5);
    RUN_TEST(test_htmlParseElement_auto_close_on_eof_non_html5);
    RUN_TEST(test_htmlParseElement_auto_close_on_eof_html5);
    return UNITY_END();
}