tests/tests_HTMLparser_htmlSkipBogusDoctype.c

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

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

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

static htmlParserCtxtPtr make_ctxt_from_str(const char *s) {
    /* Use memory-based context so we control input content precisely */
    return htmlCreateMemoryParserCtxt(s, (int)strlen(s));
}

void setUp(void) {
    /* No global setup required */
}

void tearDown(void) {
    /* No global teardown required */
}

/* Basic case: single-line input with immediate '>' */
void test_htmlSkipBogusDoctype_basic(void) {
    const char *buf = "bogus>";
    htmlParserCtxtPtr ctxt = make_ctxt_from_str(buf);
    TEST_ASSERT_NOT_NULL(ctxt);
    TEST_ASSERT_NOT_NULL(ctxt->input);

    /* Start at known position */
    ctxt->input->line = 1;
    ctxt->input->col = 1;

    test_htmlSkipBogusDoctype(ctxt);

    /* Expect cursor right after '>' */
    int offset = (int)(ctxt->input->cur - ctxt->input->base);
    TEST_ASSERT_EQUAL_INT(6, offset); /* "bogus>" is 6 chars */
    /* No newlines, so line unchanged; col counts characters processed including '>' */
    TEST_ASSERT_EQUAL_INT(1, ctxt->input->line);
    TEST_ASSERT_EQUAL_INT(7, ctxt->input->col); /* initial 1 + 6 chars processed */

    htmlFreeParserCtxt(ctxt);
}

/* Newline handling: ensure line increments and col resets to 1 after '\n' */
void test_htmlSkipBogusDoctype_with_newlines(void) {
    const char *buf = "ab\ncd>";
    htmlParserCtxtPtr ctxt = make_ctxt_from_str(buf);
    TEST_ASSERT_NOT_NULL(ctxt);
    TEST_ASSERT_NOT_NULL(ctxt->input);

    ctxt->input->line = 1;
    ctxt->input->col = 1;

    test_htmlSkipBogusDoctype(ctxt);

    int offset = (int)(ctxt->input->cur - ctxt->input->base);
    TEST_ASSERT_EQUAL_INT(6, offset); /* processed entire string up to and including '>' */

    /* One newline encountered: line increments by 1 */
    TEST_ASSERT_EQUAL_INT(2, ctxt->input->line);

    /* Col progression:
       'a' -> 2, 'b' -> 3, '\n' -> (inc to 4 then reset to 1), 'c' -> 2, 'd' -> 3, '>' -> 4
     */
    TEST_ASSERT_EQUAL_INT(4, ctxt->input->col);

    htmlFreeParserCtxt(ctxt);
}

/* No closing '>' => should advance to end-of-buffer with correct line/col */
void test_htmlSkipBogusDoctype_no_gt_reaches_eof(void) {
    const char *buf = "no gt present";
    const int len = (int)strlen(buf);
    htmlParserCtxtPtr ctxt = make_ctxt_from_str(buf);
    TEST_ASSERT_NOT_NULL(ctxt);
    TEST_ASSERT_NOT_NULL(ctxt->input);

    ctxt->input->line = 5;
    ctxt->input->col = 10;

    test_htmlSkipBogusDoctype(ctxt);

    /* Cursor at end */
    int offset = (int)(ctxt->input->cur - ctxt->input->base);
    TEST_ASSERT_EQUAL_INT(len, offset);

    /* No newlines in input; col increments by total characters processed */
    TEST_ASSERT_EQUAL_INT(5, ctxt->input->line);
    TEST_ASSERT_EQUAL_INT(10 + len, ctxt->input->col);

    htmlFreeParserCtxt(ctxt);
}

/* Zero available input at entry (cur == end) => no changes */
void test_htmlSkipBogusDoctype_zero_avail_noop(void) {
    const char *buf = ">";
    htmlParserCtxtPtr ctxt = make_ctxt_from_str(buf);
    TEST_ASSERT_NOT_NULL(ctxt);
    TEST_ASSERT_NOT_NULL(ctxt->input);

    /* Move to end so avail == 0 */
    ctxt->input->cur = ctxt->input->end;
    ctxt->input->line = 3;
    ctxt->input->col = 7;

    test_htmlSkipBogusDoctype(ctxt);

    /* Expect unchanged */
    TEST_ASSERT_TRUE(ctxt->input->cur == ctxt->input->end);
    TEST_ASSERT_EQUAL_INT(3, ctxt->input->line);
    TEST_ASSERT_EQUAL_INT(7, ctxt->input->col);

    htmlFreeParserCtxt(ctxt);
}

/* Long input to ensure scanning works when avail is large (>64) */
void test_htmlSkipBogusDoctype_long_input(void) {
    /* Create 200 'x' characters followed by '>' */
    const int xcount = 200;
    const int total = xcount + 1;
    char *buf = (char *)malloc((size_t)total + 1);
    TEST_ASSERT_NOT_NULL(buf);
    memset(buf, 'x', (size_t)xcount);
    buf[xcount] = '>';
    buf[total] = '\0';

    htmlParserCtxtPtr ctxt = make_ctxt_from_str(buf);
    TEST_ASSERT_NOT_NULL(ctxt);
    TEST_ASSERT_NOT_NULL(ctxt->input);

    ctxt->input->line = 1;
    ctxt->input->col = 1;

    test_htmlSkipBogusDoctype(ctxt);

    int offset = (int)(ctxt->input->cur - ctxt->input->base);
    TEST_ASSERT_EQUAL_INT(total, offset);    /* advanced past '>' */
    TEST_ASSERT_EQUAL_INT(1, ctxt->input->line);
    TEST_ASSERT_EQUAL_INT(1 + total, ctxt->input->col);

    htmlFreeParserCtxt(ctxt);
    free(buf);
}

int main(void) {
    UNITY_BEGIN();

    RUN_TEST(test_htmlSkipBogusDoctype_basic);
    RUN_TEST(test_htmlSkipBogusDoctype_with_newlines);
    RUN_TEST(test_htmlSkipBogusDoctype_no_gt_reaches_eof);
    RUN_TEST(test_htmlSkipBogusDoctype_zero_avail_noop);
    RUN_TEST(test_htmlSkipBogusDoctype_long_input);

    return UNITY_END();
}