tests/tests_HTMLparser_htmlCheckMeta.c

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

/* The wrapper provided in the module for the static function */
void test_htmlCheckMeta(htmlParserCtxtPtr ctxt, const xmlChar **atts);

static htmlParserCtxtPtr make_ctxt(void) {
    htmlParserCtxtPtr ctxt = htmlNewParserCtxt();
    TEST_ASSERT_NOT_NULL_MESSAGE(ctxt, "Failed to create HTML parser context");
    xmlCtxtResetLastError(ctxt);
    return ctxt;
}

void setUp(void) {
    /* Setup code here, or leave empty */
}

void tearDown(void) {
    /* Cleanup code here, or leave empty */
}

static void assert_encoding_equals_case_insensitive(const xmlChar *enc, const char *expected) {
    TEST_ASSERT_NOT_NULL_MESSAGE(enc, "Encoding was not set");
    TEST_ASSERT_EQUAL_INT_MESSAGE(0, xmlStrcasecmp(enc, BAD_CAST expected),
                                  "Declared encoding does not match expected (case-insensitive compare)");
}

void test_htmlCheckMeta_charset_utf8_sets_declared_encoding(void) {
    htmlParserCtxtPtr ctxt = make_ctxt();

    const xmlChar *atts[] = {
        BAD_CAST "charset", BAD_CAST "utf-8",
        NULL
    };

    TEST_ASSERT_NULL_MESSAGE(ctxt->encoding, "Precondition failed: encoding should start as NULL");
    xmlCtxtResetLastError(ctxt);

    test_htmlCheckMeta(ctxt, atts);

    assert_encoding_equals_case_insensitive(ctxt->encoding, "utf-8");

    xmlErrorPtr err = xmlCtxtGetLastError(ctxt);
    TEST_ASSERT_TRUE_MESSAGE(err == NULL || err->code == 0,
                             "Unexpected error after setting UTF-8 encoding");

    htmlFreeParserCtxt(ctxt);
}

void test_htmlCheckMeta_http_equiv_content_type_sets_declared_encoding(void) {
    htmlParserCtxtPtr ctxt = make_ctxt();

    const xmlChar *atts[] = {
        BAD_CAST "http-equiv", BAD_CAST "Content-Type",
        BAD_CAST "content", BAD_CAST "text/html; charset=ISO-8859-1",
        NULL
    };

    TEST_ASSERT_NULL(ctxt->encoding);
    xmlCtxtResetLastError(ctxt);

    test_htmlCheckMeta(ctxt, atts);

    assert_encoding_equals_case_insensitive(ctxt->encoding, "ISO-8859-1");

    xmlErrorPtr err = xmlCtxtGetLastError(ctxt);
    TEST_ASSERT_TRUE_MESSAGE(err == NULL || err->code == 0,
                             "Unexpected error after setting ISO-8859-1 encoding");

    htmlFreeParserCtxt(ctxt);
}

void test_htmlCheckMeta_charset_precedence_over_http_equiv(void) {
    htmlParserCtxtPtr ctxt = make_ctxt();

    const xmlChar *atts[] = {
        BAD_CAST "http-equiv", BAD_CAST "Content-Type",
        BAD_CAST "content", BAD_CAST "text/html; charset=ISO-8859-1",
        BAD_CAST "charset", BAD_CAST "utf-8",
        NULL
    };

    TEST_ASSERT_NULL(ctxt->encoding);
    xmlCtxtResetLastError(ctxt);

    test_htmlCheckMeta(ctxt, atts);

    /* charset attribute should take precedence */
    assert_encoding_equals_case_insensitive(ctxt->encoding, "utf-8");

    htmlFreeParserCtxt(ctxt);
}

void test_htmlCheckMeta_non_ascii_incompatible_emits_error_and_ignores(void) {
    htmlParserCtxtPtr ctxt = make_ctxt();

    const xmlChar *atts[] = {
        BAD_CAST "charset", BAD_CAST "utf-16",
        NULL
    };

    TEST_ASSERT_NULL(ctxt->encoding);
    xmlCtxtResetLastError(ctxt);

    test_htmlCheckMeta(ctxt, atts);

    /* Should not set encoding due to non-ASCII-compatibility */
    TEST_ASSERT_NULL_MESSAGE(ctxt->encoding, "Encoding should not be set for non-ASCII-compatible charset");

    xmlErrorPtr err = xmlCtxtGetLastError(ctxt);
    TEST_ASSERT_NOT_NULL_MESSAGE(err, "Expected an error for non-ASCII-compatible charset");
    TEST_ASSERT_EQUAL_INT_MESSAGE(XML_ERR_UNSUPPORTED_ENCODING, err->code,
                                  "Expected XML_ERR_UNSUPPORTED_ENCODING error code");

    htmlFreeParserCtxt(ctxt);
}

void test_htmlCheckMeta_null_context_no_crash(void) {
    const xmlChar *atts[] = {
        BAD_CAST "charset", BAD_CAST "utf-8",
        NULL
    };

    /* Just ensure no crash; nothing to assert on context */
    test_htmlCheckMeta(NULL, atts);

    TEST_ASSERT_TRUE(1);
}

void test_htmlCheckMeta_null_atts_no_action(void) {
    htmlParserCtxtPtr ctxt = make_ctxt();

    TEST_ASSERT_NULL(ctxt->encoding);
    xmlCtxtResetLastError(ctxt);

    test_htmlCheckMeta(ctxt, NULL);

    TEST_ASSERT_NULL_MESSAGE(ctxt->encoding, "Encoding should remain NULL when atts is NULL");

    htmlFreeParserCtxt(ctxt);
}

void test_htmlCheckMeta_content_without_http_equiv_no_action(void) {
    htmlParserCtxtPtr ctxt = make_ctxt();

    const xmlChar *atts[] = {
        BAD_CAST "content", BAD_CAST "text/html; charset=UTF-8",
        NULL
    };

    TEST_ASSERT_NULL(ctxt->encoding);
    xmlCtxtResetLastError(ctxt);

    test_htmlCheckMeta(ctxt, atts);

    TEST_ASSERT_NULL_MESSAGE(ctxt->encoding, "Encoding should not be set without http-equiv=Content-Type");

    htmlFreeParserCtxt(ctxt);
}

void test_htmlCheckMeta_case_insensitive_matching(void) {
    htmlParserCtxtPtr ctxt = make_ctxt();

    const xmlChar *atts[] = {
        BAD_CAST "HTTP-EQUIV", BAD_CAST "content-type",
        BAD_CAST "CoNtEnT", BAD_CAST "text/html; charset=iso-8859-1",
        NULL
    };

    TEST_ASSERT_NULL(ctxt->encoding);
    xmlCtxtResetLastError(ctxt);

    test_htmlCheckMeta(ctxt, atts);

    assert_encoding_equals_case_insensitive(ctxt->encoding, "iso-8859-1");

    htmlFreeParserCtxt(ctxt);
}

int main(void) {
    xmlInitParser();

    UNITY_BEGIN();
    RUN_TEST(test_htmlCheckMeta_charset_utf8_sets_declared_encoding);
    RUN_TEST(test_htmlCheckMeta_http_equiv_content_type_sets_declared_encoding);
    RUN_TEST(test_htmlCheckMeta_charset_precedence_over_http_equiv);
    RUN_TEST(test_htmlCheckMeta_non_ascii_incompatible_emits_error_and_ignores);
    RUN_TEST(test_htmlCheckMeta_null_context_no_crash);
    RUN_TEST(test_htmlCheckMeta_null_atts_no_action);
    RUN_TEST(test_htmlCheckMeta_content_without_http_equiv_no_action);
    RUN_TEST(test_htmlCheckMeta_case_insensitive_matching);
    int res = UNITY_END();

    xmlCleanupParser();
    return res;
}