libxml / tests /tests_HTMLparser_htmlCreateMemoryParserCtxtInternal.c

Upload folder using huggingface_hub

6baed57 verified 2 months ago

4.03 kB

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

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

	/* The module provides a global test wrapper for the static function. */
	extern htmlParserCtxtPtr test_htmlCreateMemoryParserCtxtInternal(const char *url,
	const char *buffer, size_t size,
	const char *encoding);

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

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

	static void free_ctxt_if_not_null(htmlParserCtxtPtr ctxt) {
	if (ctxt != NULL) {
	htmlFreeParserCtxt(ctxt);
	}
	}

	void test_htmlCreateMemoryParserCtxtInternal_returns_NULL_on_null_buffer(void) {
	htmlParserCtxtPtr ctxt = test_htmlCreateMemoryParserCtxtInternal(NULL, NULL, 0, NULL);
	TEST_ASSERT_NULL(ctxt);
	}

	void test_htmlCreateMemoryParserCtxtInternal_creates_context_with_basic_html(void) {
	const char *buf = "<!DOCTYPE html><html><head><title>T</title></head>"
	"<body>Hello</body></html>";
	size_t size = strlen(buf);

	htmlParserCtxtPtr ctxt = test_htmlCreateMemoryParserCtxtInternal(NULL, buf, size, NULL);
	TEST_ASSERT_NOT_NULL(ctxt);
	free_ctxt_if_not_null(ctxt);
	}

	void test_htmlCreateMemoryParserCtxtInternal_accepts_zero_length_buffer(void) {
	/* Non-NULL pointer but zero size should still create a context for empty input. */
	const char dummy = 0; /* Any addressable byte is fine; size is 0. */

	htmlParserCtxtPtr ctxt = test_htmlCreateMemoryParserCtxtInternal(NULL, &dummy, 0, NULL);
	TEST_ASSERT_NOT_NULL(ctxt);
	free_ctxt_if_not_null(ctxt);
	}

	void test_htmlCreateMemoryParserCtxtInternal_accepts_non_null_url(void) {
	const char *url = "http://example.com/index.html";
	const char *buf = "<html><body>URL Test</body></html>";
	size_t size = strlen(buf);

	htmlParserCtxtPtr ctxt = test_htmlCreateMemoryParserCtxtInternal(url, buf, size, NULL);
	TEST_ASSERT_NOT_NULL(ctxt);
	free_ctxt_if_not_null(ctxt);
	}

	void test_htmlCreateMemoryParserCtxtInternal_accepts_explicit_utf8_encoding(void) {
	const char *buf = "<html><body>UTF-8 ✓</body></html>";
	size_t size = strlen(buf);
	const char *enc = "UTF-8";

	htmlParserCtxtPtr ctxt = test_htmlCreateMemoryParserCtxtInternal(NULL, buf, size, enc);
	TEST_ASSERT_NOT_NULL(ctxt);
	free_ctxt_if_not_null(ctxt);
	}

	void test_htmlCreateMemoryParserCtxtInternal_accepts_iso_8859_1_encoding(void) {
	/* Buffer contains 0xE9 (é in ISO-8859-1). */
	const char latin1_html[] = "<html><body>caf\xE9</body></html>";
	size_t size = sizeof(latin1_html) - 1; /* exclude terminating NUL */
	const char *enc = "ISO-8859-1";

	htmlParserCtxtPtr ctxt = test_htmlCreateMemoryParserCtxtInternal(NULL, latin1_html, size, enc);
	TEST_ASSERT_NOT_NULL(ctxt);
	free_ctxt_if_not_null(ctxt);
	}

	void test_htmlCreateMemoryParserCtxtInternal_handles_embedded_nul_bytes(void) {
	/* Explicit size includes an embedded NUL in the buffer. */
	const char html_with_nul[] = "<html><body>abc\0def</body></html>";
	size_t size = sizeof(html_with_nul) - 1; /* include the embedded NUL, exclude trailing terminator */

	htmlParserCtxtPtr ctxt = test_htmlCreateMemoryParserCtxtInternal(NULL, html_with_nul, size, NULL);
	TEST_ASSERT_NOT_NULL(ctxt);
	free_ctxt_if_not_null(ctxt);
	}

	int main(void) {
	UNITY_BEGIN();

	RUN_TEST(test_htmlCreateMemoryParserCtxtInternal_returns_NULL_on_null_buffer);
	RUN_TEST(test_htmlCreateMemoryParserCtxtInternal_creates_context_with_basic_html);
	RUN_TEST(test_htmlCreateMemoryParserCtxtInternal_accepts_zero_length_buffer);
	RUN_TEST(test_htmlCreateMemoryParserCtxtInternal_accepts_non_null_url);
	RUN_TEST(test_htmlCreateMemoryParserCtxtInternal_accepts_explicit_utf8_encoding);
	RUN_TEST(test_htmlCreateMemoryParserCtxtInternal_accepts_iso_8859_1_encoding);
	RUN_TEST(test_htmlCreateMemoryParserCtxtInternal_handles_embedded_nul_bytes);

	return UNITY_END();
	}