libxml / tests /tests_HTMLparser_htmlParseChunk.c

Upload folder using huggingface_hub

6baed57 verified 2 months ago

4.96 kB

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

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

	/* Simple SAX state to record callbacks */
	typedef struct {
	int startCount;
	int endCount;
	int charsCount;
	char collected[512];
	size_t collectedLen;
	} TestSAXState;

	/* SAX callbacks */
	static void test_startDocument(void *ctx) {
	TestSAXState st = (TestSAXState )ctx;
	st->startCount++;
	}

	static void test_endDocument(void *ctx) {
	TestSAXState st = (TestSAXState )ctx;
	st->endCount++;
	}

	static void test_characters(void ctx, const xmlChar ch, int len) {
	TestSAXState st = (TestSAXState )ctx;
	if (st->collectedLen + (size_t)len >= sizeof(st->collected))
	len = (int)(sizeof(st->collected) - st->collectedLen - 1);
	if (len > 0) {
	memcpy(st->collected + st->collectedLen, ch, (size_t)len);
	st->collectedLen += (size_t)len;
	st->collected[st->collectedLen] = '\0';
	}
	st->charsCount++;
	}

	/* Helper to create a push parser context with our SAX handler and state */
	static htmlParserCtxtPtr create_push_ctxt(TestSAXState *state) {
	xmlSAXHandler sax;
	memset(&sax, 0, sizeof(sax));
	sax.startDocument = test_startDocument;
	sax.endDocument = test_endDocument;
	sax.characters = test_characters;

	htmlParserCtxtPtr ctxt = htmlCreatePushParserCtxt(&sax, state, NULL, 0, NULL, XML_CHAR_ENCODING_NONE);
	return ctxt;
	}

	/* Unity hooks */
	void setUp(void) {
	xmlInitParser();
	}

	void tearDown(void) {
	/* Nothing specific per-test */
	}

	/* Tests */

	static void test_htmlParseChunk_invalid_arguments(void) {
	/* NULL context */
	int ret = htmlParseChunk(NULL, "x", 1, 0);
	TEST_ASSERT_EQUAL_INT(XML_ERR_ARGUMENT, ret);

	/* Valid context for further invalid arg checks */
	TestSAXState st;
	memset(&st, 0, sizeof(st));
	htmlParserCtxtPtr ctxt = create_push_ctxt(&st);
	TEST_ASSERT_NOT_NULL(ctxt);
	TEST_ASSERT_NOT_NULL(ctxt->input);
	TEST_ASSERT_NOT_NULL(ctxt->input->buf);

	/* Positive size with NULL chunk */
	ret = htmlParseChunk(ctxt, NULL, 1, 0);
	TEST_ASSERT_EQUAL_INT(XML_ERR_ARGUMENT, ret);

	/* Negative size */
	ret = htmlParseChunk(ctxt, "x", -5, 0);
	TEST_ASSERT_EQUAL_INT(XML_ERR_ARGUMENT, ret);

	htmlFreeParserCtxt(ctxt);
	}

	static void test_htmlParseChunk_partial_then_resume_starts_document(void) {
	TestSAXState st;
	memset(&st, 0, sizeof(st));
	htmlParserCtxtPtr ctxt = create_push_ctxt(&st);
	TEST_ASSERT_NOT_NULL(ctxt);

	/* Push a single byte: not enough to leave XML_PARSER_START when terminate==0 */
	int ret = htmlParseChunk(ctxt, "<", 1, 0);
	TEST_ASSERT_EQUAL_INT(XML_ERR_OK, ret);
	TEST_ASSERT_EQUAL_INT(0, st.startCount);
	TEST_ASSERT_EQUAL_INT(0, st.endCount);

	/* Push the remainder of a simple start tag */
	ret = htmlParseChunk(ctxt, "html>", 5, 0);
	TEST_ASSERT_EQUAL_INT(XML_ERR_OK, ret);
	TEST_ASSERT_EQUAL_INT(1, st.startCount);
	TEST_ASSERT_EQUAL_INT(0, st.endCount);

	/* Push some character data to ensure characters callback fires */
	st.collectedLen = 0;
	st.collected[0] = '\0';
	ret = htmlParseChunk(ctxt, "<body>Hi", 9, 0); /* includes <body>Hi */
	TEST_ASSERT_EQUAL_INT(XML_ERR_OK, ret);
	TEST_ASSERT_TRUE(st.charsCount >= 1);
	TEST_ASSERT_NOT_NULL(strstr(st.collected, "Hi"));

	htmlFreeParserCtxt(ctxt);
	}

	static void test_htmlParseChunk_terminate_calls_endDocument(void) {
	TestSAXState st;
	memset(&st, 0, sizeof(st));
	htmlParserCtxtPtr ctxt = create_push_ctxt(&st);
	TEST_ASSERT_NOT_NULL(ctxt);

	/* Feed a minimal, not fully closed document */
	const char *chunk1 = "<!DOCTYPE html><html><body>Hello";
	int ret = htmlParseChunk(ctxt, chunk1, (int)strlen(chunk1), 0);
	TEST_ASSERT_EQUAL_INT(XML_ERR_OK, ret);
	TEST_ASSERT_EQUAL_INT(1, st.startCount);
	TEST_ASSERT_EQUAL_INT(0, st.endCount);

	/* Terminate: should auto-close and call endDocument */
	ret = htmlParseChunk(ctxt, NULL, 0, 1);
	TEST_ASSERT_EQUAL_INT(XML_ERR_OK, ret);
	TEST_ASSERT_EQUAL_INT(1, st.endCount);

	htmlFreeParserCtxt(ctxt);
	}

	static void test_htmlParseChunk_respects_stop_parser(void) {
	TestSAXState st;
	memset(&st, 0, sizeof(st));
	htmlParserCtxtPtr ctxt = create_push_ctxt(&st);
	TEST_ASSERT_NOT_NULL(ctxt);

	/* Stop the parser via public API */
	xmlStopParser(ctxt);

	int ret = htmlParseChunk(ctxt, "<html>", 6, 0);
	TEST_ASSERT_EQUAL_INT(XML_ERR_USER_STOP, ret);

	htmlFreeParserCtxt(ctxt);
	}

	int main(void) {
	UNITY_BEGIN();
	RUN_TEST(test_htmlParseChunk_invalid_arguments);
	RUN_TEST(test_htmlParseChunk_partial_then_resume_starts_document);
	RUN_TEST(test_htmlParseChunk_terminate_calls_endDocument);
	RUN_TEST(test_htmlParseChunk_respects_stop_parser);
	int rc = UNITY_END();
	xmlCleanupParser();
	return rc;
	}