libxml / tests /tests_HTMLtree_htmlDocDump.c

Upload folder using huggingface_hub

6baed57 verified 12 days ago

8.2 kB

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

	#include <libxml/tree.h>
	#include <stdio.h>
	#include <stdlib.h>
	#include <string.h>

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

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

	/* Helper: create a temporary file path and open it for writing (binary).
	* Returns 1 on success, 0 on failure. On success, outPath is set and outF is opened. /
	static int create_temp_file(char outPath, size_t outPathSize, FILE *outF) {
	if (outPath == NULL \|\| outF == NULL \|\| outPathSize == 0)
	return 0;

	/* Use tmpnam to get a unique path (adequate for test purposes). */
	char tmpbuf[L_tmpnam];
	#if defined(_MSC_VER)
	if (tmpnam_s(tmpbuf, L_tmpnam) != 0)
	return 0;
	#else
	if (tmpnam(tmpbuf) == NULL)
	return 0;
	#endif
	/* Copy to caller-provided buffer */
	strncpy(outPath, tmpbuf, outPathSize - 1);
	outPath[outPathSize - 1] = '\0';

	FILE *f = fopen(outPath, "wb");
	if (f == NULL)
	return 0;

	*outF = f;
	return 1;
	}

	/* Helper: read entire file into a newly allocated buffer; returns NULL on failure.
	* The returned buffer is NUL-terminated; length (optional) is set to the file size. */
	static char read_file_to_string(const char path, long *length) {
	if (path == NULL)
	return NULL;

	FILE *f = fopen(path, "rb");
	if (f == NULL)
	return NULL;

	if (fseek(f, 0, SEEK_END) != 0) {
	fclose(f);
	return NULL;
	}
	long size = ftell(f);
	if (size < 0) {
	fclose(f);
	return NULL;
	}
	rewind(f);

	char buf = (char )malloc((size_t)size + 1);
	if (buf == NULL) {
	fclose(f);
	return NULL;
	}

	size_t nread = fread(buf, 1, (size_t)size, f);
	fclose(f);
	buf[nread] = '\0';
	if (length)
	*length = (long)nread;
	return buf;
	}

	/* Helper: build a simple HTML document with <html><body><p>Hello</p></body></html> */
	static xmlDocPtr build_simple_html_doc(void) {
	xmlDocPtr doc = htmlNewDocNoDtD(NULL, NULL);
	if (doc == NULL)
	return NULL;

	xmlNodePtr root = htmlNewDocNode(doc, NULL, BAD_CAST "html", NULL);
	if (root == NULL) {
	xmlFreeDoc(doc);
	return NULL;
	}
	xmlDocSetRootElement(doc, root);

	xmlNodePtr body = htmlNewDocNode(doc, NULL, BAD_CAST "body", NULL);
	if (body == NULL) {
	xmlFreeDoc(doc);
	return NULL;
	}
	xmlAddChild(root, body);

	xmlNodePtr p = htmlNewDocNode(doc, NULL, BAD_CAST "p", NULL);
	if (p == NULL) {
	xmlFreeDoc(doc);
	return NULL;
	}
	xmlAddChild(body, p);

	xmlNodePtr text = xmlNewText(BAD_CAST "Hello");
	if (text == NULL) {
	xmlFreeDoc(doc);
	return NULL;
	}
	xmlAddChild(p, text);

	return doc;
	}

	/* Test: Passing NULL document should return -1 and not close the file. */
	static void test_htmlDocDump_null_doc_returns_minus1(void) {
	char path[256];
	FILE *f = NULL;
	TEST_ASSERT_TRUE_MESSAGE(create_temp_file(path, sizeof(path), &f), "Failed to create temp file");

	int ret = htmlDocDump(f, NULL);
	TEST_ASSERT_EQUAL_INT_MESSAGE(-1, ret, "htmlDocDump should return -1 for NULL doc");

	/* Since the function returned early, it should not have closed the FILE. We close it now. /
	TEST_ASSERT_EQUAL_INT_MESSAGE(0, fflush(f), "Flushing file failed");
	TEST_ASSERT_EQUAL_INT_MESSAGE(0, fclose(f), "Closing file failed");

	/* File should be empty (no output) */
	long size = 0;
	char *contents = read_file_to_string(path, &size);
	TEST_ASSERT_NOT_NULL_MESSAGE(contents, "Failed to read temp file");
	TEST_ASSERT_EQUAL_INT_MESSAGE(0, (int)size, "File should be empty when doc is NULL");
	free(contents);
	remove(path);
	}

	/* Test: Passing NULL file should return -1. */
	static void test_htmlDocDump_null_file_returns_minus1(void) {
	xmlDocPtr doc = build_simple_html_doc();
	TEST_ASSERT_NOT_NULL_MESSAGE(doc, "Failed to build HTML doc");

	int ret = htmlDocDump(NULL, doc);
	TEST_ASSERT_EQUAL_INT_MESSAGE(-1, ret, "htmlDocDump should return -1 for NULL file");

	xmlFreeDoc(doc);
	}

	/* Test: Unsupported encoding should cause htmlDocDump to return -1 and not write anything. */
	static void test_htmlDocDump_invalid_encoding_returns_minus1_and_no_write(void) {
	xmlDocPtr doc = build_simple_html_doc();
	TEST_ASSERT_NOT_NULL_MESSAGE(doc, "Failed to build HTML doc");

	/* Set an encoding that is very likely unsupported */
	doc->encoding = xmlStrdup(BAD_CAST "x-unknown-encoding-xyz-1234");

	char path[256];
	FILE *f = NULL;
	TEST_ASSERT_TRUE_MESSAGE(create_temp_file(path, sizeof(path), &f), "Failed to create temp file");

	int ret = htmlDocDump(f, doc);
	TEST_ASSERT_EQUAL_INT_MESSAGE(-1, ret, "htmlDocDump should return -1 for unsupported encoding");

	/* The function should have returned before creating/using the output buffer, so FILE* is still open */
	TEST_ASSERT_EQUAL_INT_MESSAGE(0, fflush(f), "Flushing file failed");
	TEST_ASSERT_EQUAL_INT_MESSAGE(0, fclose(f), "Closing file failed");

	long size = -1;
	char *contents = read_file_to_string(path, &size);
	TEST_ASSERT_NOT_NULL_MESSAGE(contents, "Failed to read temp file");
	TEST_ASSERT_EQUAL_INT_MESSAGE(0, (int)size, "File should be empty on encoder failure");
	free(contents);
	remove(path);

	xmlFreeDoc(doc); /* Frees doc->encoding as well */
	}

	/* Test: Successful dump with default (NULL) encoding should write content and return >= 0. */
	static void test_htmlDocDump_writes_content_with_default_encoding(void) {
	xmlDocPtr doc = build_simple_html_doc();
	TEST_ASSERT_NOT_NULL_MESSAGE(doc, "Failed to build HTML doc");

	/* Ensure default (NULL) encoding to trigger fallback to "HTML" */
	doc->encoding = NULL;

	char path[256];
	FILE *f = NULL;
	TEST_ASSERT_TRUE_MESSAGE(create_temp_file(path, sizeof(path), &f), "Failed to create temp file");

	int ret = htmlDocDump(f, doc);
	TEST_ASSERT_MESSAGE(ret >= 0, "htmlDocDump should succeed (ret >= 0) with default encoding");

	/* On success, htmlDocDump likely closed the FILE, so don't fclose(f) here. /

	long size = -1;
	char *contents = read_file_to_string(path, &size);
	TEST_ASSERT_NOT_NULL_MESSAGE(contents, "Failed to read dumped file");
	TEST_ASSERT_MESSAGE(size > 0, "Dumped file should contain data");
	TEST_ASSERT_NOT_NULL_MESSAGE(strstr(contents, "Hello"), "Dumped HTML should contain text 'Hello'");
	/* Also check that <html> tag appears */
	TEST_ASSERT_NOT_NULL_MESSAGE(strstr(contents, "<html"), "Dumped HTML should contain <html> element");

	free(contents);
	remove(path);
	xmlFreeDoc(doc);
	}

	/* Test: Successful dump with explicit UTF-8 encoding should write content and return >= 0. */
	static void test_htmlDocDump_writes_content_with_utf8_encoding(void) {
	xmlDocPtr doc = build_simple_html_doc();
	TEST_ASSERT_NOT_NULL_MESSAGE(doc, "Failed to build HTML doc");

	/* Set UTF-8 encoding explicitly */
	doc->encoding = xmlStrdup(BAD_CAST "UTF-8");

	char path[256];
	FILE *f = NULL;
	TEST_ASSERT_TRUE_MESSAGE(create_temp_file(path, sizeof(path), &f), "Failed to create temp file");

	int ret = htmlDocDump(f, doc);
	TEST_ASSERT_MESSAGE(ret >= 0, "htmlDocDump should succeed (ret >= 0) with UTF-8 encoding");

	long size = -1;
	char *contents = read_file_to_string(path, &size);
	TEST_ASSERT_NOT_NULL_MESSAGE(contents, "Failed to read dumped file");
	TEST_ASSERT_MESSAGE(size > 0, "Dumped file should contain data");
	TEST_ASSERT_NOT_NULL_MESSAGE(strstr(contents, "Hello"), "Dumped HTML should contain text 'Hello'");

	free(contents);
	remove(path);
	xmlFreeDoc(doc);
	}

	/* Main runner */
	int main(void) {
	UNITY_BEGIN();
	RUN_TEST(test_htmlDocDump_null_doc_returns_minus1);
	RUN_TEST(test_htmlDocDump_null_file_returns_minus1);
	RUN_TEST(test_htmlDocDump_invalid_encoding_returns_minus1_and_no_write);
	RUN_TEST(test_htmlDocDump_writes_content_with_default_encoding);
	RUN_TEST(test_htmlDocDump_writes_content_with_utf8_encoding);
	return UNITY_END();
	}