1 /*
2 +----------------------------------------------------------------------+
3 | Copyright (c) The PHP Group |
4 +----------------------------------------------------------------------+
5 | This source file is subject to version 3.01 of the PHP license, |
6 | that is bundled with this package in the file LICENSE, and is |
7 | available through the world-wide-web at the following url: |
8 | https://www.php.net/license/3_01.txt |
9 | If you did not receive a copy of the PHP license and are unable to |
10 | obtain it through the world-wide-web, please send a note to |
11 | license@php.net so we can mail you a copy immediately. |
12 +----------------------------------------------------------------------+
13 | Authors: Niels Dossche <nielsdos@php.net> |
14 +----------------------------------------------------------------------+
15 */
16
17 #ifdef HAVE_CONFIG_H
18 #include <config.h>
19 #endif
20
21 #include "php.h"
22 #if defined(HAVE_LIBXML) && defined(HAVE_DOM)
23 #include "php_dom.h"
24 #include "namespace_compat.h"
25 #include "private_data.h"
26 #include "xml_serializer.h"
27 #include <libxml/xmlsave.h>
28
check_options_validity(uint32_t arg_num,zend_long options)29 static bool check_options_validity(uint32_t arg_num, zend_long options)
30 {
31 const zend_long VALID_OPTIONS = XML_PARSE_RECOVER
32 | XML_PARSE_NOENT
33 #if LIBXML_VERSION >= 21300
34 | XML_PARSE_NO_XXE
35 #endif
36 | XML_PARSE_DTDLOAD
37 | XML_PARSE_DTDATTR
38 | XML_PARSE_DTDVALID
39 | XML_PARSE_NOERROR
40 | XML_PARSE_NOWARNING
41 | XML_PARSE_NOBLANKS
42 | XML_PARSE_XINCLUDE
43 | XML_PARSE_NSCLEAN
44 | XML_PARSE_NOCDATA
45 | XML_PARSE_NONET
46 | XML_PARSE_PEDANTIC
47 | XML_PARSE_COMPACT
48 | XML_PARSE_HUGE
49 | XML_PARSE_BIG_LINES;
50 if ((options & ~VALID_OPTIONS) != 0) {
51 zend_argument_value_error(arg_num, "contains invalid flags (allowed flags: "
52 "LIBXML_RECOVER, "
53 "LIBXML_NOENT, "
54 #if LIBXML_VERSION >= 21300
55 "LIBXML_NO_XXE, "
56 #endif
57 "LIBXML_DTDLOAD, "
58 "LIBXML_DTDATTR, "
59 "LIBXML_DTDVALID, "
60 "LIBXML_NOERROR, "
61 "LIBXML_NOWARNING, "
62 "LIBXML_NOBLANKS, "
63 "LIBXML_XINCLUDE, "
64 "LIBXML_NSCLEAN, "
65 "LIBXML_NOCDATA, "
66 "LIBXML_NONET, "
67 "LIBXML_PEDANTIC, "
68 "LIBXML_COMPACT, "
69 "LIBXML_PARSEHUGE, "
70 "LIBXML_BIGLINES)");
71 return false;
72 }
73 return true;
74 }
75
76 /* Living spec never creates explicit namespace declaration nodes.
77 * They are only written upon serialization but never appear in the tree.
78 * So in principle we could just ignore them outright.
79 * However, step 10 in https://html.spec.whatwg.org/multipage/parsing.html#create-an-element-for-the-token (Date 2023-12-15)
80 * requires us to have the declaration as an attribute available */
dom_mark_namespaces_as_attributes_too(php_dom_libxml_ns_mapper * ns_mapper,xmlDocPtr doc)81 void dom_mark_namespaces_as_attributes_too(php_dom_libxml_ns_mapper *ns_mapper, xmlDocPtr doc)
82 {
83 xmlNodePtr node = doc->children;
84 while (node != NULL) {
85 if (node->type == XML_ELEMENT_NODE) {
86 php_dom_ns_compat_mark_attribute_list(ns_mapper, node);
87 }
88
89 node = php_dom_next_in_tree_order(node, NULL);
90 }
91 }
92
PHP_METHOD(Dom_XMLDocument,createEmpty)93 PHP_METHOD(Dom_XMLDocument, createEmpty)
94 {
95 const char *version = NULL;
96 size_t encoding_len = strlen("UTF-8");
97 const char *encoding = "UTF-8";
98 size_t version_len;
99 if (zend_parse_parameters(ZEND_NUM_ARGS(), "|sp", &version, &version_len, &encoding, &encoding_len) == FAILURE) {
100 RETURN_THROWS();
101 }
102
103 xmlCharEncodingHandlerPtr handler = xmlFindCharEncodingHandler(encoding);
104
105 if (handler != NULL) {
106 xmlCharEncCloseFunc(handler);
107 } else {
108 zend_argument_value_error(2, "is not a valid document encoding");
109 RETURN_THROWS();
110 }
111
112 xmlDocPtr lxml_doc = xmlNewDoc((const xmlChar *) version);
113 if (UNEXPECTED(lxml_doc == NULL)) {
114 goto oom;
115 }
116
117 lxml_doc->encoding = xmlStrdup((const xmlChar *) encoding);
118
119 dom_object *intern = php_dom_instantiate_object_helper(
120 return_value,
121 dom_xml_document_class_entry,
122 (xmlNodePtr) lxml_doc,
123 NULL
124 );
125 dom_set_xml_class(intern->document);
126 intern->document->private_data = php_dom_libxml_private_data_header(php_dom_private_data_create());
127 return;
128
129 oom:
130 php_dom_throw_error(INVALID_STATE_ERR, true);
131 RETURN_THROWS();
132 }
133
load_from_helper(INTERNAL_FUNCTION_PARAMETERS,int mode)134 static void load_from_helper(INTERNAL_FUNCTION_PARAMETERS, int mode)
135 {
136 const char *source, *override_encoding = NULL;
137 size_t source_len, override_encoding_len;
138 zend_long options = 0;
139 if (zend_parse_parameters(
140 ZEND_NUM_ARGS(),
141 "s|lp!",
142 &source,
143 &source_len,
144 &options,
145 &override_encoding,
146 &override_encoding_len
147 ) == FAILURE) {
148 RETURN_THROWS();
149 }
150
151 if (!source_len) {
152 zend_argument_value_error(1, "must not be empty");
153 RETURN_THROWS();
154 }
155
156 if (ZEND_SIZE_T_INT_OVFL(source_len)) {
157 zend_argument_value_error(1, "is too long");
158 RETURN_THROWS();
159 }
160
161 /* See php_libxml_streams_IO_open_wrapper(), apparently this caused issues in the past. */
162 if (mode == DOM_LOAD_FILE && strstr(source, "%00")) {
163 zend_argument_value_error(1, "must not contain percent-encoded NUL bytes");
164 RETURN_THROWS();
165 }
166
167 if (!check_options_validity(2, options)) {
168 RETURN_THROWS();
169 }
170
171 xmlCharEncodingHandlerPtr encoding = NULL;
172 if (override_encoding != NULL) {
173 encoding = xmlFindCharEncodingHandler(override_encoding);
174 if (!encoding) {
175 zend_argument_value_error(3, "must be a valid document encoding");
176 RETURN_THROWS();
177 }
178 options |= XML_PARSE_IGNORE_ENC;
179 }
180
181 xmlDocPtr lxml_doc = dom_document_parser(NULL, mode, source, source_len, options, encoding);
182 if (UNEXPECTED(lxml_doc == NULL || lxml_doc == DOM_DOCUMENT_MALFORMED)) {
183 if (!EG(exception)) {
184 if (lxml_doc == DOM_DOCUMENT_MALFORMED) {
185 php_dom_throw_error_with_message(SYNTAX_ERR, "XML fragment is not well-formed", true);
186 } else {
187 if (mode == DOM_LOAD_FILE) {
188 zend_throw_exception_ex(NULL, 0, "Cannot open file '%s'", source);
189 } else {
190 php_dom_throw_error(INVALID_STATE_ERR, true);
191 }
192 }
193 }
194 RETURN_THROWS();
195 }
196 if (lxml_doc->encoding == NULL) {
197 if (override_encoding) {
198 lxml_doc->encoding = xmlStrdup((const xmlChar *) override_encoding);
199 } else {
200 lxml_doc->encoding = xmlStrdup((const xmlChar *) "UTF-8");
201 }
202 }
203 if (mode == DOM_LOAD_FILE && lxml_doc->URL != NULL) {
204 if (!php_is_stream_path((char *) lxml_doc->URL)) {
205 /* Check for "file:/" instead of "file://" because of libxml2 quirk */
206 if (strncmp((const char *) lxml_doc->URL, "file:/", sizeof("file:/") - 1) != 0) {
207 #ifdef PHP_WIN32
208 xmlChar *buffer = xmlStrdup((const xmlChar *) "file:///");
209 #else
210 xmlChar *buffer = xmlStrdup((const xmlChar *) "file://");
211 #endif
212 if (buffer != NULL) {
213 xmlChar *new_buffer = xmlStrcat(buffer, lxml_doc->URL);
214 if (new_buffer != NULL) {
215 xmlFree(BAD_CAST lxml_doc->URL);
216 lxml_doc->URL = new_buffer;
217 } else {
218 xmlFree(buffer);
219 }
220 }
221 } else {
222 #ifdef PHP_WIN32
223 lxml_doc->URL = php_dom_libxml_fix_file_path(BAD_CAST lxml_doc->URL);
224 #endif
225 }
226 }
227 }
228 dom_object *intern = php_dom_instantiate_object_helper(
229 return_value,
230 dom_xml_document_class_entry,
231 (xmlNodePtr) lxml_doc,
232 NULL
233 );
234 dom_set_xml_class(intern->document);
235 dom_document_convert_to_modern(intern->document, lxml_doc);
236 }
237
dom_document_convert_to_modern(php_libxml_ref_obj * document,xmlDocPtr lxml_doc)238 void dom_document_convert_to_modern(php_libxml_ref_obj *document, xmlDocPtr lxml_doc)
239 {
240 php_dom_private_data *private_data = php_dom_private_data_create();
241 php_dom_libxml_ns_mapper *ns_mapper = php_dom_ns_mapper_from_private(private_data);
242 document->private_data = php_dom_libxml_private_data_header(private_data);
243 dom_mark_namespaces_as_attributes_too(ns_mapper, lxml_doc);
244 }
245
PHP_METHOD(Dom_XMLDocument,createFromString)246 PHP_METHOD(Dom_XMLDocument, createFromString)
247 {
248 load_from_helper(INTERNAL_FUNCTION_PARAM_PASSTHRU, DOM_LOAD_STRING);
249 }
250
PHP_METHOD(Dom_XMLDocument,createFromFile)251 PHP_METHOD(Dom_XMLDocument, createFromFile)
252 {
253 load_from_helper(INTERNAL_FUNCTION_PARAM_PASSTHRU, DOM_LOAD_FILE);
254 }
255
php_new_dom_write_smart_str(void * context,const char * buffer,int len)256 static int php_new_dom_write_smart_str(void *context, const char *buffer, int len)
257 {
258 smart_str *str = context;
259 smart_str_appendl(str, buffer, len);
260 return len;
261 }
262
get_private_data_from_node(xmlNodePtr node)263 static php_dom_private_data *get_private_data_from_node(xmlNodePtr node)
264 {
265 dom_object *intern = php_dom_object_get_data(node);
266 return intern != NULL ? php_dom_get_private_data(intern) : NULL;
267 }
268
php_new_dom_dump_node_to_str_ex(xmlNodePtr node,int options,bool format,const char * encoding)269 static zend_string *php_new_dom_dump_node_to_str_ex(xmlNodePtr node, int options, bool format, const char *encoding)
270 {
271 smart_str str = {0};
272
273 int status = -1;
274 xmlSaveCtxtPtr ctxt = xmlSaveToIO(php_new_dom_write_smart_str, NULL, &str, encoding, XML_SAVE_AS_XML | options);
275 if (EXPECTED(ctxt != NULL)) {
276 xmlCharEncodingHandlerPtr handler = xmlFindCharEncodingHandler(encoding);
277 xmlOutputBufferPtr out = xmlOutputBufferCreateIO(php_new_dom_write_smart_str, NULL, &str, handler);
278 if (EXPECTED(out != NULL)) {
279 status = dom_xml_serialize(ctxt, out, node, format, false, get_private_data_from_node(node));
280 status |= xmlOutputBufferFlush(out);
281 status |= xmlOutputBufferClose(out);
282 } else {
283 xmlCharEncCloseFunc(handler);
284 }
285 (void) xmlSaveClose(ctxt);
286 }
287
288 if (UNEXPECTED(status < 0)) {
289 smart_str_free_ex(&str, false);
290 return NULL;
291 }
292
293 return smart_str_extract(&str);
294 }
295
php_new_dom_dump_node_to_str(xmlDocPtr doc,xmlNodePtr node,bool format,const char * encoding)296 static zend_string *php_new_dom_dump_node_to_str(xmlDocPtr doc, xmlNodePtr node, bool format, const char *encoding)
297 {
298 return php_new_dom_dump_node_to_str_ex(node, 0, format, encoding);
299 }
300
php_new_dom_dump_doc_to_str(xmlDocPtr doc,int options,const char * encoding)301 static zend_string *php_new_dom_dump_doc_to_str(xmlDocPtr doc, int options, const char *encoding)
302 {
303 return php_new_dom_dump_node_to_str_ex((xmlNodePtr) doc, options, options & XML_SAVE_FORMAT, encoding);
304 }
305
php_new_dom_dump_node_to_file(const char * filename,xmlDocPtr doc,xmlNodePtr node,bool format,const char * encoding)306 zend_long php_new_dom_dump_node_to_file(const char *filename, xmlDocPtr doc, xmlNodePtr node, bool format, const char *encoding)
307 {
308 xmlCharEncodingHandlerPtr handler = xmlFindCharEncodingHandler(encoding);
309 xmlOutputBufferPtr out = xmlOutputBufferCreateFilename(filename, handler, 0);
310 if (!out) {
311 xmlCharEncCloseFunc(handler);
312 return -1;
313 }
314
315 php_stream *stream = out->context;
316
317 int status = -1;
318 xmlSaveCtxtPtr ctxt = xmlSaveToIO(out->writecallback, NULL, stream, encoding, XML_SAVE_AS_XML);
319 if (EXPECTED(ctxt != NULL)) {
320 status = dom_xml_serialize(ctxt, out, node, format, false, get_private_data_from_node(node));
321 status |= xmlOutputBufferFlush(out);
322 (void) xmlSaveClose(ctxt);
323 }
324
325 size_t offset = php_stream_tell(stream);
326
327 (void) xmlOutputBufferClose(out);
328
329 return status < 0 ? status : (zend_long) offset;
330 }
331
php_new_dom_dump_doc_to_file(const char * filename,xmlDocPtr doc,bool format,const char * encoding)332 static zend_long php_new_dom_dump_doc_to_file(const char *filename, xmlDocPtr doc, bool format, const char *encoding)
333 {
334 return php_new_dom_dump_node_to_file(filename, doc, (xmlNodePtr) doc, format, encoding);
335 }
336
337 static const php_libxml_document_handlers php_new_dom_default_document_handlers = {
338 .dump_node_to_str = php_new_dom_dump_node_to_str,
339 .dump_doc_to_str = php_new_dom_dump_doc_to_str,
340 .dump_node_to_file = php_new_dom_dump_node_to_file,
341 .dump_doc_to_file = php_new_dom_dump_doc_to_file,
342 };
343
dom_set_xml_class(php_libxml_ref_obj * document)344 void dom_set_xml_class(php_libxml_ref_obj *document)
345 {
346 document->class_type = PHP_LIBXML_CLASS_MODERN;
347 document->handlers = &php_new_dom_default_document_handlers;
348 }
349
350 #endif /* HAVE_LIBXML && HAVE_DOM */
351
