Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
38.76% |
69 / 178 |
|
25.00% |
4 / 16 |
CRAP | |
0.00% |
0 / 1 |
| XmlTypeCheck | |
38.98% |
69 / 177 |
|
25.00% |
4 / 16 |
1150.56 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| newFromFilename | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| newFromString | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getRootElement | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| validateFromInput | |
70.59% |
12 / 17 |
|
0.00% |
0 / 1 |
5.64 | |||
| readNext | |
71.43% |
5 / 7 |
|
0.00% |
0 / 1 |
1.02 | |||
| validate | |
66.67% |
26 / 39 |
|
0.00% |
0 / 1 |
32.37 | |||
| getAttributesArray | |
42.86% |
3 / 7 |
|
0.00% |
0 / 1 |
4.68 | |||
| expandNS | |
40.00% |
2 / 5 |
|
0.00% |
0 / 1 |
2.86 | |||
| elementOpen | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| elementClose | |
66.67% |
6 / 9 |
|
0.00% |
0 / 1 |
3.33 | |||
| elementData | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| processingInstructionHandler | |
77.78% |
7 / 9 |
|
0.00% |
0 / 1 |
3.10 | |||
| dtdHandler | |
0.00% |
0 / 26 |
|
0.00% |
0 / 1 |
156 | |||
| checkDTDIsSafe | |
0.00% |
0 / 10 |
|
0.00% |
0 / 1 |
2 | |||
| parseDTD | |
0.00% |
0 / 38 |
|
0.00% |
0 / 1 |
210 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * @license GPL-2.0-or-later |
| 4 | * @file |
| 5 | */ |
| 6 | |
| 7 | namespace Wikimedia\Mime; |
| 8 | |
| 9 | use Exception; |
| 10 | use XMLReader; |
| 11 | |
| 12 | /** |
| 13 | * XML syntax and type checker. |
| 14 | * |
| 15 | * Since MediaWiki 1.24.2, this uses XMLReader instead of xml_parse, which gives us |
| 16 | * more control over the expansion of XML entities. When passed to the |
| 17 | * callback, entities will be fully expanded, but may report the XML is |
| 18 | * invalid if expanding the entities are likely to cause a DoS. |
| 19 | * |
| 20 | * @newable |
| 21 | * @since 1.12.0 |
| 22 | * @ingroup Mime |
| 23 | */ |
| 24 | class XmlTypeCheck { |
| 25 | /** |
| 26 | * @var bool|null Will be set to true or false to indicate whether the file is |
| 27 | * well-formed XML. Note that this doesn't check schema validity. |
| 28 | */ |
| 29 | public $wellFormed = null; |
| 30 | |
| 31 | /** |
| 32 | * @var bool Will be set to true if the optional element filter returned |
| 33 | * a match at some point. |
| 34 | */ |
| 35 | public $filterMatch = false; |
| 36 | |
| 37 | /** |
| 38 | * Will contain the type of filter hit if the optional element filter returned |
| 39 | * a match at some point. |
| 40 | * @var mixed |
| 41 | */ |
| 42 | public $filterMatchType = false; |
| 43 | |
| 44 | /** |
| 45 | * @var string Name of the document's root element, including any namespace |
| 46 | * as an expanded URL. |
| 47 | */ |
| 48 | public $rootElement = ''; |
| 49 | |
| 50 | /** |
| 51 | * @var string[] A stack of strings containing the data of each xml element as it's processed. |
| 52 | * Append data to the top string of the stack, then pop off the string and process it when the |
| 53 | * element is closed. |
| 54 | */ |
| 55 | protected $elementData = []; |
| 56 | |
| 57 | /** |
| 58 | * @var array A stack of element names and attributes, as we process them. |
| 59 | */ |
| 60 | protected $elementDataContext = []; |
| 61 | |
| 62 | /** |
| 63 | * @var int Current depth of the data stack. |
| 64 | */ |
| 65 | protected $stackDepth = 0; |
| 66 | |
| 67 | /** @var callable|null */ |
| 68 | protected $filterCallback; |
| 69 | |
| 70 | /** |
| 71 | * @var array Additional parsing options |
| 72 | */ |
| 73 | private $parserOptions = [ |
| 74 | 'processing_instruction_handler' => null, |
| 75 | 'external_dtd_handler' => '', |
| 76 | 'dtd_handler' => '', |
| 77 | 'require_safe_dtd' => true |
| 78 | ]; |
| 79 | |
| 80 | /** |
| 81 | * Allow filtering an XML file. |
| 82 | * |
| 83 | * Filters should return either true or a string to indicate something |
| 84 | * is wrong with the file. $this->filterMatch will store if the |
| 85 | * file failed validation (true = failed validation). |
| 86 | * $this->filterMatchType will contain the validation error. |
| 87 | * $this->wellFormed will contain whether the xml file is well-formed. |
| 88 | * |
| 89 | * @note If multiple filters are hit, only one of them will have the |
| 90 | * result stored in $this->filterMatchType. |
| 91 | * |
| 92 | * @param string $input a filename or string containing the XML element |
| 93 | * @param callable|null $filterCallback (optional) |
| 94 | * Function to call to do additional custom validity checks from the |
| 95 | * SAX element handler event. This gives you access to the element |
| 96 | * namespace, name, attributes, and text contents. |
| 97 | * Filter should return a truthy value describing the error. |
| 98 | * @param bool $isFile (optional) indicates if the first parameter is a |
| 99 | * filename (default, true) or if it is a string (false) |
| 100 | * @param array $options list of additional parsing options: |
| 101 | * processing_instruction_handler: Callback for xml_set_processing_instruction_handler |
| 102 | * external_dtd_handler: Callback for the url of external dtd subset |
| 103 | * dtd_handler: Callback given the full text of the <!DOCTYPE declaration. |
| 104 | * require_safe_dtd: Only allow non-recursive entities in internal dtd (default true) |
| 105 | */ |
| 106 | public function __construct( $input, $filterCallback = null, $isFile = true, $options = [] ) { |
| 107 | $this->filterCallback = $filterCallback; |
| 108 | $this->parserOptions = array_merge( $this->parserOptions, $options ); |
| 109 | $this->validateFromInput( $input, $isFile ); |
| 110 | } |
| 111 | |
| 112 | /** |
| 113 | * Alternative constructor: from filename |
| 114 | * |
| 115 | * @param string $fname the filename of an XML document |
| 116 | * @param callable|null $filterCallback (optional) |
| 117 | * Function to call to do additional custom validity checks from the |
| 118 | * SAX element handler event. This gives you access to the element |
| 119 | * namespace, name, and attributes, but not to text contents. |
| 120 | * Filter should return 'true' to toggle on $this->filterMatch |
| 121 | * @return XmlTypeCheck |
| 122 | */ |
| 123 | public static function newFromFilename( $fname, $filterCallback = null ) { |
| 124 | return new self( $fname, $filterCallback, true ); |
| 125 | } |
| 126 | |
| 127 | /** |
| 128 | * Alternative constructor: from string |
| 129 | * |
| 130 | * @param string $string a string containing an XML element |
| 131 | * @param callable|null $filterCallback (optional) |
| 132 | * Function to call to do additional custom validity checks from the |
| 133 | * SAX element handler event. This gives you access to the element |
| 134 | * namespace, name, and attributes, but not to text contents. |
| 135 | * Filter should return 'true' to toggle on $this->filterMatch |
| 136 | * @return XmlTypeCheck |
| 137 | */ |
| 138 | public static function newFromString( $string, $filterCallback = null ) { |
| 139 | return new self( $string, $filterCallback, false ); |
| 140 | } |
| 141 | |
| 142 | /** |
| 143 | * Get the root element. Simple accessor to $rootElement |
| 144 | * |
| 145 | * @return string |
| 146 | */ |
| 147 | public function getRootElement() { |
| 148 | return $this->rootElement; |
| 149 | } |
| 150 | |
| 151 | /** |
| 152 | * @param string $xml |
| 153 | * @param bool $isFile |
| 154 | */ |
| 155 | private function validateFromInput( $xml, $isFile ) { |
| 156 | // Allow text and attr nodes over 10 MB, e.g. embedded embedded raster images in SVG (T387969). |
| 157 | $xmlParseHuge = LIBXML_PARSEHUGE; |
| 158 | if ( defined( 'MW_PHPUNIT_TEST' ) ) { |
| 159 | // Use low limits while running tests, because XmlTypeCheckTest::testRecursiveEntity() |
| 160 | // requires over 1 GB of memory and over a minute of time with huge limits (T392782). |
| 161 | // Maybe this should be configurable for low-resource deployments? |
| 162 | $xmlParseHuge = 0; |
| 163 | } |
| 164 | $reader = new XMLReader(); |
| 165 | if ( $isFile ) { |
| 166 | $s = $reader->open( $xml, null, LIBXML_NOERROR | LIBXML_NOWARNING | $xmlParseHuge ); |
| 167 | } else { |
| 168 | $s = $reader->XML( $xml, null, LIBXML_NOERROR | LIBXML_NOWARNING | $xmlParseHuge ); |
| 169 | } |
| 170 | if ( $s !== true ) { |
| 171 | // Couldn't open the XML |
| 172 | $this->wellFormed = false; |
| 173 | } else { |
| 174 | // phpcs:ignore Generic.PHP.NoSilencedErrors -- suppress deprecation per T268847 |
| 175 | $oldDisable = @libxml_disable_entity_loader( true ); |
| 176 | $reader->setParserProperty( XMLReader::SUBST_ENTITIES, true ); |
| 177 | try { |
| 178 | $this->validate( $reader ); |
| 179 | } catch ( Exception $e ) { |
| 180 | // Calling this malformed, because we didn't parse the whole |
| 181 | // thing. Maybe just an external entity refernce. |
| 182 | $this->wellFormed = false; |
| 183 | throw $e; |
| 184 | } finally { |
| 185 | $reader->close(); |
| 186 | // phpcs:ignore Generic.PHP.NoSilencedErrors |
| 187 | @libxml_disable_entity_loader( $oldDisable ); |
| 188 | } |
| 189 | } |
| 190 | } |
| 191 | |
| 192 | private function readNext( XMLReader $reader ): bool { |
| 193 | set_error_handler( function ( $line, $file ) { |
| 194 | $this->wellFormed = false; |
| 195 | return true; |
| 196 | } ); |
| 197 | $ret = $reader->read(); |
| 198 | restore_error_handler(); |
| 199 | return $ret; |
| 200 | } |
| 201 | |
| 202 | private function validate( XMLReader $reader ) { |
| 203 | // First, move through anything that isn't an element, and |
| 204 | // handle any processing instructions with the callback |
| 205 | do { |
| 206 | if ( !$this->readNext( $reader ) ) { |
| 207 | // Hit the end of the document before any elements |
| 208 | $this->wellFormed = false; |
| 209 | return; |
| 210 | } |
| 211 | if ( $reader->nodeType === XMLReader::PI ) { |
| 212 | $this->processingInstructionHandler( $reader->name, $reader->value ); |
| 213 | } |
| 214 | if ( $reader->nodeType === XMLReader::DOC_TYPE ) { |
| 215 | $this->dtdHandler( $reader ); |
| 216 | } |
| 217 | } while ( $reader->nodeType != XMLReader::ELEMENT ); |
| 218 | |
| 219 | // Process the rest of the document |
| 220 | do { |
| 221 | switch ( $reader->nodeType ) { |
| 222 | case XMLReader::ELEMENT: |
| 223 | $name = $this->expandNS( |
| 224 | $reader->name, |
| 225 | $reader->namespaceURI |
| 226 | ); |
| 227 | if ( $this->rootElement === '' ) { |
| 228 | $this->rootElement = $name; |
| 229 | } |
| 230 | $empty = $reader->isEmptyElement; |
| 231 | $attrs = $this->getAttributesArray( $reader ); |
| 232 | $this->elementOpen( $name, $attrs ); |
| 233 | if ( $empty ) { |
| 234 | $this->elementClose(); |
| 235 | } |
| 236 | break; |
| 237 | |
| 238 | case XMLReader::END_ELEMENT: |
| 239 | $this->elementClose(); |
| 240 | break; |
| 241 | |
| 242 | case XMLReader::WHITESPACE: |
| 243 | case XMLReader::SIGNIFICANT_WHITESPACE: |
| 244 | case XMLReader::CDATA: |
| 245 | case XMLReader::TEXT: |
| 246 | $this->elementData( $reader->value ); |
| 247 | break; |
| 248 | |
| 249 | case XMLReader::ENTITY_REF: |
| 250 | // Unexpanded entity (maybe external?), |
| 251 | // don't send to the filter (xml_parse didn't) |
| 252 | break; |
| 253 | |
| 254 | case XMLReader::COMMENT: |
| 255 | // Don't send to the filter (xml_parse didn't) |
| 256 | break; |
| 257 | |
| 258 | case XMLReader::PI: |
| 259 | // Processing instructions can happen after the header too |
| 260 | $this->processingInstructionHandler( |
| 261 | $reader->name, |
| 262 | $reader->value |
| 263 | ); |
| 264 | break; |
| 265 | case XMLReader::DOC_TYPE: |
| 266 | // We should never see a doctype after first |
| 267 | // element. |
| 268 | $this->wellFormed = false; |
| 269 | break; |
| 270 | default: |
| 271 | // One of DOC, ENTITY, END_ENTITY, |
| 272 | // NOTATION, or XML_DECLARATION |
| 273 | // xml_parse didn't send these to the filter, so we won't. |
| 274 | } |
| 275 | } while ( $this->readNext( $reader ) ); |
| 276 | |
| 277 | if ( $this->stackDepth !== 0 ) { |
| 278 | $this->wellFormed = false; |
| 279 | } elseif ( $this->wellFormed === null ) { |
| 280 | $this->wellFormed = true; |
| 281 | } |
| 282 | } |
| 283 | |
| 284 | /** |
| 285 | * Get all of the attributes for an XMLReader's current node |
| 286 | * @param XMLReader $r |
| 287 | * @return array of attributes |
| 288 | */ |
| 289 | private function getAttributesArray( XMLReader $r ) { |
| 290 | $attrs = []; |
| 291 | while ( $r->moveToNextAttribute() ) { |
| 292 | if ( $r->namespaceURI === 'http://www.w3.org/2000/xmlns/' ) { |
| 293 | // XMLReader treats xmlns attributes as normal |
| 294 | // attributes, while xml_parse doesn't |
| 295 | continue; |
| 296 | } |
| 297 | $name = $this->expandNS( $r->name, $r->namespaceURI ); |
| 298 | $attrs[$name] = $r->value; |
| 299 | } |
| 300 | return $attrs; |
| 301 | } |
| 302 | |
| 303 | /** |
| 304 | * @param string $name element or attribute name, maybe with a full or short prefix |
| 305 | * @param string $namespaceURI |
| 306 | * @return string the name prefixed with namespaceURI |
| 307 | */ |
| 308 | private function expandNS( $name, $namespaceURI ) { |
| 309 | if ( $namespaceURI ) { |
| 310 | $parts = explode( ':', $name ); |
| 311 | $localname = array_pop( $parts ); |
| 312 | return "$namespaceURI:$localname"; |
| 313 | } |
| 314 | return $name; |
| 315 | } |
| 316 | |
| 317 | /** |
| 318 | * @param string $name |
| 319 | * @param array $attribs |
| 320 | */ |
| 321 | private function elementOpen( $name, $attribs ) { |
| 322 | $this->elementDataContext[] = [ $name, $attribs ]; |
| 323 | $this->elementData[] = ''; |
| 324 | $this->stackDepth++; |
| 325 | } |
| 326 | |
| 327 | private function elementClose() { |
| 328 | [ $name, $attribs ] = array_pop( $this->elementDataContext ); |
| 329 | $data = array_pop( $this->elementData ); |
| 330 | $this->stackDepth--; |
| 331 | $callbackReturn = false; |
| 332 | |
| 333 | if ( is_callable( $this->filterCallback ) ) { |
| 334 | $callbackReturn = ( $this->filterCallback )( $name, $attribs, $data ); |
| 335 | } |
| 336 | if ( $callbackReturn ) { |
| 337 | // Filter hit! |
| 338 | $this->filterMatch = true; |
| 339 | $this->filterMatchType = $callbackReturn; |
| 340 | } |
| 341 | } |
| 342 | |
| 343 | /** |
| 344 | * @param string $data |
| 345 | */ |
| 346 | private function elementData( $data ) { |
| 347 | // Collect any data here, and we'll run the callback in elementClose |
| 348 | $this->elementData[ $this->stackDepth - 1 ] .= trim( $data ); |
| 349 | } |
| 350 | |
| 351 | /** |
| 352 | * @param string $target |
| 353 | * @param string $data |
| 354 | */ |
| 355 | private function processingInstructionHandler( $target, $data ) { |
| 356 | $callbackReturn = false; |
| 357 | if ( $this->parserOptions['processing_instruction_handler'] ) { |
| 358 | $callbackReturn = $this->parserOptions['processing_instruction_handler']( |
| 359 | $target, |
| 360 | $data |
| 361 | ); |
| 362 | } |
| 363 | if ( $callbackReturn ) { |
| 364 | // Filter hit! |
| 365 | $this->filterMatch = true; |
| 366 | $this->filterMatchType = $callbackReturn; |
| 367 | } |
| 368 | } |
| 369 | |
| 370 | /** |
| 371 | * Handle coming across a <!DOCTYPE declaration. |
| 372 | * |
| 373 | * @param XMLReader $reader Reader currently pointing at DOCTYPE node. |
| 374 | */ |
| 375 | private function dtdHandler( XMLReader $reader ) { |
| 376 | $externalCallback = $this->parserOptions['external_dtd_handler']; |
| 377 | $generalCallback = $this->parserOptions['dtd_handler']; |
| 378 | $checkIfSafe = $this->parserOptions['require_safe_dtd']; |
| 379 | if ( !$externalCallback && !$generalCallback && !$checkIfSafe ) { |
| 380 | return; |
| 381 | } |
| 382 | $dtd = $reader->readOuterXml(); |
| 383 | $callbackReturn = false; |
| 384 | |
| 385 | if ( $generalCallback ) { |
| 386 | $callbackReturn = $generalCallback( $dtd ); |
| 387 | } |
| 388 | if ( $callbackReturn ) { |
| 389 | // Filter hit! |
| 390 | $this->filterMatch = true; |
| 391 | $this->filterMatchType = $callbackReturn; |
| 392 | $callbackReturn = false; |
| 393 | } |
| 394 | |
| 395 | $parsedDTD = $this->parseDTD( $dtd ); |
| 396 | if ( $externalCallback && isset( $parsedDTD['type'] ) ) { |
| 397 | $callbackReturn = $externalCallback( |
| 398 | $parsedDTD['type'], |
| 399 | $parsedDTD['publicid'] ?? null, |
| 400 | $parsedDTD['systemid'] ?? null |
| 401 | ); |
| 402 | } |
| 403 | if ( $callbackReturn ) { |
| 404 | // Filter hit! |
| 405 | $this->filterMatch = true; |
| 406 | $this->filterMatchType = $callbackReturn; |
| 407 | } |
| 408 | |
| 409 | if ( $checkIfSafe && isset( $parsedDTD['internal'] ) && |
| 410 | !$this->checkDTDIsSafe( $parsedDTD['internal'] ) |
| 411 | ) { |
| 412 | $this->wellFormed = false; |
| 413 | } |
| 414 | } |
| 415 | |
| 416 | /** |
| 417 | * Check if the internal subset of the DTD is safe. |
| 418 | * |
| 419 | * We allow an extremely restricted subset of DTD features. |
| 420 | * |
| 421 | * Safe is defined as: |
| 422 | * * Only contains entity definitions (e.g. No <!ATLIST ) |
| 423 | * * Entity definitions are not "system" entities |
| 424 | * * Entity definitions are not "parameter" (i.e. %) entities |
| 425 | * * Entity definitions do not reference other entities except & |
| 426 | * and quotes. Entity aliases (where the entity contains only |
| 427 | * another entity are allowed) |
| 428 | * * Entity references aren't overly long (>255 bytes). |
| 429 | * * <!ATTLIST svg xmlns:xlink CDATA #FIXED "http://www.w3.org/1999/xlink"> |
| 430 | * allowed if matched exactly for compatibility with graphviz |
| 431 | * * Comments. |
| 432 | * |
| 433 | * @param string $internalSubset The internal subset of the DTD |
| 434 | * @return bool true if safe. |
| 435 | */ |
| 436 | private function checkDTDIsSafe( $internalSubset ) { |
| 437 | $res = preg_match( |
| 438 | '/^(?:\s*<!ENTITY\s+\S+\s+' . |
| 439 | '(?:"(?:&[^"%&;]{1,64};|(?:[^"%&]|&|"){0,255})"' . |
| 440 | '|\'(?:&[^\'%&;]{1,64};|(?:[^\'%&]|&|'){0,255})\')\s*>' . |
| 441 | '|\s*<!--(?:[^-]|-[^-])*-->' . |
| 442 | '|\s*<!ATTLIST svg xmlns:xlink CDATA #FIXED ' . |
| 443 | '"http:\/\/www.w3.org\/1999\/xlink">)*\s*$/', |
| 444 | $internalSubset |
| 445 | ); |
| 446 | |
| 447 | return (bool)$res; |
| 448 | } |
| 449 | |
| 450 | /** |
| 451 | * Parse DTD into parts. |
| 452 | * |
| 453 | * If there is an error parsing the dtd, sets wellFormed to false. |
| 454 | * |
| 455 | * @param string $dtd |
| 456 | * @return array Possibly containing keys publicid, systemid, type and internal. |
| 457 | */ |
| 458 | private function parseDTD( $dtd ) { |
| 459 | $m = []; |
| 460 | $res = preg_match( |
| 461 | '/^<!DOCTYPE\s*\S+\s*' . |
| 462 | '(?:(?P<typepublic>PUBLIC)\s*' . |
| 463 | '(?:"(?P<pubquote>[^"]*)"|\'(?P<pubapos>[^\']*)\')' . // public identifer |
| 464 | '\s*"(?P<pubsysquote>[^"]*)"|\'(?P<pubsysapos>[^\']*)\'' . // system identifier |
| 465 | '|(?P<typesystem>SYSTEM)\s*' . |
| 466 | '(?:"(?P<sysquote>[^"]*)"|\'(?P<sysapos>[^\']*)\')' . |
| 467 | ')?\s*' . |
| 468 | '(?:\[\s*(?P<internal>.*)\])?\s*>$/s', |
| 469 | $dtd, |
| 470 | $m |
| 471 | ); |
| 472 | if ( !$res ) { |
| 473 | $this->wellFormed = false; |
| 474 | return []; |
| 475 | } |
| 476 | $parsed = []; |
| 477 | foreach ( $m as $field => $value ) { |
| 478 | if ( $value === '' || is_numeric( $field ) ) { |
| 479 | continue; |
| 480 | } |
| 481 | switch ( $field ) { |
| 482 | case 'typepublic': |
| 483 | case 'typesystem': |
| 484 | $parsed['type'] = $value; |
| 485 | break; |
| 486 | case 'pubquote': |
| 487 | case 'pubapos': |
| 488 | $parsed['publicid'] = $value; |
| 489 | break; |
| 490 | case 'pubsysquote': |
| 491 | case 'pubsysapos': |
| 492 | case 'sysquote': |
| 493 | case 'sysapos': |
| 494 | $parsed['systemid'] = $value; |
| 495 | break; |
| 496 | case 'internal': |
| 497 | $parsed['internal'] = $value; |
| 498 | break; |
| 499 | } |
| 500 | } |
| 501 | return $parsed; |
| 502 | } |
| 503 | } |
| 504 | |
| 505 | /** @deprecated class alias since 1.43 */ |
| 506 | class_alias( XmlTypeCheck::class, 'XmlTypeCheck' ); |