c-harris /
phpquery
This project does not seem to handle request data directly as such no vulnerable execution paths were found.
include, or for example
via PHP's auto-loading mechanism.
These results are based on our legacy PHP analysis, consider migrating to our new PHP analysis engine instead. Learn more
| 1 | <?php |
||
| 58 | $this->id = $newDocumentID ? $newDocumentID : md5(microtime()); |
||
| 59 | } |
||
| 60 | |||
| 61 | public function load($markup, $contentType = null, $newDocumentID = null) |
||
|
|
|||
| 62 | { |
||
| 63 | // PhpQuery::$documents[$id] = $this; |
||
| 64 | $this->contentType = strtolower($contentType); |
||
| 65 | if ($markup instanceof \DOMDocument) { |
||
| 66 | $this->document = $markup; |
||
| 67 | $this->root = $this->document; |
||
| 72 | $loaded = $this->loadMarkup($markup); |
||
| 73 | } |
||
| 74 | if ($loaded) { |
||
| 75 | // $this->document->formatOutput = true; |
||
| 76 | $this->document->preserveWhiteSpace = true; |
||
| 77 | $this->xpath = new \DOMXPath($this->document); |
||
| 78 | $this->afterMarkupLoad(); |
||
| 79 | return true; |
||
| 80 | // remember last loaded document |
||
| 81 | // return PhpQuery::selectDocument($id); |
||
| 82 | } |
||
| 83 | return false; |
||
| 84 | } |
||
| 145 | } |
||
| 146 | $this->document = new \DOMDocument($version, $charset); |
||
| 147 | $this->charset = $this->document->encoding; |
||
| 148 | // $this->document->encoding = $charset; |
||
| 149 | $this->document->formatOutput = true; |
||
| 150 | $this->document->preserveWhiteSpace = true; |
||
| 151 | } |
||
| 204 | // ok trust the document |
||
| 205 | PhpQuery::debug("DETECTED '$docEncoding'"); |
||
| 206 | // Detected does not match what document says... |
||
| 207 | if ($docEncoding !== $documentCharset) { |
||
| 208 | // Tricky.. |
||
| 209 | } |
||
| 210 | if ($docEncoding !== $requestedCharset) { |
||
| 217 | PhpQuery::debug("TODO: charset conversion without mbstring..."); |
||
| 218 | } |
||
| 219 | } |
||
| 220 | $return = false; |
||
| 221 | if ($this->isDocumentFragment) { |
||
| 222 | PhpQuery::debug("Full markup load (HTML), DocumentFragment detected, using charset '$charset'"); |
||
| 223 | $return = $this->documentFragmentLoadMarkup($this, $charset, $markup); |
||
| 235 | } |
||
| 236 | } |
||
| 237 | if ($return && !$this->contentType) { |
||
| 238 | $this->contentType = 'text/html'; |
||
| 239 | } |
||
| 240 | return $return; |
||
| 241 | } |
||
| 263 | $charset = null; |
||
| 264 | // charset from XML declaration @var string |
||
| 265 | $documentCharset = $this->charsetFromXML($markup); |
||
| 266 | if (!$documentCharset) { |
||
| 267 | if ($this->isXHTML) { |
||
| 268 | // this is XHTML, try to get charset from content-type meta header |
||
| 269 | $documentCharset = $this->charsetFromHTML($markup); |
||
| 270 | if ($documentCharset) { |
||
| 271 | PhpQuery::debug("Full markup load (XML), appending XHTML charset '$documentCharset'"); |
||
| 272 | $this->charsetAppendToXML($markup, $documentCharset); |
||
| 273 | $charset = $documentCharset; |
||
| 274 | } |
||
| 275 | } |
||
| 285 | if (!$charset) { |
||
| 286 | $charset = PhpQuery::$defaultCharset; |
||
| 287 | } |
||
| 288 | if ($requestedCharset && $documentCharset |
||
| 289 | && $requestedCharset != $documentCharset |
||
| 290 | ) { |
||
| 291 | // TODO place for charset conversion |
||
| 292 | // $charset = $requestedCharset; |
||
| 293 | } |
||
| 294 | $return = false; |
||
| 295 | if ($this->isDocumentFragment) { |
||
| 296 | PhpQuery::debug("Full markup load (XML), DocumentFragment detected, using charset '$charset'"); |
||
| 297 | $return = $this->documentFragmentLoadMarkup($this, $charset, $markup); |
||
| 298 | } else { |
||
| 299 | // FIXME ??? |
||
| 300 | if ($isContentTypeXHTML && !$isMarkupXHTML) { |
||
| 301 | if (!$documentCharset) { |
||
| 302 | PhpQuery::debug("Full markup load (XML), appending charset '$charset'"); |
||
| 306 | // see http://pl2.php.net/manual/en/book.dom.php#78929 |
||
| 307 | // LIBXML_DTDLOAD (>= PHP 5.1) |
||
| 308 | // does XML ctalogues works with LIBXML_NONET |
||
| 309 | // $this->document->resolveExternals = true; |
||
| 310 | // TODO test LIBXML_COMPACT for performance improvement |
||
| 311 | // create document |
||
| 312 | $this->documentCreate($charset); |
||
| 313 | if (phpversion() < 5.1) { |
||
| 314 | $this->document->resolveExternals = true; |
||
| 315 | $return = PhpQuery::$debug === 2 ? $this->document->loadXML($markup) |
||
| 316 | : @$this->document->loadXML($markup); |
||
| 317 | } else { |
||
| 318 | /** @link http://pl2.php.net/manual/en/libxml.constants.php */ |
||
| 319 | $libxmlStatic = PhpQuery::$debug === 2 ? LIBXML_DTDLOAD |
||
| 320 | | LIBXML_DTDATTR | LIBXML_NONET |
||
| 321 | : LIBXML_DTDLOAD | LIBXML_DTDATTR | LIBXML_NONET | LIBXML_NOWARNING |
||
| 322 | | LIBXML_NOERROR; |
||
| 323 | $return = $this->document->loadXML($markup, $libxmlStatic); |
||
| 324 | // if (! $return) |
||
| 325 | // $return = $this->document->loadHTML($markup); |
||
| 326 | } |
||
| 327 | if ($return) { |
||
| 328 | $this->root = $this->document; |
||
| 331 | if ($return) { |
||
| 332 | if (!$this->contentType) { |
||
| 333 | if ($this->isXHTML) { |
||
| 334 | $this->contentType = 'application/xhtml+xml'; |
||
| 335 | } else { |
||
| 336 | $this->contentType = 'text/xml'; |
||
| 337 | } |
||
| 338 | } |
||
| 339 | return $return; |
||
| 349 | } |
||
| 350 | // XXX ok ? |
||
| 351 | return strpos($markup, "<!DOCTYPE html") !== false; |
||
| 352 | // return stripos($doctype, 'xhtml') !== false; |
||
| 353 | // $doctype = isset($dom->doctype) && is_object($dom->doctype) |
||
| 354 | // ? $dom->doctype->publicId |
||
| 355 | // : self::$defaultDoctype; |
||
| 356 | } |
||
| 357 | |||
| 358 | protected function isXML($markup) |
||
| 359 | { |
||
| 360 | // return strpos($markup, '<?xml') !== false && stripos($markup, 'xhtml') === false; |
||
| 361 | return strpos(substr($markup, 0, 100), '<' . '?xml') !== false; |
||
| 362 | } |
||
| 363 | |||
| 364 | protected function contentTypeToArray($contentType) |
||
| 365 | { |
||
| 366 | $test = null; |
||
| 367 | $test = $matches = explode(';', trim(strtolower($contentType))); |
||
| 368 | if (isset($matches[1])) { |
||
| 369 | $matches[1] = explode('=', $matches[1]); |
||
| 370 | // strip 'charset=' |
||
| 411 | |||
| 412 | protected function charsetFromXML($markup) |
||
| 413 | { |
||
| 414 | $matches; |
||
| 415 | // find declaration |
||
| 416 | preg_match('@<' . '?xml[^>]+encoding\\s*=\\s*(["|\'])(.*?)\\1@i', $markup, $matches); |
||
| 417 | return isset($matches[2]) ? strtolower($matches[2]) : null; |
||
| 485 | return self::isDocumentFragmentHTML($markup); |
||
| 486 | } |
||
| 487 | |||
| 488 | public function importAttr($value) |
||
| 489 | { |
||
| 490 | // TODO |
||
| 491 | } |
||
| 506 | $source |
||
| 507 | ); |
||
| 508 | } |
||
| 509 | // if (is_array($source)) { |
||
| 510 | // foreach($source as $node) { |
||
| 511 | // if (is_string($node)) { |
||
| 512 | // // string markup |
||
| 513 | // $fake = $this->documentFragmentCreate($node, $sourceCharset); |
||
| 514 | // if ($fake === false) |
||
| 515 | // throw new \Exception("Error loading documentFragment markup"); |
||
| 516 | // else |
||
| 517 | // $return = array_merge($return, |
||
| 518 | // $this->import($fake->root->childNodes) |
||
| 519 | // ); |
||
| 520 | // } else { |
||
| 521 | // $return[] = $this->document->importNode($node, true); |
||
| 522 | // } |
||
| 523 | // } |
||
| 524 | // return $return; |
||
| 525 | // } else { |
||
| 526 | // // string markup |
||
| 527 | // $fake = $this->documentFragmentCreate($source, $sourceCharset); |
||
| 528 | // if ($fake === false) |
||
| 529 | // throw new \Exception("Error loading documentFragment markup"); |
||
| 530 | // else |
||
| 531 | // return $this->import($fake->root->childNodes); |
||
| 532 | // } |
||
| 533 | if (is_array($source) || $source instanceof \DOMNodeList) { |
||
| 534 | // dom nodes |
||
| 565 | if (!$charset) { |
||
| 566 | $charset = $this->charset; |
||
| 567 | } |
||
| 568 | // $fake->documentCreate($this->charset); |
||
| 569 | if ($source instanceof \DOMNode && !($source instanceof \DOMNodeList)) { |
||
| 570 | $source = array( |
||
| 571 | $source |
||
| 575 | // dom nodes |
||
| 576 | // load fake document |
||
| 577 | if (!$this->documentFragmentLoadMarkup($fake, $charset)) { |
||
| 578 | return false; |
||
| 579 | } |
||
| 580 | $nodes = $fake->import($source); |
||
| 581 | foreach ($nodes as $node) { |
||
| 592 | * |
||
| 593 | * @param $document DOMDocumentWrapper |
||
| 594 | * @param $markup |
||
| 595 | * @return $document |
||
| 596 | */ |
||
| 597 | private function documentFragmentLoadMarkup($fragment, $charset, $markup = null) |
||
| 598 | { |
||
| 651 | $markup = $fragment->markup(); |
||
| 652 | if ($fragment->isXML) { |
||
| 653 | $markup = substr($markup, 0, strrpos($markup, '</fake>')); |
||
| 654 | View Code Duplication | if ($fragment->isXHTML) { |
|
| 655 | $markup = substr($markup, strpos($markup, '<fake') + 43); |
||
| 656 | } else { |
||
| 657 | $markup = substr($markup, strpos($markup, '<fake>') + 6); |
||
| 658 | } |
||
| 659 | View Code Duplication | } else { |
|
| 660 | $markup = substr($markup, strpos($markup, '<body>') + 6); |
||
| 661 | $markup = substr($markup, 0, strrpos($markup, '</body>')); |
||
| 662 | } |
||
| 690 | if ($this->isDocumentFragment && !$innerMarkup) { |
||
| 691 | foreach ($nodes as $i => $node) { |
||
| 692 | if ($node->isSameNode($this->root)) { |
||
| 693 | // var_dump($node); |
||
| 694 | $nodes = array_slice($nodes, 0, $i) |
||
| 695 | + PhpQuery::DOMNodeListToArray($node->childNodes) |
||
| 696 | + array_slice($nodes, $i + 1); |
||
| 736 | if ($this->isDocumentFragment) { |
||
| 737 | // documentFragment, html only... |
||
| 738 | self::debug("Getting markup, DocumentFragment detected"); |
||
| 739 | // return $this->markup( |
||
| 740 | //// $this->document->getElementsByTagName('body')->item(0) |
||
| 741 | // $this->document->root, true |
||
| 742 | // ); |
||
| 743 | $markup = $this->documentFragmentToMarkup($this); |
||
| 744 | // no need for markupFixXHTML, as it's done thought markup($nodes) method |
||
| 803 |
Oleg Mikhailov (miholeus)
Greg Payne
Loading history...
This check looks from parameters that have been defined for a function or method, but which are not used in the method body.