voku /
simple_html_dom
| @@ 836-865 (lines=30) @@ | ||
| 833 | * |
|
| 834 | * @return HtmlDomParser |
|
| 835 | */ |
|
| 836 | public function loadHtmlFile(string $filePath, $libXMLExtraOptions = null): DomParserInterface |
|
| 837 | { |
|
| 838 | // reset |
|
| 839 | self::$domBrokenReplaceHelper = []; |
|
| 840 | ||
| 841 | if ( |
|
| 842 | !\preg_match("/^https?:\/\//i", $filePath) |
|
| 843 | && |
|
| 844 | !\file_exists($filePath) |
|
| 845 | ) { |
|
| 846 | throw new \RuntimeException("File ${filePath} not found"); |
|
| 847 | } |
|
| 848 | ||
| 849 | try { |
|
| 850 | if (\class_exists('\voku\helper\UTF8')) { |
|
| 851 | /** @noinspection PhpUndefinedClassInspection */ |
|
| 852 | $html = UTF8::file_get_contents($filePath); |
|
| 853 | } else { |
|
| 854 | $html = \file_get_contents($filePath); |
|
| 855 | } |
|
| 856 | } catch (\Exception $e) { |
|
| 857 | throw new \RuntimeException("Could not load file ${filePath}"); |
|
| 858 | } |
|
| 859 | ||
| 860 | if ($html === false) { |
|
| 861 | throw new \RuntimeException("Could not load file ${filePath}"); |
|
| 862 | } |
|
| 863 | ||
| 864 | return $this->loadHtml($html, $libXMLExtraOptions); |
|
| 865 | } |
|
| 866 | ||
| 867 | /** |
|
| 868 | * Get the HTML as XML or plain XML if needed. |
|
| @@ 441-467 (lines=27) @@ | ||
| 438 | * |
|
| 439 | * @return XmlDomParser |
|
| 440 | */ |
|
| 441 | public function loadHtmlFile(string $filePath, $libXMLExtraOptions = null): DomParserInterface |
|
| 442 | { |
|
| 443 | if ( |
|
| 444 | !\preg_match("/^https?:\/\//i", $filePath) |
|
| 445 | && |
|
| 446 | !\file_exists($filePath) |
|
| 447 | ) { |
|
| 448 | throw new \RuntimeException("File ${filePath} not found"); |
|
| 449 | } |
|
| 450 | ||
| 451 | try { |
|
| 452 | if (\class_exists('\voku\helper\UTF8')) { |
|
| 453 | /** @noinspection PhpUndefinedClassInspection */ |
|
| 454 | $html = UTF8::file_get_contents($filePath); |
|
| 455 | } else { |
|
| 456 | $html = \file_get_contents($filePath); |
|
| 457 | } |
|
| 458 | } catch (\Exception $e) { |
|
| 459 | throw new \RuntimeException("Could not load file ${filePath}"); |
|
| 460 | } |
|
| 461 | ||
| 462 | if ($html === false) { |
|
| 463 | throw new \RuntimeException("Could not load file ${filePath}"); |
|
| 464 | } |
|
| 465 | ||
| 466 | return $this->loadHtml($html, $libXMLExtraOptions); |
|
| 467 | } |
|
| 468 | ||
| 469 | /** |
|
| 470 | * @param string $selector |
|
| @@ 505-531 (lines=27) @@ | ||
| 502 | * |
|
| 503 | * @return XmlDomParser |
|
| 504 | */ |
|
| 505 | public function loadXmlFile(string $filePath, $libXMLExtraOptions = null): self |
|
| 506 | { |
|
| 507 | if ( |
|
| 508 | !\preg_match("/^https?:\/\//i", $filePath) |
|
| 509 | && |
|
| 510 | !\file_exists($filePath) |
|
| 511 | ) { |
|
| 512 | throw new \RuntimeException("File ${filePath} not found"); |
|
| 513 | } |
|
| 514 | ||
| 515 | try { |
|
| 516 | if (\class_exists('\voku\helper\UTF8')) { |
|
| 517 | /** @noinspection PhpUndefinedClassInspection */ |
|
| 518 | $xml = UTF8::file_get_contents($filePath); |
|
| 519 | } else { |
|
| 520 | $xml = \file_get_contents($filePath); |
|
| 521 | } |
|
| 522 | } catch (\Exception $e) { |
|
| 523 | throw new \RuntimeException("Could not load file ${filePath}"); |
|
| 524 | } |
|
| 525 | ||
| 526 | if ($xml === false) { |
|
| 527 | throw new \RuntimeException("Could not load file ${filePath}"); |
|
| 528 | } |
|
| 529 | ||
| 530 | return $this->loadXml($xml, $libXMLExtraOptions); |
|
| 531 | } |
|
| 532 | ||
| 533 | /** |
|
| 534 | * @param callable $callback |
|
