voku /
simple_html_dom
| @@ 437-463 (lines=27) @@ | ||
| 434 | * |
|
| 435 | * @return XmlDomParser |
|
| 436 | */ |
|
| 437 | public function loadHtmlFile(string $filePath, $libXMLExtraOptions = null): DomParserInterface |
|
| 438 | { |
|
| 439 | if ( |
|
| 440 | !\preg_match("/^https?:\/\//i", $filePath) |
|
| 441 | && |
|
| 442 | !\file_exists($filePath) |
|
| 443 | ) { |
|
| 444 | throw new \RuntimeException("File ${filePath} not found"); |
|
| 445 | } |
|
| 446 | ||
| 447 | try { |
|
| 448 | if (\class_exists('\voku\helper\UTF8')) { |
|
| 449 | /** @noinspection PhpUndefinedClassInspection */ |
|
| 450 | $html = UTF8::file_get_contents($filePath); |
|
| 451 | } else { |
|
| 452 | $html = \file_get_contents($filePath); |
|
| 453 | } |
|
| 454 | } catch (\Exception $e) { |
|
| 455 | throw new \RuntimeException("Could not load file ${filePath}"); |
|
| 456 | } |
|
| 457 | ||
| 458 | if ($html === false) { |
|
| 459 | throw new \RuntimeException("Could not load file ${filePath}"); |
|
| 460 | } |
|
| 461 | ||
| 462 | return $this->loadHtml($html, $libXMLExtraOptions); |
|
| 463 | } |
|
| 464 | ||
| 465 | /** |
|
| 466 | * @param string $selector |
|
| @@ 501-527 (lines=27) @@ | ||
| 498 | * |
|
| 499 | * @return XmlDomParser |
|
| 500 | */ |
|
| 501 | public function loadXmlFile(string $filePath, $libXMLExtraOptions = null): self |
|
| 502 | { |
|
| 503 | if ( |
|
| 504 | !\preg_match("/^https?:\/\//i", $filePath) |
|
| 505 | && |
|
| 506 | !\file_exists($filePath) |
|
| 507 | ) { |
|
| 508 | throw new \RuntimeException("File ${filePath} not found"); |
|
| 509 | } |
|
| 510 | ||
| 511 | try { |
|
| 512 | if (\class_exists('\voku\helper\UTF8')) { |
|
| 513 | /** @noinspection PhpUndefinedClassInspection */ |
|
| 514 | $xml = UTF8::file_get_contents($filePath); |
|
| 515 | } else { |
|
| 516 | $xml = \file_get_contents($filePath); |
|
| 517 | } |
|
| 518 | } catch (\Exception $e) { |
|
| 519 | throw new \RuntimeException("Could not load file ${filePath}"); |
|
| 520 | } |
|
| 521 | ||
| 522 | if ($xml === false) { |
|
| 523 | throw new \RuntimeException("Could not load file ${filePath}"); |
|
| 524 | } |
|
| 525 | ||
| 526 | return $this->loadXml($xml, $libXMLExtraOptions); |
|
| 527 | } |
|
| 528 | ||
| 529 | /** |
|
| 530 | * @param callable $callback |
|
| @@ 830-859 (lines=30) @@ | ||
| 827 | * |
|
| 828 | * @return HtmlDomParser |
|
| 829 | */ |
|
| 830 | public function loadHtmlFile(string $filePath, $libXMLExtraOptions = null): DomParserInterface |
|
| 831 | { |
|
| 832 | // reset |
|
| 833 | self::$domBrokenReplaceHelper = []; |
|
| 834 | ||
| 835 | if ( |
|
| 836 | !\preg_match("/^https?:\/\//i", $filePath) |
|
| 837 | && |
|
| 838 | !\file_exists($filePath) |
|
| 839 | ) { |
|
| 840 | throw new \RuntimeException("File ${filePath} not found"); |
|
| 841 | } |
|
| 842 | ||
| 843 | try { |
|
| 844 | if (\class_exists('\voku\helper\UTF8')) { |
|
| 845 | /** @noinspection PhpUndefinedClassInspection */ |
|
| 846 | $html = UTF8::file_get_contents($filePath); |
|
| 847 | } else { |
|
| 848 | $html = \file_get_contents($filePath); |
|
| 849 | } |
|
| 850 | } catch (\Exception $e) { |
|
| 851 | throw new \RuntimeException("Could not load file ${filePath}"); |
|
| 852 | } |
|
| 853 | ||
| 854 | if ($html === false) { |
|
| 855 | throw new \RuntimeException("Could not load file ${filePath}"); |
|
| 856 | } |
|
| 857 | ||
| 858 | return $this->loadHtml($html, $libXMLExtraOptions); |
|
| 859 | } |
|
| 860 | ||
| 861 | /** |
|
| 862 | * Get the HTML as XML or plain XML if needed. |
|
