voku /
simple_html_dom
| @@ 434-460 (lines=27) @@ | ||
| 431 | * |
|
| 432 | * @return XmlDomParser |
|
| 433 | */ |
|
| 434 | public function loadHtmlFile(string $filePath, $libXMLExtraOptions = null): DomParserInterface |
|
| 435 | { |
|
| 436 | if ( |
|
| 437 | !\preg_match("/^https?:\/\//i", $filePath) |
|
| 438 | && |
|
| 439 | !\file_exists($filePath) |
|
| 440 | ) { |
|
| 441 | throw new \RuntimeException("File ${filePath} not found"); |
|
| 442 | } |
|
| 443 | ||
| 444 | try { |
|
| 445 | if (\class_exists('\voku\helper\UTF8')) { |
|
| 446 | /** @noinspection PhpUndefinedClassInspection */ |
|
| 447 | $html = UTF8::file_get_contents($filePath); |
|
| 448 | } else { |
|
| 449 | $html = \file_get_contents($filePath); |
|
| 450 | } |
|
| 451 | } catch (\Exception $e) { |
|
| 452 | throw new \RuntimeException("Could not load file ${filePath}"); |
|
| 453 | } |
|
| 454 | ||
| 455 | if ($html === false) { |
|
| 456 | throw new \RuntimeException("Could not load file ${filePath}"); |
|
| 457 | } |
|
| 458 | ||
| 459 | return $this->loadHtml($html, $libXMLExtraOptions); |
|
| 460 | } |
|
| 461 | ||
| 462 | /** |
|
| 463 | * @param string $selector |
|
| @@ 498-524 (lines=27) @@ | ||
| 495 | * |
|
| 496 | * @return XmlDomParser |
|
| 497 | */ |
|
| 498 | public function loadXmlFile(string $filePath, $libXMLExtraOptions = null): self |
|
| 499 | { |
|
| 500 | if ( |
|
| 501 | !\preg_match("/^https?:\/\//i", $filePath) |
|
| 502 | && |
|
| 503 | !\file_exists($filePath) |
|
| 504 | ) { |
|
| 505 | throw new \RuntimeException("File ${filePath} not found"); |
|
| 506 | } |
|
| 507 | ||
| 508 | try { |
|
| 509 | if (\class_exists('\voku\helper\UTF8')) { |
|
| 510 | /** @noinspection PhpUndefinedClassInspection */ |
|
| 511 | $xml = UTF8::file_get_contents($filePath); |
|
| 512 | } else { |
|
| 513 | $xml = \file_get_contents($filePath); |
|
| 514 | } |
|
| 515 | } catch (\Exception $e) { |
|
| 516 | throw new \RuntimeException("Could not load file ${filePath}"); |
|
| 517 | } |
|
| 518 | ||
| 519 | if ($xml === false) { |
|
| 520 | throw new \RuntimeException("Could not load file ${filePath}"); |
|
| 521 | } |
|
| 522 | ||
| 523 | return $this->loadXml($xml, $libXMLExtraOptions); |
|
| 524 | } |
|
| 525 | ||
| 526 | /** |
|
| 527 | * @param callable $callback |
|
| @@ 754-783 (lines=30) @@ | ||
| 751 | * |
|
| 752 | * @return HtmlDomParser |
|
| 753 | */ |
|
| 754 | public function loadHtmlFile(string $filePath, $libXMLExtraOptions = null): DomParserInterface |
|
| 755 | { |
|
| 756 | // reset |
|
| 757 | self::$domBrokenReplaceHelper = []; |
|
| 758 | ||
| 759 | if ( |
|
| 760 | !\preg_match("/^https?:\/\//i", $filePath) |
|
| 761 | && |
|
| 762 | !\file_exists($filePath) |
|
| 763 | ) { |
|
| 764 | throw new \RuntimeException("File ${filePath} not found"); |
|
| 765 | } |
|
| 766 | ||
| 767 | try { |
|
| 768 | if (\class_exists('\voku\helper\UTF8')) { |
|
| 769 | /** @noinspection PhpUndefinedClassInspection */ |
|
| 770 | $html = UTF8::file_get_contents($filePath); |
|
| 771 | } else { |
|
| 772 | $html = \file_get_contents($filePath); |
|
| 773 | } |
|
| 774 | } catch (\Exception $e) { |
|
| 775 | throw new \RuntimeException("Could not load file ${filePath}"); |
|
| 776 | } |
|
| 777 | ||
| 778 | if ($html === false) { |
|
| 779 | throw new \RuntimeException("Could not load file ${filePath}"); |
|
| 780 | } |
|
| 781 | ||
| 782 | return $this->loadHtml($html, $libXMLExtraOptions); |
|
| 783 | } |
|
| 784 | ||
| 785 | /** |
|
| 786 | * Get the HTML as XML or plain XML if needed. |
|
