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 |
|
| @@ 813-842 (lines=30) @@ | ||
| 810 | * |
|
| 811 | * @return HtmlDomParser |
|
| 812 | */ |
|
| 813 | public function loadHtmlFile(string $filePath, $libXMLExtraOptions = null): DomParserInterface |
|
| 814 | { |
|
| 815 | // reset |
|
| 816 | self::$domBrokenReplaceHelper = []; |
|
| 817 | ||
| 818 | if ( |
|
| 819 | !\preg_match("/^https?:\/\//i", $filePath) |
|
| 820 | && |
|
| 821 | !\file_exists($filePath) |
|
| 822 | ) { |
|
| 823 | throw new \RuntimeException("File ${filePath} not found"); |
|
| 824 | } |
|
| 825 | ||
| 826 | try { |
|
| 827 | if (\class_exists('\voku\helper\UTF8')) { |
|
| 828 | /** @noinspection PhpUndefinedClassInspection */ |
|
| 829 | $html = UTF8::file_get_contents($filePath); |
|
| 830 | } else { |
|
| 831 | $html = \file_get_contents($filePath); |
|
| 832 | } |
|
| 833 | } catch (\Exception $e) { |
|
| 834 | throw new \RuntimeException("Could not load file ${filePath}"); |
|
| 835 | } |
|
| 836 | ||
| 837 | if ($html === false) { |
|
| 838 | throw new \RuntimeException("Could not load file ${filePath}"); |
|
| 839 | } |
|
| 840 | ||
| 841 | return $this->loadHtml($html, $libXMLExtraOptions); |
|
| 842 | } |
|
| 843 | ||
| 844 | /** |
|
| 845 | * Get the HTML as XML or plain XML if needed. |
|
