1. <?php
  3. /**
  4. * Collects metrics of DOM elements for sandbox usage (no async).
  5. *
  6. * @param string $html The HTML string to analyze.
  7. * @return array An array of DOM element metrics.
  8. */
  9. function collectDomMetrics(string $html): array
  10. {
  11. $dom = new DOMDocument();
  12. @$dom->loadHTML($html); // Suppress warnings for potentially invalid HTML
  14. $metrics = [];
  16. /**
  17. * Recursive function to traverse the DOM and collect metrics.
  18. *
  19. * @param DOMNode $node The current DOM node.
  20. */
  21. function traverseDom(DOMNode $node): void
  22. {
  23. switch ($node->nodeType) {
  24. case XML_ELEMENT_NODE:
  25. $element = $node as DOMElement;
  26. $metrics[] = [
  27. 'tag' => $element->tagName,
  28. 'attributes' => [],
  29. 'children_count' => $element->childCount()
  30. ];
  31. foreach ($element->attributes as $attr) {
  32. $metrics[count($metrics) - 1]['attributes'][$attr->name] = $attr->value;
  33. }
  34. traverseDom($element); // Recursive call for child elements
  35. break;
  36. case XML_TEXT_NODE:
  37. $text = $node->textContent;
  38. if (!empty($text)) {
  39. $metrics[] = [
  40. 'type' => 'text',
  41. 'content' => $text
  42. ];
  43. }
  44. break;
  45. }
  46. }
  48. $root = $dom->documentElement;
  49. if ($root) {
  50. traverseDom($root);
  51. }
  53. return $metrics;
  54. }
  56. // Example Usage (for testing)
  57. if (isset($_SERVER['argv'][1])) {
  58. $html = $_SERVER['argv'][1];
  59. $metrics = collectDomMetrics($html);
  60. print_r($metrics);
  61. } else {
  62. //Example HTML
  63. $html = '<html><body><h1>Hello World</h1><p>This is a paragraph.</p><a href="#">Link</a></body></html>';
  64. $metrics = collectDomMetrics($html);
  65. print_r($metrics);
  66. }
  69. ?>

Add your comment