<?php

/**

* Tokenizes the content of DOM elements with verbose logging.

*

* @param DOMDocument $domDocument The DOMDocument object.

* @return array An array of tokenized elements, where each element is an associative array

* containing the token type and the token value.

*/

function tokenizeDom(DOMDocument $domDocument): array

{

$tokens = [];

// Recursive function to traverse the DOM tree

function traverse(DOMNode $node, string $path = "") {

$nodeName = $node->nodeName;

$tokenType = 'NODE'; // Default token type for nodes

$tokenValue = $nodeName;

//Log the node being processed

$logMessage = "Processing node: {$path} - {$nodeName} ({$tokenType} - {$tokenValue})";

error_log($logMessage);

// Handle text nodes

if ($node->nodeType === XML_TEXT) {

$tokenType = 'TEXT';

$tokenValue = trim($node->textContent);

if ($tokenValue !== "") {

$tokens[] = ['type' => $tokenType, 'value' => $tokenValue];

}

} else {

//Handle element nodes

$tokenType = 'ELEMENT';

$tokenValue = $nodeName;

//Process attributes

if ($node->hasAttributes()) {

foreach ($node->attributes as $attribute) {

$attributeName = $attribute->name;

$attributeValue = $attribute->value;

$logMessage = " Attribute: {$path}/{$nodeName} - {$attributeName} = {$attributeValue}";

error_log($logMessage);

$tokens[] = ['type' => 'ATTRIBUTE', 'name' => $attributeName, 'value' => $attributeValue];

}

}

// Recursively process child nodes

foreach ($node->childNodes as $childNode) {

$childPath = $path . '/' . $nodeName;

traverse($childNode, $childPath);

}

}

}

// Start traversal from the root node

traverse($domDocument->documentElement);

return $tokens;

}

// Example usage (replace with your DOMDocument)

$html = '<div id="root" class="container"><p>Hello, <b>World</b>!</p><a href="https://example.com">Link</a>';

$dom = new DOMDocument();

$dom->loadHTML($html);

$tokenizedData = tokenizeDom($dom);

// Output the tokenized data

echo "<pre>";

print_r($tokenizedData);

echo "</pre>";

?>