DOMNode::C14N() is way slower on created DOMDocuments than loaded ones

c14n-slow-on-generated-dom-testcase.php

<?php
/**
 * This code demonstrates the performance difference when calling
 * DOMNode::C14N() on a generated DOMDocument versus on a DOMDocument that was
 * loaded from a string.
 *
 * Calling DOMDocument::normalizeDocument() doesn't make any difference.
 *
 * Output example:
 *  Testing with 500 items
 *  Native DOM (not normalized)… 2.55357 seconds
 *  Native DOM (normalized)… 2.52199 seconds
 *  Export and re-import DOM… 0.06444 seconds
 */

define('NUMBER_OF_ITEMS', 500);
define('XML_NAME_SPACE', 'http://example.com/xml');

echo sprintf('Testing with %d items', NUMBER_OF_ITEMS) . PHP_EOL;

echo 'Generated DOM… ';
$nativeDOM = generate_random_dom();
$s = microtime(true);
iterate_with_xpath($nativeDOM);
echo number_format(microtime(true) - $s, 5) . ' seconds' . PHP_EOL;

echo 'Generated DOM with normalizeDocument()… ';
$nativeDOMNormalized = generate_random_dom();
$s = microtime(true);
$nativeDOMNormalized->normalizeDocument();
iterate_with_xpath($nativeDOMNormalized);
echo number_format(microtime(true) - $s, 5) . ' seconds' . PHP_EOL;

echo 'Export and re-import DOM… ';
$dom3 = generate_random_dom();
$s = microtime(true);
$importedXmlString = new \DOMDocument('1.0', 'UTF-8');
$importedXmlString->loadXML($dom3->saveXML());
iterate_with_xpath($importedXmlString);
echo number_format(microtime(true) - $s, 5) . ' seconds' . PHP_EOL;

/**
 * Iterate over items' content and call C14N
 *
 * @param \DOMDocument $dom
 */
function iterate_with_xpath(\DOMDocument $dom): void
{
    $xpath = new \DOMXPath($dom);
    $xpath->registerNamespace('listElement', XML_NAME_SPACE);

    /** @var \DOMElement $item */
    foreach($xpath->query('//listElement:alpha', null, XML_NAME_SPACE) as $item) {
        $item->C14N();
    }
}

/**
 * Generate random DOMDocument
 *
 * Example:
 *  <listElement>
 *      <listItem index="0">
 *          <alpha>random value here</alpha>
 *      </listItem>
 *      ...
 *  </listElement>
 *
 * @return \DOMDocument
 */
function generate_random_dom(): \DOMDocument
{
    $dom = new \DOMDocument();
    $dom->formatOutput = true;

    /** @var \DOMElement $list */
    $list = $dom->createElementNS(XML_NAME_SPACE, 'listElement');
    $dom->appendChild($list);

    for($i = 0; $i < NUMBER_OF_ITEMS; $i++) {
        /** @var \DOMElement $listItem */
        $listItem = $dom->createElementNS(XML_NAME_SPACE, 'listItem');
        $listItem->setAttribute('index', $i);
        $list->appendChild($listItem);

        $subElement = $dom->createElementNS(XML_NAME_SPACE, 'alpha');
        $subElement->appendChild($dom->createTextNode(uniqid('a', true)));
        $listItem->appendChild($subElement);
    }

    return $dom;
}