DOMDocument::saveHTML() adds additional symbols before some UTF-8 characters in HTML

[MurzNN]

Description

The following code:

<?php
$html='<p>My Brand®</p>';
$dom = new \DOMDocument();
$dom->encoding = 'UTF-8';
$dom->loadHTML($html, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
$result = $dom->saveHTML();
print($result);

Resulted in this output:

<p>My Brand&Acirc;&reg;</p>

But I expected this output instead:

<p>My Brand®</p>

Or at least this:

<p>My Brand&reg;</p>

The same issue is with other unicode characters, more examples:

• <p>My ☆ Brand</p> >>> <p>My &acirc;&#152;&#134; Brand</p>
• <div>€ 100</div> >>> <div>&acirc;&#130;&not; 100</div>
• <p>À bientôt!</p> >>> <p>&Atilde;&#128; bient&Atilde;&acute;t!</p>
• <p>Hello 😕 there!</p> >>> <p>Hello &eth;&#159;&#152;&#149; there!</p>

Could you please explain why this happens, and how to fix this issue? And maybe suggest any workarounds?

Also, as I see, it forcibly converts all unicode characters to HTML entities, but many people prefer to keep the original formatting (keep HTML entities as entities, but UTF-8 symbols as symbols), would be good to fix this too.

PHP Version

PHP 8.4.5 (cli) (built: Mar 17 2025 20:35:32) (NTS)
Copyright (c) The PHP Group
Zend Engine v4.4.5, Copyright (c) Zend Technologies
    with Zend OPcache v8.4.5, Copyright (c), by Zend Technologies

Operating System

Ubuntu 25.04

[MurzNN]

Found a workaround for this issue:

<?php
$html='<p>Hello 😕 there!</p>';
$dom = new \DOMDocument();
$dom->encoding = 'UTF-8';
$forceEncodingPrefix = '<?xml encoding="UTF-8">';
$dom->loadHTML($forceEncodingPrefix . $html, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
$result = $dom->saveHTML();
$result = str_replace($forceEncodingPrefix, '', $result);
$result = html_entity_decode($result);
print($result);

It outputs the HTML with the original formatting:

<p>Hello 😕 there!</p>

So it adds the encoding explicitly to the HTML string. Seems the $dom->encoding = 'UTF-8' just do not work?

[nielsdos]

The encoding property is indeed ignored for parsing, it is meant for re-encoding during serialization.
The encoding defaults to iso-latin-1 which interprets the UTF-8 symbols incorrectly, causing the weird entities you see.

This is unfortunately one of the strange legacy behaviours of DOMDocument.
You can't set the encoding specifically unless you use the xml tag hack.
Changing this will cause a BC break. And as we've learned previously from fixing issues in ext/dom, these BC breaks can't be done because too many people rely on broken behaviour.

The proper solution is to use the new PHP 8.4 DOM classes that properly handle modern HTML and also allow you to set the encoding explicitly.

Example fixed (using new 8.4 classes):

<?php
$html = '<p>My Brand®</p>';
$dom = \Dom\HTMLDocument::createFromString($html, LIBXML_HTML_NOIMPLIED | LIBXML_NOERROR, overrideEncoding: 'UTF-8'); // Passing overrideEncoding: 'UTF-8' is optional because UTF-8 is the default, I just wanted to demonstrate the parameter.
$result = $dom->saveHTML();
print($result);

Gives:

<p>My Brand®</p>