Why PHP DOMDocument loadHTML doesn't work for

Question

Here is my code:

<?php

$data = <<<DATA
<div>
    <p>سلام</p>                                         // focus on this line
    <p class="myclass">Remove this one</p>
    <p>But keep this</p>
    <div style="color: red">and this</div>
    <div style="color: red">and <p>also</p> this</div>
    <div style="color: red">and this <div style="color: red">too</div></div>
</div>
DATA;

$dom = new DOMDocument();
$dom->loadHTML(mb_convert_encoding($data, 'HTML-ENTITIES', 'UTF-8'), LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
$xpath = new DOMXPath($dom);

foreach ($xpath->query("//*[@*]") as $node) {
    $parent = $node->parentNode;
    while ($node->hasChildNodes()) {
        $parent->insertBefore($node->lastChild, $node->nextSibling);
    }
    $parent->removeChild($node);
}

echo $dom->saveHTML();

As I've mentioned in the title of my question, the content of my website is Persian (not English). But code about doesn't work for Persian characters.

Current output:

.
.
    <p>&#1587;&#1604;&#1575;&#1605;</p>
.
.

Expected output:

.
.
    <p>سلام</p>
.
.

What's wrong with it and how can I fix it?

Note: Also as you see I've used mb_convert_encoding($data, 'HTML-ENTITIES', 'UTF-8') to make it correct (based on this answer) but still it doesn't work.

Answer 1

The Persian characters are being encoded as numeric character references. They'll appear appropriately in a browser or you can see the original by decoding them with html_entity_decode(), e.g.:

echo html_entity_decode("سلام");

outputs:

سلام

---

If you prefer the original characters in the output rather than numeric character references, you can change:

echo $dom->saveHTML();

to:

echo $dom->saveHTML($dom->documentElement);

This alters the serialization a bit and the result is:

<div>
    <p>سلام</p>
    Remove this one
    <p>But keep this</p>
    and this
    and <p>also</p> this
    and this too
</div>

Example.