Greg Greg - 12 days ago 5
PHP Question

PHP DomDocument failing to handle utf-8 characters (☆)

The webserver is serving responses with utf-8 encoding, all files are saved with utf-8 encoding, and everything I know of setting has been set to utf-8 encoding.

Here's a quick program, to test if the output works:

<?php
$html = <<<HTML
<!doctype html>
<html>
<head>
<meta charset="utf-8">
<title>Test!</title>
</head>
<body>
<h1>☆ Hello ☆ World ☆</h1>
</body>
</html>
HTML;

$dom = new DomDocument("1.0", "utf-8");
$dom->loadHTML($html);

header("Content-Type: text/html; charset=utf-8");
echo($dom->saveHTML());


The output of the program is:

<!DOCTYPE html>
<html><head><meta charset="utf-8"><title>Test!</title></head><body>
<h1>&acirc;&#152;&#134; Hello &acirc;&#152;&#134; World &acirc;&#152;&#134;</h1>
</body></html>


Which renders as:

☆ Hello ☆ World ☆






What could I be doing wrong? How much more specific do I have to be to tell the DomDocument to handle utf-8 properly?

Answer

DOMDocument::loadHTML() expects a HTML string.

HTML uses the ISO-8859-1 encoding (ISO Latin Alphabet No. 1) as default per it's specs. That is since longer, see 6.1. The HTML Document Character Set. In reality that is more the default support for Windows-1252 in common webbrowsers.

I go back that far because PHP's DOMDocument is based on libxml and that brings the HTMLparser which is designed for HTML 4.0.

I'd say it's safe to assume then that you can load an ISO-8859-1 encoded string.

Your string is UTF-8 encoded. Turn all characters higher than 127 / h7F into HTML Entities and you're fine. If you don't want to do that your own, that is what mb_convert_encoding with the HTML-ENTITIES target encoding does:

  • Those characters that have named entities, will get the named entitiy. € -> &euro;
  • The others get their numeric (decimal) entity, e.g. ☆ -> &#9734;

The following is a code example that makes the progress a bit more visible by using a callback function:

$html = preg_replace_callback('/[\x{80}-\x{10FFFF}]/u', function($match) {
    list($utf8) = $match;
    $entity = mb_convert_encoding($utf8, 'HTML-ENTITIES', 'UTF-8');
    printf("%s -> %s\n", $utf8, $entity);
    return $entity;
}, $html);

This exemplary outputs for your string:

☆ -> &#9734;
☆ -> &#9734;
☆ -> &#9734;

Anyway, that's just for looking deeper into your string. You want to have it either converted into an encoding loadHTML can deal with. That can be done by converting all outside of US-ASCII into HTML Entities:

$us_ascii = mb_convert_encoding($utf_8, 'HTML-ENTITIES', 'UTF-8');

Take care that your input is actually UTF-8 encoded. If you have even mixed encodings (that can happen with some inputs) mb_convert_encoding can only handle one encoding per string. I already outlined above how to more specifically do string replacements with the help of regular expressions, so I leave further details for now.

The other alternative is to hint the encoding. This can be done in your case by modifying the document and adding a

<meta http-equiv="content-type" content="text/html; charset=utf-8">

which is a Content-Type specifying a charset. That is also best practice for HTML strings that are not available via a webserver (e.g. saved on disk or inside a string like in your example). The webserver normally set's that as the response header.

If you don't care the misplaced warnings, you can just add it in front of the string:

$dom = new DomDocument();
$dom->loadHTML('<meta http-equiv="content-type" content="text/html; charset=utf-8">'.$html);

Per the HTML 2.0 specs, elements that can only appear in the <head> section of a document, will be automatically placed there. This is what happens here, too. The output (pretty-print):

<!DOCTYPE html>
<html>
  <head>
    <meta http-equiv="content-type" content="text/html; charset=utf-8">
    <meta charset="utf-8">
    <title>Test!</title>
  </head>
  <body>
    <h1>☆ Hello ☆ World ☆</h1>    
  </body>
</html>
Comments