Popular:
 Building Resilient Systems on AWS: Learn how to design and implement a resilient, highly available, fault-tolerant infrastructure on AWS.

DOMDocument and UTF-8 Problem

By  on  

A few weeks back I shared how I used PHP DOMDocument to reliably update all image URLs from standard HTTP to HTTPS.  DOMDocument made a difficult problem seem incredibly easy ... but with one side-effect that it took me a while to spot:  UTF-8 characters were being mutated into another set of characters.  I was seeing a bunch of odd characters like "ãç³" and"»ã®é" all over each blog post.

I knew the problem was happening during the DOMDocument parsing and that I need to find a fix quickly.  The solution was just a tiny bit of code:

// Create a DOMDocument instance 
$doc = new DOMDocument();

// The fix: mb_convert_encoding conversion
$doc->loadHTML(mb_convert_encoding($content, 'HTML-ENTITIES', 'UTF-8'));

After setting the character set with mb_convert_encoding, the odd characters vanished and the desired characters were back in place.  Phew!

Request Metrics real user monitoring
Request Metrics real user monitoring
Request Metrics real user monitoring
Request Metrics real user monitoring

Recent Features

Incredible Demos

Discussion

  1. Markus

    instead of converting the input string from your encoding into UTF8 you can also tell the DOMDocument with the 2nd arg in which encoding your string is.

    http://php.net/manual/en/domdocument.construct.php

    This should save you some cpu-cycles and reduce memory consumption.

  2. Whelping

    Actually I was already doing

    $html = new DOMDocument( null, 'UTF-8' );

    and still getting weird characters like  – adding your fix did the trick for me, thanks David!

    • Whelping

      It turns out that specifying the encoding argument when you instantiate a DOMDocument doesn’t encode the contents, just sets the document’s header – see this comment http://php.net/manual/en/domdocument.construct.php#78027. It probably wasn’t necessary to specify UTF-8 anyway, as that’s the default. So something like David’s fix is needed to change any unwanted encodings in the content.

  3. Raja Amer Khan

    Thanks for the tutorial. It really saved time.

  4. Dario

    Thank you very much you life saver

  5. Sam

    Thank you very for this :D
    Caught this bug too, even after instantiating it properly

  6. Rafael

    Awesome, character encoding has always been a pain the a.

  7. Jignesh

    Thank you very much David.
    This helped me a lot. because I had bad characters in wordpress the_content.
    This method removed special characters pasted from word document.
    :D

  8. Shaban

    Thanks a lot, David.
    I was about to pull my hair out :D

  9. Robert Andrews

    Whilst this sorts out the worst of the character nasties, I am still seeing instances where an apostrophe appears as a question mark.

    The character on the input end is ’ … it appears as ?

    Somtimes I think what I may be experiencing is the crime of taking articles which originated in Microsoft Word, or maybe even Google Docs – ie. curly apostrophe. But I’m not sure – there are pieces of text that I’ve written out directly, which I’m not sure could explain this problem.

    It’s frustrating that I can’t iron this out. They appear just fine if I disable my code which uses DOMDocument.

    No code that I can find online (not MS Word character-conversion functions https://stackoverflow.com/questions/1262038/how-to-replace-microsoft-encoded-quotes-in-php and not DOMDocument modifiers) seems to work.

  10. Rasso

    In PHP8.2, mb_convert_encoding is deprecated. Instead of 

    $document->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'));

    I am now doing this:

    $document->loadHTML(htmlspecialchars_decode(htmlentities($html)));

    Has been working fine in my tests

Wrap your code in <pre class="{language}"></pre> tags, link to a GitHub gist, JSFiddle fiddle, or CodePen pen to embed!