Home/ Questions/Q 6612315
In Process

The Archive Base Latest Questions

Editorial Team
  • 0
Asked: 2026-05-25T20:04:59+00:00

I’m using Mail_mimeDecode to extract attachments from incoming emails. Everything was working well for

  • 0

I’m using Mail_mimeDecode to extract attachments from incoming emails. Everything was working well for a while, until I started receiving attachments with filenames encoded in KOI8, with a section header like this:

Content-Disposition: attachment; filename="=?KOI8-R?B?8NLJzM/Wxc7JxSAudHh0?="

mimeDecode does a perfectly reasonable thing and returns the filename in KOI8:

$attachmentNameInKOI8 = $part->d_parameters['filename'];

The problem is that I need it in UTF-8. In this specific example, I can run the following to do the conversion:

$attachmentNameInUTF8 = iconv('KOI8', 'UTF-8', $attachmentNameInKOI8);

But without trying to parse the message manually, I don’t know when the name is in KOI8 and when it’s not. I’m also worried that some other encoding will come through soon, so I need a way to handle anything that might come my way.

I had read that mb_detect_encoding is not reliable, and in fact I could not get it to detect the string as KOI8.

Is there a way to tell mimeDecode to do the translation for me? I looked at the sourcecode of mimeDecode.php:_decodeHeader() and I can see that it parses the encoding but then does nothing with it, which seems a wasted opportunity.

UPDATE: To be clear, this is only a problem with headers and not with bodies because mimeDecode exposes the charset of the body, so it’s very easy to run iconv yourself like this:

$bodyutf = iconv($textpart->ctype_parameters['charset'], 'UTF-8', $textpart->body);
Share

Leave an answer

You must login to add an answer.

Need An Account,

1 Answer

  1. Editorial Team

    Adding a line to _decodeHeader() before the replace seems to do the trick:

    $text = iconv($charset, 'UTF-8', $text);
$input = str_replace($encoded, $text, $input);

    Seems weird that they didn’t build some such option into the original class, doesn’t it?

    NOTE: I’ve since noticed that Subject lines and other headers can also be encoded the same way as filenames (RFC2047). It appears that adding the iconv line into _decodeHeader addresses all these cases.

    Weird that such a feature wasn’t already built into mimeDecode–this can’t be a rare problem.

    EDIT: I now understand that the point of mimeDecode having an option for decode_headers=false is to get the raw values so you can decode them yourself. This seems such a waste given that there’s no point to having mimeDecode decode your headers ever if you can’t trust that it’s going to return a string in an expected charset (it would make more sense for it to accept a charset as a parameter to decode to; or null means no decoding… I have a feeling they’re unlikely to change it for little me.) So the point is you need to do your own decoding. Unfortunately it’s not as simple as a straight call to imap_utf8() or imap_mime_header_decode(). You could either take the _decodeHeader() function from mimeDecode and modify it or use something like this:

    http://www.php.net/manual/en/function.imap-mime-header-decode.php#71762

    EDIT #2: Unbelievably, the mimeDecode guys already incorporated my suggestion into their latest svn:

    https://pear.php.net/bugs/bug.php?id=18876

    On that version, you can now set decode_headers=’UTF-8′ and mimeDecode will do all the work for you. Wow!

    • 0