Andreas Haugstrup Pedersen | solitude.dk

Drupal is usually great at handling special characters. Unlike other systems everything has always worked out of the box for me in Drupal. Recently however I did end up with an edge case involving mail headers.

What I wanted to do was to send a nicely formatted From-header so the site name would show up instead of just the e-mail address. The header could for example look like this: From: Mÿ Fäncÿ Sitë <mail@example.com>

As e-mail headers can only contain US-ASCII characters any special characters needs to be escaped, but Drupal does this automatically, right? Yes and no. Drupal does run all mail headers through mime_header_encode() to escape non-US-ASCII characters. In our case that results in:

From: =?UTF-8?B?TcO/IEbDpG5jw78gU2l0w6sgPG1haWxAZXhhbXBsZS5jb20+?=

It's a little more garbled than it should be. A correctly formatted header in our case would be:

From: =?UTF-8?B?TcO/IEbDpG5jw78gU2l0w6s=?= <mail@example.com>

As you can see the "name" of the from-address should be escaped with all the UTF-8 magic, but the e-mail address itself should not be encoded. E-mail clients handle these malformed headers badly—often showing no From-name at all or showing a part of the following header as the From-name.

The solution is to do the encoding yourself before sending the header off to drupal_mail(). In our case it would look like this:

$from = mime_header_encode('Mÿ Fäncÿ Sitë'). ' <mail@example.com>';

This of course means that the name is runs through the encode twice—once by you and once by drupal_mail(). Fortunately, mime_header_encode() is smart enough to only encode strings which contain non-US-ASCII characters and you can safely run strings through it multiple times without risking double-encoding issues.