Using fread to detect file type

[NotionCommotion]

I had some problems detecting file type for a Microsoft Outlook document, and requinix recommended using doing the MIME magic manually instead of using finfo (reference http://forums.phpfreaks.com/topic/294730-mime-type-for-outlook-msg-file/).

 

Seemed to work fine, but recently found that more than one file type (i.e. an old Microsoft document file) has the same magic constant as an Outlook file.  How can I find the correct "magic" strings that represent different file types?

<?php
function test($file)
{
    $h = fopen($file, 'rb');
    $magic = "\xD0\xCF\x11\xE0\xA1\xB1\x1A\xE1\x00";    //How do I determine the correct string for different file types?
    $bytes = fread($h, strlen($magic));
    fclose($h);
    echo($file.' '.($magic==$bytes?'equal':'not equal').'<br>');    
}
test('test.msg');  //echos equal
test('test.DOC');  //echos equal
test('test.doc');  //echos equal
test('magic.php');  //echos not equal
?>

[requinix]

.msg and .doc and others all use the Compound File Binary File Format. They all have that same 9 byte header.

 

To properly detect each type within you'll have to parse the file and look for identifying information. For example, it seems .doc files can be identified by a 0xA5EC at a certain location I've yet to pinpoint since it probably requires knowing more about the CFB format.

[NotionCommotion]

Thanks Requinix,  This is getting a bit more complicated than I want it to be.  My main purpose for doing this is to ensure that uploaded files have extensions which match the file type so that that I could restrict files based on extension and be assured that I am really restricting that real file.  What I will probably do is if the file extension is msg, then manually check it, but if doc, then use finfo.

[requinix]

There are other problematic file types too, like DOCX and its siblings will be reported as application/zip. (Which is actually what they are.)

 

Use MIME detection for many things, MIME + file extensions for the exceptions.

[NotionCommotion]

Yea, learned early 'bot docx and cousin, but msg took me for a loop.