Created attachment 8797 [details] [review]
proposed magic

Simple proposal of a magic for application/x-cd-image, taken from /usr/share/file/magic.

Thanks for your bug report and your efforts!

I'm not totally sure about one of your two glob offset, though. I'm referring to the ISO 9660 spec [1] here:

* We have the system area which is 16 logical sectors large (16*2048 bytes=32768 bytes)
* We then have an obscure entry of "(None) optional" which I don't understand. How large is this entry supposed to be?
* After the optional entry, we have the primary volume descriptor, which will have
 * * two bytes for its type
 * * followed by its ID which is supposed to be "CD001", i.e. which we want to match

Assuming that the optional entry is not present, your matchlet with an offset of
32769
seems to be correct (offset counting starts from 0, so the two-byte type field in the beginning of the volume descriptor is properly reflected).

I don't understand why you added another matchlet with offset 32633, though.
Thanks for your work up to this point, though! :)

[1] http://users.pandora.be/it3.consultants.bvba/handouts/ISO9960.html

Basically, magic(1) has this:

# CDROM Filesystems
32769    string    CD001     ISO 9660 CD-ROM filesystem data
# "application id" which appears to be used as a volume label
>32808   string    >\0       '%s'
>34816   string    \000CD001\001EL\ TORITO\ SPECIFICATION    (bootable)
37633    string    CD001     ISO 9660 CD-ROM filesystem data (raw 2352 byte sectors)

that's why ;)

This was rejected in the past. We can't go this far in a file to check its mime-type.