Description of problem:

Sometimes g-zip archives are not detected properly even though the magic code is the expected one (\037\213 at offset 0). In such case, "application/octet-stream" is detected instead.

It seems that this issue was present, then was fixed in Fedora (file 5.17), but base is more recent (see Additional info).


Version-Release number of selected component (if applicable):

file-5.11-33.el7.x86_64 


How reproducible:

Always with archive attached to the BZ


Steps to Reproduce:
1. file -i <archive>


Actual results:

application/octet-stream; charset=binary


Expected results:

application/x-gzip; charset=binary


Additional info:

[root@rhel7u2a opt]# for i in $(ls); do echo -ne "$i"; $i/bin/file -i /root/filename ; done
file-4.26/root/filename: application/x-gzip
file-5.00/root/filename: application/octet-stream; charset=binary
file-5.01/root/filename: application/octet-stream; charset=binary
file-5.02/root/filename: application/octet-stream; charset=binary
file-5.03/root/filename: application/octet-stream; charset=binary
file-5.04/root/filename: application/octet-stream; charset=binary
file-5.05/root/filename: application/octet-stream; charset=binary
file-5.06/root/filename: application/octet-stream; charset=binary
file-5.07/root/filename: application/octet-stream; charset=binary
file-5.08/root/filename: application/octet-stream; charset=binary
file-5.09/root/filename: application/octet-stream; charset=binary
file-5.10/root/filename: application/octet-stream; charset=binary
file-5.11/root/filename: application/octet-stream; charset=binary
file-5.12/root/filename: application/octet-stream; charset=binary
file-5.13/root/filename: application/octet-stream; charset=binary
file-5.14/root/filename: application/octet-stream; charset=binary
file-5.15/root/filename: application/octet-stream; charset=unknown
file-5.16/root/filename: binary; charset=binary
file-5.17/root/filename: application/x-gzip; charset=binary

-> fixed in 5.17

Workaround:

A workaround is to copy the g-zip code from source (magic/Magdir/compress, lines 23 to 51) to /etc/magic.

I don't understand why this actually works.

Thank you for reporting the bug!

This was fixed by the following upstream commit:

    https://github.com/file/file/commit/FILE5_16-6-g5c94551

... while only the following hunk is actually needed:

--- a/magic/Magdir/compress
+++ b/magic/Magdir/compress
@@ -22,6 +22,7 @@
 #         other than 8 ("deflate", the only method defined in RFC 1952).
 0       string          \037\213        gzip compressed data
 !:mime application/x-gzip
+!:strength * 2
 >2 byte    <8    \b, reserved method
 >2 byte    >8    \b, unknown method
 >3 byte    &0x01   \b, ASCII

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHBA-2018:3079