[File] jpeg detection regression

Christoph Biedl astron.com.bwoj at manchmal.in-ulm.de
Thu Aug 26 17:09:11 UTC 2021



commit FILE5_40-25-g750e69ca

    Author: Christos Zoulas <christos at zoulas.com>
    Date:   Mon Apr 19 18:58:30 2021 +0000

        Regognize jpeg-ls and make the magic for jpeg stronger so that it does not
        need extra strength boost

unfortunatly introduced a regression, one file in my collection is now
detected as

| DOS/MBR boot sector; partition 1 : ID=0xb3, active 0xb0, start-CHS (0x2b3,204,11), end-CHS (0x148,243,45), startsector 943011019, 1208593131 sectors; partition 3 : ID=0xd7, active 0xdc, start-CHS (0x75,139,10), end-CHS (0x120,3,45), startsector 2186963805, 1290699090 sectors

instead of (correctly)

| JPEG image data, JFIF standard 1.01, resolution (DPI), density 96x96, segment length 16, baseline, precision 8, 407x374, components 3\012- data

Problem is (if I read the dumps correctly) the strength of the new magic
is rather poor:

150 old strength
131 DOS/MBR strength
 70  new strength

So indeed adding a strenght of 62 restores the desired detection. But
perhaps there is a better way to fix this.

Sample file:

md5: e46cdc3a85e8da2b0e61f55d7247276f
sha256: ddbdc5539ef1f279d848a3761eb9d99e819bd61fae48f60e7c42f6ee0932c863

Local file name is "4972961.bin.jpeg" - possibly it was part of the
report 421 in the old bugtracker, so you might have a copy of it.


More information about the File mailing list