AVI/MKV sample can't reconstruct due to false positive match offset

[Gfy]

http://www.srrdb.com/release/details/Day.Zero.LIMITED.DVDRip.XViD-BaLD

• [track_id] => 0
• [track_size] => 7456151

• [match_offset_main_movie_file] => 256648820

• [track_id] => 1

• [track_size] => 1190064
• [match_offset_main_movie_file] => 10260

video is fine, I could only see audio was the problem

It looks like it finds a match for the audio pretty early in the file. If reconstruction fails, look if there are other matches for the tracks in the file. Start with the track that has the lowest offset.

When creating the SRS with the verification option on, check if both offsets are relatively close to each other. If this isn't the case, try searching for more suitable offsets?

[Gfy]

http://www.srrdb.com/release/details/Jason.Becker.Not.Dead.Yet.LIMITED.DVDRip.XviD-BiQ

The sample does not rebuild either, but it doens't look related.

[Gfy]

Same problem on an MKV file with AC3 audio https://www.srrdb.com/release/details/Adventure.Time.With.Finn.And.Jake.S06E19.1080p.BluRay.x264-PRESENT

File Details:   Size           CRC
                -------------  --------
                   35 155 848  83F924E9


Track Details:  Track  Length         Codec
                -----  -------------  --------------------
                    1     31 868 173  V_MPEG4/ISO/AVC
                    2      3 268 608  A_AC3

Parse Details:  Metadata     Attachments   Track Data     Total
                -----------  ------------  -------------  -------------
                     19 067             0     35 136 781     35 155 848

                    [0] => Array
                        (
                            [track_id] => 1
                            [track_size] => 31868173
                            [match_offset_main_movie_file] => 262815417
                        )

                    [1] => Array
                        (
                            [track_id] => 2
                            [track_size] => 3268608
                            [match_offset_main_movie_file] => 101259722
                        )

The -k sample does not have the correct audio. The audio from early on in the sample gets taken. A lot of false positive matches show up, but not the correct offset:

        # sample-adventure.time.with.finn.and.jake.s06e19.1080p.bluray.x264-present
        if (
            tracks[2].match_offset == 101259722 or
            tracks[2].match_offset == 101272900 or
            tracks[2].match_offset == 247220114 or
            tracks[2].match_offset == 247834927 or
            tracks[2].match_offset == 248453115 or
            tracks[2].match_offset == 248473925 or
            tracks[2].match_offset == 248488582 or
            tracks[2].match_offset == 248503552 or
            tracks[2].match_offset == 262606837 or
            tracks[2].match_offset == 262744314 or
            tracks[2].match_offset == 262807006 or
            tracks[2].match_offset == 263054871 or
            tracks[2].match_offset == 263225470 or
            tracks[2].match_offset == 389521601 or
            tracks[2].match_offset == 389332383 or
            tracks[2].match_offset == 492039891):
            print(tracks[2].check_bytes)
            done = False
            tracks[2].check_bytes = b""
            tracks[2].match_offset = 0
            tracks[2].match_length = 0

Increasing the signature size does not help with finding the correct audio location:

Track Location Complete...    Elapsed Time: 0.88s
1: 262815417
2: 433192277
Track Extraction Complete...  Elapsed Time: 1.56s
Rebuild Complete...           Elapsed Time: 1.73s

File Details:   Size           CRC
                -------------  --------
Expected    :      35 155 848  83F924E9
Actual      :      35 155 848  AFED660C

This is a known issue for LOL xvid releases.

Rebuild failed for sample: sample-adventure.time.with.finn.and.jake.s06e19.1080p.bluray.x264-present.mkv

The .ac3 fingerprint has a lot of zero bytes and contains a repeating pattern: b'\x0bw1\xed\x1e@K\x7f\xf3\x06\xa0\xb8a\xff:\xbe|\xf9\xf3\xe7\xcf\x9f>|\xf9\xf3\xe7\xcf\x9f>|\xf9\xf3\xe7\xcf\x9f>|\xf9\xf3\xe7\xcf\x9f>|\xf9\xf3\xe7\xcf\x9f>|\xf9\xf3\xe7\xcf\x9f>|\xf9\xf3\xe7\xcf\x9f>|\xf9\xf3\xe7\xcf\x9f>|\xf9\xf3\xe7\xdf\xfc\xea\xf9\xf3\xe7\xcf\x9f>|\xf9\xf3\xe7\xcf\x9f>|\xf9\xf3\xe7\xcf\x9f>|\xf9\xf3\xe7\xcf\x9f>|\xf9\xf3\xe7\xcf\x9f>|\xf9\xf3\xe7\xcf\x9f>|\xf9\xf3\xe7\xcf\x9f>|\xf9\xf3\xe7\xcf\x9f>|\xf9\xf3\xe7\xcf\x9fyO\xa4 @\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'

[Gfy]

Verifying all the sample track data before storing the match offset?