#10 Bad times from mpeg2 transport stream

[Brian Enigma]

I have a transport stream that VLC seems to play and caption fine, and for which ccextractor can pull out text, but the start and end time codes for individual captions are impossible values: end times before start times, overlapping, etc. The final summary reports a min PTS of negative 6 minutes (my understanding is that PTS is supposed to be an unsigned value, so this seems wrong). In stepping through the code, I can see fts_now occasionally incrementing (and sometimes decrementing) in much larger steps than I'd expect, but I'm having difficulty determining why (fts_now is calculated from a few variables which are each, in turn, calculated from a few variables and conditions and I'm having difficulty keeping track of everything that's influencing fts). The clips I've seen this on so far are pulldown and frequently report (incorrectly, according to my stream analyzer) having broken GOPs.

Sample file: http://netninja.com/junk/OR08-prg2.ts

Total frames time: 00:03:53:032 (6984 frames at 29.97fps)
incl. pulldown frames: 00:00:09:142 (274 frames at 29.97fps)

Min PTS: -00:06:02:410
(Reference clock was reset at some point, Min PTS is approximated)
Max PTS: 00:03:58:195
Length: 00:10:00:605

Initial GOP time: 21:36:28:363
Final GOP time: 21:40:19:400 +1F
Diff. GOP length: 00:03:51:037 +1F (00:03:51:070)

[Volker Quetschke]

I had a look at the OR08-prg2.ts file and found that it is barely an acceptable file. You stripped out the audio track and the PTS information of that file - so min and max PTS information are meaningless anyway for this file.

On the other hand, using the --goptime switch should have worked, but it fails. I will look into fixing this.