replaced regex for turning list of matched frames -> start, end frames.
Replaced by the frame matching logic having the frames just keep track of their own number.
This should result in far less errors, and a marginal speedup
catch errors where a frame image either doesn't exist or can't be opened/read/turned into a PIL image.
This is handled by ending the addition of new segments to the fingerprint, potentially leaving a partial fingerprint.
The upside of that is it's possible the file could still be processed with a partial fingerprint.
The downside is that it could allow other issues to go unnoticed.
Issues
the regex used to take the matched frames and turn them into start and end frames was buggy and often resulted in the wrong start time.
Changes
replaced regex for turning list of matched frames -> start, end frames. Replaced by the frame matching logic having the frames just keep track of their own number. This should result in far less errors, and a marginal speedup
catch errors where a frame image either doesn't exist or can't be opened/read/turned into a PIL image. This is handled by ending the addition of new segments to the fingerprint, potentially leaving a partial fingerprint. The upside of that is it's possible the file could still be processed with a partial fingerprint. The downside is that it could allow other issues to go unnoticed.
Issues