BirdNET-Analyzer icon indicating copy to clipboard operation
BirdNET-Analyzer copied to clipboard

analyze.py seems to skip many files if it encounters a corrupt file

Open kami2012 opened this issue 1 year ago • 4 comments

I analyzed about 2.4 million .wav files in junk files of about 250.000 to 300.000 files. In two of these junks there were 4 corrupted files. Because of these 4 corrupted files, about 6550 other uncorrupted files could not be analyzed. It seems that analyze.py gets some sort of "hiccup" when it encounters a corrupt file and skips hundreds of subsequent files that have no problem.

Is this something that could be optimized in a newer version?

kami2012 avatar May 22 '24 08:05 kami2012

If the Analyzer encounters an error or corrupt file, the file just gets skipped. Could you check your error_log.txt and log.txt for error codes if this happens again?

Josef-Haupt avatar May 28 '24 16:05 Josef-Haupt

I encountered a similar issue. I had a folder with 25 audio files, but none of the files were being uploaded for analysis. After doing some checks, I discovered that one of the files was corrupt. The corrupt file was not being skipped, and as a result, the other files were not uploaded for analysis either. Once I deleted the corrupt file, the entire folder uploaded successfully with all the remaining files.

GrunCrow avatar Jun 12 '24 08:06 GrunCrow

As I said, please check your error_log.txt and log.txt in the install directory for error codes if this happens again, and paste the contents into this issue. Or provide us with a corrupt file, for us to run tests.

Josef-Haupt avatar Jun 17 '24 10:06 Josef-Haupt

As I said, please check your error_log.txt and log.txt in the install directory for error codes if this happens again, and paste the contents into this issue. Or provide us with a corrupt file, for us to run tests.

This is for example a corrupt file: Corrupt_Audio.zip

GrunCrow avatar Jun 18 '24 07:06 GrunCrow

I just inserted your corrupt file into a batch, but the current repo correctly skips the file, but analyses the rest. Is this still an issure for you in the current main?

Josef-Haupt avatar Oct 29 '24 10:10 Josef-Haupt