Error Report hscommon\gui\progress_window.py

[bdworking]

Error Report from crash.

Debug Log

Application Name: dupeGuru
Version: 4.0.4 RC
Windows

Traceback (most recent call last):
File "hscommon\gui\progress_window.py", line 101, in pulse
File "core\app.py", line 323, in _job_error
File "hscommon\jobprogress\performer.py", line 43, in _async_run
File "core\app.py", line 780, in do
File "core\scanner.py", line 137, in get_dupe_groups
File "core\pe\scanner.py", line 31, in _getmatches
File "core\pe\matchblock.py", line 167, in getmatches
File "core\pe\matchblock.py", line 83, in prepare_pictures
File "core\pe\photo.py", line 105, in get_blocks
File "qt\pe\photo.py", line 36, in _plat_get_blocks
TypeError: '<=' not supported between instances of 'int' and 'str'

[JensDA]

I've got a similar failure on Linux while scanning in picture mode with Match pictures of different dimensions enabled and a filter hardness of 93.

Application Name: dupeGuru
Version: 4.0.4 RC

Traceback (most recent call last):
  File "/usr/share/dupeguru/hscommon/gui/progress_window.py", line 101, in pulse
    should_continue = self._error_func(self.jobid, err)
  File "/usr/share/dupeguru/core/app.py", line 323, in _job_error
    raise err
  File "/usr/share/dupeguru/hscommon/jobprogress/performer.py", line 43, in _async_run
    target(*args)
  File "/usr/share/dupeguru/core/app.py", line 780, in do
    self.results.groups = scanner.get_dupe_groups(files, self.ignore_list, j)
  File "/usr/share/dupeguru/core/scanner.py", line 137, in get_dupe_groups
    matches = self._getmatches(files, j)
  File "/usr/share/dupeguru/core/pe/scanner.py", line 26, in _getmatches
    return matchblock.getmatches(
  File "/usr/share/dupeguru/core/pe/matchblock.py", line 167, in getmatches
    pictures = prepare_pictures(pictures, cache_path, with_dimensions=not match_scaled, j=j)
  File "/usr/share/dupeguru/core/pe/matchblock.py", line 83, in prepare_pictures
    blocks = picture.get_blocks(BLOCK_COUNT_PER_SIDE)
  File "/usr/share/dupeguru/core/pe/photo.py", line 105, in get_blocks
    return self._plat_get_blocks(block_count_per_side, self._get_orientation())
  File "/usr/share/dupeguru/qt/pe/photo.py", line 36, in _plat_get_blocks
    if 2 <= orientation <= 8:
TypeError: '<=' not supported between instances of 'int' and 'str'

[bdworking]

I was moving files and a folder while dupguru was processing those files/folder.
Suspect the code is not paranoid enough and does not handle small failures gracefully.

[glubsy]

If you can reproduce the bug, please enable logging in the Preferences and post the log here.

This has something to do with EXIF data. The error probably comes from core/pe/exif.py (in get_fields() or more specifically in dump_IFD())or core/pe/photo.py.

[JensDA]

On reproduction, debug log showed this:

2020-08-06 03:17:18,236 - DEBUG - Exif header length: 15952 bytes
2020-08-06 03:17:18,237 - DEBUG - Motorola format
2020-08-06 03:17:18,237 - DEBUG - IFD at offset 8
2020-08-06 03:17:18,237 - DEBUG - Entries for IFD 8: 12
2020-08-06 03:17:18,238 - DEBUG - Exif SubIFD at offset 240:
2020-08-06 03:17:18,238 - DEBUG - Entries for IFD 240: 34
2020-08-06 03:17:18,242 - DEBUG - Exif Interoperability SubSubIFD at offset 13002:
2020-08-06 03:17:18,242 - DEBUG - Entries for IFD 13002: 1

[JensDA]

exiftool reports Orientation : Horizontal (normal)

[glubsy]

Thanks for the report. I'm assuming this is a TIFF file. If you want to test it further you could run my workaround branch (in debug mode still) and see the exact output.
Or if you know which file is causing this and don't mind sharing it, I could run the test for you.

[JensDA]

I tested the workaround branch and can confirm that the failure no longer occurs:
WARNING - Orientation for file '/home/jens/test/test.jpg' was a str '1', not an int!

Unrelated, but is it normal for the scan in picture mode to take multiple minutes to scan a directory containing only 3 pictures using Sqlite cache mode?

[glubsy]

Thank you very much for testing the patch and reporting the log. Looks like it might be safe to just cast the string to an int (hopefully).

Unrelated, but is it normal for the scan in picture mode to take multiple minutes to scan a directory containing only 3 pictures using Sqlite cache mode?

This is not normal, unless maybe the files are exceptionally huge? Not familiar with the code to be honest but this might warrant a separate issue report.