Python, GExiv2, and unicode file names

[jim-easterbrook]

I've run into a problem using GExiv2 from Python and I'm not sure exactly where the problem lies. A short script shows the problem:

import sys

import gi
gi.require_version('GExiv2', '0.10')
from gi.repository import GExiv2

def main():
    print('GExiv2 version', GExiv2.MAJOR_VERSION, GExiv2.MINOR_VERSION,
          GExiv2.MICRO_VERSION)
    for path in sys.argv[1:]:
        print(path)
        md = GExiv2.Metadata()
        md.open_path(path)
        print(md.get_exif_tags())
    return 0

if __name__ == "__main__":
    sys.exit(main())

When run on a non-ascii file name this happens:

$ python gexiv2_test.py /c/Users/Jim/Pictures/fünny_name.JPG
GExiv2 version 0 12 1
C:/Users/Jim/Pictures/fünny_name.JPG
Traceback (most recent call last):
  File "gexiv2_test.py", line 20, in <module>
    sys.exit(main())
  File "gexiv2_test.py", line 15, in main
    md.open_path(path)
  File "C:/msys64/mingw64/lib/python3.8/site-packages/gi/overrides/GExiv2.py", line 44, in open_path
    super(Metadata, self).open_path(path)
gi.repository.GLib.Error: GExiv2: C:/Users/Jim/Pictures/fünny_name.JPG: Failed to open the data source: No such file or directory (errno = 2) (9)

Exiv2 opens the file without a problem:

$ exiv2 -pa /c/Users/Jim/Pictures/fünny_name.JPG
Error: Directory Canon with 6144 entries considered invalid; not read.
Exif.Image.ProcessingSoftware                Ascii      26  Photini editor v2020.10.0
Exif.Image.ImageDescription                  Ascii      78  Château-Gontier, France
Exif.Image.Make                              Ascii       6  Canon
Exif.Image.Model                             Ascii      25  Canon PowerShot A1100 IS
Exif.Image.Orientation                       Short       1  top, left
...

I've seen that #2837 fixes exiv2. Does gexiv2 need fixing?

[lazka]

From what I see exiv2 exposes extra unicode API using wstring when built with EXV_UNICODE_PATH. gexiv currently passes the paths using the glib path encoding (utf-8 on Windows) to the ANSI API which is wrong, and just happens to work with ASCII.

gexiv2 needs to convert paths from utf-8 to utf-16-le, then to wstring, and then pass it to the exiv unicode API. So yes, this needs work with gexiv2. I'd suggest filing a bug there.

[jim-easterbrook]

I've found this GLib bug which appears to be related to the problem https://bugzilla.gnome.org/show_bug.cgi?id=795006. It's marked "resolved obsolete" though, as they expect to use streams rather than files in future. The Python interface to GExiv2 has a method to read from a stream but not to write to one.

[lazka]

yeah, LRN more or less comes to the same conclusion there.

[jim-easterbrook]

Bug filed with GExiv2: https://gitlab.gnome.org/GNOME/gexiv2/-/issues/59

[lazka]

Thanks