Queueing for review and possible fix in the next release.

Created attachment 540 [details]
Fix non-utf8 locale support

This is hack'ish replacement for not proper working "isinstance(s, unicode)" expression. Fixes file and folder names when non-utf8 locale is used.

(In reply to comment #0)
> gpodder 2.7 hangs when trying to sync (^S) to a directory that contains a
> non-ascii character.  The directory name is from the podcast's title.
> 
> gpodder 2.4 does not have this problem.  I did not test 2.5 or 2.6.
> 
> Example feed: http://omegataupodcast.net/category/podcast-en/feed/
> 
> The relevant entry from an exported .opml:
> 
> <outline text="Wissenschaft und Technik im Kopfhoerer / Science and Technology
> in your Headphones" title="omega tau » podcast (en)" type="rss"
> xmlUrl="http://omegataupodcast.net/category/podcast-en/feed/"/>
> 
> Note that title contains a '»' (U+00BB / RIGHT-POINTING DOUBLE ANGLE QUOTATION
> MARK / angle quotation mark (right) / &raquo; / &#187; / raquo) character.
> 
> Error message:
> 
> Exception in thread Thread-8:
> Traceback (most recent call last):
>   File "/usr/lib/python2.6/threading.py", line 525, in __bootstrap_inner
>     self.run()                            
>   File "/usr/lib/python2.6/threading.py", line 477, in run
>     self.__target(*self.__args, **self.__kwargs)
>   File ".../gpodder-2.7/src/gpodder/gtkui/desktop/sync.py", line 186, in
> sync_thread_func
>     device.add_tracks(episodes)
>   File ".../gpodder-2.7/src/gpodder/sync.py", line 200, in add_tracks
>     added = self.add_track(track)
>   File "/home/chkno/local/src/gpodder-2.7/src/gpodder/sync.py", line 597, in
> add_track
>     log('Copying %s => %s', os.path.basename(from_file),
> to_file.decode(util.encoding), sender=self)
>   File ".../gpodder-2.7/src/gpodder/liblogger.py", line 49, in log
>     print (('[%8.3f] ' % (time.time()-first_time)) + message) % args
> UnicodeEncodeError: 'ascii' codec can't encode characters in position 115-116:
> ordinal not in range(128)
> 
> 
> On startup, gpodder says:
> [   2.294] Using ISO-8859-15 as encoding. If this
> [   2.294] is incorrect, please set your $LANG variable.

I have similar (locale-related) problem with 2.5 and 2.7 under Windows.
My system uses (like other Russian Windows do) ru_RU.CP1251 locale.
If i start gPodder 'as is' it uses utf8 for  filenames and folders (both downloaded episodes and synced filenames). Thus a have to start gPodder with "set LANG=ru_RU.cp1251" in batch file.

Some investigation show me that "isinstance(filename, unicode)" expression does not work as  expected under Windows. I'm not familiar with python/unicode so can't find proper solution, attached  patch works for me, even if it really hack'ish. Perhaps, it will help you too and help devs to find/fix  the bug.

(In reply to comment #3)
> If i start gPodder 'as is' it uses utf8 for  filenames and folders (both
> downloaded episodes and synced filenames). Thus a have to start gPodder with
> "set LANG=ru_RU.cp1251" in batch file.

Do you think we can add code to gPodder that can detect the "LANG" variable based on some Windows registry setting or something?

(In reply to comment #4)
> (In reply to comment #3)
> > If i start gPodder 'as is' it uses utf8 for  filenames and folders (both
> > downloaded episodes and synced filenames). Thus a have to start gPodder with
> > "set LANG=ru_RU.cp1251" in batch file.
> 
> Do you think we can add code to gPodder that can detect the "LANG" variable
> based on some Windows registry setting or something?

Quick google'ing didn't show me any reliable methods. 

Om my machine i find tiny solution (patch below) instead of setting LANG variable.
Unfortunately i tested it only with  "filesystem sync" under win64 and
can't say what will happen under other OS/locales.  

And i want to point again: setting LANG was only the part of the solution.
There is something wrong with unicode strings (and therefore "isunicode" routine) inside gPodder code. As far as i can understood python does not treat
some of the strings passed to sanitize_filename as unicode.

Here is tiny patch: 

diff --git a/lib/site-packages/gpodder/util.py b/lib/site-packages/gpodder/util.py
index 320afd8..5232f4a 100644
--- a/lib/site-packages/gpodder/util.py
+++ b/lib/site-packages/gpodder/util.py
@@ -68,7 +68,8 @@ N_ = gpodder.ngettext
 if gpodder.ui.maemo:
     encoding = 'utf8'
 else:
-    encoding = 'iso-8859-15'
+    import sys
+    encoding = sys.getfilesystemencoding()
 
 if 'LANG' in os.environ and '.' in os.environ['LANG']:
     lang = os.environ['LANG']

Fixed: http://gpodder.org/commit/060ba862

Please test and re-open if this bug still exists in the next release. Thanks!