Fails with unicode characters in filename

I get some breakage on dapper - no error message but no conversion either.

Which characters were you having problems with?

The first file it barfed on had a single e with acute accent (french music)

Worse -- I have now tried the program on a single file with no accents, and (separate bug!?) I get stderr output:

using encoder: vorbisenc (Ogg Vorbis encoder (deprecated))
Writing to: 'file:///home/cpbl/music/brigitte/dumas/le cours des jours/02 le cours des jours.ogg'

and the main window says "Conversion done, in 0 s", but no file is created!

This program does nothing for me...

Also, I do not see the relevance of the debbug 385181 to my original problem.

That bug fix didn't help.

The Debian bug which is now fixed is NOT this problem.
 I have installed the fix for that bug, and this one still exists in the very latest version.

Hum, here I can use unicode filenames and tags...
can you please send at <kassoulet gmail com> one example file which don't work ?

I sent it.
By the way, if I "add file" and, rather than clicking through directories, I type the path (using tab completion) in the file selection dialog, I get an error dialog box:

Could not retrieve information about the file
error accessing 'file:///home/cpbl/music/brigitte/dumas/le%20cours%20des%20jours/01%20h%EF%BF%BDlium.wma': File not found

even though the file is obviously there: i got to the path by tab completion .

C

To be explicit: I am using:
SoundConverter 0.8.7

Hi,

On 9/27/06, Christopher Barrington-Leigh <email address hidden> wrote:
> I sent it.

I'm not able to reproduce the error.
I attach a new version that you can test if you want, hopefully you
will get better messages.
please send me the output of the program and/or screenshot of the error.

to install, unpack and run "sudo make install"

thank you for your help !

--
Gautier Portet <kassoulet gmail com>
  --==-- http://kassoulet.free.fr --==--

I am not seeing the attachment.
Thanks!
c

On 9/28/06, Christopher Barrington-Leigh <email address hidden> wrote:
> I am not seeing the attachment.
> Thanks!

[Gautier sent me a link, privately]

Behaviour fails on several levels. I can detail others separately, but if I select "single file" and give it the same one as before (by clicking through menus), the app does nothing, but says:

[1301][cpbl@:~/tmp/dl/soundconverter-0.8.8]$ ./soundconverter.py
SoundConverter 0.8.8
  Gstreamer version: 0.8.11, Python binding version: 0.8.3
locale charset: UTF-8
  using gnomevfssrc
Traceback (most recent call last):
  File "./soundconverter.py", line 426, in do_work
    if self.work():
  File "./soundconverter.py", line 500, in work
    self.tasks[0].finish()
  File "./soundconverter.py", line 629, in finish
    self.found_type_hook(self.sound_file, self.found_type)
  File "./soundconverter.py", line 931, in found_type
    self.append_file(sound_file)
  File "./soundconverter.py", line 1000, in append_file
    self.model.set(i, 0, self.format_cell(sound_file))
  File "./soundconverter.py", line 991, in format_cell
    error.show(_("Invalid character in filename:\n'%s'\n(%s)") % (s, e))
TypeError: show() takes exactly 3 arguments (2 given)

Hey, want the music, too?
This is not the whole directory tree, but contains the examples above:
http://grad.econ.ubc.ca/cpbl/lptmp/

This is a temporary location; please destroy the music when the bug is fixed. :)

On 9/29/06, Christopher Barrington-Leigh <email address hidden> wrote:
> [Gautier sent me a link, privately]
>
> Behaviour fails on several levels. I can detail others separately, but
> if I select "single file" and give it the same one as before (by
> clicking through menus), the app does nothing, but says:
> TypeError: show() takes exactly 3 arguments (2 given)

oh I see, I've updated the file, hopefully it will get a nice error
message this time.

this is strange, since filenames are handled in unicode, and I'd
tested with kanji filenames...
maybe I misunderstood the whole gtk unicode handling...

--
Gautier Portet <kassoulet gmail com>
  --==-- http://kassoulet.free.fr --==--

Gautier,
I am not sure whether you intended me to download a new version and carry on testing, but I have provided sets of music which seem to give a reproducible error here. I would point out that the functionality for me fails differently if I (1) select a file by typing its path, (2) select a directory containing a file with foreign characters. I have provided my directory structure in the .tar files.

Using your latest version, I got the following:

SoundConverter 0.8.8
  Gstreamer version: 0.8.11, Python binding version: 0.8.3
locale charset: UTF-8
  using gnomevfssrc
[0517][cpbl@:~/tmp/dl/soundconverter-0.8.8]$ Traceback (most recent call last):
  File "./soundconverter.py", line 426, in do_work
    if self.work():
  File "./soundconverter.py", line 500, in work
    self.tasks[0].finish()
  File "./soundconverter.py", line 629, in finish
    self.found_type_hook(self.sound_file, self.found_type)
  File "./soundconverter.py", line 931, in found_type
    self.append_file(sound_file)
  File "./soundconverter.py", line 1001, in append_file
    self.model.set(i, 0, self.format_cell(sound_file))
  File "./soundconverter.py", line 991, in format_cell
    sys.stderr.write(_("Invalid character in filename:\n'%s'\n(%s)") % (urllib.unquote(sound_file.get_uri()), e) )
UnicodeDecodeError: 'utf8' codec can't decode bytes in position 62-64: invalid data

Hi,

On 9/30/06, Christopher Barrington-Leigh <email address hidden> wrote:
> Gautier,
> I am not sure whether you intended me to download a new version and carry on testing, but I have provided sets of music which seem to give a reproducible error here. I would point out that the functionality for me fails differently if I (1) select a file by typing its path, (2) select a directory containing a file with foreign characters. I have provided my directory structure in the .tar files.

I just wanted you to test a version with a better error message,
anyway, with the files you sent me, I can reproduce the error and thus
fix it, so thank you so much for your help.

This is in fact a file system error: the files on your system should
be in encoded utf-8, but these album filenames are encoded in
iso8859-1. In my system, the filenames appear with a "?" symbol on
every accented letter.

I think I know how to fix this, so expect a new version soon, I will
recontact you as soon as I release it.

And really thank you so much !

--
Gautier Portet <kassoulet gmail com>
  --==-- http://kassoulet.free.fr --==--

This seems to be fixed in 0.9.2. Can you confirm, Gautier? If so, I'll attempt to get this into Edgy.

On 10/16/06, William Grant <email address hidden> wrote:
> This seems to be fixed in 0.9.2. Can you confirm, Gautier? If so, I'll
> attempt to get this into Edgy.

Yes, this should be fixed now.
I hope Christopher will confim...

--
Gautier Portet <kassoulet gmail com>
  --==-- http://kassoulet.free.fr --==--

Thanks.
I couldn't find a .deb, so compiled it.
Great! It seems not to get upset at unicode characters anymore, unless the following error output is related:
(There do seem to be various other troubles so that I wouldn't list this as a releasable app. I might play with it more and make another bug report(s).)

SoundConverter 0.9.2
  using Gstreamer version: 0.10.6, Python binding version: 0.10.4
  using gnomevfssrc
Queue done in 0s
error:Could not determine type of stream. (balagane/thumbs.db)
Mime type skipped: image/jpeg (mail us if this is an error)
Mime type skipped: image/jpeg (mail us if this is an error)
Mime type skipped: image/jpeg (mail us if this is an error)
Mime type skipped: text/plain (mail us if this is an error)
Mime type skipped: image/jpeg (mail us if this is an error)
Mime type skipped: text/plain (mail us if this is an error)
error:Could not determine type of stream. (dumas/le cours des jours/thumbs.db)
Mime type skipped: image/jpeg (mail us if this is an error)
Mime type skipped: image/jpeg (mail us if this is an error)
Mime type skipped: image/jpeg (mail us if this is an error)
Mime type skipped: text/plain (mail us if this is an error)
Mime type skipped: image/jpeg (mail us if this is an error)
error:Could not determine type of stream. (daniel lavoie/live au divan vert/thumbs.db)
Mime type skipped: image/jpeg (mail us if this is an error)
Mime type skipped: image/jpeg (mail us if this is an error)
Mime type skipped: text/plain (mail us if this is an error)
Mime type skipped: image/jpeg (mail us if this is an error)
Mime type skipped: image/jpeg (mail us if this is an error)
Mime type skipped: text/plain (mail us if this is an error)
Mime type skipped: text/plain (mail us if this is an error)
Mime type skipped: text/plain (mail us if this is an error)
error:Could not determine type of stream. (carla bruni/quelqu'un m'a dit/thumbs.db)
Mime type skipped: image/jpeg (mail us if this is an error)
Mime type skipped: image/jpeg (mail us if this is an error)
Mime type skipped: text/plain (mail us if this is an error)
Mime type skipped: image/jpeg (mail us if this is an error)
Mime type skipped: image/jpeg (mail us if this is an error)
Traceback (most recent call last):
  File "/usr/local/bin/soundconverter", line 1241, in append_file_tags
    self.model.set(sound_file.model, 0, self.format_cell(sound_file))
  File "/usr/local/bin/soundconverter", line 1203, in format_cell
    params[item] = markup_escape(sound_file.get_tag(item))
  File "/usr/local/bin/soundconverter", line 223, in markup_escape
    str = "&amp;".join(str.split("&"))
AttributeError: 'list' object has no attribute 'split'
error:Could not determine type of stream. (bnabar/couche-tard et lve tt/desktop.ini)
error:Could not determine type of stream. (bnabar/desktop.ini)
Mime type skipped: text/plain (mail us if this is an error)
Mime type skipped: text/plain (mail us if this is an error)
Queue done in 22s
Queue done in 25s

recent version fixes this.

Crash report

I'm a newbie, so don't hurt me if i did something wrong by posting this ;)

Thanks much. I really enjoy Ubuntu GNU/Linux.