Multilingual support

[anonymous]

Using libtorrent (0.6.7) and rtorrent (0.2.7), it seems that it does not support any torrent using non-ASCII filenames. It uses something that is weird as the filename instead.

[rakshasa]

Currently it faithfully uses whatever filename the torrent encodes. I'm not familiar with how non-ascii filenames are encoded and how it should be converted.

Though atleast in rtorrent characters that can't be displayed should perhaps be converted to hex or something.

[rakshasa]

Ticket #33 (Closed: duplicate)

Description by Bernaard:

Rtorrent doesn't make folders containing æøå and sets torrent status to inactive. This happens on a SuSE 9.2-box with a unicoded JFS filesystem.

Modified by rakshasa

• resolution set to duplicate
• status changed from new to closed

I tested it out by making a torrent from a directory named "åøæ". When trying to seed it, it created a directory with a mangled file name. Looks like this bug is caused by #17.

[Bernaard]

Well, it didn't even create the directory on my box and went "closed" and "inactive"!

[rakshasa]

Which is propably due to the directory names being mangled. Different environments can react differently to this problem.

[Bernaard]

I see... Thanks for the good work with rtorrent/libtorrent. Bernaard

[anonymous]

BitComet? (and probably some other clients) insert a key named "encoding" into the metadata .torrent file. You could use iconv(3) to recode the file names to either UTF-8 or system locale.

[anonymous]

Many metainfo making programs include a "name.utf-8" field next to the "name" field, or for batch torrents, a "path.utf-8" field. If the system locale is set to UTF-8, you could set libtorrent default to the utf-8 filenames.

[rakshasa]

Until then, you should use the badly named "encoding_list = utf-8" option.

[zap]

But if I use encoding_list=utf-8 it will name files in UTF-8, right? I mean, I'll get garbage names instead of names in my system charset encoding.

[anonymous]

It would be great to pick up name.utf-8 and path.utf-8 from torrent files.

That way, torrents that don't specify an encoding can still be handled correctly, without creating filenames with ? characters. The utf-8 name should be iconv-ed to the system's filename encoding, which one can get from nl_langinfo(CODESET). An rtorrent config option could override that filename encoding.

[crystal.shine.ua@gmail.com]

Hi, I also had same problem. But the good thing is that it uses whatever filename the torrent encodes. I'm not quite aware of how non-ascii filenames are encoded and how it should be converted, and I looked for some articles on this topic at  http://filecraft.com/ but I think it should be somehow fixed.

Edward E

[rakshasa]

Milestone 1.0 Soggy Rabbit deleted