Bug#502879: azureus: has problems decoding file names in some torrents

[Christian Ohm]

On Friday, 25 September 2009 at 13:26, Adrian Perez wrote:
> The file shows me ok, the question marks, in The_8472 and Nolar words,
> means you don't have the font/character-set installed.
> There's nothing wrong with the code know, as far as we can see.

It's not a font issue, the Japanese filenames that are decoded correctly are
all shown correctly as well in Azureus, and in contrast, when I download this
torrent with azureus, the filename is wrong in other programs as well.

When I load this torrent, I get the attached result, where the first file
should be "(ドラマ)エジソンの母 第01話...", but each kana (not kanji) is
replaced with a question mark. Kana and kanji are two types of Japanese
characters, where kana are only used in Japan, and kanji are adapted from
Chinese.

Looking at the torrent file itself (with less on an utf-8 terminal), it seems
it contains the filenames twice, once as "path" in (I guess, it is shown as
"encoding" before) big5 (which is a Chinese encoding and so probably cannot
represent the Japanese-only kana), and once as "path.utf-8" which is shown
correctly. But even when I set "Files - Torrents - Decoding Character Sets -
Default torrent encoding when selection required to UTF-8, Azureus still shows
question marks - looks to me like it just uses the first filename it finds,
while qbittorrent and transmission use the UTF-8 name.

Best regards,
Christian Ohm