unicode isn't used for filenames

[avorobey]

I'm running calibre 0.5.14 on Windows XP. When I import an LRF book in Russian, everything works fine but its author and title, in the storage tree of calibre, are mangled into (mostly) underscores. I see that calibre took the utf-8 author/title and ascii-fied them this way, replacing all non-ascii chars with underscores. The names are all unintelligible and it's impossible to find a file manually if I need to for whatever reason.

I'm storing the books on an NTFS disk that has no problem working with unicode dirnames/filenames, so it'd be great if calibre could just do that. If that's impossible for some reason, maybe a transliteration scheme for a few major alphabets?

[kovidgoyal]

Supporting unicode file names across 3 operating systems and a dozen different file systems is impossible I'm afraid.

[avorobey]

Could it be an option in preferences, defaulting to off? If on, just don't underscorize the names and let python do what it can with utf-8 filenames.

I know how unicode filenames can be horrible on different platforms. But consider someone like me who wants to build a library of 100 or 500 non-English books in calibre. The disk hierarchy will be a completely useless mess so the user is locked into calibre and can't get their files out in any meaningful way. That's an uncomfortable feeling.

[kovidgoyal]

Does the save to disk function in calibre also mangle the names? I don't recall if I escape non ascii characters in that function or not.

[avorobey]

It also does, yes.

[yaroslav]

I would like to second this request, since I've run into this problem too. In addition to being confusing, especially when I want to remove books from the device, I expect it to cause conflicts for short author-title pairs and large library, since the only difference between different title is number of underscores. Even worse for authors - I expect there are quite a few authors with the same name length. Because of this problem, I only use command line tools from calibre distribution for books conversion and don't use it for library management at all.

[alleycat]

I am running calibre on English Windows XP with Russian Cyrillic set as default non-ascii encoding. I have a mixture of English, Russian, French, German and Swedish books, more than 1200, and all works fine, i.e. I can see Russian books with Cyillic file names in the file system as well as in Calibre. Everything is consistent, until I take the database to work, where my PC "does not know Cyrillic". When I manipulate Latin books from calibre, all is well, Cyrillic ones look fine, too, until I click on them, which gives this:

[Error 123] The filename, directory name, or volume label syntax is incorrect: u'H:
MY DOCUMENTS
ML
????? ????????' Detailed traceback: Traceback (most recent call last):

File "calibre\gui2\library.pyo", line 350, in current_changed File "calibre\gui2\library.pyo", line 318, in get_book_display_info File "calibre\library\database2.pyo", line 490, in abspath File "os.pyo", line 150, in makedirs File "os.pyo", line 157, in makedirs

Windows Error?: [Error 123] The filename, directory name, or volume label syntax is incorrect: u'H:
MY DOCUMENTS
ML
????? ????????'

However, looking at the same record in SQLite (table "books"):

"47","Призрак Александра Вольфа","Призрак Александра Вольфа","2009-03-28 15:14:11","NULL","1","Газданов; Гайто","","Гайто Газданов/Призрак? Александра Вольфа (47)"

and similarly fine in calibre GUI, of course.

Confusing, isn't it?

[kovidgoyal]

In calibre 0.6.1 all filesnames will be "intelligently" converted to ASCII which should fix these problems.

[bascorp]

 красивые картинки