Wrong filename encoding in premium autobackup?
I use quite some unicode characters in my project and task names, and I see that these are converted to latin1 in the backup archives that are created automatically.
I think this is a bug.
Can you confirm please?
Asked by Miklós Bagi on Aug 21, 2013 - 07:19
Thanks for the quick resopnse.
Is that "import from template" you're referring to? I haven't found other options to import on the UI.
I downloaded the file "287f2399ba93088740183e7cff7c2198.zip" from Options/Backup section. When I unzip that I get files like: KarbantartÃ¡s.txt and I would expect that to be Karbantartás.txt. It's getting weirder, i.e. LÃ©pcsÅ‘korlÃ¡tok and stuff :-)
My filesystem supports unicode correctly:
user@host ~/Downloads/test $ mkdir "(ლಠ益ಠ)ლ Y U NO WORK"
user@host ~/Downloads/test $ ls -la
drwxr-xr-x 3 user group 4096 Aug 21 15:16 .
drwxr-xr-x 7 user group 12288 Aug 21 15:13 ..
-rw-r--r-- 1 user group 41501 Aug 21 14:20 287f2399ba93088740183e7cff7c2198.zip
drwxr-xr-x 2 user group 4096 Aug 21 15:16 (ლಠ益ಠ)ლ Y U NO WORK
Thank you for the confirmation, we'll look into it and see if we can allow this file naming. It may be risky as these are file names they could in rare cases affect the ability to unzip the archive if you try to do this on a system that won't support some characters, but we'll see what can be done.
I took a closer look and it seems like this is double utf8 encoded, not latin as I suspected in my first post.
This usually happens when a routine encodes an already utf8 encoded text to utf8 again.
In the example I used above: KarbantartÃ¡s, it seems that the á character is ending up %25C3%25A1 instead of %C3%A1.
For your reference: http://www.i18nqa.com/debug/utf8-debug.html