[Bf-committers] Arabic translation support in the UI script

Dalai Felinto dfelinto at gmail.com
Thu Oct 6 09:22:21 CEST 2011

I talked with Sergey and this is the current workflow. We can add it to
//po/README.txt if other translators will run into that. But the basic
information is there already:

(1) run update_msg.py (it needs compiled Blender)
(2) run update_pot.py - gather the latest strings from Blender
(3) run update_po.py - sync in the new changes from trunk
(4) run clean_po.py - remove the no longer needed msgstr/msgid
(5) run merge_po.py to bring the new .po file into trunk's .po

If you actually want to discard the old translations from trunk's ar.po it's
better to cook your own .py for that (trivial to adapt the arabic_to_utf.py
for that as you know). Given that this is the exception and not the rule I
don't think we need this officially in trunk. This would happen between (4)
and (5).

* Note: you likely want to run clean_po.py in your new .po to avoid
translate strings that are not needed.

(can someone please share this in the translators mailing-list?)


2011/10/5 Dalai Felinto <dfelinto at gmail.com>

> Hi Yousef ,
> Blender is no longer storing the .pot file in trunk. The reason is that
> this file changes a lot, and is really easy to rebuild running the scripts
> in the .po folder.
> Would it help if I send you the latest blender.pot and the ar.po resulting
> from that? You would still need to run a .po merge.
> Maybe we could take the resulting trunk ar.po, clean all the msgstr and
> then use the merge_po.py scrip.
> The script was written by Sergey, CC here.
> @Sergey, is that right? any other suggestions here?
> By the way, here [1] you can find my new take in the arabic_to_utf script.
> There are a few changes making it to look more python-like (I got some of
> your changes there). The result will still differ from your take on it. The
> reason(s) is: (1) I believe I need to check for both side-connections, not
> only the start. (2) since your 1st script it seems you didn't check whether
> the char itself can connect to the left/right (always checking only the next
> chars). (3) if you don't increase (g) inside the loop for
> the laaam/laaaa/laaahz/.... you will end up processing the following char
> twice.
> * I'm not so sure about the (3), but the (1) and (2) reflects the little I
> understand of the arabic language (reading your code + wikipedia + help from
> a friend)
> Regards,
> Dalai
> 1- http://dl.dropbox.com/u/3292898/arabic/arabic_to_utf.py
> 2011/10/4 Yousef Hurfoush <bat3a at msn.com>
>> hi
>> @ Dalai
>> i finished translating the 1st milestone in ar.po (~1000 phrase), and i'd
>> like to update it in the trunk:
>> the problem is that it is based on pot file which was before the new
>> generated ar.po
>> another issue is that this we are re-translating the file from zero,
>> and the one in the trunk is about 5000 fuzzy translated misplaced words
>> so i need an empty .pot template and i can't generate it the old way!
>> can you provide me with an empty ar.po file i could work on?
>> and can i merge my translated file into it without errors
>> here is the new translation:
>> https://sourceforge.net/p/batp/code/44/tree/done/ar.po
>> and a screen shot in blender:
>> https://sourceforge.net/p/batp/screenshot/1st~1000.jpg
>> thanks in advance
>> Yousef Harfoush

More information about the Bf-committers mailing list