[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [ESD-translators] state of the template manager -- kitchen

From: Thérèse Godefroy
Subject: Re: [ESD-translators] state of the template manager -- kitchen
Date: Mon, 28 Mar 2016 21:02:57 +0200
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Icedove/38.7.0

Hello Tomas, hello all,

Le 28/03/2016 05:47, Stary, Tomas a écrit :
> I have now started using the po4a and it works
> very good for me. I think it should be the preferred way of working for
> anyone who is starting a new translation from scratch. I can also see
> benefits of the system for existing translation, but it is up to you
> if you decide to migrate to this system. If yes, I have described the
> process here.

That's a very good summary. has a short but detailed howto for POedit [3]. We could probably
adapt it to our workflow.

> To make it more simple, I have prepared Makefile with all the required
> commands for the creation of empty template POT files, extraction of
> PO files with strings from an existing translation, and update of the
> PO file for a new version. The generation of the translated HTML is
> done by commands in Makefile.gen.


Here are some variations that you might consider:

1. In Makefile (po4a-gettextize):
Disable extraction of <script> (assuming nobody will want to translate
the javascript comments), enable extraction of <meta> content, + a few
cosmetic things -- "porefs=none" disables the list of occurrences above
each string; these lists may be useful at the beginning, but will become
bothersome in the long run.

PO4A_FLAGS=-M utf-8 -f xhtml -o porefs=none \
           -o 'untranslated=<script>' \
           -o 'attributes=<meta>content' \
           --package-name emailselfdefense --package-version $version

2. In Makefile.gen, it may be useful to relax the default condition for
actual regeneration of translated files (option "keep"). The default is
80%. It means you can't generate a translated page before 80% of the
strings are translated. That's a little too high IMO. You may want to
generate pages that are 30% translated (or even less) for proofreading.

PO4A_FLAGS=-M utf-8 -f xhtml --keep 30

3. The language list shouldn't be translated, so we might as well
replace it with a placeholder in the original. The real thing would be
added and customized with the "current" class after the translated pages
are generated (that goes for the English version as well).

4. The html tag needs to be customized, at least for RTL languages.
This could be done with sed, after regeneration. Alternatively,
<!DOCTYPE>, <html>, <head> and a few untranslatable tags could be
replaced with a placeholder:
   <html><head> <!-- HTML header -->
(the 2 tags are needed to keep po4a happy).

5. The system could be further refined with 2 external scripts:

- gnun-add-fuzzy-diff (from the GNUN package [4]). Very useful to locate
a change within a string. It runs after po4a-updatepo.

- some sort of reformatting script to get rid of the extra spaces, tabs,
linefeeds, etc., to make sure that a spurious difference between
index.html, mac.html and windows.html isn't going to add useless strings
to an existing POT file, or make the POs fuzzy for no reason. There is
such a script in the kitchen (reformat-html), but I am currently using a
simpler one (reformat-original, attached).

Basically, the ingredients to build translated pages would be (1) the PO
file, (2) English originals with placeholders, (3) language list, (4)
header. The final processing could be done just before uploading to the
live repo.

> I have tried to put a lot of comments in the Makefiles, but perhaps
> some bits might not be completely clear. I have found the manual
> pages[2] of po4a quite helpful (especially the Graphical overview).

Your comments seem clear.


> [0]
> [1]
> [2]

Attachment: reformat-original.gz
Description: application/gzip

reply via email to

[Prev in Thread] Current Thread [Next in Thread]