tiens, vu que t'as pas de reponse ... je te balance le contenu du po/readme de mon webmail (horde/IMP, une merveille)
ca va p'tet t'aider ...
---
$Horde: horde/po/README,v 1.1.2.9 2002/03/24 13:36:08 jan Exp $
Horde Translation Guide
Copyright 2000-2002 Joris Braakman <jbraakman@yahoo.com>
Copyright 2001-2002 Chuck Hagenbuch <chuck@horde.org>
Copyright 2001-2002 Jan Schneider <jan@horde.org>
Contents
========
- GNU gettext, PHP and Horde
- Starting a new translation
- Example messages header
- Updating a existing translation
- Debugging
- Solaris
- Changing the English standard texts
GNU gettext, PHP and Horde
--------------------------
Horde uses GNU gettext for internationalization (i18n) and localization
(l10n). The manual at http://www.gnu.org/manual/gettext/index.html is
biased against C and using Emacs. This is more for Horde.
There is a good explanation for PHP and gettext at:
http://www.faqts.com/knowledge-bas [...] 53/fid/422
People seem to like learning from examples better, so I have used
dutch (nl_NL) as an example everywhere.
Starting a new translation
--------------------------
- run xgettext.sh (in this directory - horde/po/), this generates a
messages.po file.
- rename it to the language you are translating to, e.g. nl_NL.po
- adjust the header of nl_NL.po, then remove the #, fuzzy line.
Emacs has support for .po files, but I used vi
- translate, the time consuming part.
- convert the nl_NL.po to horde.mo and put the file in place:
$ make install
Compiling locale nl_NL:
261 translated messages, 21 untranslated messages.
... done
- modify the horde/config/lang.php to use nl_NL as the default language, if you
wish.
- Make sure your language exists in the horde/config/lang.php file. Also, if
the translation uses a character set that isn't ascii, make sure the
appropriate charset is defined in $nls['charsets']. You should also define
an alias for the language because some browsers only send a two letter code
in their Accept-Charset header.
And then it works.
Example messages header
-----------------------
# Dutch translation for Horde.
# Joris Braakman <jbraakman@yahoo.com>, 2000.
#
msgid ""
msgstr ""
"Project-Id-Version: Horde 2.3\n"
"POT-Creation-Date: 2000-08-14 10:30+0200\n"
"PO-Revision-Date: 2000-08-14 17:17+02:00\n"
"Last-Translator: Joris Braakman <jbraakman@yahoo.com>\n"
"Language-Team: Dutch <dev@lists.horde.org>\n"
"MIME-Version: 1.0\n"
"Content-Type: text/plain; charset=ISO-8859-1\n"
"Content-Transfer-Encoding: 8-bit\n"
Updating a existing translation
-------------------------------
As soon as you check out a new version you have to check the translations
again, messages might have been changed after all. If you don't, you
can get strange effects. Information is cached every where so as a no
brain solutions I sometimes also restart the webbrowser and webserver
to make sure everything is fresh. msgmerge will mark translations that
have changed a bit to fuzzy.
If you want to update an existing translation and contribute it to the Horde
project you should first contact the last translator to see if he is still
working on it, doesn't have time to further maintain it or is even unreachable.
- run xgettext.sh again, this generates a messages.po file.
- merge it with the old .po file:
$ msgmerge nl_NL.po messages.po > nl_NL-new.po
- translate/update the strings that are new or have become fuzzy.
- delete the '#, fuzzy' lines.
- move the nl_NL-new.po to nl_NL.po
- put the new translations in effect
$ make install
Compiling locale nl_NL:
261 translated messages, 21 untranslated messages.
... done
Debugging
---------
Is this locale (nl_NL) installed at all?
$ locale -a
should list all locales installed on your system.
On Debian not all locales may be enabled by default. Edit /etc/locale.gen and
run locale-gen if you changed the list of enabled locales.
Do you have any .mo files? Usually in /usr/share/locale/
e.g. /usr/share/locale/nl/LC_MESSAGES/tar.mo
Does gettext even work?
Get a string to translate,
$ strings /bin/tar | grep Memory
Memory exhausted
$ (LANG=nl_NL; LANGUAGE=nl_NL; LC_MESSAGES=nl_NL; gettext tar "Memory exhausted" )
Geheugen uitgeput
Does the local Horde file work?
Assuming that you have put the translated Horde file in
/data/www/horde/locale/nl_NL/LC_MESSAGES/horde.mo
$ export TEXTDOMAINDIR=/data/www/horde/locale
$ (LANG=nl_NL; LANGUAGE=nl_NL; LC_MESSAGES=nl_NL; gettext horde "Message" )
Bericht
Create a file in the horde directory, langtest.php:
<?php
setlocale(LC_MESSAGES, 'nl_NL';
putenv('LANG=nl_NL';
putenv('LANGUAGE=nl_NL';
// use the tar test.
echo dgettext('tar', 'Memory exhausted';
echo '<br />';
// Specify location of translation tables
bindtextdomain('horde', './locale';
// Choose domain
textdomain('horde';
// Print the already tested message
echo _("Message" );
echo '<br />';
// this should print the same.
echo dgettext('horde', 'Message';
?>
OUTPUT web browser:
Geheugen uitgeput
Bericht
Bericht
Solaris
-------
Since the .mo files are binary, they are platform specific. You
have to rerun make in all po directories.
You have not installed the Supplementary Partial Locales (SUNWploc1) if
you get:
$ LANG=nl_NL
couldn't set locale correctly
This is what it should say.
$ pkginfo | grep ploc
system SUNWploc Partial Locales
system SUNWploc1 Supplementary Partial Locales
The stuff is installed in /usr/lib/locale
$ ls /usr/lib/locale/nl
LC_COLLATE LC_CTYPE LC_MESSAGES LC_MONETARY LC_NUMERIC LC_TIME nl.so.1
We assume that the Solaris gettext implementation is installed when we compile
the translations. Thus the make process will fail if the installed gettext is
from GNU. If anyone knows how to determine if the installed gettext is from
Solaris or GNU on Solaris systems, please send us a note.
Changing the English standard texts
-----------------------------------
You can also use gettext to change some of the English texts to your own taste.
- Create a message.po file as described in "Starting a new translation"
- Edit the msgstr entry of the strings you want to change. You can leave all
other msgstr entries empty.
- Run "make install"