[Koha-translate] [important]NEW TRANSLATION TOOL

[paul POULAIN]

I've just commited a new translation tool.
Feel free to test it AND IMPROVE IT.

TRANSLATION TOOL :
============
This transation tool should greatly help Koha translators.
It's composed of 2 script :
* test-extract.pl, that extracts the texts in a template. It is called by
* tmpl_process.
tmpl process can do 3 things :
- create a file with all sentences all files in a directory (& it's 
subdirectories)
- update an existing translation file.
- rebuild translated templates from english & translation file.

Call tmpl_process --help to get full explanations.

HOW TO TRANSLATE Koha :
============
1- create your translation file :
./tmpl_process.pl update -i 
/home/paul/koha.dev/koha/koha-tmpl/opac-tmpl/default/en/ -s opac.fr -r
2- translate your opac.fr file (english & french being separated by a tab
3- create your translated version :
./tmpl_process.pl install -i 
/home/paul/koha.dev/koha/koha-tmpl/opac-tmpl/default/en/ -o 
/home/paul/koha.dev/koha/koha-tmpl/opac-tmpl/default/fr2/ -s opac.fr -r
4- copy images in your new directory (as they are NOT moved by 
tmpl_process.pl install)

If something changes in english version, go back to pt1 & repeat.

WEAKNESSES
=======
I've found some weaknesses (some solvable probably, but maybe some 
unsolvable)
- HTML::Templates pb :
if there is only a <TMPL* > tag between 2 strings, text-extract 
sometimes merges both strings, and so can't replace them in the 
translated template. I've solved all those cases with a <br/> or another 
HTML statement.

- SHORT STATEMENTS
short statements (like "in") can be replaced stupidly (like in cgi-bin 
=> gives cgi-bdans in french). the workaround has been to replace only 
complete words (
line 185 of tmpl_process :
$line =~ s/(\W)$text(\W)/$1$subst$2/g;
)
Don't seem to have any side effect.

- WORD exist in FILENAME
words that are in the templates and in a perl script name are replaced !
For example : Search is replace by Recherche in french, thus,giving <a 
href="cgi-bin/opac-Recherche.pl">, which is wrong...
The only way I've found to solve this is to ignore any translation whose 
translation string is IGNORE
(search is the only cas i've found in french opac)

- MISSING STRINGS
Maybe some strings that should be translated are missing.
You can add whatever you want in text-extract.pl.
For instance, it extracts :
- standard text
- text in alt="SOMETEXT"
- meta html encoding (8859-1) to enable non european translations.

COPYRIGHT
======
@ paul poulain (paul.poulain address@hidden free.fr) & Jerome Vizcaino vizcainj address@hidden 
esiee.fr

--
Paul POULAIN
Consultant indépendant en logiciels libres
responsable francophone de koha (SIGB libre http://www.koha-fr.org)