In the lot-1.0.zip:
build.xml main ant build script containing main targets definitions
build.*.xml ant build script files loaded from build.xml
config/* config files to control how the script works
# selected according to properties 'mode',
'package', 'locale' etc.
# config/*.conf and config/mozilla1.8/* are for 1.5.0.x
filters/* filters files, that define replacement of variables
in the src locale files
# as you can expect, ja.filters is for ja and
ja-JP-mac.filters is for ja-JP-mac
# common.filters etc are used to make maintenance easy
src/* src locale files (both dtd and properties are in utf-8)
# src/mozilla1.8/ja/* are for 1.5.0.x ja/ja-JP-mac
I use this script for all localization works (get diff of CVS, get
locale files from CVS, make langpack, compare-locales, re-package win
installer, generate diff html table, etc...) and the script is a
little complicated. But the flow to generate (split) ja and ja-JP-mac
The script will replace variables like '@@varname@@' to platform
dependent strings according to defined in the filters files and output
ja/ja-JP-mac files under 'loose' directory.
# In this time, properties files are converted from UTF-8 to Unicode escape.
That is, we maintain one set of files and in case we have to split
translation (words, expression, window size, setting...), we'll define
variables in the filters and use that variables in the src locale
# Files commited into L10N CVS are the files generated under 'loose' directory.
To know usage of Localization Tools, exec:
ant help -Dlang=English
To generage ja/ja-JP-mac files and lang pack, exec:
And you can see other command line examples with:
In case of Japanese, we have two translation for the same English word
(or sentence, etc...) according to each platform's custom/guideline.
It sometimes depends on the context and it cannot be automatically
generated from one to the another.
In case of Serbian, they need not two translation. They just needs two
same translation written in different scripts (alphabet) and it can be
automatically generated from one (cyrillic) to the another (laten).
For difference like window size, Serbian also be able to do same thing
(or just our tool with a little customization to exec their sed
If the build script on mozilla.org support variable to split
translation, we don't need to generate locally and other locales will
be able to do same or similer things.