Internationalization

This article deals with the localisation/translation of both TDM itself, as well as Fan Missions.

The I18N support in TDM will only work with v1.07 or newer.

I18N, L10N?

I18N stands for internationalisation (the 18 means there are 18 letters missing between I and N), and L10N stands for localisation.

In TDM this means that both the menu, the HUD and all the readable content (shown in in-game GUIs) in a mission can be shown in the language the user selects.

Note: There is not yet support for subtitles. This will come with the next version.

For Users

You can switch the language of TDM in the Settings menu under Video, General. The language is stored in the CVAR sys_lang (Note: in v1.07, it was stored in tdm_lang).

Switching the language does not require a restart, except for switching between Russian and Western languages. There the game will inform you that you need to restart it manually, or the display will not look correct.

Changing the language will change the language of the HUD, as well as the inventory immidiately. It can also be done during a mission.

Note: Not all FMs are yet multi-lingual, and for these certain inventory items, the objectives, or readables might appear in English. See the list of already internationalized FMs and their status.

For Mappers

As a mapper, you want to reach an audience as wide as possible with your FM. One way is to provide alternative language versions of your map. There are two ways that can be achieved:

You can ignore the issue, and hope that somebody else modifies your FM and provides a translation. The past shows that the changes that someone does this are slim, tho. Only a handful of FMs were every translated, most of them only into one other language. That is because modifying and translating an FM are quite a lot of work, and translators would rather just translate things than to muck around with your PK4 file.
You can (from v1.07 on) build in support for translation from the beginning, and make the work for translators easier.

The second point can achieved in two ways:

You build the FM with hard-coded strings like 'Silver Key', and then use I18N.pl to transform the FM into a package that can have external dictionaries. That might be easier to work with in DR (as DR itself has no support for dictionaries yet), but the script might not be 100% correct and you may need to manually help it out. However, this is a step that needs to be done only once.
You build your FM with string templates like "#str_23456" (use the range 20000 … 89999) instead of hard-coded names, and supply an external dictionary with these strings.

In both cases, translating your mission becomes super easy, there is only one file with all the strings in place, and this can be easily edited.

FM specific dictionaries

Please see the main article about about translating FMs for details.

Choosing Names

The TDM core dictionary contains a lot of strings, both for inventory items like keys, for locations, monster names, jobs, titles and descriptions. If possible, use one of these names. That has the advantage that the translation is already provided in many languages. For instance, instead of "a silvery key", use if possible "Silver Key" (if you plan on converting your mission with I18N.pl) or "#str_10000" (if you are doing it manually).

Browse through strings/all.lang to see which strings are available.

Inventory Item Names with multiple lines

If you have an inventory item name which is very long, you can insert \n to split it up into two (but not more!) lines:

"#str_20000" "This is a very very\nlong item name."

You should try in the game to see how it looks, so that both lines are roughly the same length.

Support in DR

Darkradiant currently does not support the dictionaries, so strings will look like "#str_07118" instead of the "OK". This is tracked here: https://bugs.thedarkmod.com/view.php?id=2845

For Developers

CVAR tdm_lang

tdm_lang was only used before TDM 2.0 and is obsolete.

CVAR sys_lang

The CVAR sys_lang contains the current language as English lower-case string, e.g. "english", "german", "portuguese" etc. This variable is controlled by the GUI via a call back into the SDK code, which switches the language underneath, and also reloads the GUI.

If you change the CVAR manually (via the console f.i.), you need to restart TDM so it works!

Code

The main translation object is of the class CI18N (which is short for "Class Internationalization") and lives in framework/I18N.cpp and framework/I18N.h (in v1.07, it was in Darkmod/I18N.cpp and Darkmod/I18N.h).

Charset

The TDM (old D3) code that handles the GUI bitmap font can only load a specific range of bytes as characters. To get the most out of the available entries, an special charset (based on ISO 8859-1 with modifications) is used. The details can be found on this page.

New Strings

If you need to add a new string, follow these steps:

search for a fitting, free ID in strings/all.lang.
Add your string there. If possible, add any translation you can, or ask the translators to translate it for you.
re-generate the language files with perl devel/gen_lang.pl

Note that if the translations for the new strings (or any other string) are missing, the script will substitute the English variant - so you should always have at least the English variant.

The other translations are optional, but if they are added later, you need to re-run perl devel/gen_lang.pl! See also the next point:

Always edit all.lang with a UTF-8 editor and keep it in UTF-8.

Never edit the other language files manually, they will be regenerated and your edits will be lost.

Make sure you use the latest versions of all.lang and gen_lang.pl. If you don't have SVN access, work with someone who does and can provide you with the latest versions. When your changes are ready to be checked in, make sure that not only the new *.lang files get checked in, but also the new gen_lang.pl

Adding a new i18n language

The TDM i18n system works by having the translated strings given individual numbers. These numbers are included from separate language files at compile time, and then replaced with the language strings at runtime.

Relevant files

all.lang - all the languages
devel/gen_lang.pl - perl script for breaking up the all.lang master file
(language).lang - the individual language files, eg english.lang or german.lang
(language).map - a character mapping file, eg german.map

Very basic overview of the process

Edit the file "all.lang". Just copy the english language section to the end of the file and rename the new section. Make sure you use a Unicode editor and all.lang stays in UTF-8!
Add the correct entry for the language name in strings/all.lang after the names in the range of 2460 .. 2500 (unless it already exists). The language name is to be expected in the local name, e.g. it is "Deutsch" in the english translation, not "German". If the language name contains characters that cannot be represented in the current charset of the language, please use either the special characters from ISO-8859-1, or an translitertion like "Bulgarian"
if possible, add the same entries in languages that do not use ISO-8859-1
Run the perl script devel/gen_lang.pl To do this, you need to have commit access to the developer SVN. This will "break up" the all.lang file into separate files for each translated language.
If nec, create a new ".map" file by copying one of the existing .map files in "strings/".
Add the new ".lang" file (and possible ".map" file) to SVN.

The new strings should now be accessible at the next start of TDM.

Missing Translations

The following will generate files called "missing_language.txt" in the darkmod directory:

perl devel/gen_lang.pl --missing

Note that the script can no differentiate between a string that was not translated, and a string that is the same even after translation. An example is "Normal" in English, and German. This will be listed as missing string in missing_german.txt.

To avoid that, add the following comment:

"#str_03005"    "Normal"                        // stays the same

to each language variant where the string is unchanged after translation.

Since it can be a bother to ask translators for a new string translation every other day, the current solution is to gather strings for a while, then regenerate the missing_xyz.txt files, and then ask the translators for a translation.

Relocating String IDs

Note: Relocation of string IDs can only be done either before release, or when it is clear that this string is not used by any FM, otherwise you break FMs!

perl devel/rename_string.pl FIRSTOLDID FRISTNEWID COUNT

Example:

perl devel/rename_string.pl 2236 2298 2

GUI

In v1.07 there was a bug in choiceDef inside the GUI, which did not allow other characters than a-z. To work around this choiceDef bug, two new main menu commands were added, called "initChoice" and "stepChoice". However, for v1.08, the bug is fixed in the main code (due to the source code now being available), so the work-arounds were removed.

Known bugs

At the moment only a few things are still not working:

Not all fonts contain accented characters yet.
DarkRadiant does not support the #str_01234 strings yet.
Doom3 does not render/support all characters (f.i. 0xFF in Russian, 0xF7 .. 0x9F in Western). To work around this, the characters are remapped (moved) during load. See I18N - Charset for details on this.