Checksum tools includes basic localization support.
Localization is performed by calling the Localize function with a hash of translations of a string for all supported languages. These hashes have the following structure:
{ language_code => translation, ... }
An example might be:
{ 'en' => 'Hi!', 'de' => 'Hallo!', 'es' => '¡Hola!',
'fr' => 'Salut!', 'nl' => 'Hai!', 'sv' => 'Hej!' }
If the hash reference shown above were stored in a variable named $greeting then an appropriately-translated string could be displayed by calling:
print Localize( $greeting ) . "\n";
Most of these hashes are stored in one of three global hashes: %AVAILABLE_COMMANDS, %AVAILABLE_OPTIONS, and %STRINGS (the first two also contain non-localization information). These are initialized by the GetAvailableCommands, GetAvailableOptions, and GetStrings functions, respectively. The Usage function also contains additional strings that need to be localized; look for calls similar to:
print Localize( { 'en' => <<EOF
English translation
over
multiple lines
EOF
} ); # END localize
To add a French translation for the above, one might change that statement to read:
print Localize( {
'en' => <<EOF
English translation
over
multiple lines
EOF
, 'fr' => <<EOF
Traduction française
sur
plusieurs lignes
EOF
} ); # END localize
Note that here documents are used for the multi-line strings. One quirk is that one may need to place two newlines at the end of these strings (as shown above) since Localize will always strip the final one.
The hashes in the GetAvailableCommands, GetAvailableOptions, and GetStrings functions are similar. For example, to add a translation for the help command one might change (in GetAvailableCommands):
'usage' => { 'en' => <<EOF
-h | --help Show brief usage message
EOF
},
to:
'usage' => {
'en' => <<EOF
-h | --help Show brief usage message
EOF
, 'nl' => <<EOF
-h | --help Toon kort gebruiksbericht
EOF
},
Note that only the message has been changed, and not the option names. This is done to ensure compatibility of commandline options for all languages.
When choosing which string to use, Localize first determines the current language from (in order of preference) the LANG, LANGUAGE, LC_ALL, and KDE_LANG environment variables; if none of these is set then the string 'en' (for English) is used. Next, an appropriate translated string is sought by looking for the full locale, then the locale less any character encoding, then also less any region code. If no match is found then 'en' (for English) is used. For example, if LANG is fr_CA.UTF-8 then first a translation for fr_CA.UTF-8 is sought, then fr_CA, then fr, and lastly en.
All strings should be formatted to 80 characters or less in width. This is simply to make things look nice on-screen and is not a technical limitation.
The current system may lead to a lot of bloat if many translations are added but it's a start...