Subversion Repositories SvarDOS

              SVARLANG.LIB - THE SVARDOS TRANSLATION C LIBRARY

                   Copyright (C) 2021-2024 Mateusz Viste



SvarLANG is a library and toolset for enabling SvarDOS applications to easily
support multiple languages. It is part of the SvarDOS project.

Homepage: http://svardos.org/svarlang/


### PREPARING TRANSLATION FILES ###############################################

The translation files must be CATS-style text files in the usual format:

1.1:Hello, World!
1.2:Help screen
2.0:Type /? for more options

The files must be named as EN.TXT, DE.TXT, FR.TXT, etc. Then, they must be
converted into SvarLANG's binary format using the TLUMACZ tool:

tlumacz en fr pl (...)

The first language provided in the command line is the reference language and
is used both as the default (embedded in the application) language, as well as
to substitute messages missing in other languages.

TLUMACZ computes two files:

 * OUT.LNG   - the binary file that contains all translations
 * DEFLANG.C - the default translations that will be embedded into the program

Then, DEFLANG.C must be compiled and linked to your program along with
SVARLNGx.LIB. From there you will be able to use SvarLANG calls, like this
very basic example:

  svarlang_load("myprogram.lng", "pl");  /* load PL lang from myprogram.lng */
  puts(svarlang_str(2, 0));              /* display the string with id 2.0 */

A more practical, real-world example would probably be this one:

  svarlang_autoload_exepath(argv[0], getenv("LANG"));
  puts(svarlang_str(2, 0));

Read svarlang.h for more information about available functions.


### ESCAPED CHARACTERS ########################################################

Translation strings may contain some escaped characters. At this time only the
following escaped characters are supported: \e \r \n \t and \\


### DIRTY STRINGS #############################################################

In the CATS-style source translation, lines may be prefixed with a '?' sign:

?1.1:Hello, World!

Such string is used by tlumacz like any other, but it is also reported on the
command-line with a warning about the line being "dirty" (that is, requiring
to be reviewed by a translator).


### ENVIRONMENT ###############################################################

The program translation file should be named "PROGNAME.LNG", where PROGNAME
is the program's name. This file should be placed in a well-known location,
typically the program's own directory.

The %LANG% environment variable usually defines what language should be loaded,
albeit the program can just as well provide its own controls for language
selection and pass this information to svarlang_load() accordingly.


### WHY IS IT BETTER THAN CATS? ###############################################

The CATS library is heavier and slower, as it embeds a text-file parser.
Translations also take more disk space since each language is stored in a
separate file, leading to cluster waste. Finally, CATS requires default strings
to be part of the application's source code, while SvarLANG keeps all strings
in TXT files and embedds the default one inside the application in an automated
way at compile time.

There is also a licensing issue: CATS/Kitten libraries are published under the
terms of a viral, corrosive license. SvarLANG, on the other hand, is published
under a truly free, liberal MIT license.


### FILE FORMAT ###############################################################

File =
    magic         : Char[4] := "SvL\x1a" (ie. "SvL" followed with a 0x1a char)
                  ; 0x1a is an end-of-file marker that prevents TYPE garbage
    num_strings   : U16
    languages     : array[num_languages] of Language

Language =
    lang_id       : Char[2]
    len_strings   : U16 := SizeOf(strings)
    dictionary    : StringDict
    strings       : array[File.num_strings] of StringZ (possibly compressed)

StringDict =
    elements : array[File.num_strings] of DictEntry
             ; sorted by DictEntry.Id

DictEntry =
    id     : U16
    offset : U16
           ; relative to Language.strings[0]

StringZ = array[?] of Char ; zero-terminated string


NOTE 1: All numeric values are stored in x86 (little endian) order.

NOTE 2: A language with compressed strings is flagged with bit 0x8000 set in
        lang_id. The strings are then compressed with the MVCOMP algorithm,
        see https://mvcomp.sourceforge.io for more details about MVCOMP.


####################################################################### EOF ###