~4Dgifts/toolbox/src/exampleCode/i18n README
This directory contains example source code dealing with
internationalization and localization as well as draft
copies of the IRIX Internationalization chapter (Primer.ps)
for what will eventually be the Internationalization chapter
of the "IRIX Systems Programming Guide" as well as a Nutshell
(Nutshell.ps) guide as an attempt to supply a high-level
description of the issues involved in internationalizing
software.
To successfully run the programs in this directory, you
need to be running an O.S level of 5.* or higher, and
need to have Motif 1.2 running and X11 R5.
Some examples will also require installed images of
intl.sw. See below for more details.
See the Frequently Asked Questions i18n file located at
~4Dgifts/toolbox/FAQs/netfaqs/i18n-faq.
xpg3_wchar :
this is a sample code which shows two things
1) use of wide characters for string manipulation
2) XPG/3 way of doing message catalogs (the other
interface of cataloging is MNLS).
the sample code is supplied with Japanese and German
catalogs under the directory jp and de respectively.
to run the Japanese version :
setenv XMODIFIERS @im=_XWNMO
run xwnmo
run iwsh or exterm
setenv NLSPATH jp
setenv LANG ja_JP.EUC
xpg3_wchar japanese_input_string
and for German
setenv NLSPATH de
setenv LANG de
xpg3_wchar german_input_string
do man catopen for full details on the location of catalogs.
do man gencat for help on generating message catalogs.
if the NLSPATH environment variable is not set, it
will run in the default english mode.
the program takes the input string and outputs two strings
which are made of even and odd alphabets from the original
string (not useful in real life, but shows how to manipulate
wide character strings).
mnls_wchar :
this is similar to xpg3_wchar except that it uses mnls
way of cataloging.
the sample code is supplied with Japanese and German
catalogs under the directory jp and de respectively.
you will have to move (or copy) the file mnls_wchar.cat from these
respective directories to
/usr/lib/locale/de/LC_MESSAGES for German
and to /usr/lib/locale/ja_JP.EUC/LC_MESSAGES for Japanese
to run the Japanese version :
setenv LANG ja_JP.EUC
mnls_wchar japanese_input_string
and for German
setenv LANG de
mnls_wchar german_input_string
=======================================================================
Message Catalogs for Internationalization :
-------------------------------------------
Apart from the text which appears on the user interface
widgets, an internationalized application must display
other texts like prompts, error messages in the native
language. Message cataloging facility is commonly used for this
type of intertionalization of strings.
Message catalogs are compiled databases of strings. The application
extracts all the strings from a locale dependent message catalog.
Although these strings are commonly used for communication
in the native language, they can be used for any purpose.
On Irix, there are two different interfaces to message catalogs:
MNLS and XPG/3. MNLS is AT&T's Multi National Language Support and
XPG/3 is the X/Open Portability Guide, Issue 3, by X/Open, Ltd.
Developers working on base system utilities will probably be
using MNLS, whereas developers working on independent projects will
be using XPG/3. In order to maximize portability, developers are
encouraged to use XPG/3 if given a choice.
Catalog generation is exclusively the task of localizers. But the
developer must know the contents of the catalog in order to specify
the number of a message to be used in the application. So in order
to inform the localizer which strings to translate and how they
should create a catalog, the application developer should provide
a master catalog for the developer's locale.
MNLS message catalogs
---------------------
MNLS message catalogs are merely list of strings in a file, separated
by new-lines. Strings are referenced by line number in the original
source file. Adding strings to an existing catalog should only be done
at the end of the file. Otherwise, applications already using this catalog
will end up accessing different strings. The three functions
pfmt, gettxt, and printf are sufficient for most applications in
this interface. fmtmsg() a comprehensive formatter is also available.
MNLS catalogs do not need to be specifically opened. The catalog can be
set explicitly once, or it can be specified everytime a string is
needed.
When a file of message strings is ready to be compiled, run mkmsgs()
and save the file in /usr/lib/locale/<locale>/LC_MESSAGES
(for example <locale> for german could be de).
following is an example German mnls message strings which needs to
be compiled using mkmsgs
Benutzung: %s Zeichenkette\n
Kein freier Speicher mehr\n
Fehler beim Erzeugen eines 'wide characters'\n
Hier ist die Ausgabe :
mnls_wchar is an example source which uses MNLS interface of catalog.
XPG/3 message catalogs
----------------------
XPG/3 message catalogs contains sets of numbered messages. The
three functions catopen, catgets, and catclose, provide a simple
mechanism for retrieving numbered strings from a plain text file.
This interface requires that a catalog be opened before it is read.
The XPG/3 catalogs are located using the environment variable NLSPATH.
The man page on catopen() gives full details on the path.
After creating the file with message strings, you need to compile them
into a binary form using gencat() and save the file in the location
where catopen() can find it (NLSPATH mentioned above).
following is an example German xpg3 message strings which needs to
be compiled using gencat
$set 1
1 Benutzung: %s Zeichenkette\n
2 Kein freier Speicher mehr\n
3 Fehler beim Erzeugen eines 'wide characters'\n
4 Hier ist die Ausgabe :
xpg3_wchar is an example source which uses XPG/3 interface of catalog.
Wide Characters
---------------
The example source also demonstrates another very useful concept
in internationalized applications: the wide character.
A wide character is a data object of type wchar_t, which holds the
system's largest numerical code for a character. On Irix 5.* and
beyond, the sizeof(wchar_t) is 4.
Since the strings made of wide characters have all the characters of
the same size, the string can be treated as an array, and the application
can simply index into the array in order to manipulate it.
Wide characters are not as space efficient as multibyte strings and
applications which do not need to perform string editing may never
use wchars.
-----------------
Yusuf Attarwala
SGI - Applications.
x 3-3342
Source
Documentation
Reference
Subdirectories
OR, ...