<sect><heading>user-ja  --  two sets of messages in ASCII and native codeset in the same language</heading>

<P>
The author of this section is Tomohiro KUBOTA
(<email>kubota@debian.or.jp</email>).
</P>


<sect1><heading>Introduction</heading>

<P>
<package>user-ja</package> is a Debian-specific software which establishes basic
settings for Japanese-speaking beginners.
<package>user-ja</package> does not automatically establishes the settings.
A user who needs Japanese environment has to invoke <prgn>user-ja-conf</prgn>.
</P>

<P>
Since <prgn>user-ja-conf</prgn> is a software to establish Japanese
environment, the environment where user-ja runs may
be poor Japanese environment.  For example, <prgn>user-ja-conf</prgn>
must not assume that Japanese character can be displayed.
However, Japanese character should be used in environments
where it is possible.
</P>

<P>
<package>user-ja</package> is a simple example which switches two sets of 
messages, one is written using ASCII characters and the 
other Japanese characters.  Note that both of them are 
written in Japanese language.  This is beyond what gettext 
can do.
</P>

<P>
Though <package>user-ja</package> is a Japanese-specific software, this problem
of ability to display non-ASCII character is common to non-ASCII
languages.
</P>


<sect1><heading>Strategy</heading>

<P>
The following environments can display Japanese characters:
<prgn>kon</prgn> (Kanji Console), <prgn>kterm</prgn>, and 
<prgn>krxvt</prgn> (in <package>rxvt-ml</package> package).
And more, telnet softwares for Windows and so on may be able 
to display Japanese characters.
</P>

<P>
At first, <prgn>user-ja-conf</prgn> detects the environment.  If it can
display Japanese characters, go ahead.  If not, try to establish 
a new environment and invoke itself in it.
If detection is failed, display Japanese characters and ask
the user whether he/she can read it.
</P>


<sect1><heading>Implementation</heading>

<P>
<prgn>user-ja-conf</prgn> is a perl script.  Here shows a function which
check whether Japanese native characters can be displayed or not
and try to establish an environment where native characters
can be displayed, if not.
<example>
sub isNC($$)
{
        my ($KANJI, $TTY, $TERM, $DISPLAY, $WHICH);
        $TTY = `/usr/bin/tty`;
        $TERM = $ENV{TERM};
        $DISPLAY = $ENV{DISPLAY};
        $WHICH = '/usr/bin/which';
        $THIS = $_[0];
        $OPT = $_[1];

        if ($TERM eq 'kon' || $TERM eq 'kterm') {
                $KANJI=1;
        } elsif ($DISPLAY ne '' &amp;&amp; system("$WHICH kterm &gt;/dev/null")==0) {
                exec("kterm -km euc -e $THIS $OPT");
        } elsif ($DISPLAY ne '' &amp;&amp; system("$WHICH krxvt &gt;/dev/null")==0) {
                exec("krxvt -km euc -e $THIS $OPT");
        } else {
                print STDERR &amp;sourceset2displayset(
                  "Japanese sentence in Japanese characters 'Can you read this sentence?'\n");
                print STDERR 
                  "Japanese sentence in ASCII characters 'Can you read the above sentence written in Kanji? [y/N] ";
                $a = &lt;&gt;;
                if ($a =~ /y|Y/) {
                        $KANJI=1;
                } elsif ($TTY =~ m#/dev/tty[0-9]+#) {
                        print STDERR
                          "Japanese sentence in ASCII characters 'Shall I invoke \'KON\'? [Y/n] ";
                        $a = &lt;&gt;;
                        exec("kon -e $THIS $OPT") if ($a !~ /n|N/);
                        $KANJI=0;
                } else {
                        $KANJI=0;
                }
        }
        $KANJI;
}
</example>
</P>

<P>
<tt>&amp;sourceset2displayset($)</tt> is a function to convert a string from 
codeset for source code into codeset for display.
This is needed because codeset for program source (in this case, 
perl script) and dotfiles may be different.
<footnote>
There are three popular codesets for Japanese --- ISO-2022-JP, EUC-JP,
and SHIFT-JIS.  EUC-JP should be used for perl source code because 
all non-ASCII characters in EUC-JP do not have values in 0x21 - 0x7e.
However, ISO-2022-JP is the safest codeset to display because EUC-JP
and SHIFT-JIS have to be used exclusively.  However, ISO-2022-JP is the
most difficult codeset to implement and there may be a terminal 
environment which does not understand ISO-2022-JP (for example, Minicom).
On the other hand, dotfiles may be written in any codesets, according
to one's favorite and purpose.
</footnote>
</P>

<P>
The following function is prepared to display messages in
appropriate codeset.  Don't care '<tt>Lang::</tt>' package.
<example>
sub disp ($$) {
        if ($NC) {print STDERR &amp;Lang::sourceset2displayset($_[1]);}
        else {print STDERR $_[0];}
}
</example>
</P>

<P>
This is an example how the <TT>disp</TT> function is used.
<example>
sub disp_finish()
{
        &amp;Sub::disp(&lt;&lt;EOF1,&lt;&lt;EOF2);

[Enter] key WO OSUTO KONO user-ja-conf HA SYUURYOU SHIMASU.
EOF1

Japanese sentence in Japanese characters 'Push [Enter] key to finish.'
EOF2
}
</example>
Here the sentence '[Enter] key WO OSUTO...' is the Latin alphabet expression
of Japanese.
</P>

<P>
Thus almost all messages are duplicated using <tt>disp</tt> function.
</P>