Changeset 3816

Timestamp:

12/30/07 19:55:27 (11 months ago)

Author:

tokuhirom

Message:

Encode-JP-Mobile: ucm version convert-pictogram branch.

Location:

lang/perl/Encode-JP-Mobile/branches/feature-convert-pictogram-ucm

Files:

• Makefile.PL (modified) (1 diff)
• lib/Encode/JP/Mobile.pm (modified) (3 diffs)
• lib/Encode/JP/Mobile/UTF8.pm (deleted)
• t/emoji_convert_utf8.t (modified) (2 diffs)
• tools/make-pictogram-ucm.pl (modified) (5 diffs)
• ucm/x-utf8-docomo-pictogram.ucm (deleted)
• ucm/x-utf8-docomo.ucm (added)
• ucm/x-utf8-kddi-pictogram.ucm (deleted)
• ucm/x-utf8-kddi.ucm (added)
• ucm/x-utf8-softbank-pictogram.ucm (deleted)
• ucm/x-utf8-softbank.ucm (added)

lang/perl/Encode-JP-Mobile/branches/feature-convert-pictogram-ucm/Makefile.PL

@@ -15 +15 @@
 my @ucms = qw( 
     sjis-imode sjis-kddi sjis-kddi-auto sjis-softbank-auto
-    utf8-docomo-pictogram utf8-kddi-pictogram utf8-softbank-pictogram
+    utf8-docomo utf8-kddi utf8-softbank
 );
 my %tables = (

lang/perl/Encode-JP-Mobile/branches/feature-convert-pictogram-ucm/lib/Encode/JP/Mobile.pm

@@ -11 +11 @@
 
 use Encode::Alias;
+# sjis
 define_alias('x-sjis-docomo' => 'x-sjis-imode');
 define_alias('x-sjis-ezweb' => 'x-sjis-kddi');
@@ -17 +18 @@
 define_alias('x-sjis-airh' => 'cp932');
 define_alias('x-sjis-vodafone-auto' => 'x-sjis-softbank-auto');
+
+# utf8
+define_alias( 'x-utf8-imode'    => 'x-utf8-docomo' );
+define_alias( 'x-utf8-ezweb'    => 'x-utf8-kddi' );
+define_alias( 'x-utf8-vodafone' => 'x-utf8-softbank' );
 
 # backward compatiblity
@@ -31 +37 @@
 use Encode::JP::Mobile::KDDIJIS;
 
-use Encode::JP::Mobile::UTF8;
+# use Encode::JP::Mobile::UTF8;
 
 sub InDoCoMoPictograms {

lang/perl/Encode-JP-Mobile/branches/feature-convert-pictogram-ucm/t/emoji_convert_utf8.t

@@ -12 +12 @@
     my ($docomo, $kddi, $softbank) = @_;
     
-    is encode('x-utf8-docomo',   decode('utf-8', $docomo)), $docomo;
-    is encode('x-utf8-kddi',     decode('utf-8', $docomo)), $kddi;
-    is encode('x-utf8-softbank', decode('utf-8', $docomo)), $softbank;
+    is encode('x-utf8-docomo',   decode('utf-8', $docomo)), $docomo, 'I => I';
+    is encode('x-utf8-kddi',     decode('utf-8', $docomo)), $kddi, 'I => E';
+    is encode('x-utf8-softbank', decode('utf-8', $docomo)), $softbank, 'I => V';
     
     is encode('x-utf8-docomo',   decode('utf-8', $kddi)), $docomo;
-    is encode('x-utf8-kddi',     decode('utf-8', $kddi)), $kddi;
+    is encode('x-utf8-kddi',     decode('utf-8', $kddi)), $kddi, 'E => E';
     is encode('x-utf8-softbank', decode('utf-8', $kddi)), $softbank;
     
-    is encode('x-utf8-docomo',   decode('utf-8', $softbank)), $docomo;
+    is encode('x-utf8-docomo',   decode('utf-8', $softbank)), $docomo, 'V => I';
     is encode('x-utf8-kddi',     decode('utf-8', $softbank)), $kddi;
     is encode('x-utf8-softbank', decode('utf-8', $softbank)), $softbank;
@@ -35 +35 @@
     
     is encode('x-utf8-docomo',   decode('utf-8', $kddi)), $docomo;
-    is encode('x-utf8-kddi',     decode('utf-8', $kddi)), $kddi;
+    is encode('x-utf8-kddi',     decode('utf-8', $kddi)), $kddi, 'E => E';
     is encode('x-utf8-softbank', decode('utf-8', $kddi)), $softbank;
     
     is encode('x-utf8-docomo',   decode('utf-8', $softbank)), $docomo;
-    is encode('x-utf8-kddi',     decode('utf-8', $softbank)), $kddi;
+    is encode('x-utf8-kddi',     decode('utf-8', $softbank)), $kddi, 'V => E';
     is encode('x-utf8-softbank', decode('utf-8', $softbank)), $softbank;
 }

lang/perl/Encode-JP-Mobile/branches/feature-convert-pictogram-ucm/tools/make-pictogram-ucm.pl

@@ -3 +3 @@
 use YAML;
 use Encode;
+use Encode::JP::Mobile;
 
 my $map = YAML::LoadFile 'dat/convert-map-utf8.yaml';
@@ -12 +13 @@
 );
 
+my $uni_range_for = {
+    docomo   => Encode::JP::Mobile::InDoCoMoPictograms(),
+    kddi     => Encode::JP::Mobile::InKDDIPictograms(),
+    softbank => Encode::JP::Mobile::InSoftBankPictograms(),
+};
+
+my $cp932_ucm = shift @ARGV or die "Usage: $0 cp932.ucm\n";
+
 for my $to (qw( docomo kddi softbank )) {
-    open my $fh, '>', "ucm/x-utf8-$to-pictogram.ucm";
+    open my $fh, '>', "ucm/x-utf8-$to.ucm" or die $!;
     print {$fh} <<HEAD
-<code_set_name> "x-utf8-$to-pictogram"
-<code_set_alias> "x-utf8-$alias{$to}-pictogram"
+<code_set_name> "x-utf8-$to"
+<code_set_alias> "x-utf8-$alias{$to}"
 <mb_cur_min> 1
 <mb_cur_max> 2
@@ -24 +33 @@
     ;
 
+    # basic map
+    print {$fh} unicode_ucm($cp932_ucm);
+
+    # convert map
     for my $from (qw( docomo kddi softbank )) {
         next if $from eq $to;
 
-        print {$fh} "# pictogram convert map ($from => $to)\n";
+        print {$fh} "\n\n# pictogram convert map ($from => $to)\n";
 
         for my $srcuni (sort keys %{$map->{$from}}) {
@@ -33 +46 @@
             printf {$fh} "<U%s> %s |1 # %s\n", $srcuni, unihex2utf8hex($dstuni), comment_for($from);
         }
+    }
+
+    if ($to eq 'kddi') {
+        # ura-kddi
+        range_each($to, sub {
+            my $unicode = shift;
+            my $sjiscode = hex unpack "H*", encode('x-sjis-kddi', chr $unicode);
+            my $unihex = sprintf '%X', $sjiscode - 0x0700;
+            print {$fh} sprintf "<U%s> %s |1 # %s\n", $unihex, unihex2utf8hex($unihex), "UraKDDI pictogram";
+        });
+    } else {
+        range_each($to, sub {
+            my $unicode = shift;
+            my $unihex = sprintf '%X', $unicode;
+            print {$fh} sprintf "<U%s> %s |1 # %s\n", $unihex, unihex2utf8hex($unihex), "$to pictogram";
+        });
     }
 
@@ -55 +84 @@
 }
 
+sub unicode_ucm {
+    my $cp932_ucm = shift;
+    my $res = '';
+    open my $fh, '<', $cp932_ucm or die "$!: $cp932_ucm";
+    while (my $line = <$fh>) {
+        if ($line =~ /^<U(.{4})> \S+ \|0 # (.+)$/) {
+            my ($unihex, $comment) = ($1, $2);
+
+            # for FallBack.
+            next if $comment eq 'PRIVATE USE AREA';
+
+            $res .= sprintf "<U%s> %s |1 # %s\n", $unihex, unihex2utf8hex($unihex), $comment;
+        }
+    }
+    $res;
+}
+
+sub range_each {
+    my ($carrier, $code) = @_;
+
+    my $map = $uni_range_for->{$carrier};
+    for my $range (split /\n/, $map) {
+        my ($min, $max) = map { hex $_ } split /\t/, $range;
+        my $i = $min;
+        while ($i <= $max) {
+            $code->( $i );
+            $i++;
+        }
+    }
+}
+