Changeset 21522

Timestamp:

10/17/08 20:28:41 (5 years ago)

Author:

tokuhirom

Message:

implement :)

Location:

lang/perl/Lingua-JA-Regular-Unicode/trunk

Files:

• lib/Lingua/JA/Regular/Unicode.pm (modified) (4 diffs)
• t/01_alnum_z2h.t (added)
• t/02_hira2kata.t (added)
• t/03_space_z2h.t (added)
• tools (added)
• tools/createmap.pl (added)

lang/perl/Lingua-JA-Regular-Unicode/trunk/lib/Lingua/JA/Regular/Unicode.pm

@@ -6 +6 @@
 use Exporter 'import';
 
-our @EXPORT = qw/ hira2kata alnum_z2h space_z2h /;
+our @EXPORT = qw/ hiragana2katakana alnum_z2h space_z2h /;
+
+# regexp is generated by tools/createmap.pl
+
+sub alnum_z2h {
+    local $_ = shift;
+    tr/\x{FF07}\x{FF3B}\x{FF56}\x{FF4F}\x{FF0C}\x{FF5E}\x{FF14}\x{FF43}\x{FF19}\x{FF26}\x{FF55}\x{FF3F}\x{FF2D}\x{FF27}\x{FF34}\x{FF37}\x{FF30}\x{FF51}\x{FFE3}\x{FF60}\x{FF36}\x{FF49}\x{FF29}\x{FF52}\x{FF1A}\x{FF3A}\x{FF38}\x{FF3D}\x{FF4C}\x{FF1E}\x{FF5D}\x{FFE6}\x{FF01}\x{FF5C}\x{FF58}\x{FF16}\x{FF05}\x{FF54}\x{FF3E}\x{FF18}\x{FF45}\x{FF24}\x{FF2B}\x{FF15}\x{FF4A}\x{FF0D}\x{FFE0}\x{FF48}\x{FF11}\x{FF5B}\x{FF35}\x{FF12}\x{FF2E}\x{FF28}\x{FF06}\x{FF10}\x{FF03}\x{FF2F}\x{FF4E}\x{FFE2}\x{FF20}\x{FF5F}\x{FF46}\x{FF13}\x{FF31}\x{FF41}\x{FF50}\x{FF2A}\x{FFE5}\x{FF1F}\x{FF21}\x{FF57}\x{FF3C}\x{FF04}\x{FF02}\x{FF22}\x{FF4D}\x{FF23}\x{FF17}\x{FF1B}\x{FFE4}\x{FF1D}\x{FF59}\x{FF0B}\x{FF47}\x{FF39}\x{FF32}\x{FF42}\x{FF2C}\x{FF4B}\x{FF09}\x{FF33}\x{FF40}\x{FF25}\x{FF08}\x{FFE1}\x{FF0A}\x{FF0E}\x{FF5A}\x{FF53}\x{FF0F}\x{FF1C}\x{FF44}/\x{0027}\x{005B}\x{0076}\x{006F}\x{002C}\x{007E}\x{0034}\x{0063}\x{0039}\x{0046}\x{0075}\x{005F}\x{004D}\x{0047}\x{0054}\x{0057}\x{0050}\x{0071}\x{00AF}\x{2986}\x{0056}\x{0069}\x{0049}\x{0072}\x{003A}\x{005A}\x{0058}\x{005D}\x{006C}\x{003E}\x{007D}\x{20A9}\x{0021}\x{007C}\x{0078}\x{0036}\x{0025}\x{0074}\x{005E}\x{0038}\x{0065}\x{0044}\x{004B}\x{0035}\x{006A}\x{002D}\x{00A2}\x{0068}\x{0031}\x{007B}\x{0055}\x{0032}\x{004E}\x{0048}\x{0026}\x{0030}\x{0023}\x{004F}\x{006E}\x{00AC}\x{0040}\x{2985}\x{0066}\x{0033}\x{0051}\x{0061}\x{0070}\x{004A}\x{00A5}\x{003F}\x{0041}\x{0077}\x{005C}\x{0024}\x{0022}\x{0042}\x{006D}\x{0043}\x{0037}\x{003B}\x{00A6}\x{003D}\x{0079}\x{002B}\x{0067}\x{0059}\x{0052}\x{0062}\x{004C}\x{006B}\x{0029}\x{0053}\x{0060}\x{0045}\x{0028}\x{00A3}\x{002A}\x{002E}\x{007A}\x{0073}\x{002F}\x{003C}\x{0064}/;
+    $_;
+}
+
+sub hiragana2katakana {
+    local $_ = shift;
+    tr/\x{3077}\x{3094}\x{306B}\x{3080}\x{3066}\x{3044}\x{3067}\x{3079}\x{309E}\x{3090}\x{3075}\x{3068}\x{304A}\x{308A}\x{3052}\x{305D}\x{3065}\x{3088}\x{306F}\x{3064}\x{3056}\x{3057}\x{3083}\x{306E}\x{3063}\x{306D}\x{3072}\x{3043}\x{305F}\x{3087}\x{3051}\x{307E}\x{308C}\x{3073}\x{3084}\x{304C}\x{307D}\x{306C}\x{307A}\x{304F}\x{305E}\x{3071}\x{3054}\x{3092}\x{3078}\x{305A}\x{304B}\x{3074}\x{3085}\x{308E}\x{3042}\x{304D}\x{3096}\x{3047}\x{3069}\x{3060}\x{308D}\x{3082}\x{3048}\x{308F}\x{3093}\x{3076}\x{305C}\x{3081}\x{306A}\x{3061}\x{3070}\x{3062}\x{308B}\x{3059}\x{3041}\x{3095}\x{307C}\x{3089}\x{3049}\x{309D}\x{3050}\x{307B}\x{3055}\x{3091}\x{304E}\x{307F}\x{305B}\x{3058}\x{3053}\x{3045}\x{3086}\x{3046}/\x{30D7}\x{30F4}\x{30CB}\x{30E0}\x{30C6}\x{30A4}\x{30C7}\x{30D9}\x{30FE}\x{30F0}\x{30D5}\x{30C8}\x{30AA}\x{30EA}\x{30B2}\x{30BD}\x{30C5}\x{30E8}\x{30CF}\x{30C4}\x{30B6}\x{30B7}\x{30E3}\x{30CE}\x{30C3}\x{30CD}\x{30D2}\x{30A3}\x{30BF}\x{30E7}\x{30B1}\x{30DE}\x{30EC}\x{30D3}\x{30E4}\x{30AC}\x{30DD}\x{30CC}\x{30DA}\x{30AF}\x{30BE}\x{30D1}\x{30B4}\x{30F2}\x{30D8}\x{30BA}\x{30AB}\x{30D4}\x{30E5}\x{30EE}\x{30A2}\x{30AD}\x{30F6}\x{30A7}\x{30C9}\x{30C0}\x{30ED}\x{30E2}\x{30A8}\x{30EF}\x{30F3}\x{30D6}\x{30BC}\x{30E1}\x{30CA}\x{30C1}\x{30D0}\x{30C2}\x{30EB}\x{30B9}\x{30A1}\x{30F5}\x{30DC}\x{30E9}\x{30A9}\x{30FD}\x{30B0}\x{30DB}\x{30B5}\x{30F1}\x{30AE}\x{30DF}\x{30BB}\x{30B8}\x{30B3}\x{30A5}\x{30E6}\x{30A6}/;
+    $_;
+}
+
+sub space_z2h {
+    local $_ = shift;
+    tr/\x{3000}/\x{0020}/; # convert \x{IDEOGRAPHIC SPACE} to \x{SPACE}
+    $_;
+}
 
 1;
@@ -15 +35 @@
 =head1 NAME
 
-Lingua::JA::Regular::Unicode -
+Lingua::JA::Regular::Unicode - convert japanese chars.
 
 =head1 SYNOPSIS
@@ -21 +41 @@
     use Lingua::JA::Regular::Unicode qw/regular/;
     alnum_z2h("Ａ１");                                        # => "A1"
-    hira2kata("ほげ");                                        # => "ホゲ"
-    space_z2h("\x{FULL WIDTH WHITESPACE}");                   # => \x{HALF WIDTH WHITESPACE}
+    hiragana2katakana("ほげ");                                # => "ホゲ"
+    space_z2h("\x{0300}");                                    # => 半角スペース
 
 =head1 DESCRIPTION
 
-Lingua::JA::Regular::Unicode is
+Lingua::JA::Regular::Unicode is regularizer.
+
+=over 4
+
+=item alnum_z2h
+
+convert alphabet and numbers ZENKAKU to HANKAKU.
+
+=item space_z2h
+
+convert spaces ZENKAKU to HANKAKU.
+
+=item hiragana2katakana
+
+convert HIRAGANA to KATAKANA.
+
+=back
 
 =head1 AUTHOR
@@ -34 +70 @@
 =head1 SEE ALSO
 
+L<Lingua::JA::Regular>
+
 =head1 LICENSE