Changeset 7900 for lang/perl/Acme-Shukugawa-Atom

Timestamp:

03/13/08 19:51:16 (5 years ago)

Author:

daisuke

Message:

lang/perl/Acme-Shukugawa-Atom; Add OO interface that allows passing custom word lists

Location:

lang/perl/Acme-Shukugawa-Atom/trunk

Files:

• Changes (modified) (1 diff)
• Makefile.PL (modified) (1 diff)
• lib/Acme/Shukugawa/Atom.pm (modified) (7 diffs)

lang/perl/Acme-Shukugawa-Atom/trunk/Changes

@@ -3 +3 @@
 
 0.00004 -
+  * Add OO interface
   * Work with new tests (otsune)
   * Add t/02_shisu.t

lang/perl/Acme-Shukugawa-Atom/trunk/Makefile.PL

@@ -4 +4 @@
 name('Acme-Shukugawa-Atom');
 all_from('lib/Acme/Shukugawa/Atom.pm');
+requires('Class::Accessor::Fast');
 requires('Text::MeCab', '0.20007');
 requires('Encode');

lang/perl/Acme-Shukugawa-Atom/trunk/lib/Acme/Shukugawa/Atom.pm

@@ -4 +4 @@
 use strict;
 use warnings;
+use base qw(Class::Accessor::Fast);
 use utf8;
 use Encode qw(decode_utf8);
@@ -10 +11 @@
 our $VERSION = '0.00003';
 
-sub translate
-{
-    my $self   = shift;
-    my $string = decode_utf8(shift);
-
-    $self->preprocess(\$string);
-    $self->runthrough(\$string);
-    $self->postprocess(\$string);
-
-    return $string;
-}
+__PACKAGE__->mk_accessors($_) for qw(custom_words);
 
 # Special case handling -- this could be optimized further
 # put it in a sharefile later
-my (@SPECIAL, $EXCEPTION, $RE_SMALL, $RE_SYLLABLE, $RE_NBAR);
+our (@DEFAULT_WORDS, $RE_EXCEPTION, $RE_SMALL, $RE_SYLLABLE, $RE_NBAR);
 BEGIN
 {
@@ -30 +21 @@
     $RE_SYLLABLE = decode_utf8("(?:.$RE_SMALL?)");
     $RE_NBAR     = decode_utf8("^ンー");
-    @SPECIAL = (
+    @DEFAULT_WORDS = (
         '急ぎで|急いでる?' => '巻きで',
         '小飼弾|(?i)dankogai|(?i)kogaidan' => 'ガイダンコ',
@@ -50 +41 @@
         '片付け|かたづけ' => 'わらう'
     );
-    $EXCEPTION = decode_utf8(join("|",
-        map { $SPECIAL[$_ * 2 + 1] } (0..$#SPECIAL/2) ));
+}
+
+sub _create_exception_re
+{
+    my $self = shift;
+    my $custom = $self->custom_words;
+
+    return decode_utf8(join("|",
+        map { $custom->[$_ * 2 + 1] } (0..(scalar(@$custom) - 1)/2) ));
+}
+
+sub translate
+{
+    my $self   = shift;
+    my $string = decode_utf8(shift);
+
+    if (! ref $self) {
+        $self = $self->new({ custom_words => \@DEFAULT_WORDS, @_ });
+    }
+
+    # Create local RE_EXCEPTION
+    local $RE_EXCEPTION = $self->_create_exception_re;
+
+    $self->preprocess(\$string);
+    $self->runthrough(\$string);
+    $self->postprocess(\$string);
+
+    return $string;
 }
 
@@ -57 +74 @@
 {
     my ($self, $strref) = @_;
-
-    for(0..$#SPECIAL/2) {
-        my $pattern = $SPECIAL[$_ * 2];
-        my $replace = $SPECIAL[$_ * 2 + 1];
+    my $custom = $self->custom_words;
+
+    for(0..(scalar(@$custom) - 1)/2) {
+        my $pattern = $custom->[$_ * 2];
+        my $replace = $custom->[$_ * 2 + 1];
         $$strref =~ s/$pattern/$replace/g;
     }
@@ -75 +93 @@
     my $ret = '';
 
-    foreach my $text (split(/($EXCEPTION)/, $$strref)) {
-        if ($text =~ /$EXCEPTION/) {
+    foreach my $text (split(/($RE_EXCEPTION)/, $$strref)) {
+        if ($text =~ /$RE_EXCEPTION/) {
             $ret .= $text;
             next;
@@ -222 +240 @@
   my $newstring = Acme::Shukugawa::Atom->translate($string);
 
+  my $atom = Acme::Shukugawa::Atom->new(
+    # The default values are stored in @Acme::Shukugawa::Atom::DEFAULT_WORDS
+    custom_words => [
+      'regexp1' => 'replacement1'
+      'regexp2' => 'replacement2'
+      'regexp3' => 'replacement3'
+      'regexp4' => 'replacement4'
+      ....
+    ]
+  );
+  my $newstring = $atom->translate($string);
+
+  # shorter way
+  my $newstring = Acme::Shukugawa::Atom->translate($string,
+    custom_words => [ ... ]
+  );
+
 =head1 DESCRIPTION