Changeset 1574 for lang/perl/Geography-AddressExtract-Japan/trunk/lib/Geography/AddressExtract/Japan.pm

Timestamp:

11/15/07 16:03:31 (12 months ago)

Author:

yappo

Message:

r12@haruna (orig r10): ko | 2006-06-21 23:16:40 +0900
add normalize rules
add city patterns

Files:

• lang/perl/Geography-AddressExtract-Japan/trunk/lib/Geography/AddressExtract/Japan.pm (modified) (2 diffs)

lang/perl/Geography-AddressExtract-Japan/trunk/lib/Geography/AddressExtract/Japan.pm

@@ -75 +75 @@
     my($self, $right, $opt) = @_;
 
+    # address position normalize
     if ($opt->{number} && $opt->{number} =~ /^([���������岼])/) {
         my $prefix = $1;
@@ -84 +85 @@
             $opt->{match_text} .= $append;
         }
+    }
+
+    if ($opt->{aza} && $opt->{aza} =~ /^([\p{Hiragana}\p{Katakana}])/) {
+        my $prefix = $1;
+        $opt->{aza} = '';
+        $opt->{number} = '';
+        $opt->{match_text} =~ s/^($opt->{city})\s*$prefix.+$/$1/;
+    }
+
+    if ($opt->{city} =~ /([\p{Hiragana}\p{Katakana}]+)$/) {
+        my $kana = $1;
+        $opt->{city} =~ s/$kana//;
+        $opt->{aza} = $kana;
+    }
+
+    if ($opt->{city} =~ /([\p{Hiragana}\p{Katakana}]+)$/) {
+        my $kana = $1;
+        $opt->{city} =~ s/$kana//;
+        $opt->{aza} = $kana;
+    }
+
+    if ($opt->{aza} && $opt->{aza} =~ /^[\p{Hiragana}\p{Katakana}]/ && !$opt->{number}) {
+        if ($right =~ /^[-���ݤΥ�(
+            (?:(?:(?:[�����͸����Ȭ����)?[�����͸����Ȭ�塻]+|\d+)|[a-zA-Z����)
+            (?:(?:��?)?[-���ݤΥ�?(?:(?:(?:[�����͸����Ȭ����)?[�����͸����Ȭ�塻]+|\d+)|[a-zA-Z����)��?
+            )/x) {
+            $opt->{number} = $1;
+            $opt->{match_text} .= $&;
+        }
+    }
+
+    if ($opt->{aza} && $opt->{aza} =~ /\p{Han}+([�Ƥˤ�ؤΤˤ���].*)$/) {
+        my $prefix = $1;
+        $opt->{aza} =~ s/$prefix$//;
+        $opt->{number} = '';
+        $opt->{match_text} =~ s/($opt->{aza}).+$/$1/;
     }