Changeset 4346 for lang/perl/Acme-Actors-JA

Timestamp:

01/10/08 12:49:58 (11 months ago)

Author:

daisuke

Message:

lang/perl/Text-MeCab?; ad ability to scrape for yomi

Location:

lang/perl/Acme-Actors-JA/trunk

Files:

• eg (added)
• eg/scrape_details.pl (added)
• lib/Acme/Actors/JA/Scraper.pm (modified) (2 diffs)

lang/perl/Acme-Actors-JA/trunk/lib/Acme/Actors/JA/Scraper.pm

@@ -25 +25 @@
 sub scrape_actors_jp
 {
+    my $class = shift;
     my $uri = URI->new("http://ja.wikipedia.org/wiki/%E6%97%A5%E6%9C%AC%E3%81%AE%E7%94%B7%E5%84%AA%E4%B8%80%E8%A6%A7");
 
     my $scraper = scraper {
-        process "#bodyContent table li>a", 'names[]' => 'TEXT' ;
+        process "#bodyContent table li>a",
+            'actors[]' => {
+                'name'      => 'TEXT',
+                'wikipedia' => '@href',
+            }
+        ;
     };
     $scraper->scrape($uri);
@@ -63 +69 @@
 }
 
+sub scrape_details
+{
+    my $class = shift;
+    my $actor = shift;
+
+    my $detail = scraper {
+        process q|//table[@class = 'infobox']/tr[position() = 1]/td/small|,
+            yomi_infobox => 'TEXT'
+        ;
+
+        # This is where the first name usually sits
+        process q|//div[@id = 'bodyContent']/p|,
+            'yomi_text' => [ 'TEXT', sub {
+                use utf8;
+                if (m{[\(（]([^、　]+)}) {
+                    $_ = $1;
+                } else {
+                    $_ = '';
+                }
+            } ]
+        ;
+    }->scrape($actor->{wikipedia});
+
+    $actor->{yomi} ||= $detail->{yomi_infobox} || $detail->{yomi_text};
+    $actor->{yomi} =~ s/[　\s]+//g;
+
+    $actor;
+}
+
 1;