Changeset 4347 for lang/perl/Acme-Actors-JA

Timestamp:

01/10/08 13:09:39 (11 months ago)

Author:

daisuke

Message:

lang/perl/Acme-Actors-JA; change return values

Location:

lang/perl/Acme-Actors-JA/trunk

Files:

• eg/scrape_details.pl (modified) (1 diff)
• lib/Acme/Actors/JA/Scraper.pm (modified) (7 diffs)

lang/perl/Acme-Actors-JA/trunk/eg/scrape_details.pl

@@ -5 +5 @@
 use URI;
 
-my $h = Acme::Actors::JA::Scraper->scrape_actors_jp();
+my $actors = [];
 
-foreach my $actor (@{ $h->{actors} }) {
+push @$actors, @{Acme::Actors::JA::Scraper->scrape_actress_jp()};
+push @$actors, @{Acme::Actors::JA::Scraper->scrape_actors_jp()};
+
+foreach my $actor (@$actors) {
     Acme::Actors::JA::Scraper->scrape_details($actor);
     print Dump($actor);

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

@@ -5 +5 @@
 
 package Acme::Actors::JA::Scraper;
+use utf8;
 use strict;
 use warnings;
@@ -36 +37 @@
         ;
     };
-    $scraper->scrape($uri);
+    $scraper->scrape($uri)->{actors};
 }
 
@@ -44 +45 @@
 
     my $scraper = scraper {
-        process "#bodyContent table li>a", 'names[]' => 'TEXT' ;
+        process "#bodyContent table li>a",
+            'actors[]' => {
+                'name'      => 'TEXT',
+                'wikipedia' => '@href'
+            }
+        ;
     };
-    $scraper->scrape($uri);
+    $scraper->scrape($uri)->{actors};
 }
 
@@ -54 +60 @@
 
     my $scraper = scraper {
-        process q|//div[@id = 'bodyContent']/ul[position() > 2 and position() != last()]/li/a|, 'names[]' => 'TEXT' ;
+        process q|//div[@id = 'bodyContent']/ul[position() > 2 and position() != last()]/li/a|,
+            'actors[]' => {
+                'name'      => 'TEXT',
+                'wikipedia' => '@href',
+            }
+        ;
     };
-    $scraper->scrape($uri);
+    $scraper->scrape($uri)->{actors};
 }
 
@@ -64 +75 @@
 
     my $scraper = scraper {
-        process q|//div[@id = 'bodyContent']/ul[position() > 2 and position() != last()]/li/a|, 'names[]' => 'TEXT' ;
+        process q|//div[@id = 'bodyContent']/ul[position() > 2 and position() != last()]/li/a|,
+            'actors[]' => {
+                'name'      => 'TEXT',
+                'wikipedia' => '@href'
+            }
+        ;
     };
-    $scraper->scrape($uri);
+    $scraper->scrape($uri)->{actors};
 }
 
@@ -82 +98 @@
         process q|//div[@id = 'bodyContent']/p|,
             'yomi_text' => [ 'TEXT', sub {
-                use utf8;
-                if (m{[\(（]([^、　]+)}) {
+                if (m{[\(（]([^、　）]+)}) {
                     $_ = $1;
                 } else {
@@ -93 +108 @@
 
     $actor->{yomi} ||= $detail->{yomi_infobox} || $detail->{yomi_text};
-    $actor->{yomi} =~ s/[　\s]+//g;
+    if ($actor->{yomi}) {
+        $actor->{yomi} =~ s/[　\s]+//g;
+    }
 
     $actor;