Changeset 1922 for lang/perl/WWW-CommentGetter/trunk

Timestamp:

11/23/07 13:26:24 (6 years ago)

Author:

tokuhirom

Message:

WWW-CommentGetter: support new hatena style.

Files:

• lang/perl/WWW-CommentGetter/trunk/WWW-CommentGetter/lib/WWW/CommentGetter/Plugin/HatenaDiary.pm (modified) (1 diff)

lang/perl/WWW-CommentGetter/trunk/WWW-CommentGetter/lib/WWW/CommentGetter/Plugin/HatenaDiary.pm

@@ -11 +11 @@
     return unless $url =~ m{^http://d.hatena.ne.jp.*};
 
-    # tokuhirom \x{300e}\x{305d}\x{3093}\x{306a}\x{3082}\x{3093}\x{304b}\x{306d}\x{3002}\x{300f} (2007/10/22 07:29)
-
     my $ret = scraper {
-        process 'div.commentshort > p', 'bodies[]' => sub {
-            my $elem = shift;
-            if ($elem->as_text =~ m{^(.+) \x{300e}(.+)\x{300f} \((\d{4}/\d\d/\d\d \d\d:\d\d)\)$}) {
-                return +{ author => $self->trim($1), body => $2, date => $3 };
-            } else {
-                return;
-            }
+        process '//div[@class="commentshort"]/p[@class!="commentmessage"]', 'comments[]' => scraper {
+            process '//span[@class="commentator"]', 'author', 'TEXT';
+            process '//span[@class="commentbody"]', 'body',   'TEXT';
+            process '//span[@class="timestamp"]',   'date',   'TEXT';
         };
     }->scrape(URI->new($url));
 
-    return [
-        grep { $_ } @{ $ret->{bodies} }
-    ];
+    return [ @{ $ret->{comments} } ];
 }