Changeset 18428 for lang/perl/Net-DMM/trunk/lib/Net/DMM/Scraper/SearchResult.pm

Timestamp:

08/29/08 16:42:16 (4 months ago)

Author:

bayashi

Message:

いろいろと手を加えました

Files:

• lang/perl/Net-DMM/trunk/lib/Net/DMM/Scraper/SearchResult.pm (modified) (3 diffs)

lang/perl/Net-DMM/trunk/lib/Net/DMM/Scraper/SearchResult.pm

@@ -1 +1 @@
 package Net::DMM::Scraper::SearchResult;
-#
-# 検索結果ページをスクレイピング
-#
-
-use base qw(Class::Accessor);
-__PACKAGE__->mk_accessors(
-        qw( txt_list img_list title link )
-);
 
 use strict;
@@ -13 +5 @@
 #use utf8;
 
-use URI::Fetch;
-use Cache::FileCache;
-use Encode qw(encode decode);
-use URI;
 use Web::Scraper;
+use Net::DMM::Util qw( _fetch _rm_space );
 
-our $VERSION = '0.0.2';
+use base qw(Class::Accessor);
+__PACKAGE__->mk_accessors(
+    qw( txt_list img_list title link )
+);
 
 sub new {
-        my ($class, $opt) = @_;
-        my $self = bless {
-                url   => $opt->{url},
-                opt   => $opt->{opt},
-                title => [],
-                link  => [],
-        }, $class;
-        $self->_scrape if $self->{url};
-        $self;
+    my ($class, $opt) = @_;
+    my $self = bless {
+        url   => $opt->{url},
+        opt   => $opt->{opt},
+        title => [],
+        link  => [],
+    }, $class;
+    $self->_scrape if $self->{url};
+    $self;
 }
 
 sub _scrape {
-        my $self = shift;
-        my $url  = URI->new($self->{url});
-        my $result = scraper {
-                process '/html/body/table/tr/td[3]/table[3]/tr/td/table/tr/td[5]/a',
-                'txt_list' => '@href';
-                process '/html/body/table/tr/td[3]/table[3]/tr/td/table/tr/td[6]/a',
-                'img_list' => '@href';
-                process '/html/body/table/tr/td/table/tr/td/table/tr/td[2]/a',
-                'title[]' => 'TEXT',
-                'link[]'  => '@href';
-        }->scrape($self->_fetch($url));
+    my $self = shift;
+    my $result = scraper {
+        process '/html/body/table/tr/td[3]/table[3]/tr/td/table/tr/td[5]/a',
+        'txt_list' => '@href';
+        process '/html/body/table/tr/td[3]/table[3]/tr/td/table/tr/td[6]/a',
+        'img_list' => '@href';
+        process '/html/body/table/tr/td/table/tr/td/table/tr/td[2]/a',
+        'title[]' => 'TEXT',
+        'link[]'  => '@href';
+    }->scrape($self->_fetch($self->{url}));
 
-        foreach my $key (keys %$result){
-                if($key =~ /^(title|link)$/){
-                        foreach my $i (@{$result->{$key}}){
-                                push(@{$self->{$1}}, $i);
-                        }
-                }else{
-                        $self->set($key, $self->_delete_space($result->{$key}));
-                }
-        }
-}
-
-sub _delete_space {
-        my ($self, $str) = @_;
-        $str =~ s/^\s*(.*?)\s*$/$1/;
-        return $str;
-}
-
-sub _fetch {
-        my $self = shift || return;
-        my $url  = shift || return;
-        my $page = '';
-        if($self->{opt}->{fetch}){
-                $page = URI::Fetch->fetch(
-                        $url,
-                        UserAgent => $self->{opt}->{ua},
-                        Cache     => $self->{opt}->{fetch}->{cache},
-                        NoNetwork => $self->{opt}->{fetch}->{nonetwork},
-                );
-        }else{
-                $page = URI::Fetch->fetch(
-                        $url,
-                        UserAgent => $self->{opt}->{ua},
-                );
-        }
-        return encode('utf8', decode('euc-jp', $page->content));
+    foreach my $key (keys %$result){
+        if($key =~ /^(title|link)$/){
+            foreach my $i (@{$result->{$key}}){
+                push(@{$self->{$1}}, $i);
+            }
+        }else{
+            $self->set($key, $self->_rm_space($result->{$key}));
+        }
+    }
 }
 
@@ -91 +56 @@
 Net::DMM::Scraper::SearchResult
 
+=head1 METHOD
 
-=head1 VERSION
+=over
 
-This document describes Net::DMM::Scraper::SearchResult version 0.0.2
+=item new
 
-
-=head1 SYNOPSIS
-
-    use Net::DMM::Scraper::SearchResult;
+=back
 
 =head1 AUTHOR