Grep/Source/Feed.pm

# Dobrica Pavlinusic, <dpavlin@rot13.org> 02/22/07 19:42:14 CET

use strict;
use warnings;

package Grep::Source::Feed;

use base 'Jifty::Object';

use Data::Dump qw/dump/;

=head2 fetch

  Grep::Source::Feed->fetch( $source_obj );

=cut

sub fetch {
        my $self = shift;

        my $parent = shift || $self->log->confess('no parent?');
        my ($feed,$uri) = ($parent->feed, $parent->uri);
        die "no uri" unless ($uri);
        die "feed is not a Grep::Model::Feed but ", ref $feed unless $feed->isa('Grep::Model::Feed');

        my $ua = LWP::UserAgent->new;
        $ua->default_header( 'Cookie' => $feed->cookie );
        my $r = $ua->get( $uri );
        die $feed->title . " returned " . $r->status_line . " for $uri\n" unless ( $r->is_success );
        my $content = $r->content;

        $parent->save('feed.xml', $content);

        die "No content returned from $uri\n" unless length( $content ) > 1;

        my $xml_feed = XML::Feed->parse( \$content )
                or die $feed->title, " returned ", XML::Feed->errstr, "\n";

        $self->log->debug("getting ", scalar $xml_feed->entries, " entries from ", $xml_feed->format, " feed ", $xml_feed->title);

        for my $entry ($xml_feed->entries) {

                $parent->add_record(
                        in_feed => $feed,
                        title => $entry->title || $entry->link,
                        link => $entry->link,
                        content => $entry->content->body,
                        summary => $entry->summary->body,
                        category => $entry->category,
                        author => $entry->author,
                        created_on => $entry->issued ? $entry->issued->strftime("%Y-%m-%d %H:%M:%S") : undef,
                        last_update => $entry->modified ? $entry->modified->strftime("%Y-%m-%d %H:%M:%S") : undef,
                );

        }

}

1;
1	dpavlin	72	# Dobrica Pavlinusic, <dpavlin@rot13.org> 02/22/07 19:42:14 CET
2
3			use strict;
4			use warnings;
5
6			package Grep::Source::Feed;
7
8	dpavlin	104	use base 'Jifty::Object';
9
10	dpavlin	72	use Data::Dump qw/dump/;
11
12			=head2 fetch
13
14	dpavlin	110	Grep::Source::Feed->fetch( $source_obj );
15	dpavlin	72
16			=cut
17
18			sub fetch {
19			my $self = shift;
20
21	dpavlin	110	my $parent = shift \|\| $self->log->confess('no parent?');
22	dpavlin	72	my ($feed,$uri) = ($parent->feed, $parent->uri);
23			die "no uri" unless ($uri);
24			die "feed is not a Grep::Model::Feed but ", ref $feed unless $feed->isa('Grep::Model::Feed');
25
26			my $ua = LWP::UserAgent->new;
27			$ua->default_header( 'Cookie' => $feed->cookie );
28			my $r = $ua->get( $uri );
29			die $feed->title . " returned " . $r->status_line . " for $uri\n" unless ( $r->is_success );
30			my $content = $r->content;
31
32	dpavlin	99	$parent->save('feed.xml', $content);
33	dpavlin	86
34	dpavlin	72	die "No content returned from $uri\n" unless length( $content ) > 1;
35
36			my $xml_feed = XML::Feed->parse( \$content )
37			or die $feed->title, " returned ", XML::Feed->errstr, "\n";
38
39	dpavlin	110	$self->log->debug("getting ", scalar $xml_feed->entries, " entries from ", $xml_feed->format, " feed ", $xml_feed->title);
40	dpavlin	72
41			for my $entry ($xml_feed->entries) {
42
43			$parent->add_record(
44			in_feed => $feed,
45	dpavlin	99	title => $entry->title \|\| $entry->link,
46	dpavlin	72	link => $entry->link,
47			content => $entry->content->body,
48			summary => $entry->summary->body,
49			category => $entry->category,
50			author => $entry->author,
51	dpavlin	169	created_on => $entry->issued ? $entry->issued->strftime("%Y-%m-%d %H:%M:%S") : undef,
52			last_update => $entry->modified ? $entry->modified->strftime("%Y-%m-%d %H:%M:%S") : undef,
53	dpavlin	72	);
54
55			}
56
57			}
58
59			1;