| 1 |
use strict; |
| 2 |
use warnings; |
| 3 |
|
| 4 |
=head1 NAME |
| 5 |
|
| 6 |
Grep::Action::AddFeed |
| 7 |
|
| 8 |
=cut |
| 9 |
|
| 10 |
package Grep::Action::AddFeed; |
| 11 |
use base qw/Grep::Action::CreateFeed/; |
| 12 |
|
| 13 |
use Feed::Find; |
| 14 |
use LWP::UserAgent; |
| 15 |
use Data::Dump qw/dump/; |
| 16 |
|
| 17 |
=head2 canonicalize_uri |
| 18 |
|
| 19 |
Replace C<grep>' with C<%s> in URI arguments |
| 20 |
|
| 21 |
=cut |
| 22 |
|
| 23 |
sub canonicalize_uri { |
| 24 |
my $self = shift; |
| 25 |
my $value = shift; |
| 26 |
warn "uri: $value"; |
| 27 |
if ($value =~ s/\bgrep\b/%s/) { |
| 28 |
$self->canonicalization_note( uri => 'Replaced grep with %s' ); |
| 29 |
} |
| 30 |
return $value; |
| 31 |
} |
| 32 |
|
| 33 |
=head2 canonicalize_cookie |
| 34 |
|
| 35 |
Remove C<Cookie:> header from beginning and replace EOL with space. |
| 36 |
|
| 37 |
=cut |
| 38 |
|
| 39 |
# disabled for now |
| 40 |
sub xx_canonicalize_cookie { |
| 41 |
my $self = shift; |
| 42 |
my $value = shift; |
| 43 |
|
| 44 |
warn "cookie: $value"; |
| 45 |
|
| 46 |
$self->canonicalization_note( uri => 'Removed Cookie: header' ) |
| 47 |
if ($value =~ s/^Cookie:\s+//); |
| 48 |
|
| 49 |
$self->canonicalization_note( uri => 'Converted EOL to space' ) |
| 50 |
if ($value =~ s/[\n\r]/ /gs); |
| 51 |
|
| 52 |
return $value; |
| 53 |
} |
| 54 |
=head2 take_action |
| 55 |
|
| 56 |
=cut |
| 57 |
|
| 58 |
sub take_action { |
| 59 |
my $self = shift; |
| 60 |
|
| 61 |
my @ARGS = @_; |
| 62 |
|
| 63 |
# Custom action code |
| 64 |
|
| 65 |
my $ua = LWP::UserAgent->new; |
| 66 |
|
| 67 |
my $cookie = $self->argument_value('cookie'); |
| 68 |
if ($cookie =~ s/{x!(26|3b)}/chr(hex($1))/gei) { |
| 69 |
$self->argument_value('cookie', $cookie); |
| 70 |
} |
| 71 |
|
| 72 |
Jifty->log->debug("using cookie: $cookie"); |
| 73 |
$ua->default_header( 'Cookie' => $cookie ); |
| 74 |
|
| 75 |
my $search_moniker = 'grep'; |
| 76 |
|
| 77 |
my $uri = $self->argument_value('uri'); |
| 78 |
$uri =~ s/{x!(26|3b)}/chr(hex($1))/gei; |
| 79 |
|
| 80 |
Jifty->log->debug("trying to find feed on $uri"); |
| 81 |
|
| 82 |
my $r = $ua->get( sprintf( $uri, $search_moniker ) ); |
| 83 |
|
| 84 |
return $self->result->error( $r->status_line . " from $uri" ) unless ( $r->is_success ); |
| 85 |
|
| 86 |
my $ct = $r->header('Content-type') or warn "can't get Content-type"; |
| 87 |
my $content = $r->content; |
| 88 |
|
| 89 |
Jifty->log->debug("got ", length( $content ), " bytes $ct"); |
| 90 |
|
| 91 |
if ( $ct =~ /xml/ ) { |
| 92 |
$self->result->message( "Assuming $uri is feed from $ct" ); |
| 93 |
return $self->SUPER::take_action( @ARGS ); |
| 94 |
} |
| 95 |
|
| 96 |
my $base_uri = $uri; |
| 97 |
$base_uri =~ s!/[^/]+$!!; |
| 98 |
|
| 99 |
my @feeds = Feed::Find->find_in_html( \$content, $base_uri ); |
| 100 |
|
| 101 |
if (@feeds) { |
| 102 |
|
| 103 |
Jifty->log->info("found possible feeds: ", dump( @feeds )); |
| 104 |
|
| 105 |
my @search_feeds = map { |
| 106 |
my $t = $_; |
| 107 |
$t =~ s/\b$search_moniker\b/%s/; |
| 108 |
$t |
| 109 |
} grep(/\b$search_moniker\b/,@feeds); |
| 110 |
|
| 111 |
|
| 112 |
if ( my $feed_uri = shift @search_feeds ) { |
| 113 |
|
| 114 |
$self->result->message('Found ' . @feeds . " feeds, using first: $feed_uri" ); |
| 115 |
$self->argument_value('uri', $feed_uri); |
| 116 |
|
| 117 |
Jifty->log->debug("calling parent take_action with new uri $feed_uri"); |
| 118 |
|
| 119 |
return $self->SUPER::take_action( @ARGS ); |
| 120 |
|
| 121 |
} else { |
| 122 |
Jifty->log->debug("found feeds ", dump( @feeds ), " but none of them has search moniker!"); |
| 123 |
} |
| 124 |
} |
| 125 |
|
| 126 |
Jifty->log->debug("no feeds found, trying content_class detection"); |
| 127 |
|
| 128 |
my $source = Grep::Source->new(); |
| 129 |
|
| 130 |
if ( my $class = $source->content_class( $content ) ) { |
| 131 |
|
| 132 |
Jifty->log->debug("$class registred for feed $uri"); |
| 133 |
|
| 134 |
$self->argument_value('source', $source); |
| 135 |
$self->result->message("Found $class scraper for $uri" ); |
| 136 |
return $self->SUPER::take_action( @ARGS ); |
| 137 |
|
| 138 |
} else { |
| 139 |
|
| 140 |
$self->result->error('No feeds found on supplied URI'); |
| 141 |
return 0; |
| 142 |
|
| 143 |
} |
| 144 |
|
| 145 |
} |
| 146 |
|
| 147 |
1; |
| 148 |
|
Parent Directory
| 
Revision Log