12 |
|
|
13 |
use Feed::Find; |
use Feed::Find; |
14 |
use LWP::UserAgent; |
use LWP::UserAgent; |
15 |
|
|
16 |
use Data::Dump qw/dump/; |
use Data::Dump qw/dump/; |
17 |
|
|
18 |
=head2 canonicalize_uri |
=head2 canonicalize_uri |
24 |
sub canonicalize_uri { |
sub canonicalize_uri { |
25 |
my $self = shift; |
my $self = shift; |
26 |
my $value = shift; |
my $value = shift; |
27 |
warn "uri: $value"; |
$self->log->debug("canonicalize uri $value"); |
28 |
if ($value =~ s/\bgrep\b/%s/) { |
if ($value =~ s/\bgrep\b/%s/) { |
29 |
$self->canonicalization_note( uri => 'Replaced grep with %s' ); |
$self->canonicalization_note( uri => 'Replaced grep with %s' ); |
30 |
} |
} |
42 |
my $self = shift; |
my $self = shift; |
43 |
my $value = shift; |
my $value = shift; |
44 |
|
|
45 |
warn "cookie: $value"; |
#warn "cookie: $value"; |
46 |
|
|
47 |
$self->canonicalization_note( uri => 'Removed Cookie: header' ) |
$self->canonicalization_note( uri => 'Removed Cookie: header' ) |
48 |
if ($value =~ s/^Cookie:\s+//); |
if ($value =~ s/^Cookie:\s+//); |
84 |
|
|
85 |
return $self->result->error( $r->status_line . " from $uri" ) unless ( $r->is_success ); |
return $self->result->error( $r->status_line . " from $uri" ) unless ( $r->is_success ); |
86 |
|
|
87 |
if ($r->header('Content-type') =~ /xml/) { |
my $ct = $r->header('Content-type') or warn "can't get Content-type"; |
88 |
$self->result->message( "Assuming $uri is feed and using it" ); |
my $content = $r->content; |
89 |
|
|
90 |
|
Jifty->log->debug("got ", length( $content ), " bytes $ct"); |
91 |
|
|
92 |
|
if ( $ct =~ /xml/ ) { |
93 |
|
Grep::Source->save( 'addfeed.xml', $content ); |
94 |
|
$self->result->message( "Assuming $uri is feed from $ct" ); |
95 |
return $self->SUPER::take_action( @ARGS ); |
return $self->SUPER::take_action( @ARGS ); |
96 |
} |
} |
97 |
|
|
98 |
my @feeds = Feed::Find->find_in_html( $r->content ); |
Grep::Source->save( 'addfeed.html', $content ); |
99 |
|
|
100 |
|
my $base_uri = $uri; |
101 |
|
$base_uri =~ s!/[^/]+$!!; |
102 |
|
|
103 |
|
my @feeds = Feed::Find->find_in_html( \$content, $base_uri ); |
104 |
|
|
105 |
if (@feeds) { |
if (@feeds) { |
106 |
|
|
107 |
Jifty->log->info("found possible feeds: ", dump( @feeds )); |
Jifty->log->info("found possible feeds: ", dump( @feeds )); |
108 |
|
|
109 |
@feeds = map { |
my @search_feeds = map { |
110 |
my $t = $_; |
my $t = $_; |
111 |
$t =~ s/\b$search_moniker\b/%s/; |
$t =~ s/\b$search_moniker\b/%s/; |
112 |
$t |
$t |
113 |
} grep(/\b$search_moniker\b/,@feeds); |
} grep(/\b$search_moniker\b/,@feeds); |
114 |
|
|
115 |
|
|
116 |
my $feed_uri = shift @feeds; |
if ( my $feed_uri = shift @search_feeds ) { |
117 |
|
|
118 |
|
$self->result->message('Found ' . @feeds . " feeds, using first: $feed_uri" ); |
119 |
|
$self->argument_value('uri', $feed_uri); |
120 |
|
|
121 |
return $self->result->error("Can't find any feed at $uri") unless ( $feed_uri ); |
Jifty->log->debug("calling parent take_action with new uri $feed_uri"); |
122 |
|
|
123 |
$self->result->message('Found ' . @feeds . " feeds, using first: $feed_uri" ); |
return $self->SUPER::take_action( @ARGS ); |
|
$self->argument_value('uri', $feed_uri); |
|
124 |
|
|
125 |
Jifty->log->debug("calling parent take_action with new uri $feed_uri"); |
} else { |
126 |
|
Jifty->log->debug("found feeds ", dump( @feeds ), " but none of them has search moniker!"); |
127 |
|
} |
128 |
|
} |
129 |
|
|
130 |
|
Jifty->log->debug("no feeds found, trying content_class detection"); |
131 |
|
|
132 |
|
my $source = Grep::Source->new(); |
133 |
|
|
134 |
|
if ( my $class = $source->content_class( $content ) ) { |
135 |
|
|
136 |
|
Jifty->log->debug("$class registred for feed $uri"); |
137 |
|
|
138 |
|
$self->argument_value('source', "$class" ); |
139 |
|
$self->result->message("Found $class scraper for $uri" ); |
140 |
return $self->SUPER::take_action( @ARGS ); |
return $self->SUPER::take_action( @ARGS ); |
141 |
|
|
142 |
} else { |
} else { |
143 |
|
|
|
warn "no feeds in ", $r->content; |
|
|
|
|
144 |
$self->result->error('No feeds found on supplied URI'); |
$self->result->error('No feeds found on supplied URI'); |
145 |
return 0; |
return 0; |
146 |
|
|
147 |
} |
} |
148 |
|
|
149 |
} |
} |