Updated changed for 0.45 (oops) and 0.46.
[catagits/XML-Feed.git] / lib / XML / Feed / Format / RSS.pm
CommitLineData
3353d70c 1# $Id$
0d5e38d1 2
729cd7a8 3package XML::Feed::Format::RSS;
0d5e38d1 4use strict;
5
6use base qw( XML::Feed );
0d5e38d1 7use DateTime::Format::Mail;
8use DateTime::Format::W3CDTF;
723eaa15 9use XML::Atom::Util qw(iso2dt);
7b5fccb1 10use XML::Feed::Enclosure;
0d5e38d1 11
973e1f9e 12our $PREFERRED_PARSER = "XML::RSS";
13
9b6bc912 14
15sub identify {
16 my $class = shift;
17 my $xml = shift;
18 my $tag = $class->_get_first_tag($xml);
19 return ($tag eq 'rss' || $tag eq 'RDF');
20}
21
973e1f9e 22sub init_empty {
4e9c4625 23 my ($feed, %args) = @_;
24 $args{'version'} ||= '2.0';
973e1f9e 25 eval "use $PREFERRED_PARSER"; die $@ if $@;
4e9c4625 26 $feed->{rss} = $PREFERRED_PARSER->new(%args);
fe3b3201 27 $feed->{rss}->add_module(prefix => "content", uri => 'http://purl.org/rss/1.0/modules/content/');
813f78d8 28 $feed->{rss}->add_module(prefix => "dcterms", uri => 'http://purl.org/dc/terms/');
9a36f82c 29 $feed->{rss}->add_module(prefix => "atom", uri => 'http://www.w3.org/2005/Atom');
30 $feed->{rss}->add_module(prefix => "geo", uri => 'http://www.w3.org/2003/01/geo/wgs84_pos#');
973e1f9e 31 $feed;
32}
33
0d5e38d1 34sub init_string {
35 my $feed = shift;
36 my($str) = @_;
973e1f9e 37 $feed->init_empty;
b1aa7a62 38 my $opts = {
39 hashrefs_instead_of_strings => 1,
40 };
41 $opts->{allow_multiple} = [ 'enclosure' ] if $XML::Feed::MULTIPLE_ENCLOSURES;
0d5e38d1 42 if ($str) {
b1aa7a62 43 $feed->{rss}->parse($$str, $opts );
0d5e38d1 44 }
45 $feed;
46}
47
48sub format { 'RSS ' . $_[0]->{rss}->{'version'} }
49
50## The following elements are the same in all versions of RSS.
973e1f9e 51sub title { shift->{rss}->channel('title', @_) }
52sub link { shift->{rss}->channel('link', @_) }
53sub description { shift->{rss}->channel('description', @_) }
e23387f1 54sub updated { shift->modified(@_) }
0d5e38d1 55
813f78d8 56# This doesn't exist in RSS
57sub id { }
58
0d5e38d1 59## This is RSS 2.0 only--what's the equivalent in RSS 1.0?
973e1f9e 60sub copyright { shift->{rss}->channel('copyright', @_) }
0d5e38d1 61
5383a560 62sub base {
63 my $feed = shift;
64 if (@_) {
65 $feed->{rss}->{'xml:base'} = $_[0];
66 } else {
67 $feed->{rss}->{'xml:base'};
68 }
69}
70
0d5e38d1 71## The following all work transparently in any RSS version.
72sub language {
973e1f9e 73 my $feed = shift;
74 if (@_) {
75 $feed->{rss}->channel('language', $_[0]);
76 $feed->{rss}->channel->{dc}{language} = $_[0];
77 } else {
78 $feed->{rss}->channel('language') ||
79 $feed->{rss}->channel->{dc}{language};
80 }
0d5e38d1 81}
82
9a36f82c 83sub self_link {
84 my $feed = shift;
85
86 if (@_) {
87 my $uri = shift;
88
89 $feed->{rss}->channel->{'atom'}{'link'} =
90 {
91 rel => "self",
92 href => $uri,
93 type => "application/rss+xml",
94 };
95 }
96
97 return $feed->{rss}->channel->{'atom'}{'link'};
98}
99
100
0d5e38d1 101sub generator {
973e1f9e 102 my $feed = shift;
103 if (@_) {
104 $feed->{rss}->channel('generator', $_[0]);
105 $feed->{rss}->channel->{'http://webns.net/mvcb/'}{generatorAgent} =
106 $_[0];
107 } else {
108 $feed->{rss}->channel('generator') ||
109 $feed->{rss}->channel->{'http://webns.net/mvcb/'}{generatorAgent};
110 }
0d5e38d1 111}
112
113sub author {
973e1f9e 114 my $feed = shift;
115 if (@_) {
116 $feed->{rss}->channel('webMaster', $_[0]);
117 $feed->{rss}->channel->{dc}{creator} = $_[0];
118 } else {
119 $feed->{rss}->channel('webMaster') ||
120 $feed->{rss}->channel->{dc}{creator};
121 }
0d5e38d1 122}
123
124sub modified {
973e1f9e 125 my $rss = shift->{rss};
126 if (@_) {
127 $rss->channel('pubDate',
128 DateTime::Format::Mail->format_datetime($_[0]));
129 ## XML::RSS is so weird... if I set this, it will try to use
130 ## the value for the lastBuildDate, which I don't want--because
131 ## this date is formatted for an RSS 1.0 feed. So it's commented out.
132 #$rss->channel->{dc}{date} =
133 # DateTime::Format::W3CDTF->format_datetime($_[0]);
fe3b3201 134 } else {
135 my $date;
136 eval {
137 if (my $ts = $rss->channel('pubDate')) {
138 $date = DateTime::Format::Mail->parse_datetime($ts);
139 } elsif ($ts = $rss->channel->{dc}{date}) {
140 $date = DateTime::Format::W3CDTF->parse_datetime($ts);
141 }
142 };
143 return $date;
0d5e38d1 144 }
145}
146
147sub entries {
148 my $rss = $_[0]->{rss};
149 my @entries;
150 for my $item (@{ $rss->{items} }) {
729cd7a8 151 push @entries, XML::Feed::Entry::Format::RSS->wrap($item);
3bdbab6f 152 $entries[-1]->{_version} = $rss->{'version'};
0d5e38d1 153 }
154 @entries;
155}
156
973e1f9e 157sub add_entry {
33d4cb3f 158 my $feed = shift;
159 my $entry = shift || return;
160 $entry = $feed->_convert_entry($entry);
973e1f9e 161 $feed->{rss}->add_item(%{ $entry->unwrap });
162}
163
164sub as_xml { $_[0]->{rss}->as_string }
165
729cd7a8 166package XML::Feed::Entry::Format::RSS;
0d5e38d1 167use strict;
168
3bdbab6f 169sub format { 'RSS ' . $_[0]->{'_version'} }
170
a749d9b9 171use XML::Feed::Content;
172
0d5e38d1 173use base qw( XML::Feed::Entry );
174
973e1f9e 175sub init_empty { $_[0]->{entry} = { } }
176
5383a560 177sub base {
178 my $entry = shift;
179 @_ ? $entry->{entry}->{'xml:base'} = $_[0] : $entry->{entry}->{'xml:base'};
180}
181
973e1f9e 182sub title {
183 my $entry = shift;
184 @_ ? $entry->{entry}{title} = $_[0] : $entry->{entry}{title};
185}
186
187sub link {
188 my $entry = shift;
189 if (@_) {
190 $entry->{entry}{link} = $_[0];
191 ## For RSS 2.0 output from XML::RSS. Sigh.
192 $entry->{entry}{permaLink} = $_[0];
193 } else {
7b5fccb1 194 $entry->{entry}{link} ||
195 $entry->{entry}{permaLink} ||
196 $entry->{entry}{guid};
973e1f9e 197 }
198}
a749d9b9 199
200sub summary {
973e1f9e 201 my $item = shift->{entry};
202 if (@_) {
203 $item->{description} = ref($_[0]) eq 'XML::Feed::Content' ?
204 $_[0]->body : $_[0];
205 ## Because of the logic below, we need to add some dummy content,
206 ## so that we'll properly recognize the description we enter as
207 ## the summary.
fe3b3201 208 if (!$item->{content}{encoded} &&
973e1f9e 209 !$item->{'http://www.w3.org/1999/xhtml'}{body}) {
fe3b3201 210 $item->{content}{encoded} = ' ';
973e1f9e 211 }
212 } else {
213 ## Some RSS feeds use <description> for a summary, and some use it
214 ## for the full content. Pretty gross. We don't want to return the
215 ## full content if the caller expects a summary, so the heuristic is:
216 ## if the <entry> contains both a <description> and one of the elements
217 ## typically used for the full content, use <description> as summary.
218 my $txt;
219 if ($item->{description} &&
fe3b3201 220 ($item->{content}{encoded} ||
973e1f9e 221 $item->{'http://www.w3.org/1999/xhtml'}{body})) {
222 $txt = $item->{description};
723eaa15 223 ## Blogspot's 'short' RSS feeds do this in the Atom namespace
224 ## for no obviously good reason.
225 } elsif ($item->{'http://www.w3.org/2005/Atom'}{summary}) {
226 $txt = $item->{'http://www.w3.org/2005/Atom'}{summary};
973e1f9e 227 }
228 XML::Feed::Content->wrap({ type => 'text/plain', body => $txt });
a749d9b9 229 }
a749d9b9 230}
0d5e38d1 231
232sub content {
973e1f9e 233 my $item = shift->{entry};
234 if (@_) {
e3b96b9c 235 my $c;
236 if (ref($_[0]) eq 'XML::Feed::Content') {
237 if (defined $_[0]->base) {
238 $c = { 'content' => $_[0]->body, 'xml:base' => $_[0]->base };
239 } else {
240 $c = $_[0]->body;
241 }
242 } else {
243 $c = $_[0];
244 }
fe3b3201 245 $item->{content}{encoded} = $c;
973e1f9e 246 } else {
7c89ffff 247 my $base;
973e1f9e 248 my $body =
fe3b3201 249 $item->{content}{encoded} ||
973e1f9e 250 $item->{'http://www.w3.org/1999/xhtml'}{body} ||
e3b96b9c 251 $item->{description};
252 if ('HASH' eq ref($body)) {
253 $base = $body->{'xml:base'};
254 $body = $body->{content};
255 }
7c89ffff 256 XML::Feed::Content->wrap({ type => 'text/html', body => $body, base => $base });
973e1f9e 257 }
0d5e38d1 258}
259
260sub category {
a0cca2a4 261 my $entry = shift;
262 my $item = $entry->{entry};
973e1f9e 263 if (@_) {
a0cca2a4 264 my @tmp = ($entry->category, @_);
265 $item->{category} = [@tmp];
266 $item->{dc}{subject} = [@tmp];
973e1f9e 267 } else {
a0cca2a4 268 my $r = $item->{category} || $item->{dc}{subject};
e5a65e5e 269 my @r = ref($r) eq 'ARRAY' ? @$r : defined $r? ($r) : ();
a0cca2a4 270 return wantarray? @r : $r[0];
973e1f9e 271 }
0d5e38d1 272}
273
274sub author {
973e1f9e 275 my $item = shift->{entry};
276 if (@_) {
277 $item->{author} = $item->{dc}{creator} = $_[0];
278 } else {
279 $item->{author} || $item->{dc}{creator};
280 }
0d5e38d1 281}
282
283## XML::RSS doesn't give us access to the rdf:about for the <item>,
284## so we have to fall back to the <link> element in RSS 1.0 feeds.
285sub id {
973e1f9e 286 my $item = shift->{entry};
287 if (@_) {
288 $item->{guid} = $_[0];
289 } else {
290 $item->{guid} || $item->{link};
291 }
0d5e38d1 292}
293
294sub issued {
973e1f9e 295 my $item = shift->{entry};
296 if (@_) {
297 $item->{dc}{date} = DateTime::Format::W3CDTF->format_datetime($_[0]);
298 $item->{pubDate} = DateTime::Format::Mail->format_datetime($_[0]);
299 } else {
ecac864a 300 ## Either of these could die if the format is invalid.
301 my $date;
302 eval {
303 if (my $ts = $item->{pubDate}) {
304 my $parser = DateTime::Format::Mail->new;
305 $parser->loose;
306 $date = $parser->parse_datetime($ts);
daba7240 307 } elsif ($ts = $item->{dc}{date} or $ts = $item->{dcterms}{date}) {
308 $date = DateTime::Format::W3CDTF->parse_datetime($ts);
ecac864a 309 }
310 };
311 return $date;
0d5e38d1 312 }
313}
314
315sub modified {
973e1f9e 316 my $item = shift->{entry};
317 if (@_) {
fe3b3201 318 $item->{dcterms}{modified} =
973e1f9e 319 DateTime::Format::W3CDTF->format_datetime($_[0]);
320 } else {
723eaa15 321 if (my $ts = $item->{dcterms}{modified} || $item->{'http://www.w3.org/2005/Atom'}{updated}) {
322 return eval { DateTime::Format::W3CDTF->parse_datetime($ts) } || eval { XML::Atom::Util::iso2dt($ts) };
323 }
0d5e38d1 324 }
325}
326
9a36f82c 327sub lat {
328 my $item = shift->{entry};
329 if (@_) {
bd19b444 330 $item->{geo}{lat} = $_[0];
9a36f82c 331 } else {
bd19b444 332 return $item->{geo}{lat};
9a36f82c 333 }
334}
335
336sub long {
337 my $item = shift->{entry};
338 if (@_) {
bd19b444 339 $item->{geo}{long} = $_[0];
9a36f82c 340 } else {
bd19b444 341 return $item->{geo}{long};
9a36f82c 342 }
343}
344
12a4079f 345sub enclosure {
346 my $entry = shift;
347
348 if (@_) {
349 my $enclosure = shift;
b1aa7a62 350 my $val = {
af6b00a4 351 url => $enclosure->{url},
352 type => $enclosure->{type},
353 length => $enclosure->{length}
b1aa7a62 354 };
355 if ($XML::Feed::MULTIPLE_ENCLOSURES) {
356 push @{$entry->{entry}->{enclosure}}, $val;
357 } else {
358 $entry->{entry}->{enclosure} = $val;
359 }
12a4079f 360 } else {
b1aa7a62 361 my $tmp = $entry->{entry}->{enclosure};
7b5fccb1 362 if (defined $tmp) {
363 my @encs = map { XML::Feed::Enclosure->new($_) }
364 (ref $tmp eq 'ARRAY')? @$tmp : ($tmp);
365 return ($XML::Feed::MULTIPLE_ENCLOSURES)? @encs : $encs[-1];
366 }
367 return;
12a4079f 368 }
af6b00a4 369}
9a36f82c 370
0d5e38d1 3711;