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