Commit | Line | Data |
fe3b3201 |
1 | # $Id: RSS.pm 1934 2006-04-22 05:13:55Z btrott $ |
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 | $feed->init_empty; |
0d5e38d1 |
36 | if ($str) { |
f62087b4 |
37 | $feed->{rss}->parse($$str, { hashrefs_instead_of_strings => 1 } ); |
0d5e38d1 |
38 | } |
39 | $feed; |
40 | } |
41 | |
42 | sub format { 'RSS ' . $_[0]->{rss}->{'version'} } |
43 | |
44 | ## The following elements are the same in all versions of RSS. |
973e1f9e |
45 | sub title { shift->{rss}->channel('title', @_) } |
46 | sub link { shift->{rss}->channel('link', @_) } |
47 | sub description { shift->{rss}->channel('description', @_) } |
0d5e38d1 |
48 | |
813f78d8 |
49 | # This doesn't exist in RSS |
50 | sub id { } |
51 | |
0d5e38d1 |
52 | ## This is RSS 2.0 only--what's the equivalent in RSS 1.0? |
973e1f9e |
53 | sub copyright { shift->{rss}->channel('copyright', @_) } |
0d5e38d1 |
54 | |
5383a560 |
55 | sub base { |
56 | my $feed = shift; |
57 | if (@_) { |
58 | $feed->{rss}->{'xml:base'} = $_[0]; |
59 | } else { |
60 | $feed->{rss}->{'xml:base'}; |
61 | } |
62 | } |
63 | |
0d5e38d1 |
64 | ## The following all work transparently in any RSS version. |
65 | sub language { |
973e1f9e |
66 | my $feed = shift; |
67 | if (@_) { |
68 | $feed->{rss}->channel('language', $_[0]); |
69 | $feed->{rss}->channel->{dc}{language} = $_[0]; |
70 | } else { |
71 | $feed->{rss}->channel('language') || |
72 | $feed->{rss}->channel->{dc}{language}; |
73 | } |
0d5e38d1 |
74 | } |
75 | |
9a36f82c |
76 | sub self_link { |
77 | my $feed = shift; |
78 | |
79 | if (@_) { |
80 | my $uri = shift; |
81 | |
82 | $feed->{rss}->channel->{'atom'}{'link'} = |
83 | { |
84 | rel => "self", |
85 | href => $uri, |
86 | type => "application/rss+xml", |
87 | }; |
88 | } |
89 | |
90 | return $feed->{rss}->channel->{'atom'}{'link'}; |
91 | } |
92 | |
93 | |
0d5e38d1 |
94 | sub generator { |
973e1f9e |
95 | my $feed = shift; |
96 | if (@_) { |
97 | $feed->{rss}->channel('generator', $_[0]); |
98 | $feed->{rss}->channel->{'http://webns.net/mvcb/'}{generatorAgent} = |
99 | $_[0]; |
100 | } else { |
101 | $feed->{rss}->channel('generator') || |
102 | $feed->{rss}->channel->{'http://webns.net/mvcb/'}{generatorAgent}; |
103 | } |
0d5e38d1 |
104 | } |
105 | |
106 | sub author { |
973e1f9e |
107 | my $feed = shift; |
108 | if (@_) { |
109 | $feed->{rss}->channel('webMaster', $_[0]); |
110 | $feed->{rss}->channel->{dc}{creator} = $_[0]; |
111 | } else { |
112 | $feed->{rss}->channel('webMaster') || |
113 | $feed->{rss}->channel->{dc}{creator}; |
114 | } |
0d5e38d1 |
115 | } |
116 | |
117 | sub modified { |
973e1f9e |
118 | my $rss = shift->{rss}; |
119 | if (@_) { |
120 | $rss->channel('pubDate', |
121 | DateTime::Format::Mail->format_datetime($_[0])); |
122 | ## XML::RSS is so weird... if I set this, it will try to use |
123 | ## the value for the lastBuildDate, which I don't want--because |
124 | ## this date is formatted for an RSS 1.0 feed. So it's commented out. |
125 | #$rss->channel->{dc}{date} = |
126 | # DateTime::Format::W3CDTF->format_datetime($_[0]); |
fe3b3201 |
127 | } else { |
128 | my $date; |
129 | eval { |
130 | if (my $ts = $rss->channel('pubDate')) { |
131 | $date = DateTime::Format::Mail->parse_datetime($ts); |
132 | } elsif ($ts = $rss->channel->{dc}{date}) { |
133 | $date = DateTime::Format::W3CDTF->parse_datetime($ts); |
134 | } |
135 | }; |
136 | return $date; |
0d5e38d1 |
137 | } |
138 | } |
139 | |
140 | sub entries { |
141 | my $rss = $_[0]->{rss}; |
142 | my @entries; |
143 | for my $item (@{ $rss->{items} }) { |
729cd7a8 |
144 | push @entries, XML::Feed::Entry::Format::RSS->wrap($item); |
3bdbab6f |
145 | $entries[-1]->{_version} = $rss->{'version'}; |
0d5e38d1 |
146 | } |
147 | @entries; |
148 | } |
149 | |
973e1f9e |
150 | sub add_entry { |
33d4cb3f |
151 | my $feed = shift; |
152 | my $entry = shift || return; |
153 | $entry = $feed->_convert_entry($entry); |
973e1f9e |
154 | $feed->{rss}->add_item(%{ $entry->unwrap }); |
155 | } |
156 | |
157 | sub as_xml { $_[0]->{rss}->as_string } |
158 | |
729cd7a8 |
159 | package XML::Feed::Entry::Format::RSS; |
0d5e38d1 |
160 | use strict; |
161 | |
3bdbab6f |
162 | sub format { 'RSS ' . $_[0]->{'_version'} } |
163 | |
a749d9b9 |
164 | use XML::Feed::Content; |
165 | |
0d5e38d1 |
166 | use base qw( XML::Feed::Entry ); |
167 | |
973e1f9e |
168 | sub init_empty { $_[0]->{entry} = { } } |
169 | |
5383a560 |
170 | sub base { |
171 | my $entry = shift; |
172 | @_ ? $entry->{entry}->{'xml:base'} = $_[0] : $entry->{entry}->{'xml:base'}; |
173 | } |
174 | |
973e1f9e |
175 | sub title { |
176 | my $entry = shift; |
177 | @_ ? $entry->{entry}{title} = $_[0] : $entry->{entry}{title}; |
178 | } |
179 | |
180 | sub link { |
181 | my $entry = shift; |
182 | if (@_) { |
183 | $entry->{entry}{link} = $_[0]; |
184 | ## For RSS 2.0 output from XML::RSS. Sigh. |
185 | $entry->{entry}{permaLink} = $_[0]; |
186 | } else { |
187 | $entry->{entry}{link} || $entry->{entry}{guid}; |
188 | } |
189 | } |
a749d9b9 |
190 | |
191 | sub summary { |
973e1f9e |
192 | my $item = shift->{entry}; |
193 | if (@_) { |
194 | $item->{description} = ref($_[0]) eq 'XML::Feed::Content' ? |
195 | $_[0]->body : $_[0]; |
196 | ## Because of the logic below, we need to add some dummy content, |
197 | ## so that we'll properly recognize the description we enter as |
198 | ## the summary. |
fe3b3201 |
199 | if (!$item->{content}{encoded} && |
973e1f9e |
200 | !$item->{'http://www.w3.org/1999/xhtml'}{body}) { |
fe3b3201 |
201 | $item->{content}{encoded} = ' '; |
973e1f9e |
202 | } |
203 | } else { |
204 | ## Some RSS feeds use <description> for a summary, and some use it |
205 | ## for the full content. Pretty gross. We don't want to return the |
206 | ## full content if the caller expects a summary, so the heuristic is: |
207 | ## if the <entry> contains both a <description> and one of the elements |
208 | ## typically used for the full content, use <description> as summary. |
209 | my $txt; |
210 | if ($item->{description} && |
fe3b3201 |
211 | ($item->{content}{encoded} || |
973e1f9e |
212 | $item->{'http://www.w3.org/1999/xhtml'}{body})) { |
213 | $txt = $item->{description}; |
214 | } |
215 | XML::Feed::Content->wrap({ type => 'text/plain', body => $txt }); |
a749d9b9 |
216 | } |
a749d9b9 |
217 | } |
0d5e38d1 |
218 | |
219 | sub content { |
973e1f9e |
220 | my $item = shift->{entry}; |
221 | if (@_) { |
e3b96b9c |
222 | my $c; |
223 | if (ref($_[0]) eq 'XML::Feed::Content') { |
224 | if (defined $_[0]->base) { |
225 | $c = { 'content' => $_[0]->body, 'xml:base' => $_[0]->base }; |
226 | } else { |
227 | $c = $_[0]->body; |
228 | } |
229 | } else { |
230 | $c = $_[0]; |
231 | } |
fe3b3201 |
232 | $item->{content}{encoded} = $c; |
973e1f9e |
233 | } else { |
7c89ffff |
234 | my $base; |
973e1f9e |
235 | my $body = |
fe3b3201 |
236 | $item->{content}{encoded} || |
973e1f9e |
237 | $item->{'http://www.w3.org/1999/xhtml'}{body} || |
e3b96b9c |
238 | $item->{description}; |
239 | if ('HASH' eq ref($body)) { |
240 | $base = $body->{'xml:base'}; |
241 | $body = $body->{content}; |
242 | } |
7c89ffff |
243 | XML::Feed::Content->wrap({ type => 'text/html', body => $body, base => $base }); |
973e1f9e |
244 | } |
0d5e38d1 |
245 | } |
246 | |
247 | sub category { |
973e1f9e |
248 | my $item = shift->{entry}; |
249 | if (@_) { |
250 | $item->{category} = $item->{dc}{subject} = $_[0]; |
251 | } else { |
252 | $item->{category} || $item->{dc}{subject}; |
253 | } |
0d5e38d1 |
254 | } |
255 | |
256 | sub author { |
973e1f9e |
257 | my $item = shift->{entry}; |
258 | if (@_) { |
259 | $item->{author} = $item->{dc}{creator} = $_[0]; |
260 | } else { |
261 | $item->{author} || $item->{dc}{creator}; |
262 | } |
0d5e38d1 |
263 | } |
264 | |
265 | ## XML::RSS doesn't give us access to the rdf:about for the <item>, |
266 | ## so we have to fall back to the <link> element in RSS 1.0 feeds. |
267 | sub id { |
973e1f9e |
268 | my $item = shift->{entry}; |
269 | if (@_) { |
270 | $item->{guid} = $_[0]; |
271 | } else { |
272 | $item->{guid} || $item->{link}; |
273 | } |
0d5e38d1 |
274 | } |
275 | |
276 | sub issued { |
973e1f9e |
277 | my $item = shift->{entry}; |
278 | if (@_) { |
279 | $item->{dc}{date} = DateTime::Format::W3CDTF->format_datetime($_[0]); |
280 | $item->{pubDate} = DateTime::Format::Mail->format_datetime($_[0]); |
281 | } else { |
ecac864a |
282 | ## Either of these could die if the format is invalid. |
283 | my $date; |
284 | eval { |
285 | if (my $ts = $item->{pubDate}) { |
286 | my $parser = DateTime::Format::Mail->new; |
287 | $parser->loose; |
288 | $date = $parser->parse_datetime($ts); |
289 | } elsif ($ts = $item->{dc}{date}) { |
290 | $date = DateTime::Format::W3CDTF->parse_datetime($ts); |
291 | } |
292 | }; |
293 | return $date; |
0d5e38d1 |
294 | } |
295 | } |
296 | |
297 | sub modified { |
973e1f9e |
298 | my $item = shift->{entry}; |
299 | if (@_) { |
fe3b3201 |
300 | $item->{dcterms}{modified} = |
973e1f9e |
301 | DateTime::Format::W3CDTF->format_datetime($_[0]); |
302 | } else { |
fe3b3201 |
303 | if (my $ts = $item->{dcterms}{modified}) { |
ecac864a |
304 | return eval { DateTime::Format::W3CDTF->parse_datetime($ts) }; |
973e1f9e |
305 | } |
0d5e38d1 |
306 | } |
307 | } |
308 | |
9a36f82c |
309 | sub lat { |
310 | my $item = shift->{entry}; |
311 | if (@_) { |
bd19b444 |
312 | $item->{geo}{lat} = $_[0]; |
9a36f82c |
313 | } else { |
bd19b444 |
314 | return $item->{geo}{lat}; |
9a36f82c |
315 | } |
316 | } |
317 | |
318 | sub long { |
319 | my $item = shift->{entry}; |
320 | if (@_) { |
bd19b444 |
321 | $item->{geo}{long} = $_[0]; |
9a36f82c |
322 | } else { |
bd19b444 |
323 | return $item->{geo}{long}; |
9a36f82c |
324 | } |
325 | } |
326 | |
327 | |
0d5e38d1 |
328 | 1; |