| @@ -1,39 +1,98 @@ | ||
| 1 | 1 | <?php |
| 2 | +require_once(dirname(__FILE__).'/feedtime.class.php'); | |
| 3 | + | |
| 4 | +/** | |
| 5 | + * class SyndicatedPost: FeedWordPress uses to manage the conversion of | |
| 6 | + * incoming items from the feed parser into posts for the WordPress | |
| 7 | + * database. It contains several internal management methods primarily | |
| 8 | + * of interest to someone working on the FeedWordPress source, as well | |
| 9 | + * as some utility methods for extracting useful data from many | |
| 10 | + * different feed formats, which may be useful to FeedWordPress users | |
| 11 | + * who make use of feed data in PHP add-ons and filters. | |
| 12 | + * | |
| 13 | + * @version 2010.0531 | |
| 14 | + */ | |
| 2 | 15 | class SyndicatedPost { |
| 3 | - var $item = null; | |
| 4 | - | |
| 16 | + var $item = null; // MagpieRSS representation | |
| 17 | + var $entry = null; // SimplePie_Item representation | |
| 18 | + | |
| 5 | 19 | var $link = null; |
| 6 | 20 | var $feed = null; |
| 7 | 21 | var $feedmeta = null; |
| 8 | 22 | |
| 23 | + var $xmlns = array (); | |
| 24 | + | |
| 9 | 25 | var $post = array (); |
| 10 | 26 | |
| 11 | 27 | var $_freshness = null; |
| 12 | 28 | var $_wp_id = null; |
| 13 | 29 | |
| 14 | - function SyndicatedPost ($item, $link) { | |
| 30 | + /** | |
| 31 | + * SyndicatedPost constructor: Given a feed item and the source from | |
| 32 | + * which it was taken, prepare a post that can be inserted into the | |
| 33 | + * WordPress database on request, or updated in place if it has already | |
| 34 | + * been syndicated. | |
| 35 | + * | |
| 36 | + * @param array $item The item syndicated from the feed. | |
| 37 | + * @param SyndicatedLink $source The feed it was syndicated from. | |
| 38 | + */ | |
| 39 | + function SyndicatedPost ($item, $source) { | |
| 15 | 40 | global $wpdb; |
| 16 | 41 | |
| 17 | - $this->link = $link; | |
| 18 | - $feedmeta = $link->settings; | |
| 19 | - $feed = $link->magpie; | |
| 42 | + if (is_array($item) | |
| 43 | + and isset($item['simplepie']) | |
| 44 | + and isset($item['magpie'])) : | |
| 45 | + $this->entry = $item['simplepie']; | |
| 46 | + $this->item = $item['magpie']; | |
| 47 | + $item = $item['magpie']; | |
| 48 | + else : | |
| 49 | + $this->item = $item; | |
| 50 | + endif; | |
| 20 | 51 | |
| 21 | - # This is ugly as all hell. I'd like to use apply_filters()'s | |
| 22 | - # alleged support for a variable argument count, but this seems | |
| 23 | - # to have been broken in WordPress 1.5. It'll be fixed somehow | |
| 24 | - # in WP 1.5.1, but I'm aiming at WP 1.5 compatibility across | |
| 25 | - # the board here. | |
| 52 | + FeedWordPress::diagnostic('feed_items', 'Considering item ['.$this->guid().'] "'.$this->entry->get_title().'"'); | |
| 53 | + | |
| 54 | + $this->link = $source; | |
| 55 | + $this->feed = $source->magpie; | |
| 56 | + $this->feedmeta = $source->settings; | |
| 57 | + | |
| 58 | + # Dealing with namespaces can get so fucking fucked. | |
| 59 | + $this->xmlns['forward'] = $source->magpie->_XMLNS_FAMILIAR; | |
| 60 | + $this->xmlns['reverse'] = array(); | |
| 61 | + foreach ($this->xmlns['forward'] as $url => $ns) : | |
| 62 | + if (!isset($this->xmlns['reverse'][$ns])) : | |
| 63 | + $this->xmlns['reverse'][$ns] = array(); | |
| 64 | + endif; | |
| 65 | + $this->xmlns['reverse'][$ns][] = $url; | |
| 66 | + endforeach; | |
| 67 | + | |
| 68 | + // Fucking SimplePie. | |
| 69 | + $this->xmlns['reverse']['rss'][] = ''; | |
| 70 | + | |
| 71 | + # These globals were originally an ugly kludge around a bug in | |
| 72 | + # apply_filters from WordPress 1.5. The bug was fixed in 1.5.1, | |
| 73 | + # and I sure hope at this point that nobody writing filters for | |
| 74 | + # FeedWordPress is still relying on them. | |
| 26 | 75 | # |
| 27 | - # Cf.: <http://mosquito.wordpress.org/view.php?id=901> | |
| 76 | + # Anyway, I hereby declare them DEPRECATED as of 8 February | |
| 77 | + # 2010. I'll probably remove the globals within 1-2 releases in | |
| 78 | + # the interests of code hygiene and memory usage. If you | |
| 79 | + # currently use them in your filters, I advise you switch off to | |
| 80 | + # accessing the public members SyndicatedPost::feed and | |
| 81 | + # SyndicatedPost::feedmeta. | |
| 82 | + | |
| 28 | 83 | global $fwp_channel, $fwp_feedmeta; |
| 29 | - $fwp_channel = $feed; $fwp_feedmeta = $feedmeta; | |
| 84 | + $fwp_channel = $this->feed; $fwp_feedmeta = $this->feedmeta; | |
| 30 | 85 | |
| 31 | - $this->feed = $feed; | |
| 32 | - $this->feedmeta = $feedmeta; | |
| 33 | - | |
| 34 | - $this->item = $item; | |
| 86 | + // Trigger global syndicated_item filter. | |
| 35 | 87 | $this->item = apply_filters('syndicated_item', $this->item, $this); |
| 88 | + | |
| 89 | + // Allow for feed-specific syndicated_item filters. | |
| 90 | + $this->item = apply_filters( | |
| 91 | + "syndicated_item_".$source->uri(), | |
| 92 | + $this->item, | |
| 93 | + $this | |
| 94 | + ); | |
| 36 | 95 | |
| 37 | 96 | # Filters can halt further processing by returning NULL |
| 38 | 97 | if (is_null($this->item)) : |
| 39 | 98 | $this->post = NULL; |
| @@ -42,51 +101,31 @@ | ||
| 42 | 101 | # That's deliberate. The escaping is done at the point |
| 43 | 102 | # of insertion, not here, to avoid double-escaping and |
| 44 | 103 | # to avoid screwing with syndicated_post filters |
| 45 | 104 | |
| 46 | - $this->post['post_title'] = apply_filters('syndicated_item_title', $this->item['title'], $this); | |
| 105 | + $this->post['post_title'] = apply_filters( | |
| 106 | + 'syndicated_item_title', | |
| 107 | + $this->entry->get_title(), $this | |
| 108 | + ); | |
| 47 | 109 | |
| 48 | - // This just gives us an alphanumeric representation of | |
| 49 | - // the author. We will look up (or create) the numeric | |
| 50 | - // ID for the author in SyndicatedPost::add() | |
| 51 | - $this->post['named']['author'] = apply_filters('syndicated_item_author', $this->author(), $this); | |
| 110 | + $this->post['named']['author'] = apply_filters( | |
| 111 | + 'syndicated_item_author', | |
| 112 | + $this->author(), $this | |
| 113 | + ); | |
| 114 | + // This just gives us an alphanumeric name for the author. | |
| 115 | + // We look up (or create) the numeric ID for the author | |
| 116 | + // in SyndicatedPost::add(). | |
| 52 | 117 | |
| 53 | - # Identify content and sanitize it. | |
| 54 | - # --------------------------------- | |
| 55 | - if (isset($this->item['atom_content'])) : | |
| 56 | - $content = $this->item['atom_content']; | |
| 57 | - elseif (isset($this->item['xhtml']['body'])) : | |
| 58 | - $content = $this->item['xhtml']['body']; | |
| 59 | - elseif (isset($this->item['xhtml']['div'])) : | |
| 60 | - $content = $this->item['xhtml']['div']; | |
| 61 | - elseif (isset($this->item['content']['encoded']) and $this->item['content']['encoded']): | |
| 62 | - $content = $this->item['content']['encoded']; | |
| 63 | - else: | |
| 64 | - $content = $this->item['description']; | |
| 65 | - endif; | |
| 66 | - $this->post['post_content'] = apply_filters('syndicated_item_content', $content, $this); | |
| 67 | - | |
| 68 | - # Identify and sanitize excerpt | |
| 69 | - $excerpt = NULL; | |
| 70 | - if ( isset($this->item['description']) and $this->item['description'] ) : | |
| 71 | - $excerpt = $this->item['description']; | |
| 72 | - elseif ( isset($content) and $content ) : | |
| 73 | - $excerpt = strip_tags($content); | |
| 74 | - if (strlen($excerpt) > 255) : | |
| 75 | - $excerpt = substr($excerpt,0,252).'...'; | |
| 76 | - endif; | |
| 77 | - endif; | |
| 78 | - $excerpt = apply_filters('syndicated_item_excerpt', $excerpt, $this); | |
| 79 | - | |
| 80 | - if (!is_null($excerpt)): | |
| 118 | + $this->post['post_content'] = apply_filters( | |
| 119 | + 'syndicated_item_content', | |
| 120 | + $this->content(), $this | |
| 121 | + ); | |
| 122 | + | |
| 123 | + $excerpt = apply_filters('syndicated_item_excerpt', $this->excerpt(), $this); | |
| 124 | + if (!empty($excerpt)): | |
| 81 | 125 | $this->post['post_excerpt'] = $excerpt; |
| 82 | 126 | endif; |
| 83 | 127 | |
| 84 | - // This is unnecessary if we use wp_insert_post | |
| 85 | - if (!$this->use_api('wp_insert_post')) : | |
| 86 | - $this->post['post_name'] = sanitize_title($this->post['post_title']); | |
| 87 | - endif; | |
| 88 | - | |
| 89 | 128 | $this->post['epoch']['issued'] = apply_filters('syndicated_item_published', $this->published(), $this); |
| 90 | 129 | $this->post['epoch']['created'] = apply_filters('syndicated_item_created', $this->created(), $this); |
| 91 | 130 | $this->post['epoch']['modified'] = apply_filters('syndicated_item_updated', $this->updated(), $this); |
| 92 | 131 | |
| @@ -91,12 +130,12 @@ | ||
| 91 | 130 | $this->post['epoch']['modified'] = apply_filters('syndicated_item_updated', $this->updated(), $this); |
| 92 | 131 | |
| 93 | 132 | // Dealing with timestamps in WordPress is so fucking fucked. |
| 94 | 133 | $offset = (int) get_option('gmt_offset') * 60 * 60; |
| 95 | - $this->post['post_date'] = gmdate('Y-m-d H:i:s', $this->published() + $offset); | |
| 96 | - $this->post['post_modified'] = gmdate('Y-m-d H:i:s', $this->updated() + $offset); | |
| 97 | - $this->post['post_date_gmt'] = gmdate('Y-m-d H:i:s', $this->published()); | |
| 98 | - $this->post['post_modified_gmt'] = gmdate('Y-m-d H:i:s', $this->updated()); | |
| 134 | + $this->post['post_date'] = gmdate('Y-m-d H:i:s', apply_filters('syndicated_item_published', $this->published(/*fallback=*/ true, /*default=*/ -1), $this) + $offset); | |
| 135 | + $this->post['post_modified'] = gmdate('Y-m-d H:i:s', apply_filters('syndicated_item_updated', $this->updated(/*fallback=*/ true, /*default=*/ -1), $this) + $offset); | |
| 136 | + $this->post['post_date_gmt'] = gmdate('Y-m-d H:i:s', apply_filters('syndicated_item_published', $this->published(/*fallback=*/ true, /*default=*/ -1), $this)); | |
| 137 | + $this->post['post_modified_gmt'] = gmdate('Y-m-d H:i:s', apply_filters('syndicated_item_updated', $this->updated(/*fallback=*/ true, /*default=*/ -1), $this)); | |
| 99 | 138 | |
| 100 | 139 | // Use feed-level preferences or the global default. |
| 101 | 140 | $this->post['post_status'] = $this->link->syndicated_status('post', 'publish'); |
| 102 | 141 | $this->post['comment_status'] = $this->link->syndicated_status('comment', 'closed'); |
| @@ -120,20 +159,49 @@ | ||
| 120 | 159 | endif; |
| 121 | 160 | if (!is_array($custom_settings)) : |
| 122 | 161 | $custom_settings = array(); |
| 123 | 162 | endif; |
| 124 | - $this->post['meta'] = array_merge($default_custom_settings, $custom_settings); | |
| 163 | + | |
| 164 | + $postMetaIn = array_merge($default_custom_settings, $custom_settings); | |
| 165 | + $postMetaOut = array(); | |
| 125 | 166 | |
| 167 | + // Big ugly fuckin loop to do any element substitutions | |
| 168 | + // that we may need. | |
| 169 | + foreach ($postMetaIn as $key => $values) : | |
| 170 | + if (is_string($values)) : $values = array($values); endif; | |
| 171 | + | |
| 172 | + $postMetaOut[$key] = array(); | |
| 173 | + foreach ($values as $value) : | |
| 174 | + if (preg_match('/\$\( ([^)]+) \)/x', $value, $ref)) : | |
| 175 | + $elements = $this->query($ref[1]); | |
| 176 | + foreach ($elements as $element) : | |
| 177 | + $postMetaOut[$key][] = str_replace( | |
| 178 | + $ref[0], | |
| 179 | + $element, | |
| 180 | + $value | |
| 181 | + ); | |
| 182 | + endforeach; | |
| 183 | + else : | |
| 184 | + $postMetaOut[$key][] = $value; | |
| 185 | + endif; | |
| 186 | + endforeach; | |
| 187 | + endforeach; | |
| 188 | + | |
| 189 | + foreach ($postMetaOut as $key => $values) : | |
| 190 | + $this->post['meta'][$key] = array(); | |
| 191 | + foreach ($values as $value) : | |
| 192 | + $this->post['meta'][$key][] = apply_filters("syndicated_post_meta_{$key}", $value, $this); | |
| 193 | + endforeach; | |
| 194 | + endforeach; | |
| 195 | + | |
| 126 | 196 | // RSS 2.0 / Atom 1.0 enclosure support |
| 127 | - if ( isset($this->item['enclosure#']) ) : | |
| 128 | - for ($i = 1; $i <= $this->item['enclosure#']; $i++) : | |
| 129 | - $eid = (($i > 1) ? "#{$id}" : ""); | |
| 130 | - $this->post['meta']['enclosure'][] = | |
| 131 | - apply_filters('syndicated_item_enclosure_url', $this->item["enclosure{$eid}@url"], $this)."\n". | |
| 132 | - apply_filters('syndicated_item_enclosure_length', $this->item["enclosure{$eid}@length"], $this)."\n". | |
| 133 | - apply_filters('syndicated_item_enclosure_type', $this->item["enclosure{$eid}@type"], $this); | |
| 134 | - endfor; | |
| 135 | - endif; | |
| 197 | + $enclosures = $this->entry->get_enclosures(); | |
| 198 | + if (is_array($enclosures)) : foreach ($enclosures as $enclosure) : | |
| 199 | + $this->post['meta']['enclosure'][] = | |
| 200 | + apply_filters('syndicated_item_enclosure_url', $enclosure->get_link(), $this)."\n". | |
| 201 | + apply_filters('syndicated_item_enclosure_length', $enclosure->get_length(), $this)."\n". | |
| 202 | + apply_filters('syndicated_item_enclosure_type', $enclosure->get_type(), $this); | |
| 203 | + endforeach; endif; | |
| 136 | 204 | |
| 137 | 205 | // In case you want to point back to the blog this was syndicated from |
| 138 | 206 | if (isset($this->feed->channel['title'])) : |
| 139 | 207 | $this->post['meta']['syndication_source'] = apply_filters('syndicated_item_source_title', $this->feed->channel['title'], $this); |
| @@ -156,33 +224,26 @@ | ||
| 156 | 224 | $this->post['meta']['syndication_source_id_original'] = $this->item['source_id']; |
| 157 | 225 | endif; |
| 158 | 226 | |
| 159 | 227 | // Store information on human-readable and machine-readable comment URIs |
| 160 | - if (isset($this->item['comments'])) : | |
| 161 | - $this->post['meta']['rss:comments'] = apply_filters('syndicated_item_comments', $this->item['comments']); | |
| 162 | - endif; | |
| 163 | 228 | |
| 164 | - // RSS 2.0 comment feeds extension | |
| 165 | - if (isset($this->item['wfw']['commentrss'])) : | |
| 166 | - $this->post['meta']['wfw:commentRSS'] = apply_filters('syndicated_item_commentrss', $this->item['wfw']['commentrss']); | |
| 167 | - endif; | |
| 229 | + // Human-readable comment URI | |
| 230 | + $commentLink = apply_filters('syndicated_item_comments', $this->comment_link(), $this); | |
| 231 | + if (!is_null($commentLink)) : $this->post['meta']['rss:comments'] = $commentLink; endif; | |
| 168 | 232 | |
| 169 | - // Atom 1.0 comment feeds link-rel | |
| 170 | - if (isset($this->item['link_replies'])) : | |
| 171 | - // There may be multiple <link rel="replies"> elements; feeds have a feed MIME type | |
| 172 | - $N = isset($this->item['link_replies#']) ? $this->item['link_replies#'] : 1; | |
| 173 | - for ($i = 1; $i <= $N; $i++) : | |
| 174 | - $currentElement = 'link_replies'.(($i > 1) ? '#'.$i : ''); | |
| 175 | - if (isset($this->item[$currentElement.'@type']) | |
| 176 | - and preg_match("\007application/(atom|rss|rdf)\+xml\007i", $this->item[$currentElement.'@type'])) : | |
| 177 | - $this->post['meta']['wfw:commentRSS'] = apply_filters('syndicated_item_commentrss', $this->item[$currentElement]); | |
| 178 | - endif; | |
| 179 | - endfor; | |
| 180 | - endif; | |
| 233 | + // Machine-readable content feed URI | |
| 234 | + $commentFeed = apply_filters('syndicated_item_commentrss', $this->comment_feed(), $this); | |
| 235 | + if (!is_null($commentFeed)) : $this->post['meta']['wfw:commentRSS'] = $commentFeed; endif; | |
| 236 | + // Yeah, yeah, now I know that it's supposed to be | |
| 237 | + // wfw:commentRss. Oh well. Path dependence, sucka. | |
| 181 | 238 | |
| 182 | 239 | // Store information to identify the feed that this came from |
| 183 | - $this->post['meta']['syndication_feed'] = $this->feedmeta['link/uri']; | |
| 184 | - $this->post['meta']['syndication_feed_id'] = $this->feedmeta['link/id']; | |
| 240 | + if (isset($this->feedmeta['link/uri'])) : | |
| 241 | + $this->post['meta']['syndication_feed'] = $this->feedmeta['link/uri']; | |
| 242 | + endif; | |
| 243 | + if (isset($this->feedmeta['link/id'])) : | |
| 244 | + $this->post['meta']['syndication_feed_id'] = $this->feedmeta['link/id']; | |
| 245 | + endif; | |
| 185 | 246 | |
| 186 | 247 | if (isset($this->item['source_link_self'])) : |
| 187 | 248 | $this->post['meta']['syndication_feed_original'] = $this->item['source_link_self']; |
| 188 | 249 | endif; |
| @@ -187,20 +248,10 @@ | ||
| 187 | 248 | $this->post['meta']['syndication_feed_original'] = $this->item['source_link_self']; |
| 188 | 249 | endif; |
| 189 | 250 | |
| 190 | 251 | // In case you want to know the external permalink... |
| 191 | - if (isset($this->item['link'])) : | |
| 192 | - $permalink = $this->item['link']; | |
| 252 | + $this->post['meta']['syndication_permalink'] = apply_filters('syndicated_item_link', $this->permalink()); | |
| 193 | 253 | |
| 194 | - // No <link> element. See if this feed has <guid isPermalink="true"> .... | |
| 195 | - elseif (isset($this->item['guid'])) : | |
| 196 | - if (isset($this->item['guid@ispermalink']) and strtolower(trim($this->item['guid@ispermalink'])) != 'false') : | |
| 197 | - $permalink = $this->item['guid']; | |
| 198 | - endif; | |
| 199 | - endif; | |
| 200 | - | |
| 201 | - $this->post['meta']['syndication_permalink'] = apply_filters('syndicated_item_link', $permalink); | |
| 202 | - | |
| 203 | 254 | // Store a hash of the post content for checking whether something needs to be updated |
| 204 | 255 | $this->post['meta']['syndication_item_hash'] = $this->update_hash(); |
| 205 | 256 | |
| 206 | 257 | // Feed-by-feed options for author and category creation |
| @@ -248,14 +299,685 @@ | ||
| 248 | 299 | $this->post['tags_input'] = array_merge($this->post['tags_input'], $this->feedmeta['tags']); |
| 249 | 300 | endif; |
| 250 | 301 | $this->post['tags_input'] = apply_filters('syndicated_item_tags', $this->post['tags_input'], $this); |
| 251 | 302 | endif; |
| 252 | - } // SyndicatedPost::SyndicatedPost() | |
| 303 | + } /* SyndicatedPost::SyndicatedPost() */ | |
| 253 | 304 | |
| 305 | + ##################################### | |
| 306 | + #### EXTRACT DATA FROM FEED ITEM #### | |
| 307 | + ##################################### | |
| 308 | + | |
| 309 | + /** | |
| 310 | + * SyndicatedPost::query uses an XPath-like syntax to query arbitrary | |
| 311 | + * elements within the syndicated item. | |
| 312 | + * | |
| 313 | + * @param string $path | |
| 314 | + * @returns array of string values representing contents of matching | |
| 315 | + * elements or attributes | |
| 316 | + */ | |
| 317 | + function query ($path) { | |
| 318 | + $urlHash = array(); | |
| 319 | + | |
| 320 | + // Allow {url} notation for namespaces. URLs will contain : and /, so... | |
| 321 | + preg_match_all('/{([^}]+)}/', $path, $match, PREG_SET_ORDER); | |
| 322 | + foreach ($match as $ref) : | |
| 323 | + $urlHash[md5($ref[1])] = $ref[1]; | |
| 324 | + endforeach; | |
| 325 | + | |
| 326 | + foreach ($urlHash as $hash => $url) : | |
| 327 | + $path = str_replace('{'.$url.'}', '{#'.$hash.'}', $path); | |
| 328 | + endforeach; | |
| 329 | + | |
| 330 | + $path = explode('/', $path); | |
| 331 | + foreach ($path as $index => $node) : | |
| 332 | + if (preg_match('/{#([^}]+)}/', $node, $ref)) : | |
| 333 | + if (isset($urlHash[$ref[1]])) : | |
| 334 | + $path[$index] = str_replace( | |
| 335 | + '{#'.$ref[1].'}', | |
| 336 | + '{'.$urlHash[$ref[1]].'}', | |
| 337 | + $node | |
| 338 | + ); | |
| 339 | + endif; | |
| 340 | + endif; | |
| 341 | + endforeach; | |
| 342 | + | |
| 343 | + // Start out with a get_item_tags query. | |
| 344 | + $node = ''; | |
| 345 | + while (strlen($node)==0 and !is_null($node)) : | |
| 346 | + $node = array_shift($path); | |
| 347 | + endwhile; | |
| 348 | + | |
| 349 | + switch ($node) : | |
| 350 | + case 'feed' : | |
| 351 | + case 'channel' : | |
| 352 | + $method = "get_${node}_tags"; | |
| 353 | + $node = array_shift($path); | |
| 354 | + break; | |
| 355 | + case 'item' : | |
| 356 | + $node = array_shift($path); | |
| 357 | + default : | |
| 358 | + $method = NULL; | |
| 359 | + endswitch; | |
| 360 | + | |
| 361 | + $data = array(); | |
| 362 | + if (!is_null($node)) : | |
| 363 | + list($namespaces, $element) = $this->xpath_extended_name($node); | |
| 364 | + | |
| 365 | + $matches = array(); | |
| 366 | + foreach ($namespaces as $ns) : | |
| 367 | + if (!is_null($method)) : | |
| 368 | + $el = $this->link->simplepie->{$method}($ns, $element); | |
| 369 | + else : | |
| 370 | + $el = $this->entry->get_item_tags($ns, $element); | |
| 371 | + endif; | |
| 372 | + | |
| 373 | + if (!is_null($el)) : | |
| 374 | + $matches = array_merge($matches, $el); | |
| 375 | + endif; | |
| 376 | + endforeach; | |
| 377 | + $data = $matches; | |
| 378 | + | |
| 379 | + $node = array_shift($path); | |
| 380 | + endif; | |
| 381 | + | |
| 382 | + while (!is_null($node)) : | |
| 383 | + if (strlen($node) > 0) : | |
| 384 | + $matches = array(); | |
| 385 | + | |
| 386 | + list($ns, $element) = $this->xpath_extended_name($node); | |
| 387 | + | |
| 388 | + if (preg_match('/^@(.*)$/', $element, $ref)) : | |
| 389 | + $element = $ref[1]; | |
| 390 | + $axis = 'attribs'; | |
| 391 | + else : | |
| 392 | + $axis = 'child'; | |
| 393 | + endif; | |
| 394 | + | |
| 395 | + foreach ($data as $datum) : | |
| 396 | + foreach ($namespaces as $ns) : | |
| 397 | + if (!is_string($datum) | |
| 398 | + and isset($datum[$axis][$ns][$element])) : | |
| 399 | + if (is_string($datum[$axis][$ns][$element])) : | |
| 400 | + $matches[] = $datum[$axis][$ns][$element]; | |
| 401 | + else : | |
| 402 | + $matches = array_merge($matches, $datum[$axis][$ns][$element]); | |
| 403 | + endif; | |
| 404 | + endif; | |
| 405 | + endforeach; | |
| 406 | + endforeach; | |
| 407 | + | |
| 408 | + $data = $matches; | |
| 409 | + endif; | |
| 410 | + $node = array_shift($path); | |
| 411 | + endwhile; | |
| 412 | + | |
| 413 | + $matches = array(); | |
| 414 | + foreach ($data as $datum) : | |
| 415 | + if (is_string($datum)) : | |
| 416 | + $matches[] = $datum; | |
| 417 | + elseif (isset($datum['data'])) : | |
| 418 | + $matches[] = $datum['data']; | |
| 419 | + endif; | |
| 420 | + endforeach; | |
| 421 | + return $matches; | |
| 422 | + } /* SyndicatedPost::query() */ | |
| 423 | + | |
| 424 | + function xpath_default_namespace () { | |
| 425 | + // Get the default namespace. | |
| 426 | + $type = $this->link->simplepie->get_type(); | |
| 427 | + if ($type & SIMPLEPIE_TYPE_ATOM_10) : | |
| 428 | + $defaultNS = SIMPLEPIE_NAMESPACE_ATOM_10; | |
| 429 | + elseif ($type & SIMPLEPIE_TYPE_ATOM_03) : | |
| 430 | + $defaultNS = SIMPLEPIE_NAMESPACE_ATOM_03; | |
| 431 | + elseif ($type & SIMPLEPIE_TYPE_RSS_090) : | |
| 432 | + $defaultNS = SIMPLEPIE_NAMESPACE_RSS_090; | |
| 433 | + elseif ($type & SIMPLEPIE_TYPE_RSS_10) : | |
| 434 | + $defaultNS = SIMPLEPIE_NAMESPACE_RSS_10; | |
| 435 | + elseif ($type & SIMPLEPIE_TYPE_RSS_20) : | |
| 436 | + $defaultNS = SIMPLEPIE_NAMESPACE_RSS_20; | |
| 437 | + else : | |
| 438 | + $defaultNS = SIMPLEPIE_NAMESPACE_RSS_20; | |
| 439 | + endif; | |
| 440 | + return $defaultNS; | |
| 441 | + } /* SyndicatedPost::xpath_default_namespace() */ | |
| 442 | + | |
| 443 | + function xpath_extended_name ($node) { | |
| 444 | + $ns = NULL; $element = NULL; | |
| 445 | + | |
| 446 | + if (substr($node, 0, 1)=='@') : | |
| 447 | + $attr = '@'; $node = substr($node, 1); | |
| 448 | + else : | |
| 449 | + $attr = ''; | |
| 450 | + endif; | |
| 451 | + | |
| 452 | + if (preg_match('/^{([^}]*)}(.*)$/', $node, $ref)) : | |
| 453 | + $ns = array($ref[1]); $element = $ref[2]; | |
| 454 | + elseif (strpos($node, ':') !== FALSE) : | |
| 455 | + list($xmlns, $element) = explode(':', $node, 2); | |
| 456 | + if (isset($this->xmlns['reverse'][$xmlns])) : | |
| 457 | + $ns = $this->xmlns['reverse'][$xmlns]; | |
| 458 | + else : | |
| 459 | + $ns = array($xmlns); | |
| 460 | + endif; | |
| 461 | + | |
| 462 | + // Fucking SimplePie. For attributes in default xmlns. | |
| 463 | + if ($xmlns==$this->xmlns['forward'][$defaultNS[0]]) : | |
| 464 | + $ns[] = ''; | |
| 465 | + endif; | |
| 466 | + else : | |
| 467 | + // Often in SimplePie, the default namespace gets stored | |
| 468 | + // as an empty string rather than a URL. | |
| 469 | + $ns = array($this->xpath_default_namespace(), ''); | |
| 470 | + $element = $node; | |
| 471 | + endif; | |
| 472 | + return array(array_unique($ns), $attr.$element); | |
| 473 | + } /* SyndicatedPost::xpath_extended_name () */ | |
| 474 | + | |
| 475 | + function content () { | |
| 476 | + $content = NULL; | |
| 477 | + if (isset($this->item['atom_content'])) : | |
| 478 | + $content = $this->item['atom_content']; | |
| 479 | + elseif (isset($this->item['xhtml']['body'])) : | |
| 480 | + $content = $this->item['xhtml']['body']; | |
| 481 | + elseif (isset($this->item['xhtml']['div'])) : | |
| 482 | + $content = $this->item['xhtml']['div']; | |
| 483 | + elseif (isset($this->item['content']['encoded']) and $this->item['content']['encoded']): | |
| 484 | + $content = $this->item['content']['encoded']; | |
| 485 | + else: | |
| 486 | + $content = $this->item['description']; | |
| 487 | + endif; | |
| 488 | + return $content; | |
| 489 | + } /* SyndicatedPost::content() */ | |
| 490 | + | |
| 491 | + function excerpt () { | |
| 492 | + # Identify and sanitize excerpt: atom:summary, or rss:description | |
| 493 | + $excerpt = $this->entry->get_description(); | |
| 494 | + | |
| 495 | + # Many RSS feeds use rss:description, inadvisably, to | |
| 496 | + # carry the entire post (typically with escaped HTML). | |
| 497 | + # If that's what happened, we don't want the full | |
| 498 | + # content for the excerpt. | |
| 499 | + $content = $this->content(); | |
| 500 | + | |
| 501 | + // Ignore whitespace, case, and tag cruft. | |
| 502 | + $theExcerpt = preg_replace('/\s+/', '', strtolower(strip_tags($excerpt))); | |
| 503 | + $theContent = preg_replace('/\s+/', '', strtolower(strip_Tags($content))); | |
| 504 | + | |
| 505 | + if ( empty($excerpt) or $theExcerpt == $theContent ) : | |
| 506 | + # If content is available, generate an excerpt. | |
| 507 | + if ( strlen(trim($content)) > 0 ) : | |
| 508 | + $excerpt = strip_tags($content); | |
| 509 | + if (strlen($excerpt) > 255) : | |
| 510 | + $excerpt = substr($excerpt,0,252).'...'; | |
| 511 | + endif; | |
| 512 | + endif; | |
| 513 | + endif; | |
| 514 | + return $excerpt; | |
| 515 | + } /* SyndicatedPost::excerpt() */ | |
| 516 | + | |
| 517 | + function permalink () { | |
| 518 | + // Handles explicit <link> elements and also RSS 2.0 cases with | |
| 519 | + // <guid isPermaLink="true">, etc. Hooray! | |
| 520 | + $permalink = $this->entry->get_link(); | |
| 521 | + return $permalink; | |
| 522 | + } | |
| 523 | + | |
| 524 | + function created () { | |
| 525 | + $date = ''; | |
| 526 | + if (isset($this->item['dc']['created'])) : | |
| 527 | + $date = $this->item['dc']['created']; | |
| 528 | + elseif (isset($this->item['dcterms']['created'])) : | |
| 529 | + $date = $this->item['dcterms']['created']; | |
| 530 | + elseif (isset($this->item['created'])): // Atom 0.3 | |
| 531 | + $date = $this->item['created']; | |
| 532 | + endif; | |
| 533 | + | |
| 534 | + $epoch = new FeedTime($date); | |
| 535 | + return $epoch->timestamp(); | |
| 536 | + } /* SyndicatedPost::created() */ | |
| 537 | + | |
| 538 | + function published ($fallback = true, $default = NULL) { | |
| 539 | + $date = ''; | |
| 540 | + | |
| 541 | + # RSS is a fucking mess. Figure out whether we have a date in | |
| 542 | + # <dc:date>, <issued>, <pubDate>, etc., and get it into Unix | |
| 543 | + # epoch format for reformatting. If we can't find anything, | |
| 544 | + # we'll use the last-updated time. | |
| 545 | + if (isset($this->item['dc']['date'])): // Dublin Core | |
| 546 | + $date = $this->item['dc']['date']; | |
| 547 | + elseif (isset($this->item['dcterms']['issued'])) : // Dublin Core extensions | |
| 548 | + $date = $this->item['dcterms']['issued']; | |
| 549 | + elseif (isset($this->item['published'])) : // Atom 1.0 | |
| 550 | + $date = $this->item['published']; | |
| 551 | + elseif (isset($this->item['issued'])): // Atom 0.3 | |
| 552 | + $date = $this->item['issued']; | |
| 553 | + elseif (isset($this->item['pubdate'])): // RSS 2.0 | |
| 554 | + $date = $this->item['pubdate']; | |
| 555 | + endif; | |
| 556 | + | |
| 557 | + if (strlen($date) > 0) : | |
| 558 | + $time = new FeedTime($date); | |
| 559 | + $epoch = $time->timestamp(); | |
| 560 | + elseif ($fallback) : // Fall back to <updated> / <modified> if present | |
| 561 | + $epoch = $this->updated(/*fallback=*/ false, /*default=*/ $default); | |
| 562 | + endif; | |
| 563 | + | |
| 564 | + # If everything failed, then default to the current time. | |
| 565 | + if (is_null($epoch)) : | |
| 566 | + if (-1 == $default) : | |
| 567 | + $epoch = time(); | |
| 568 | + else : | |
| 569 | + $epoch = $default; | |
| 570 | + endif; | |
| 571 | + endif; | |
| 572 | + | |
| 573 | + return $epoch; | |
| 574 | + } /* SyndicatedPost::published() */ | |
| 575 | + | |
| 576 | + function updated ($fallback = true, $default = -1) { | |
| 577 | + $date = ''; | |
| 578 | + | |
| 579 | + # As far as I know, only dcterms and Atom have reliable ways to | |
| 580 | + # specify when something was *modified* last. If neither is | |
| 581 | + # available, then we'll try to get the time of publication. | |
| 582 | + if (isset($this->item['dc']['modified'])) : // Not really correct | |
| 583 | + $date = $this->item['dc']['modified']; | |
| 584 | + elseif (isset($this->item['dcterms']['modified'])) : // Dublin Core extensions | |
| 585 | + $date = $this->item['dcterms']['modified']; | |
| 586 | + elseif (isset($this->item['modified'])): // Atom 0.3 | |
| 587 | + $date = $this->item['modified']; | |
| 588 | + elseif (isset($this->item['updated'])): // Atom 1.0 | |
| 589 | + $date = $this->item['updated']; | |
| 590 | + endif; | |
| 591 | + | |
| 592 | + if (strlen($date) > 0) : | |
| 593 | + $time = new FeedTime($date); | |
| 594 | + $epoch = $time->timestamp(); | |
| 595 | + elseif ($fallback) : // Fall back to issued / dc:date | |
| 596 | + $epoch = $this->published(/*fallback=*/ false, /*default=*/ $default); | |
| 597 | + endif; | |
| 598 | + | |
| 599 | + # If everything failed, then default to the current time. | |
| 600 | + if (is_null($epoch)) : | |
| 601 | + if (-1 == $default) : | |
| 602 | + $epoch = time(); | |
| 603 | + else : | |
| 604 | + $epoch = $default; | |
| 605 | + endif; | |
| 606 | + endif; | |
| 607 | + | |
| 608 | + return $epoch; | |
| 609 | + } /* SyndicatedPost::updated() */ | |
| 610 | + | |
| 611 | + function update_hash () { | |
| 612 | + return md5(serialize($this->item)); | |
| 613 | + } /* SyndicatedPost::update_hash() */ | |
| 614 | + | |
| 615 | + function guid () { | |
| 616 | + $guid = null; | |
| 617 | + if (isset($this->item['id'])): // Atom 0.3 / 1.0 | |
| 618 | + $guid = $this->item['id']; | |
| 619 | + elseif (isset($this->item['atom']['id'])) : // Namespaced Atom | |
| 620 | + $guid = $this->item['atom']['id']; | |
| 621 | + elseif (isset($this->item['guid'])) : // RSS 2.0 | |
| 622 | + $guid = $this->item['guid']; | |
| 623 | + elseif (isset($this->item['dc']['identifier'])) :// yeah, right | |
| 624 | + $guid = $this->item['dc']['identifier']; | |
| 625 | + else : | |
| 626 | + // The feed does not seem to have provided us with a | |
| 627 | + // unique identifier, so we'll have to cobble together | |
| 628 | + // a tag: URI that might work for us. The base of the | |
| 629 | + // URI will be the host name of the feed source ... | |
| 630 | + $bits = parse_url($this->feedmeta['link/uri']); | |
| 631 | + $guid = 'tag:'.$bits['host']; | |
| 632 | + | |
| 633 | + // If we have a date of creation, then we can use that | |
| 634 | + // to uniquely identify the item. (On the other hand, if | |
| 635 | + // the feed producer was consicentious enough to | |
| 636 | + // generate dates of creation, she probably also was | |
| 637 | + // conscientious enough to generate unique identifiers.) | |
| 638 | + if (!is_null($this->created())) : | |
| 639 | + $guid .= '://post.'.date('YmdHis', $this->created()); | |
| 640 | + | |
| 641 | + // Otherwise, use both the URI of the item, *and* the | |
| 642 | + // item's title. We have to use both because titles are | |
| 643 | + // often not unique, and sometimes links aren't unique | |
| 644 | + // either (e.g. Bitch (S)HITLIST, Mozilla Dot Org news, | |
| 645 | + // some podcasts). But it's rare to have *both* the same | |
| 646 | + // title *and* the same link for two different items. So | |
| 647 | + // this is about the best we can do. | |
| 648 | + else : | |
| 649 | + $guid .= '://'.md5($this->item['link'].'/'.$this->item['title']); | |
| 650 | + endif; | |
| 651 | + endif; | |
| 652 | + return $guid; | |
| 653 | + } /* SyndicatedPost::guid() */ | |
| 654 | + | |
| 655 | + function author () { | |
| 656 | + $author = array (); | |
| 657 | + | |
| 658 | + if (isset($this->item['author_name'])): | |
| 659 | + $author['name'] = $this->item['author_name']; | |
| 660 | + elseif (isset($this->item['dc']['creator'])): | |
| 661 | + $author['name'] = $this->item['dc']['creator']; | |
| 662 | + elseif (isset($this->item['dc']['contributor'])): | |
| 663 | + $author['name'] = $this->item['dc']['contributor']; | |
| 664 | + elseif (isset($this->feed->channel['dc']['creator'])) : | |
| 665 | + $author['name'] = $this->feed->channel['dc']['creator']; | |
| 666 | + elseif (isset($this->feed->channel['dc']['contributor'])) : | |
| 667 | + $author['name'] = $this->feed->channel['dc']['contributor']; | |
| 668 | + elseif (isset($this->feed->channel['author_name'])) : | |
| 669 | + $author['name'] = $this->feed->channel['author_name']; | |
| 670 | + elseif ($this->feed->is_rss() and isset($this->item['author'])) : | |
| 671 | + // The author element in RSS is allegedly an | |
| 672 | + // e-mail address, but lots of people don't use | |
| 673 | + // it that way. So let's make of it what we can. | |
| 674 | + $author = parse_email_with_realname($this->item['author']); | |
| 675 | + | |
| 676 | + if (!isset($author['name'])) : | |
| 677 | + if (isset($author['email'])) : | |
| 678 | + $author['name'] = $author['email']; | |
| 679 | + else : | |
| 680 | + $author['name'] = $this->feed->channel['title']; | |
| 681 | + endif; | |
| 682 | + endif; | |
| 683 | + else : | |
| 684 | + $author['name'] = $this->feed->channel['title']; | |
| 685 | + endif; | |
| 686 | + | |
| 687 | + if (isset($this->item['author_email'])): | |
| 688 | + $author['email'] = $this->item['author_email']; | |
| 689 | + elseif (isset($this->feed->channel['author_email'])) : | |
| 690 | + $author['email'] = $this->feed->channel['author_email']; | |
| 691 | + endif; | |
| 692 | + | |
| 693 | + if (isset($this->item['author_url'])): | |
| 694 | + $author['uri'] = $this->item['author_url']; | |
| 695 | + elseif (isset($this->feed->channel['author_url'])) : | |
| 696 | + $author['uri'] = $this->item['author_url']; | |
| 697 | + elseif (isset($this->feed->channel['link'])) : | |
| 698 | + $author['uri'] = $this->feed->channel['link']; | |
| 699 | + endif; | |
| 700 | + | |
| 701 | + return $author; | |
| 702 | + } /* SyndicatedPost::author() */ | |
| 703 | + | |
| 704 | + /** | |
| 705 | + * SyndicatedPost::isTaggedAs: Test whether a feed item is | |
| 706 | + * tagged / categorized with a given string. Case and leading and | |
| 707 | + * trailing whitespace are ignored. | |
| 708 | + * | |
| 709 | + * @param string $tag Tag to check for | |
| 710 | + * | |
| 711 | + * @return bool Whether or not at least one of the categories / tags on | |
| 712 | + * $this->item is set to $tag (modulo case and leading and trailing | |
| 713 | + * whitespace) | |
| 714 | + */ | |
| 715 | + function isTaggedAs ($tag) { | |
| 716 | + $desiredTag = strtolower(trim($tag)); // Normalize case and whitespace | |
| 717 | + | |
| 718 | + // Check to see if this is tagged with $tag | |
| 719 | + $currentCategory = 'category'; | |
| 720 | + $currentCategoryNumber = 1; | |
| 721 | + | |
| 722 | + // If we have the new MagpieRSS, the number of category elements | |
| 723 | + // on this item is stored under index "category#". | |
| 724 | + if (isset($this->item['category#'])) : | |
| 725 | + $numberOfCategories = (int) $this->item['category#']; | |
| 726 | + | |
| 727 | + // We REALLY shouldn't have the old and busted MagpieRSS, but in | |
| 728 | + // case we do, it doesn't support multiple categories, but there | |
| 729 | + // might still be a single value under the "category" index. | |
| 730 | + elseif (isset($this->item['category'])) : | |
| 731 | + $numberOfCategories = 1; | |
| 732 | + | |
| 733 | + // No standard category or tag elements on this feed item. | |
| 734 | + else : | |
| 735 | + $numberOfCategories = 0; | |
| 736 | + | |
| 737 | + endif; | |
| 738 | + | |
| 739 | + $isSoTagged = false; // Innocent until proven guilty | |
| 740 | + | |
| 741 | + // Loop through category elements; if there are multiple | |
| 742 | + // elements, they are indexed as category, category#2, | |
| 743 | + // category#3, ... category#N | |
| 744 | + while ($currentCategoryNumber <= $numberOfCategories) : | |
| 745 | + if ($desiredTag == strtolower(trim($this->item[$currentCategory]))) : | |
| 746 | + $isSoTagged = true; // Got it! | |
| 747 | + break; | |
| 748 | + endif; | |
| 749 | + | |
| 750 | + $currentCategoryNumber += 1; | |
| 751 | + $currentCategory = 'category#'.$currentCategoryNumber; | |
| 752 | + endwhile; | |
| 753 | + | |
| 754 | + return $isSoTagged; | |
| 755 | + } /* SyndicatedPost::isTaggedAs() */ | |
| 756 | + | |
| 757 | + /** | |
| 758 | + * SyndicatedPost::enclosures: returns an array with any enclosures | |
| 759 | + * that may be attached to this syndicated item. | |
| 760 | + * | |
| 761 | + * @param string $type If you only want enclosures that match a certain | |
| 762 | + * MIME type or group of MIME types, you can limit the enclosures | |
| 763 | + * that will be returned to only those with a MIME type which | |
| 764 | + * matches this regular expression. | |
| 765 | + * @return array | |
| 766 | + */ | |
| 767 | + function enclosures ($type = '/.*/') { | |
| 768 | + $enclosures = array(); | |
| 769 | + | |
| 770 | + if (isset($this->item['enclosure#'])) : | |
| 771 | + // Loop through enclosure, enclosure#2, enclosure#3, .... | |
| 772 | + for ($i = 1; $i <= $this->item['enclosure#']; $i++) : | |
| 773 | + $eid = (($i > 1) ? "#{$id}" : ""); | |
| 774 | + | |
| 775 | + // Does it match the type we want? | |
| 776 | + if (preg_match($type, $this->item["enclosure{$eid}@type"])) : | |
| 777 | + $enclosures[] = array( | |
| 778 | + "url" => $this->item["enclosure{$eid}@url"], | |
| 779 | + "type" => $this->item["enclosure{$eid}@type"], | |
| 780 | + "length" => $this->item["enclosure{$eid}@length"], | |
| 781 | + ); | |
| 782 | + endif; | |
| 783 | + endfor; | |
| 784 | + endif; | |
| 785 | + return $enclosures; | |
| 786 | + } /* SyndicatedPost::enclosures() */ | |
| 787 | + | |
| 788 | + function comment_link () { | |
| 789 | + $url = null; | |
| 790 | + | |
| 791 | + // RSS 2.0 has a standard <comments> element: | |
| 792 | + // "<comments> is an optional sub-element of <item>. If present, | |
| 793 | + // it is the url of the comments page for the item." | |
| 794 | + // <http://cyber.law.harvard.edu/rss/rss.html#ltcommentsgtSubelementOfLtitemgt> | |
| 795 | + if (isset($this->item['comments'])) : | |
| 796 | + $url = $this->item['comments']; | |
| 797 | + endif; | |
| 798 | + | |
| 799 | + // The convention in Atom feeds is to use a standard <link> | |
| 800 | + // element with @rel="replies" and @type="text/html". | |
| 801 | + // Unfortunately, SimplePie_Item::get_links() allows us to filter | |
| 802 | + // by the value of @rel, but not by the value of @type. *sigh* | |
| 803 | + | |
| 804 | + // Try Atom 1.0 first | |
| 805 | + $linkElements = $this->entry->get_item_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'link'); | |
| 806 | + | |
| 807 | + // Fall back and try Atom 0.3 | |
| 808 | + if (is_null($linkElements)) : $linkElements = $this->entry->get_item_tags(SIMPLEPIE_NAMESPACE_ATOM_03, 'link'); endif; | |
| 809 | + | |
| 810 | + // Now loop through the elements, screening by @rel and @type | |
| 811 | + if (is_array($linkElements)) : foreach ($linkElements as $link) : | |
| 812 | + $rel = (isset($link['attribs']['']['rel']) ? $link['attribs']['']['rel'] : 'alternate'); | |
| 813 | + $type = (isset($link['attribs']['']['type']) ? $link['attribs']['']['type'] : NULL); | |
| 814 | + $href = (isset($link['attribs']['']['href']) ? $link['attribs']['']['href'] : NULL); | |
| 815 | + | |
| 816 | + if (strtolower($rel)=='replies' and $type=='text/html' and !is_null($href)) : | |
| 817 | + $url = $href; | |
| 818 | + endif; | |
| 819 | + endforeach; endif; | |
| 820 | + | |
| 821 | + return $url; | |
| 822 | + } | |
| 823 | + | |
| 824 | + function comment_feed () { | |
| 825 | + $feed = null; | |
| 826 | + | |
| 827 | + // Well Formed Web comment feeds extension for RSS 2.0 | |
| 828 | + // <http://www.sellsbrothers.com/spout/default.aspx?content=archive.htm#exposingRssComments> | |
| 829 | + // | |
| 830 | + // N.B.: Correct capitalization is wfw:commentRss, but | |
| 831 | + // wfw:commentRSS is common in the wild (partly due to a typo in | |
| 832 | + // the original spec). In any case, our item array is normalized | |
| 833 | + // to all lowercase anyways. | |
| 834 | + if (isset($this->item['wfw']['commentrss'])) : | |
| 835 | + $feed = $this->item['wfw']['commentrss']; | |
| 836 | + endif; | |
| 837 | + | |
| 838 | + // In Atom 1.0, the convention is to use a standard link element | |
| 839 | + // with @rel="replies". Sometimes this is also used to pass a | |
| 840 | + // link to the human-readable comments page, so we also need to | |
| 841 | + // check link/@type for a feed MIME type. | |
| 842 | + // | |
| 843 | + // Which is why I'm not using the SimplePie_Item::get_links() | |
| 844 | + // method here, incidentally: it doesn't allow you to filter by | |
| 845 | + // @type. *sigh* | |
| 846 | + if (isset($this->item['link_replies'])) : | |
| 847 | + // There may be multiple <link rel="replies"> elements; feeds have a feed MIME type | |
| 848 | + $N = isset($this->item['link_replies#']) ? $this->item['link_replies#'] : 1; | |
| 849 | + for ($i = 1; $i <= $N; $i++) : | |
| 850 | + $currentElement = 'link_replies'.(($i > 1) ? '#'.$i : ''); | |
| 851 | + if (isset($this->item[$currentElement.'@type']) | |
| 852 | + and preg_match("\007application/(atom|rss|rdf)\+xml\007i", $this->item[$currentElement.'@type'])) : | |
| 853 | + $feed = $this->item[$currentElement]; | |
| 854 | + endif; | |
| 855 | + endfor; | |
| 856 | + endif; | |
| 857 | + return $feed; | |
| 858 | + } /* SyndicatedPost::comment_feed() */ | |
| 859 | + | |
| 860 | + ################################## | |
| 861 | + #### BUILT-IN CONTENT FILTERS #### | |
| 862 | + ################################## | |
| 863 | + | |
| 864 | + var $uri_attrs = array ( | |
| 865 | + array('a', 'href'), | |
| 866 | + array('applet', 'codebase'), | |
| 867 | + array('area', 'href'), | |
| 868 | + array('blockquote', 'cite'), | |
| 869 | + array('body', 'background'), | |
| 870 | + array('del', 'cite'), | |
| 871 | + array('form', 'action'), | |
| 872 | + array('frame', 'longdesc'), | |
| 873 | + array('frame', 'src'), | |
| 874 | + array('iframe', 'longdesc'), | |
| 875 | + array('iframe', 'src'), | |
| 876 | + array('head', 'profile'), | |
| 877 | + array('img', 'longdesc'), | |
| 878 | + array('img', 'src'), | |
| 879 | + array('img', 'usemap'), | |
| 880 | + array('input', 'src'), | |
| 881 | + array('input', 'usemap'), | |
| 882 | + array('ins', 'cite'), | |
| 883 | + array('link', 'href'), | |
| 884 | + array('object', 'classid'), | |
| 885 | + array('object', 'codebase'), | |
| 886 | + array('object', 'data'), | |
| 887 | + array('object', 'usemap'), | |
| 888 | + array('q', 'cite'), | |
| 889 | + array('script', 'src') | |
| 890 | + ); /* var SyndicatedPost::$uri_attrs */ | |
| 891 | + | |
| 892 | + var $_base = null; | |
| 893 | + | |
| 894 | + function resolve_single_relative_uri ($refs) { | |
| 895 | + $tag = FeedWordPressHTML::attributeMatch($refs); | |
| 896 | + $url = Relative_URI::resolve($tag['value'], $this->_base); | |
| 897 | + return $tag['prefix'] . $url . $tag['suffix']; | |
| 898 | + } /* function SyndicatedPost::resolve_single_relative_uri() */ | |
| 899 | + | |
| 900 | + function resolve_relative_uris ($content, $obj) { | |
| 901 | + $set = $obj->link->setting('resolve relative', 'resolve_relative', 'yes'); | |
| 902 | + if ($set and $set != 'no') : | |
| 903 | + // Fallback: if we don't have anything better, use the | |
| 904 | + // item link from the feed | |
| 905 | + $obj->_base = $obj->item['link']; // Reset the base for resolving relative URIs | |
| 906 | + | |
| 907 | + // What we should do here, properly, is to use | |
| 908 | + // SimplePie_Item::get_base() -- but that method is | |
| 909 | + // currently broken. Or getting down and dirty in the | |
| 910 | + // SimplePie representation of the content tags and | |
| 911 | + // grabbing the xml_base member for the content element. | |
| 912 | + // Maybe someday... | |
| 913 | + | |
| 914 | + foreach ($obj->uri_attrs as $pair) : | |
| 915 | + list($tag, $attr) = $pair; | |
| 916 | + $pattern = FeedWordPressHTML::attributeRegex($tag, $attr); | |
| 917 | + $content = preg_replace_callback ( | |
| 918 | + $pattern, | |
| 919 | + array(&$obj, 'resolve_single_relative_uri'), | |
| 920 | + $content | |
| 921 | + ); | |
| 922 | + endforeach; | |
| 923 | + endif; | |
| 924 | + | |
| 925 | + return $content; | |
| 926 | + } /* function SyndicatedPost::resolve_relative_uris () */ | |
| 927 | + | |
| 928 | + var $strip_attrs = array ( | |
| 929 | + array('[a-z]+', 'target'), | |
| 930 | +// array('[a-z]+', 'style'), | |
| 931 | +// array('[a-z]+', 'on[a-z]+'), | |
| 932 | + ); | |
| 933 | + | |
| 934 | + function strip_attribute_from_tag ($refs) { | |
| 935 | + $tag = FeedWordPressHTML::attributeMatch($refs); | |
| 936 | + return $tag['before_attribute'].$tag['after_attribute']; | |
| 937 | + } | |
| 938 | + | |
| 939 | + function sanitize_content ($content, $obj) { | |
| 940 | + # This kind of sucks. I intend to replace it with | |
| 941 | + # lib_filter sometime soon. | |
| 942 | + foreach ($obj->strip_attrs as $pair): | |
| 943 | + list($tag,$attr) = $pair; | |
| 944 | + $pattern = FeedWordPressHTML::attributeRegex($tag, $attr); | |
| 945 | + | |
| 946 | + $content = preg_replace_callback ( | |
| 947 | + $pattern, | |
| 948 | + array(&$obj, 'strip_attribute_from_tag'), | |
| 949 | + $content | |
| 950 | + ); | |
| 951 | + endforeach; | |
| 952 | + return $content; | |
| 953 | + } /* SyndicatedPost::sanitize() */ | |
| 954 | + | |
| 955 | + ##################### | |
| 956 | + #### POST STATUS #### | |
| 957 | + ##################### | |
| 958 | + | |
| 959 | + /** | |
| 960 | + * SyndicatedPost::filtered: check whether or not this post has been | |
| 961 | + * screened out by a registered filter. | |
| 962 | + * | |
| 963 | + * @return bool TRUE iff post has been filtered out by a previous filter | |
| 964 | + */ | |
| 254 | 965 | function filtered () { |
| 255 | 966 | return is_null($this->post); |
| 256 | - } | |
| 967 | + } /* SyndicatedPost::filtered() */ | |
| 257 | 968 | |
| 969 | + /** | |
| 970 | + * SyndicatedPost::freshness: check whether post is a new post to be | |
| 971 | + * inserted, a previously syndicated post that needs to be updated to | |
| 972 | + * match the latest revision, or a previously syndicated post that is | |
| 973 | + * still up-to-date. | |
| 974 | + * | |
| 975 | + * @return int A status code representing the freshness of the post | |
| 976 | + * 0 = post already syndicated; no update needed | |
| 977 | + * 1 = post already syndicated, but needs to be updated to latest | |
| 978 | + * 2 = post has not yet been syndicated; needs to be created | |
| 979 | + */ | |
| 258 | 980 | function freshness () { |
| 259 | 981 | global $wpdb; |
| 260 | 982 | |
| 261 | 983 | if ($this->filtered()) : // This should never happen. |
| @@ -277,9 +999,9 @@ | ||
| 277 | 999 | if (count($stored_update_hashes) > 0) : |
| 278 | 1000 | $stored_update_hash = $stored_update_hashes[0]; |
| 279 | 1001 | $update_hash_changed = ($stored_update_hash != $this->update_hash()); |
| 280 | 1002 | else : |
| 281 | - $update_hash_changed = false; | |
| 1003 | + $update_hash_changed = true; // Can't find syndication meta-data | |
| 282 | 1004 | endif; |
| 283 | 1005 | |
| 284 | 1006 | preg_match('/([0-9]+)-([0-9]+)-([0-9]+) ([0-9]+):([0-9]+):([0-9]+)/', $result->post_modified_gmt, $backref); |
| 285 | 1007 | |
| @@ -284,13 +1006,30 @@ | ||
| 284 | 1006 | preg_match('/([0-9]+)-([0-9]+)-([0-9]+) ([0-9]+):([0-9]+):([0-9]+)/', $result->post_modified_gmt, $backref); |
| 285 | 1007 | |
| 286 | 1008 | $last_rev_ts = gmmktime($backref[4], $backref[5], $backref[6], $backref[2], $backref[3], $backref[1]); |
| 287 | 1009 | $updated_ts = $this->updated(/*fallback=*/ true, /*default=*/ NULL); |
| 288 | - $updated = (( | |
| 1010 | + | |
| 1011 | + $frozen_values = get_post_custom_values('_syndication_freeze_updates', $result->id); | |
| 1012 | + $frozen_post = (count($frozen_values) > 0 and 'yes' == $frozen_values[0]); | |
| 1013 | + $frozen_feed = ('yes' == $this->link->setting('freeze updates', 'freeze_updates', NULL)); | |
| 1014 | + | |
| 1015 | + // Check timestamps... | |
| 1016 | + $updated = ( | |
| 289 | 1017 | !is_null($updated_ts) |
| 290 | 1018 | and ($updated_ts > $last_rev_ts) |
| 291 | - ) or $update_hash_changed); | |
| 292 | - | |
| 1019 | + ); | |
| 1020 | + | |
| 1021 | + | |
| 1022 | + // Or the hash... | |
| 1023 | + $updated = ($updated or $update_hash_changed); | |
| 1024 | + | |
| 1025 | + // But only if the post is not frozen. | |
| 1026 | + $updated = ( | |
| 1027 | + $updated | |
| 1028 | + and !$frozen_post | |
| 1029 | + and !$frozen_feed | |
| 1030 | + ); | |
| 1031 | + | |
| 293 | 1032 | if ($updated) : |
| 294 | 1033 | $this->_freshness = 1; // Updated content |
| 295 | 1034 | $this->_wp_id = $result->id; |
| 296 | 1035 | else : |
| @@ -301,8 +1040,12 @@ | ||
| 301 | 1040 | endif; |
| 302 | 1041 | return $this->_freshness; |
| 303 | 1042 | } |
| 304 | 1043 | |
| 1044 | + ################################################# | |
| 1045 | + #### INTERNAL STORAGE AND MANAGEMENT METHODS #### | |
| 1046 | + ################################################# | |
| 1047 | + | |
| 305 | 1048 | function wp_id () { |
| 306 | 1049 | if ($this->filtered()) : // This should never happen. |
| 307 | 1050 | FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__); |
| 308 | 1051 | endif; |
| @@ -364,30 +1107,55 @@ | ||
| 364 | 1107 | |
| 365 | 1108 | if (!$this->filtered() and $freshness > 0) : |
| 366 | 1109 | unset($this->post['named']); |
| 367 | 1110 | $this->post = apply_filters('syndicated_post', $this->post, $this); |
| 1111 | + | |
| 1112 | + // Allow for feed-specific syndicated_post filters. | |
| 1113 | + $this->post = apply_filters( | |
| 1114 | + "syndicated_post_".$this->link->uri(), | |
| 1115 | + $this->post, | |
| 1116 | + $this | |
| 1117 | + ); | |
| 368 | 1118 | endif; |
| 369 | 1119 | |
| 1120 | + // Hook in early to make sure these get inserted if at all possible | |
| 1121 | + add_action( | |
| 1122 | + /*hook=*/ 'transition_post_status', | |
| 1123 | + /*callback=*/ array(&$this, 'add_rss_meta'), | |
| 1124 | + /*priority=*/ -10000, /* very early */ | |
| 1125 | + /*arguments=*/ 3 | |
| 1126 | + ); | |
| 1127 | + | |
| 370 | 1128 | if (!$this->filtered() and $freshness == 2) : |
| 371 | 1129 | // The item has not yet been added. So let's add it. |
| 1130 | + FeedWordPress::diagnostic('syndicated_posts', 'Inserting new post "'.$this->post['post_title'].'"'); | |
| 1131 | + | |
| 372 | 1132 | $this->insert_new(); |
| 373 | - $this->add_rss_meta(); | |
| 374 | - do_action('post_syndicated_item', $this->wp_id()); | |
| 1133 | + do_action('post_syndicated_item', $this->wp_id(), $this); | |
| 375 | 1134 | |
| 376 | 1135 | $ret = 'new'; |
| 377 | 1136 | elseif (!$this->filtered() and $freshness == 1) : |
| 1137 | + FeedWordPress::diagnostic('syndicated_posts', 'Updating existing post # '.$this->wp_id().', "'.$this->post['post_title'].'"'); | |
| 1138 | + | |
| 378 | 1139 | $this->post['ID'] = $this->wp_id(); |
| 379 | 1140 | $this->update_existing(); |
| 380 | - $this->add_rss_meta(); | |
| 381 | - do_action('update_syndicated_item', $this->wp_id()); | |
| 1141 | + do_action('update_syndicated_item', $this->wp_id(), $this); | |
| 382 | 1142 | |
| 383 | 1143 | $ret = 'updated'; |
| 384 | 1144 | else : |
| 385 | 1145 | $ret = false; |
| 386 | 1146 | endif; |
| 387 | - | |
| 1147 | + | |
| 1148 | + // Remove add_rss_meta hook | |
| 1149 | + remove_action( | |
| 1150 | + /*hook=*/ 'transition_post_status', | |
| 1151 | + /*callback=*/ array(&$this, 'add_rss_meta'), | |
| 1152 | + /*priority=*/ -10000, /* very early */ | |
| 1153 | + /*arguments=*/ 3 | |
| 1154 | + ); | |
| 1155 | + | |
| 388 | 1156 | return $ret; |
| 389 | - } // function SyndicatedPost::store () | |
| 1157 | + } /* function SyndicatedPost::store () */ | |
| 390 | 1158 | |
| 391 | 1159 | function insert_new () { |
| 392 | 1160 | global $wpdb, $wp_db_version; |
| 393 | 1161 | |
| @@ -668,12 +1436,14 @@ | ||
| 668 | 1436 | // for by standard WP meta-data (i.e., any interesting data about the |
| 669 | 1437 | // syndicated post other than author, title, timestamp, categories, and |
| 670 | 1438 | // guid). It's also used to hook into WordPress's support for |
| 671 | 1439 | // enclosures. |
| 672 | - function add_rss_meta () { | |
| 1440 | + function add_rss_meta ($new_status, $old_status, $post) { | |
| 1441 | + FeedWordPress::diagnostic('syndicated_posts:meta_data', 'Adding post meta-data: {'.implode(", ", array_keys($this->post['meta'])).'}'); | |
| 1442 | + | |
| 673 | 1443 | global $wpdb; |
| 674 | 1444 | if ( is_array($this->post) and isset($this->post['meta']) and is_array($this->post['meta']) ) : |
| 675 | - $postId = $this->wp_id(); | |
| 1445 | + $postId = $post->ID; | |
| 676 | 1446 | |
| 677 | 1447 | // Aggregated posts should NOT send out pingbacks. |
| 678 | 1448 | // WordPress 2.1-2.2 claim you can tell them not to |
| 679 | 1449 | // using $post_pingback, but they don't listen, so we |
| @@ -683,35 +1453,22 @@ | ||
| 683 | 1453 | WHERE post_id='$postId' AND meta_key='_pingme' |
| 684 | 1454 | "); |
| 685 | 1455 | |
| 686 | 1456 | foreach ( $this->post['meta'] as $key => $values ) : |
| 1457 | + $eKey = $wpdb->escape($key); | |
| 687 | 1458 | |
| 688 | - $key = $wpdb->escape($key); | |
| 689 | - | |
| 690 | 1459 | // If this is an update, clear out the old |
| 691 | 1460 | // values to avoid duplication. |
| 692 | 1461 | $result = $wpdb->query(" |
| 693 | 1462 | DELETE FROM $wpdb->postmeta |
| 694 | - WHERE post_id='$postId' AND meta_key='$key' | |
| 1463 | + WHERE post_id='$postId' AND meta_key='$eKey' | |
| 695 | 1464 | "); |
| 696 | 1465 | |
| 697 | 1466 | // Allow for either a single value or an array |
| 698 | 1467 | if (!is_array($values)) $values = array($values); |
| 699 | 1468 | foreach ( $values as $value ) : |
| 700 | - $value = $wpdb->escape($value); | |
| 701 | - $result = $wpdb->query(" | |
| 702 | - INSERT INTO $wpdb->postmeta | |
| 703 | - SET | |
| 704 | - post_id='$postId', | |
| 705 | - meta_key='$key', | |
| 706 | - meta_value='$value' | |
| 707 | - "); | |
| 708 | - if (!$result) : | |
| 709 | - $err = mysql_error(); | |
| 710 | - if (FEEDWORDPRESS_DEBUG) : | |
| 711 | - echo "[DEBUG:".date('Y-m-d H:i:S')."][feedwordpress]: post metadata insertion FAILED for field '$key' := '$value': [$err]"; | |
| 712 | - endif; | |
| 713 | - endif; | |
| 1469 | + FeedWordPress::diagnostic('syndicated_posts:meta_data', "Adding post meta-datum to post [$postId]: [$key] = ".FeedWordPress::val($value, /*no newlines=*/ true)); | |
| 1470 | + add_post_meta($postId, $key, $value, /*unique=*/ false); | |
| 714 | 1471 | endforeach; |
| 715 | 1472 | endforeach; |
| 716 | 1473 | endif; |
| 717 | 1474 | } /* SyndicatedPost::add_rss_meta () */ |
| @@ -722,10 +1479,10 @@ | ||
| 722 | 1479 | global $wpdb; |
| 723 | 1480 | |
| 724 | 1481 | $a = $this->author(); |
| 725 | 1482 | $author = $a['name']; |
| 726 | - $email = $a['email']; | |
| 727 | - $url = $a['uri']; | |
| 1483 | + $email = (isset($a['email']) ? $a['email'] : NULL); | |
| 1484 | + $url = (isset($a['uri']) ? $a['uri'] : NULL); | |
| 728 | 1485 | |
| 729 | 1486 | $match_author_by_email = !('yes' == get_option("feedwordpress_do_not_match_author_by_email")); |
| 730 | 1487 | if ($match_author_by_email and !FeedWordPress::is_null_email($email)) : |
| 731 | 1488 | $test_email = $email; |
| @@ -763,65 +1520,36 @@ | ||
| 763 | 1520 | |
| 764 | 1521 | else : |
| 765 | 1522 | // Check the database for an existing author record that might fit |
| 766 | 1523 | |
| 767 | - #-- WordPress 2.0+ | |
| 768 | - if (fwp_test_wp_version(FWP_SCHEMA_HAS_USERMETA)) : | |
| 769 | - | |
| 770 | - // First try the user core data table. | |
| 1524 | + // First try the user core data table. | |
| 1525 | + $id = $wpdb->get_var( | |
| 1526 | + "SELECT ID FROM $wpdb->users | |
| 1527 | + WHERE | |
| 1528 | + TRIM(LCASE(user_login)) = TRIM(LCASE('$login')) | |
| 1529 | + OR ( | |
| 1530 | + LENGTH(TRIM(LCASE(user_email))) > 0 | |
| 1531 | + AND TRIM(LCASE(user_email)) = TRIM(LCASE('$test_email')) | |
| 1532 | + ) | |
| 1533 | + OR TRIM(LCASE(user_nicename)) = TRIM(LCASE('$nice_author')) | |
| 1534 | + "); | |
| 1535 | + | |
| 1536 | + // If that fails, look for aliases in the user meta data table | |
| 1537 | + if (is_null($id)) : | |
| 771 | 1538 | $id = $wpdb->get_var( |
| 772 | - "SELECT ID FROM $wpdb->users | |
| 1539 | + "SELECT user_id FROM $wpdb->usermeta | |
| 773 | 1540 | WHERE |
| 774 | - TRIM(LCASE(user_login)) = TRIM(LCASE('$login')) | |
| 1541 | + (meta_key = 'description' AND TRIM(LCASE(meta_value)) = TRIM(LCASE('$author'))) | |
| 775 | 1542 | OR ( |
| 776 | - LENGTH(TRIM(LCASE(user_email))) > 0 | |
| 777 | - AND TRIM(LCASE(user_email)) = TRIM(LCASE('$test_email')) | |
| 778 | - ) | |
| 779 | - OR TRIM(LCASE(user_nicename)) = TRIM(LCASE('$nice_author')) | |
| 780 | - "); | |
| 781 | - | |
| 782 | - // If that fails, look for aliases in the user meta data table | |
| 783 | - if (is_null($id)) : | |
| 784 | - $id = $wpdb->get_var( | |
| 785 | - "SELECT user_id FROM $wpdb->usermeta | |
| 786 | - WHERE | |
| 787 | - (meta_key = 'description' AND TRIM(LCASE(meta_value)) = TRIM(LCASE('$author'))) | |
| 788 | - OR ( | |
| 789 | - meta_key = 'description' | |
| 790 | - AND TRIM(LCASE(meta_value)) | |
| 791 | - RLIKE CONCAT( | |
| 792 | - '(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*', | |
| 793 | - TRIM(LCASE('$reg_author')), | |
| 794 | - '( |\\t|\\r)*(\\n|\$)' | |
| 795 | - ) | |
| 796 | - ) | |
| 797 | - "); | |
| 798 | - endif; | |
| 799 | - | |
| 800 | - #-- WordPress 1.5.x | |
| 801 | - else : | |
| 802 | - $id = $wpdb->get_var( | |
| 803 | - "SELECT ID from $wpdb->users | |
| 804 | - WHERE | |
| 805 | - TRIM(LCASE(user_login)) = TRIM(LCASE('$login')) OR | |
| 806 | - ( | |
| 807 | - LENGTH(TRIM(LCASE(user_email))) > 0 | |
| 808 | - AND TRIM(LCASE(user_email)) = TRIM(LCASE('$test_email')) | |
| 809 | - ) OR | |
| 810 | - TRIM(LCASE(user_firstname)) = TRIM(LCASE('$author')) OR | |
| 811 | - TRIM(LCASE(user_nickname)) = TRIM(LCASE('$author')) OR | |
| 812 | - TRIM(LCASE(user_nicename)) = TRIM(LCASE('$nice_author')) OR | |
| 813 | - TRIM(LCASE(user_description)) = TRIM(LCASE('$author')) OR | |
| 814 | - ( | |
| 815 | - LOWER(user_description) | |
| 1543 | + meta_key = 'description' | |
| 1544 | + AND TRIM(LCASE(meta_value)) | |
| 816 | 1545 | RLIKE CONCAT( |
| 817 | 1546 | '(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*', |
| 818 | - LCASE('$reg_author'), | |
| 1547 | + TRIM(LCASE('$reg_author')), | |
| 819 | 1548 | '( |\\t|\\r)*(\\n|\$)' |
| 820 | 1549 | ) |
| 821 | 1550 | ) |
| 822 | 1551 | "); |
| 823 | - | |
| 824 | 1552 | endif; |
| 825 | 1553 | |
| 826 | 1554 | // ... if you don't find one, then do what you need to do |
| 827 | 1555 | if (is_null($id)) : |
| @@ -827,8 +1555,20 @@ | ||
| 827 | 1555 | if (is_null($id)) : |
| 828 | 1556 | if ($unfamiliar_author === 'create') : |
| 829 | 1557 | $userdata = array(); |
| 830 | 1558 | |
| 1559 | + // WordPress 3 is going to pitch a fit if we attempt to register | |
| 1560 | + // more than one user account with an empty e-mail address, so we | |
| 1561 | + // need *something* here. Ugh. | |
| 1562 | + if (strlen($email) == 0 or FeedWordPress::is_null_email($email)) : | |
| 1563 | + $hostUrl = $this->link->homepage(); | |
| 1564 | + if (is_null($hostUrl) or (strlen($hostUrl) < 0)) : | |
| 1565 | + $hostUrl = $this->link->uri(); | |
| 1566 | + endif; | |
| 1567 | + $url = parse_url($hostUrl); | |
| 1568 | + $email = $nice_author.'@'.$url['host']; | |
| 1569 | + endif; | |
| 1570 | + | |
| 831 | 1571 | #-- user table data |
| 832 | 1572 | $userdata['ID'] = NULL; // new user |
| 833 | 1573 | $userdata['user_login'] = $login; |
| 834 | 1574 | $userdata['user_nicename'] = $nice_author; |
| @@ -835,9 +1575,9 @@ | ||
| 835 | 1575 | $userdata['user_pass'] = substr(md5(uniqid(microtime())), 0, 6); // just something random to lock it up |
| 836 | 1576 | $userdata['user_email'] = $email; |
| 837 | 1577 | $userdata['user_url'] = $url; |
| 838 | 1578 | $userdata['display_name'] = $author; |
| 839 | - | |
| 1579 | + | |
| 840 | 1580 | $id = wp_insert_user($userdata); |
| 841 | 1581 | elseif (is_numeric($unfamiliar_author) and get_userdata((int) $unfamiliar_author)) : |
| 842 | 1582 | $id = (int) $unfamiliar_author; |
| 843 | 1583 | elseif ($unfamiliar_author === 'default') : |
| @@ -968,311 +1708,6 @@ | ||
| 968 | 1708 | endswitch; |
| 969 | 1709 | return $ret; |
| 970 | 1710 | } // function SyndicatedPost::use_api () |
| 971 | 1711 | |
| 972 | - #### EXTRACT DATA FROM FEED ITEM #### | |
| 973 | - | |
| 974 | - function created () { | |
| 975 | - $epoch = null; | |
| 976 | - if (isset($this->item['dc']['created'])) : | |
| 977 | - $epoch = @parse_w3cdtf($this->item['dc']['created']); | |
| 978 | - elseif (isset($this->item['dcterms']['created'])) : | |
| 979 | - $epoch = @parse_w3cdtf($this->item['dcterms']['created']); | |
| 980 | - elseif (isset($this->item['created'])): // Atom 0.3 | |
| 981 | - $epoch = @parse_w3cdtf($this->item['created']); | |
| 982 | - endif; | |
| 983 | - return $epoch; | |
| 984 | - } | |
| 985 | - function published ($fallback = true) { | |
| 986 | - $epoch = null; | |
| 987 | - | |
| 988 | - # RSS is a fucking mess. Figure out whether we have a date in | |
| 989 | - # <dc:date>, <issued>, <pubDate>, etc., and get it into Unix | |
| 990 | - # epoch format for reformatting. If we can't find anything, | |
| 991 | - # we'll use the last-updated time. | |
| 992 | - if (isset($this->item['dc']['date'])): // Dublin Core | |
| 993 | - $epoch = @parse_w3cdtf($this->item['dc']['date']); | |
| 994 | - elseif (isset($this->item['dcterms']['issued'])) : // Dublin Core extensions | |
| 995 | - $epoch = @parse_w3cdtf($this->item['dcterms']['issued']); | |
| 996 | - elseif (isset($this->item['published'])) : // Atom 1.0 | |
| 997 | - $epoch = @parse_w3cdtf($this->item['published']); | |
| 998 | - elseif (isset($this->item['issued'])): // Atom 0.3 | |
| 999 | - $epoch = @parse_w3cdtf($this->item['issued']); | |
| 1000 | - elseif (isset($this->item['pubdate'])): // RSS 2.0 | |
| 1001 | - $epoch = strtotime($this->item['pubdate']); | |
| 1002 | - elseif ($fallback) : // Fall back to <updated> / <modified> if present | |
| 1003 | - $epoch = $this->updated(/*fallback=*/ false); | |
| 1004 | - endif; | |
| 1005 | - | |
| 1006 | - # If everything failed, then default to the current time. | |
| 1007 | - if (is_null($epoch)) : | |
| 1008 | - if (-1 == $default) : | |
| 1009 | - $epoch = time(); | |
| 1010 | - else : | |
| 1011 | - $epoch = $default; | |
| 1012 | - endif; | |
| 1013 | - endif; | |
| 1014 | - | |
| 1015 | - return $epoch; | |
| 1016 | - } | |
| 1017 | - function updated ($fallback = true, $default = -1) { | |
| 1018 | - $epoch = null; | |
| 1019 | - | |
| 1020 | - # As far as I know, only dcterms and Atom have reliable ways to | |
| 1021 | - # specify when something was *modified* last. If neither is | |
| 1022 | - # available, then we'll try to get the time of publication. | |
| 1023 | - if (isset($this->item['dc']['modified'])) : // Not really correct | |
| 1024 | - $epoch = @parse_w3cdtf($this->item['dc']['modified']); | |
| 1025 | - elseif (isset($this->item['dcterms']['modified'])) : // Dublin Core extensions | |
| 1026 | - $epoch = @parse_w3cdtf($this->item['dcterms']['modified']); | |
| 1027 | - elseif (isset($this->item['modified'])): // Atom 0.3 | |
| 1028 | - $epoch = @parse_w3cdtf($this->item['modified']); | |
| 1029 | - elseif (isset($this->item['updated'])): // Atom 1.0 | |
| 1030 | - $epoch = @parse_w3cdtf($this->item['updated']); | |
| 1031 | - elseif ($fallback) : // Fall back to issued / dc:date | |
| 1032 | - $epoch = $this->published(/*fallback=*/ false, /*default=*/ $default); | |
| 1033 | - endif; | |
| 1034 | - | |
| 1035 | - # If everything failed, then default to the current time. | |
| 1036 | - if (is_null($epoch)) : | |
| 1037 | - if (-1 == $default) : | |
| 1038 | - $epoch = time(); | |
| 1039 | - else : | |
| 1040 | - $epoch = $default; | |
| 1041 | - endif; | |
| 1042 | - endif; | |
| 1043 | - | |
| 1044 | - return $epoch; | |
| 1045 | - } | |
| 1046 | - | |
| 1047 | - function update_hash () { | |
| 1048 | - return md5(serialize($this->item)); | |
| 1049 | - } | |
| 1050 | - | |
| 1051 | - function guid () { | |
| 1052 | - $guid = null; | |
| 1053 | - if (isset($this->item['id'])): // Atom 0.3 / 1.0 | |
| 1054 | - $guid = $this->item['id']; | |
| 1055 | - elseif (isset($this->item['atom']['id'])) : // Namespaced Atom | |
| 1056 | - $guid = $this->item['atom']['id']; | |
| 1057 | - elseif (isset($this->item['guid'])) : // RSS 2.0 | |
| 1058 | - $guid = $this->item['guid']; | |
| 1059 | - elseif (isset($this->item['dc']['identifier'])) :// yeah, right | |
| 1060 | - $guid = $this->item['dc']['identifier']; | |
| 1061 | - else : | |
| 1062 | - // The feed does not seem to have provided us with a | |
| 1063 | - // unique identifier, so we'll have to cobble together | |
| 1064 | - // a tag: URI that might work for us. The base of the | |
| 1065 | - // URI will be the host name of the feed source ... | |
| 1066 | - $bits = parse_url($this->feedmeta['link/uri']); | |
| 1067 | - $guid = 'tag:'.$bits['host']; | |
| 1068 | - | |
| 1069 | - // If we have a date of creation, then we can use that | |
| 1070 | - // to uniquely identify the item. (On the other hand, if | |
| 1071 | - // the feed producer was consicentious enough to | |
| 1072 | - // generate dates of creation, she probably also was | |
| 1073 | - // conscientious enough to generate unique identifiers.) | |
| 1074 | - if (!is_null($this->created())) : | |
| 1075 | - $guid .= '://post.'.date('YmdHis', $this->created()); | |
| 1076 | - | |
| 1077 | - // Otherwise, use both the URI of the item, *and* the | |
| 1078 | - // item's title. We have to use both because titles are | |
| 1079 | - // often not unique, and sometimes links aren't unique | |
| 1080 | - // either (e.g. Bitch (S)HITLIST, Mozilla Dot Org news, | |
| 1081 | - // some podcasts). But it's rare to have *both* the same | |
| 1082 | - // title *and* the same link for two different items. So | |
| 1083 | - // this is about the best we can do. | |
| 1084 | - else : | |
| 1085 | - $guid .= '://'.md5($this->item['link'].'/'.$this->item['title']); | |
| 1086 | - endif; | |
| 1087 | - endif; | |
| 1088 | - return $guid; | |
| 1089 | - } | |
| 1090 | - | |
| 1091 | - function author () { | |
| 1092 | - $author = array (); | |
| 1093 | - | |
| 1094 | - if (isset($this->item['author_name'])): | |
| 1095 | - $author['name'] = $this->item['author_name']; | |
| 1096 | - elseif (isset($this->item['dc']['creator'])): | |
| 1097 | - $author['name'] = $this->item['dc']['creator']; | |
| 1098 | - elseif (isset($this->item['dc']['contributor'])): | |
| 1099 | - $author['name'] = $this->item['dc']['contributor']; | |
| 1100 | - elseif (isset($this->feed->channel['dc']['creator'])) : | |
| 1101 | - $author['name'] = $this->feed->channel['dc']['creator']; | |
| 1102 | - elseif (isset($this->feed->channel['dc']['contributor'])) : | |
| 1103 | - $author['name'] = $this->feed->channel['dc']['contributor']; | |
| 1104 | - elseif (isset($this->feed->channel['author_name'])) : | |
| 1105 | - $author['name'] = $this->feed->channel['author_name']; | |
| 1106 | - elseif ($this->feed->is_rss() and isset($this->item['author'])) : | |
| 1107 | - // The author element in RSS is allegedly an | |
| 1108 | - // e-mail address, but lots of people don't use | |
| 1109 | - // it that way. So let's make of it what we can. | |
| 1110 | - $author = parse_email_with_realname($this->item['author']); | |
| 1111 | - | |
| 1112 | - if (!isset($author['name'])) : | |
| 1113 | - if (isset($author['email'])) : | |
| 1114 | - $author['name'] = $author['email']; | |
| 1115 | - else : | |
| 1116 | - $author['name'] = $this->feed->channel['title']; | |
| 1117 | - endif; | |
| 1118 | - endif; | |
| 1119 | - else : | |
| 1120 | - $author['name'] = $this->feed->channel['title']; | |
| 1121 | - endif; | |
| 1122 | - | |
| 1123 | - if (isset($this->item['author_email'])): | |
| 1124 | - $author['email'] = $this->item['author_email']; | |
| 1125 | - elseif (isset($this->feed->channel['author_email'])) : | |
| 1126 | - $author['email'] = $this->feed->channel['author_email']; | |
| 1127 | - endif; | |
| 1128 | - | |
| 1129 | - if (isset($this->item['author_url'])): | |
| 1130 | - $author['uri'] = $this->item['author_url']; | |
| 1131 | - elseif (isset($this->feed->channel['author_url'])) : | |
| 1132 | - $author['uri'] = $this->item['author_url']; | |
| 1133 | - else: | |
| 1134 | - $author['uri'] = $this->feed->channel['link']; | |
| 1135 | - endif; | |
| 1136 | - | |
| 1137 | - return $author; | |
| 1138 | - } // SyndicatedPost::author() | |
| 1139 | - | |
| 1140 | - /** | |
| 1141 | - * SyndicatedPost::isTaggedAs: Test whether a feed item is | |
| 1142 | - * tagged / categorized with a given string. Case and leading and | |
| 1143 | - * trailing whitespace are ignored. | |
| 1144 | - * | |
| 1145 | - * @param string $tag Tag to check for | |
| 1146 | - * | |
| 1147 | - * @return bool Whether or not at least one of the categories / tags on | |
| 1148 | - * $this->item is set to $tag (modulo case and leading and trailing | |
| 1149 | - * whitespace) | |
| 1150 | - */ | |
| 1151 | - function isTaggedAs ($tag) { | |
| 1152 | - $desiredTag = strtolower(trim($tag)); // Normalize case and whitespace | |
| 1153 | - | |
| 1154 | - // Check to see if this is tagged with $tag | |
| 1155 | - $currentCategory = 'category'; | |
| 1156 | - $currentCategoryNumber = 1; | |
| 1157 | - | |
| 1158 | - // If we have the new MagpieRSS, the number of category elements | |
| 1159 | - // on this item is stored under index "category#". | |
| 1160 | - if (isset($this->item['category#'])) : | |
| 1161 | - $numberOfCategories = (int) $this->item['category#']; | |
| 1162 | - | |
| 1163 | - // We REALLY shouldn't have the old and busted MagpieRSS, but in | |
| 1164 | - // case we do, it doesn't support multiple categories, but there | |
| 1165 | - // might still be a single value under the "category" index. | |
| 1166 | - elseif (isset($this->item['category'])) : | |
| 1167 | - $numberOfCategories = 1; | |
| 1168 | - | |
| 1169 | - // No standard category or tag elements on this feed item. | |
| 1170 | - else : | |
| 1171 | - $numberOfCategories = 0; | |
| 1172 | - | |
| 1173 | - endif; | |
| 1174 | - | |
| 1175 | - $isSoTagged = false; // Innocent until proven guilty | |
| 1176 | - | |
| 1177 | - // Loop through category elements; if there are multiple | |
| 1178 | - // elements, they are indexed as category, category#2, | |
| 1179 | - // category#3, ... category#N | |
| 1180 | - while ($currentCategoryNumber <= $numberOfCategories) : | |
| 1181 | - if ($desiredTag == strtolower(trim($this->item[$currentCategory]))) : | |
| 1182 | - $isSoTagged = true; // Got it! | |
| 1183 | - break; | |
| 1184 | - endif; | |
| 1185 | - | |
| 1186 | - $currentCategoryNumber += 1; | |
| 1187 | - $currentCategory = 'category#'.$currentCategoryNumber; | |
| 1188 | - endwhile; | |
| 1189 | - | |
| 1190 | - return $isSoTagged; | |
| 1191 | - } /* SyndicatedPost::isTaggedAs() */ | |
| 1192 | - | |
| 1193 | - var $uri_attrs = array ( | |
| 1194 | - array('a', 'href'), | |
| 1195 | - array('applet', 'codebase'), | |
| 1196 | - array('area', 'href'), | |
| 1197 | - array('blockquote', 'cite'), | |
| 1198 | - array('body', 'background'), | |
| 1199 | - array('del', 'cite'), | |
| 1200 | - array('form', 'action'), | |
| 1201 | - array('frame', 'longdesc'), | |
| 1202 | - array('frame', 'src'), | |
| 1203 | - array('iframe', 'longdesc'), | |
| 1204 | - array('iframe', 'src'), | |
| 1205 | - array('head', 'profile'), | |
| 1206 | - array('img', 'longdesc'), | |
| 1207 | - array('img', 'src'), | |
| 1208 | - array('img', 'usemap'), | |
| 1209 | - array('input', 'src'), | |
| 1210 | - array('input', 'usemap'), | |
| 1211 | - array('ins', 'cite'), | |
| 1212 | - array('link', 'href'), | |
| 1213 | - array('object', 'classid'), | |
| 1214 | - array('object', 'codebase'), | |
| 1215 | - array('object', 'data'), | |
| 1216 | - array('object', 'usemap'), | |
| 1217 | - array('q', 'cite'), | |
| 1218 | - array('script', 'src') | |
| 1219 | - ); /* var SyndicatedPost::$uri_attrs */ | |
| 1220 | - | |
| 1221 | - var $_base = null; | |
| 1222 | - | |
| 1223 | - function resolve_single_relative_uri ($refs) { | |
| 1224 | - $tag = FeedWordPressHTML::attributeMatch($refs); | |
| 1225 | - $url = Relative_URI::resolve($tag['value'], $this->_base); | |
| 1226 | - return $tag['prefix'] . $url . $tag['suffix']; | |
| 1227 | - } /* function SyndicatedPost::resolve_single_relative_uri() */ | |
| 1228 | - | |
| 1229 | - function resolve_relative_uris ($content, $obj) { | |
| 1230 | - # The MagpieRSS upgrade has some `xml:base` support baked in. | |
| 1231 | - # However, sometimes people do silly things, like putting | |
| 1232 | - # relative URIs out on a production RSS 2.0 feed or other feeds | |
| 1233 | - # with no good support for `xml:base`. So we'll do our best to | |
| 1234 | - # try to catch any remaining relative URIs and resolve them as | |
| 1235 | - # best we can. | |
| 1236 | - $obj->_base = $obj->item['link']; // Reset the base for resolving relative URIs | |
| 1237 | - | |
| 1238 | - foreach ($obj->uri_attrs as $pair) : | |
| 1239 | - list($tag, $attr) = $pair; | |
| 1240 | - $pattern = FeedWordPressHTML::attributeRegex($tag, $attr); | |
| 1241 | - $content = preg_replace_callback ( | |
| 1242 | - $pattern, | |
| 1243 | - array(&$obj, 'resolve_single_relative_uri'), | |
| 1244 | - $content | |
| 1245 | - ); | |
| 1246 | - endforeach; | |
| 1247 | - | |
| 1248 | - return $content; | |
| 1249 | - } /* function SyndicatedPost::resolve_relative_uris () */ | |
| 1250 | - | |
| 1251 | - var $strip_attrs = array ( | |
| 1252 | - array('[a-z]+', 'target'), | |
| 1253 | -// array('[a-z]+', 'style'), | |
| 1254 | -// array('[a-z]+', 'on[a-z]+'), | |
| 1255 | - ); | |
| 1256 | - | |
| 1257 | - function strip_attribute_from_tag ($refs) { | |
| 1258 | - $tag = FeedWordPressHTML::attributeMatch($refs); | |
| 1259 | - return $tag['before_attribute'].$tag['after_attribute']; | |
| 1260 | - } | |
| 1261 | - | |
| 1262 | - function sanitize_content ($content, $obj) { | |
| 1263 | - # This kind of sucks. I intend to replace it with | |
| 1264 | - # lib_filter sometime soon. | |
| 1265 | - foreach ($obj->strip_attrs as $pair): | |
| 1266 | - list($tag,$attr) = $pair; | |
| 1267 | - $pattern = FeedWordPressHTML::attributeRegex($tag, $attr); | |
| 1268 | - | |
| 1269 | - $content = preg_replace_callback ( | |
| 1270 | - $pattern, | |
| 1271 | - array(&$obj, 'strip_attribute_from_tag'), | |
| 1272 | - $content | |
| 1273 | - ); | |
| 1274 | - endforeach; | |
| 1275 | - return $content; | |
| 1276 | - } | |
| 1277 | -} // class SyndicatedPost | |
| 1712 | +} /* class SyndicatedPost */ | |
| 1278 | 1713 | |
