| 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 |
*/ |
| 15 |
class SyndicatedPost { |
| 16 |
var $item = null; // MagpieRSS representation |
| 17 |
var $entry = null; // SimplePie_Item representation |
| 18 |
|
| 19 |
var $link = null; |
| 20 |
var $feed = null; |
| 21 |
var $feedmeta = null; |
| 22 |
|
| 23 |
var $xmlns = array (); |
| 24 |
|
| 25 |
var $post = array (); |
| 26 |
|
| 27 |
var $_freshness = null; |
| 28 |
var $_wp_id = null; |
| 29 |
|
| 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) { |
| 40 |
global $wpdb; |
| 41 |
|
| 42 |
if (is_array($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 = $item; |
| 50 |
endif; |
| 51 |
|
| 52 |
FeedWordPress::diagnostic('feed_items', 'Considering item ['.$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. |
| 75 |
# |
| 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 |
|
| 83 |
global $fwp_channel, $fwp_feedmeta; |
| 84 |
$fwp_channel = $this->feed; $fwp_feedmeta = $this->feedmeta; |
| 85 |
|
| 86 |
// Trigger global syndicated_item filter. |
| 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 |
$this |
| 94 |
); |
| 95 |
|
| 96 |
# Filters can halt further processing by returning NULL |
| 97 |
if (is_null($this->item)) : |
| 98 |
$this->post = NULL; |
| 99 |
else : |
| 100 |
# Note that nothing is run through $wpdb->escape() here. |
| 101 |
# That's deliberate. The escaping is done at the point |
| 102 |
# of insertion, not here, to avoid double-escaping and |
| 103 |
# to avoid screwing with syndicated_post filters |
| 104 |
|
| 105 |
$this->post['post_title'] = apply_filters( |
| 106 |
'syndicated_item_title', |
| 107 |
$this->entry->get_title(), $this |
| 108 |
); |
| 109 |
|
| 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(). |
| 117 |
|
| 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(), $this); |
| 124 |
if (!is_null($excerpt)): |
| 125 |
$this->post['post_excerpt'] = $excerpt; |
| 126 |
endif; |
| 127 |
|
| 128 |
$this->post['epoch']['issued'] = apply_filters('syndicated_item_published', $this->published(), $this); |
| 129 |
$this->post['epoch']['created'] = apply_filters('syndicated_item_created', $this->created(), $this); |
| 130 |
$this->post['epoch']['modified'] = apply_filters('syndicated_item_updated', $this->updated(), $this); |
| 131 |
|
| 132 |
// Dealing with timestamps in WordPress is so fucking fucked. |
| 133 |
$offset = (int) get_option('gmt_offset') * 60 * 60; |
| 134 |
$this->post['post_date'] = gmdate('Y-m-d H:i:s', $this->published(/*fallback=*/ true, /*default=*/ -1) + $offset); |
| 135 |
$this->post['post_modified'] = gmdate('Y-m-d H:i:s', $this->updated(/*fallback=*/ true, /*default=*/ -1) + $offset); |
| 136 |
$this->post['post_date_gmt'] = gmdate('Y-m-d H:i:s', $this->published(/*fallback=*/ true, /*default=*/ -1)); |
| 137 |
$this->post['post_modified_gmt'] = gmdate('Y-m-d H:i:s', $this->updated(/*fallback=*/ true, /*default=*/ -1)); |
| 138 |
|
| 139 |
// Use feed-level preferences or the global default. |
| 140 |
$this->post['post_status'] = $this->link->syndicated_status('post', 'publish'); |
| 141 |
$this->post['comment_status'] = $this->link->syndicated_status('comment', 'closed'); |
| 142 |
$this->post['ping_status'] = $this->link->syndicated_status('ping', 'closed'); |
| 143 |
|
| 144 |
// Unique ID (hopefully a unique tag: URI); failing that, the permalink |
| 145 |
$this->post['guid'] = apply_filters('syndicated_item_guid', $this->guid(), $this); |
| 146 |
|
| 147 |
// User-supplied custom settings to apply to each post. Do first so that FWP-generated custom settings will overwrite if necessary; thus preventing any munging |
| 148 |
$default_custom_settings = get_option('feedwordpress_custom_settings'); |
| 149 |
if ($default_custom_settings and !is_array($default_custom_settings)) : |
| 150 |
$default_custom_settings = unserialize($default_custom_settings); |
| 151 |
endif; |
| 152 |
if (!is_array($default_custom_settings)) : |
| 153 |
$default_custom_settings = array(); |
| 154 |
endif; |
| 155 |
|
| 156 |
$custom_settings = (isset($this->link->settings['postmeta']) ? $this->link->settings['postmeta'] : null); |
| 157 |
if ($custom_settings and !is_array($custom_settings)) : |
| 158 |
$custom_settings = unserialize($custom_settings); |
| 159 |
endif; |
| 160 |
if (!is_array($custom_settings)) : |
| 161 |
$custom_settings = array(); |
| 162 |
endif; |
| 163 |
|
| 164 |
$postMetaIn = array_merge($default_custom_settings, $custom_settings); |
| 165 |
$postMetaOut = array(); |
| 166 |
|
| 167 |
// Big ugly fuckin loop to do any element substitutions |
| 168 |
// that we may need. |
| 169 |
foreach ($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 |
|
| 196 |
// RSS 2.0 / Atom 1.0 enclosure support |
| 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(), $this); |
| 203 |
endforeach; endif; |
| 204 |
|
| 205 |
// In case you want to point back to the blog this was syndicated from |
| 206 |
if (isset($this->feed->channel['title'])) : |
| 207 |
$this->post['meta']['syndication_source'] = apply_filters('syndicated_item_source_title', $this->feed->channel['title'], $this); |
| 208 |
endif; |
| 209 |
|
| 210 |
if (isset($this->feed->channel['link'])) : |
| 211 |
$this->post['meta']['syndication_source_uri'] = apply_filters('syndicated_item_source_link', $this->feed->channel['link'], $this); |
| 212 |
endif; |
| 213 |
|
| 214 |
// Make use of atom:source data, if present in an aggregated feed |
| 215 |
if (isset($this->item['source_title'])) : |
| 216 |
$this->post['meta']['syndication_source_original'] = $this->item['source_title']; |
| 217 |
endif; |
| 218 |
|
| 219 |
if (isset($this->item['source_link'])) : |
| 220 |
$this->post['meta']['syndication_source_uri_original'] = $this->item['source_link']; |
| 221 |
endif; |
| 222 |
|
| 223 |
if (isset($this->item['source_id'])) : |
| 224 |
$this->post['meta']['syndication_source_id_original'] = $this->item['source_id']; |
| 225 |
endif; |
| 226 |
|
| 227 |
// Store information on human-readable and machine-readable comment URIs |
| 228 |
|
| 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; |
| 232 |
|
| 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'] = $commentFeed; endif; |
| 236 |
// Yeah, yeah, now I know that it's supposed to be |
| 237 |
// wfw:commentRss. Oh well. Path dependence, sucka. |
| 238 |
|
| 239 |
// Store information to identify the feed that this came from |
| 240 |
if (isset($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'] = $this->feedmeta['link/id']; |
| 245 |
endif; |
| 246 |
|
| 247 |
if (isset($this->item['source_link_self'])) : |
| 248 |
$this->post['meta']['syndication_feed_original'] = $this->item['source_link_self']; |
| 249 |
endif; |
| 250 |
|
| 251 |
// In case you want to know the external permalink... |
| 252 |
$this->post['meta']['syndication_permalink'] = apply_filters('syndicated_item_link', $this->permalink()); |
| 253 |
|
| 254 |
// Store a hash of the post content for checking whether something needs to be updated |
| 255 |
$this->post['meta']['syndication_item_hash'] = $this->update_hash(); |
| 256 |
|
| 257 |
// Feed-by-feed options for author and category creation |
| 258 |
$this->post['named']['unfamiliar']['author'] = (isset($this->feedmeta['unfamiliar author']) ? $this->feedmeta['unfamiliar author'] : null); |
| 259 |
$this->post['named']['unfamiliar']['category'] = (isset($this->feedmeta['unfamiliar category']) ? $this->feedmeta['unfamiliar category'] : null); |
| 260 |
|
| 261 |
// Categories: start with default categories, if any |
| 262 |
$fc = get_option("feedwordpress_syndication_cats"); |
| 263 |
if ($fc) : |
| 264 |
$this->post['named']['preset/category'] = explode("\n", $fc); |
| 265 |
else : |
| 266 |
$this->post['named']['preset/category'] = array(); |
| 267 |
endif; |
| 268 |
|
| 269 |
if (isset($this->feedmeta['cats']) and is_array($this->feedmeta['cats'])) : |
| 270 |
$this->post['named']['preset/category'] = array_merge($this->post['named']['preset/category'], $this->feedmeta['cats']); |
| 271 |
endif; |
| 272 |
|
| 273 |
// Now add categories from the post, if we have 'em |
| 274 |
$this->post['named']['category'] = array(); |
| 275 |
if ( isset($this->item['category#']) ) : |
| 276 |
for ($i = 1; $i <= $this->item['category#']; $i++) : |
| 277 |
$cat_idx = (($i > 1) ? "#{$i}" : ""); |
| 278 |
$cat = $this->item["category{$cat_idx}"]; |
| 279 |
|
| 280 |
if ( isset($this->feedmeta['cat_split']) and strlen($this->feedmeta['cat_split']) > 0) : |
| 281 |
$pcre = "\007".$this->feedmeta['cat_split']."\007"; |
| 282 |
$this->post['named']['category'] = array_merge($this->post['named']['category'], preg_split($pcre, $cat, -1 /*=no limit*/, PREG_SPLIT_NO_EMPTY)); |
| 283 |
else : |
| 284 |
$this->post['named']['category'][] = $cat; |
| 285 |
endif; |
| 286 |
endfor; |
| 287 |
endif; |
| 288 |
$this->post['named']['category'] = apply_filters('syndicated_item_categories', $this->post['named']['category'], $this); |
| 289 |
|
| 290 |
// Tags: start with default tags, if any |
| 291 |
$ft = get_option("feedwordpress_syndication_tags"); |
| 292 |
if ($ft) : |
| 293 |
$this->post['tags_input'] = explode(FEEDWORDPRESS_CAT_SEPARATOR, $ft); |
| 294 |
else : |
| 295 |
$this->post['tags_input'] = array(); |
| 296 |
endif; |
| 297 |
|
| 298 |
if (isset($this->feedmeta['tags']) and is_array($this->feedmeta['tags'])) : |
| 299 |
$this->post['tags_input'] = array_merge($this->post['tags_input'], $this->feedmeta['tags']); |
| 300 |
endif; |
| 301 |
$this->post['tags_input'] = apply_filters('syndicated_item_tags', $this->post['tags_input'], $this); |
| 302 |
endif; |
| 303 |
} /* SyndicatedPost::SyndicatedPost() */ |
| 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 = $this->content(); |
| 500 |
if ( is_null($excerpt) or $excerpt == $content ) : |
| 501 |
# If content is available, generate an excerpt. |
| 502 |
if ( strlen(trim($content)) > 0 ) : |
| 503 |
$excerpt = strip_tags($content); |
| 504 |
if (strlen($excerpt) > 255) : |
| 505 |
$excerpt = substr($excerpt,0,252).'...'; |
| 506 |
endif; |
| 507 |
endif; |
| 508 |
endif; |
| 509 |
return $excerpt; |
| 510 |
} /* SyndicatedPost::excerpt() */ |
| 511 |
|
| 512 |
function permalink () { |
| 513 |
// Handles explicit <link> elements and also RSS 2.0 cases with |
| 514 |
// <guid isPermaLink="true">, etc. Hooray! |
| 515 |
$permalink = $this->entry->get_link(); |
| 516 |
return $permalink; |
| 517 |
} |
| 518 |
|
| 519 |
function created () { |
| 520 |
$date = ''; |
| 521 |
if (isset($this->item['dc']['created'])) : |
| 522 |
$date = $this->item['dc']['created']; |
| 523 |
elseif (isset($this->item['dcterms']['created'])) : |
| 524 |
$date = $this->item['dcterms']['created']; |
| 525 |
elseif (isset($this->item['created'])): // Atom 0.3 |
| 526 |
$date = $this->item['created']; |
| 527 |
endif; |
| 528 |
|
| 529 |
$epoch = new FeedTime($date); |
| 530 |
return $epoch->timestamp(); |
| 531 |
} /* SyndicatedPost::created() */ |
| 532 |
|
| 533 |
function published ($fallback = true, $default = NULL) { |
| 534 |
$date = ''; |
| 535 |
|
| 536 |
# RSS is a fucking mess. Figure out whether we have a date in |
| 537 |
# <dc:date>, <issued>, <pubDate>, etc., and get it into Unix |
| 538 |
# epoch format for reformatting. If we can't find anything, |
| 539 |
# we'll use the last-updated time. |
| 540 |
if (isset($this->item['dc']['date'])): // Dublin Core |
| 541 |
$date = $this->item['dc']['date']; |
| 542 |
elseif (isset($this->item['dcterms']['issued'])) : // Dublin Core extensions |
| 543 |
$date = $this->item['dcterms']['issued']; |
| 544 |
elseif (isset($this->item['published'])) : // Atom 1.0 |
| 545 |
$date = $this->item['published']; |
| 546 |
elseif (isset($this->item['issued'])): // Atom 0.3 |
| 547 |
$date = $this->item['issued']; |
| 548 |
elseif (isset($this->item['pubdate'])): // RSS 2.0 |
| 549 |
$date = $this->item['pubdate']; |
| 550 |
endif; |
| 551 |
|
| 552 |
if (strlen($date) > 0) : |
| 553 |
$time = new FeedTime($date); |
| 554 |
$epoch = $time->timestamp(); |
| 555 |
elseif ($fallback) : // Fall back to <updated> / <modified> if present |
| 556 |
$epoch = $this->updated(/*fallback=*/ false, /*default=*/ $default); |
| 557 |
endif; |
| 558 |
|
| 559 |
# If everything failed, then default to the current time. |
| 560 |
if (is_null($epoch)) : |
| 561 |
if (-1 == $default) : |
| 562 |
$epoch = time(); |
| 563 |
else : |
| 564 |
$epoch = $default; |
| 565 |
endif; |
| 566 |
endif; |
| 567 |
|
| 568 |
return $epoch; |
| 569 |
} /* SyndicatedPost::published() */ |
| 570 |
|
| 571 |
function updated ($fallback = true, $default = -1) { |
| 572 |
$date = ''; |
| 573 |
|
| 574 |
# As far as I know, only dcterms and Atom have reliable ways to |
| 575 |
# specify when something was *modified* last. If neither is |
| 576 |
# available, then we'll try to get the time of publication. |
| 577 |
if (isset($this->item['dc']['modified'])) : // Not really correct |
| 578 |
$date = $this->item['dc']['modified']; |
| 579 |
elseif (isset($this->item['dcterms']['modified'])) : // Dublin Core extensions |
| 580 |
$date = $this->item['dcterms']['modified']; |
| 581 |
elseif (isset($this->item['modified'])): // Atom 0.3 |
| 582 |
$date = $this->item['modified']; |
| 583 |
elseif (isset($this->item['updated'])): // Atom 1.0 |
| 584 |
$date = $this->item['updated']; |
| 585 |
endif; |
| 586 |
|
| 587 |
if (strlen($date) > 0) : |
| 588 |
$time = new FeedTime($date); |
| 589 |
$epoch = $time->timestamp(); |
| 590 |
elseif ($fallback) : // Fall back to issued / dc:date |
| 591 |
$epoch = $this->published(/*fallback=*/ false, /*default=*/ $default); |
| 592 |
endif; |
| 593 |
|
| 594 |
# If everything failed, then default to the current time. |
| 595 |
if (is_null($epoch)) : |
| 596 |
if (-1 == $default) : |
| 597 |
$epoch = time(); |
| 598 |
else : |
| 599 |
$epoch = $default; |
| 600 |
endif; |
| 601 |
endif; |
| 602 |
|
| 603 |
return $epoch; |
| 604 |
} /* SyndicatedPost::updated() */ |
| 605 |
|
| 606 |
function update_hash () { |
| 607 |
return md5(serialize($this->item)); |
| 608 |
} /* SyndicatedPost::update_hash() */ |
| 609 |
|
| 610 |
function guid () { |
| 611 |
$guid = null; |
| 612 |
if (isset($this->item['id'])): // Atom 0.3 / 1.0 |
| 613 |
$guid = $this->item['id']; |
| 614 |
elseif (isset($this->item['atom']['id'])) : // Namespaced Atom |
| 615 |
$guid = $this->item['atom']['id']; |
| 616 |
elseif (isset($this->item['guid'])) : // RSS 2.0 |
| 617 |
$guid = $this->item['guid']; |
| 618 |
elseif (isset($this->item['dc']['identifier'])) :// yeah, right |
| 619 |
$guid = $this->item['dc']['identifier']; |
| 620 |
else : |
| 621 |
// The feed does not seem to have provided us with a |
| 622 |
// unique identifier, so we'll have to cobble together |
| 623 |
// a tag: URI that might work for us. The base of the |
| 624 |
// URI will be the host name of the feed source ... |
| 625 |
$bits = parse_url($this->feedmeta['link/uri']); |
| 626 |
$guid = 'tag:'.$bits['host']; |
| 627 |
|
| 628 |
// If we have a date of creation, then we can use that |
| 629 |
// to uniquely identify the item. (On the other hand, if |
| 630 |
// the feed producer was consicentious enough to |
| 631 |
// generate dates of creation, she probably also was |
| 632 |
// conscientious enough to generate unique identifiers.) |
| 633 |
if (!is_null($this->created())) : |
| 634 |
$guid .= '://post.'.date('YmdHis', $this->created()); |
| 635 |
|
| 636 |
// Otherwise, use both the URI of the item, *and* the |
| 637 |
// item's title. We have to use both because titles are |
| 638 |
// often not unique, and sometimes links aren't unique |
| 639 |
// either (e.g. Bitch (S)HITLIST, Mozilla Dot Org news, |
| 640 |
// some podcasts). But it's rare to have *both* the same |
| 641 |
// title *and* the same link for two different items. So |
| 642 |
// this is about the best we can do. |
| 643 |
else : |
| 644 |
$guid .= '://'.md5($this->item['link'].'/'.$this->item['title']); |
| 645 |
endif; |
| 646 |
endif; |
| 647 |
return $guid; |
| 648 |
} /* SyndicatedPost::guid() */ |
| 649 |
|
| 650 |
function author () { |
| 651 |
$author = array (); |
| 652 |
|
| 653 |
if (isset($this->item['author_name'])): |
| 654 |
$author['name'] = $this->item['author_name']; |
| 655 |
elseif (isset($this->item['dc']['creator'])): |
| 656 |
$author['name'] = $this->item['dc']['creator']; |
| 657 |
elseif (isset($this->item['dc']['contributor'])): |
| 658 |
$author['name'] = $this->item['dc']['contributor']; |
| 659 |
elseif (isset($this->feed->channel['dc']['creator'])) : |
| 660 |
$author['name'] = $this->feed->channel['dc']['creator']; |
| 661 |
elseif (isset($this->feed->channel['dc']['contributor'])) : |
| 662 |
$author['name'] = $this->feed->channel['dc']['contributor']; |
| 663 |
elseif (isset($this->feed->channel['author_name'])) : |
| 664 |
$author['name'] = $this->feed->channel['author_name']; |
| 665 |
elseif ($this->feed->is_rss() and isset($this->item['author'])) : |
| 666 |
// The author element in RSS is allegedly an |
| 667 |
// e-mail address, but lots of people don't use |
| 668 |
// it that way. So let's make of it what we can. |
| 669 |
$author = parse_email_with_realname($this->item['author']); |
| 670 |
|
| 671 |
if (!isset($author['name'])) : |
| 672 |
if (isset($author['email'])) : |
| 673 |
$author['name'] = $author['email']; |
| 674 |
else : |
| 675 |
$author['name'] = $this->feed->channel['title']; |
| 676 |
endif; |
| 677 |
endif; |
| 678 |
else : |
| 679 |
$author['name'] = $this->feed->channel['title']; |
| 680 |
endif; |
| 681 |
|
| 682 |
if (isset($this->item['author_email'])): |
| 683 |
$author['email'] = $this->item['author_email']; |
| 684 |
elseif (isset($this->feed->channel['author_email'])) : |
| 685 |
$author['email'] = $this->feed->channel['author_email']; |
| 686 |
endif; |
| 687 |
|
| 688 |
if (isset($this->item['author_url'])): |
| 689 |
$author['uri'] = $this->item['author_url']; |
| 690 |
elseif (isset($this->feed->channel['author_url'])) : |
| 691 |
$author['uri'] = $this->item['author_url']; |
| 692 |
else: |
| 693 |
$author['uri'] = $this->feed->channel['link']; |
| 694 |
endif; |
| 695 |
|
| 696 |
return $author; |
| 697 |
} /* SyndicatedPost::author() */ |
| 698 |
|
| 699 |
/** |
| 700 |
* SyndicatedPost::isTaggedAs: Test whether a feed item is |
| 701 |
* tagged / categorized with a given string. Case and leading and |
| 702 |
* trailing whitespace are ignored. |
| 703 |
* |
| 704 |
* @param string $tag Tag to check for |
| 705 |
* |
| 706 |
* @return bool Whether or not at least one of the categories / tags on |
| 707 |
* $this->item is set to $tag (modulo case and leading and trailing |
| 708 |
* whitespace) |
| 709 |
*/ |
| 710 |
function isTaggedAs ($tag) { |
| 711 |
$desiredTag = strtolower(trim($tag)); // Normalize case and whitespace |
| 712 |
|
| 713 |
// Check to see if this is tagged with $tag |
| 714 |
$currentCategory = 'category'; |
| 715 |
$currentCategoryNumber = 1; |
| 716 |
|
| 717 |
// If we have the new MagpieRSS, the number of category elements |
| 718 |
// on this item is stored under index "category#". |
| 719 |
if (isset($this->item['category#'])) : |
| 720 |
$numberOfCategories = (int) $this->item['category#']; |
| 721 |
|
| 722 |
// We REALLY shouldn't have the old and busted MagpieRSS, but in |
| 723 |
// case we do, it doesn't support multiple categories, but there |
| 724 |
// might still be a single value under the "category" index. |
| 725 |
elseif (isset($this->item['category'])) : |
| 726 |
$numberOfCategories = 1; |
| 727 |
|
| 728 |
// No standard category or tag elements on this feed item. |
| 729 |
else : |
| 730 |
$numberOfCategories = 0; |
| 731 |
|
| 732 |
endif; |
| 733 |
|
| 734 |
$isSoTagged = false; // Innocent until proven guilty |
| 735 |
|
| 736 |
// Loop through category elements; if there are multiple |
| 737 |
// elements, they are indexed as category, category#2, |
| 738 |
// category#3, ... category#N |
| 739 |
while ($currentCategoryNumber <= $numberOfCategories) : |
| 740 |
if ($desiredTag == strtolower(trim($this->item[$currentCategory]))) : |
| 741 |
$isSoTagged = true; // Got it! |
| 742 |
break; |
| 743 |
endif; |
| 744 |
|
| 745 |
$currentCategoryNumber += 1; |
| 746 |
$currentCategory = 'category#'.$currentCategoryNumber; |
| 747 |
endwhile; |
| 748 |
|
| 749 |
return $isSoTagged; |
| 750 |
} /* SyndicatedPost::isTaggedAs() */ |
| 751 |
|
| 752 |
/** |
| 753 |
* SyndicatedPost::enclosures: returns an array with any enclosures |
| 754 |
* that may be attached to this syndicated item. |
| 755 |
* |
| 756 |
* @param string $type If you only want enclosures that match a certain |
| 757 |
* MIME type or group of MIME types, you can limit the enclosures |
| 758 |
* that will be returned to only those with a MIME type which |
| 759 |
* matches this regular expression. |
| 760 |
* @return array |
| 761 |
*/ |
| 762 |
function enclosures ($type = '/.*/') { |
| 763 |
$enclosures = array(); |
| 764 |
|
| 765 |
if (isset($this->item['enclosure#'])) : |
| 766 |
// Loop through enclosure, enclosure#2, enclosure#3, .... |
| 767 |
for ($i = 1; $i <= $this->item['enclosure#']; $i++) : |
| 768 |
$eid = (($i > 1) ? "#{$id}" : ""); |
| 769 |
|
| 770 |
// Does it match the type we want? |
| 771 |
if (preg_match($type, $this->item["enclosure{$eid}@type"])) : |
| 772 |
$enclosures[] = array( |
| 773 |
"url" => $this->item["enclosure{$eid}@url"], |
| 774 |
"type" => $this->item["enclosure{$eid}@type"], |
| 775 |
"length" => $this->item["enclosure{$eid}@length"], |
| 776 |
); |
| 777 |
endif; |
| 778 |
endfor; |
| 779 |
endif; |
| 780 |
return $enclosures; |
| 781 |
} /* SyndicatedPost::enclosures() */ |
| 782 |
|
| 783 |
function comment_link () { |
| 784 |
$url = null; |
| 785 |
|
| 786 |
// RSS 2.0 has a standard <comments> element: |
| 787 |
// "<comments> is an optional sub-element of <item>. If present, |
| 788 |
// it is the url of the comments page for the item." |
| 789 |
// <http://cyber.law.harvard.edu/rss/rss.html#ltcommentsgtSubelementOfLtitemgt> |
| 790 |
if (isset($this->item['comments'])) : |
| 791 |
$url = $this->item['comments']; |
| 792 |
endif; |
| 793 |
|
| 794 |
// The convention in Atom feeds is to use a standard <link> |
| 795 |
// element with @rel="replies" and @type="text/html". |
| 796 |
// Unfortunately, SimplePie_Item::get_links() allows us to filter |
| 797 |
// by the value of @rel, but not by the value of @type. *sigh* |
| 798 |
|
| 799 |
// Try Atom 1.0 first |
| 800 |
$linkElements = $this->entry->get_item_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'link'); |
| 801 |
|
| 802 |
// Fall back and try Atom 0.3 |
| 803 |
if (is_null($linkElements)) : $linkElements = $this->entry->get_item_tags(SIMPLEPIE_NAMESPACE_ATOM_03, 'link'); endif; |
| 804 |
|
| 805 |
// Now loop through the elements, screening by @rel and @type |
| 806 |
if (is_array($linkElements)) : foreach ($linkElements as $link) : |
| 807 |
$rel = (isset($link['attribs']['']['rel']) ? $link['attribs']['']['rel'] : 'alternate'); |
| 808 |
$type = (isset($link['attribs']['']['type']) ? $link['attribs']['']['type'] : NULL); |
| 809 |
$href = (isset($link['attribs']['']['href']) ? $link['attribs']['']['href'] : NULL); |
| 810 |
|
| 811 |
if (strtolower($rel)=='replies' and $type=='text/html' and !is_null($href)) : |
| 812 |
$url = $href; |
| 813 |
endif; |
| 814 |
endforeach; endif; |
| 815 |
|
| 816 |
return $url; |
| 817 |
} |
| 818 |
|
| 819 |
function comment_feed () { |
| 820 |
$feed = null; |
| 821 |
|
| 822 |
// Well Formed Web comment feeds extension for RSS 2.0 |
| 823 |
// <http://www.sellsbrothers.com/spout/default.aspx?content=archive.htm#exposingRssComments> |
| 824 |
// |
| 825 |
// N.B.: Correct capitalization is wfw:commentRss, but |
| 826 |
// wfw:commentRSS is common in the wild (partly due to a typo in |
| 827 |
// the original spec). In any case, our item array is normalized |
| 828 |
// to all lowercase anyways. |
| 829 |
if (isset($this->item['wfw']['commentrss'])) : |
| 830 |
$feed = $this->item['wfw']['commentrss']; |
| 831 |
endif; |
| 832 |
|
| 833 |
// In Atom 1.0, the convention is to use a standard link element |
| 834 |
// with @rel="replies". Sometimes this is also used to pass a |
| 835 |
// link to the human-readable comments page, so we also need to |
| 836 |
// check link/@type for a feed MIME type. |
| 837 |
// |
| 838 |
// Which is why I'm not using the SimplePie_Item::get_links() |
| 839 |
// method here, incidentally: it doesn't allow you to filter by |
| 840 |
// @type. *sigh* |
| 841 |
if (isset($this->item['link_replies'])) : |
| 842 |
// There may be multiple <link rel="replies"> elements; feeds have a feed MIME type |
| 843 |
$N = isset($this->item['link_replies#']) ? $this->item['link_replies#'] : 1; |
| 844 |
for ($i = 1; $i <= $N; $i++) : |
| 845 |
$currentElement = 'link_replies'.(($i > 1) ? '#'.$i : ''); |
| 846 |
if (isset($this->item[$currentElement.'@type']) |
| 847 |
and preg_match("\007application/(atom|rss|rdf)\+xml\007i", $this->item[$currentElement.'@type'])) : |
| 848 |
$feed = $this->item[$currentElement]; |
| 849 |
endif; |
| 850 |
endfor; |
| 851 |
endif; |
| 852 |
return $feed; |
| 853 |
} /* SyndicatedPost::comment_feed() */ |
| 854 |
|
| 855 |
################################## |
| 856 |
#### BUILT-IN CONTENT FILTERS #### |
| 857 |
################################## |
| 858 |
|
| 859 |
var $uri_attrs = array ( |
| 860 |
array('a', 'href'), |
| 861 |
array('applet', 'codebase'), |
| 862 |
array('area', 'href'), |
| 863 |
array('blockquote', 'cite'), |
| 864 |
array('body', 'background'), |
| 865 |
array('del', 'cite'), |
| 866 |
array('form', 'action'), |
| 867 |
array('frame', 'longdesc'), |
| 868 |
array('frame', 'src'), |
| 869 |
array('iframe', 'longdesc'), |
| 870 |
array('iframe', 'src'), |
| 871 |
array('head', 'profile'), |
| 872 |
array('img', 'longdesc'), |
| 873 |
array('img', 'src'), |
| 874 |
array('img', 'usemap'), |
| 875 |
array('input', 'src'), |
| 876 |
array('input', 'usemap'), |
| 877 |
array('ins', 'cite'), |
| 878 |
array('link', 'href'), |
| 879 |
array('object', 'classid'), |
| 880 |
array('object', 'codebase'), |
| 881 |
array('object', 'data'), |
| 882 |
array('object', 'usemap'), |
| 883 |
array('q', 'cite'), |
| 884 |
array('script', 'src') |
| 885 |
); /* var SyndicatedPost::$uri_attrs */ |
| 886 |
|
| 887 |
var $_base = null; |
| 888 |
|
| 889 |
function resolve_single_relative_uri ($refs) { |
| 890 |
$tag = FeedWordPressHTML::attributeMatch($refs); |
| 891 |
$url = Relative_URI::resolve($tag['value'], $this->_base); |
| 892 |
return $tag['prefix'] . $url . $tag['suffix']; |
| 893 |
} /* function SyndicatedPost::resolve_single_relative_uri() */ |
| 894 |
|
| 895 |
function resolve_relative_uris ($content, $obj) { |
| 896 |
$set = $obj->link->setting('resolve relative', 'resolve_relative', 'yes'); |
| 897 |
if ($set and $set != 'no') : |
| 898 |
// Fallback: if we don't have anything better, use the |
| 899 |
// item link from the feed |
| 900 |
$obj->_base = $obj->item['link']; // Reset the base for resolving relative URIs |
| 901 |
|
| 902 |
// What we should do here, properly, is to use |
| 903 |
// SimplePie_Item::get_base() -- but that method is |
| 904 |
// currently broken. Or getting down and dirty in the |
| 905 |
// SimplePie representation of the content tags and |
| 906 |
// grabbing the xml_base member for the content element. |
| 907 |
// Maybe someday... |
| 908 |
|
| 909 |
foreach ($obj->uri_attrs as $pair) : |
| 910 |
list($tag, $attr) = $pair; |
| 911 |
$pattern = FeedWordPressHTML::attributeRegex($tag, $attr); |
| 912 |
$content = preg_replace_callback ( |
| 913 |
$pattern, |
| 914 |
array(&$obj, 'resolve_single_relative_uri'), |
| 915 |
$content |
| 916 |
); |
| 917 |
endforeach; |
| 918 |
endif; |
| 919 |
|
| 920 |
return $content; |
| 921 |
} /* function SyndicatedPost::resolve_relative_uris () */ |
| 922 |
|
| 923 |
var $strip_attrs = array ( |
| 924 |
array('[a-z]+', 'target'), |
| 925 |
// array('[a-z]+', 'style'), |
| 926 |
// array('[a-z]+', 'on[a-z]+'), |
| 927 |
); |
| 928 |
|
| 929 |
function strip_attribute_from_tag ($refs) { |
| 930 |
$tag = FeedWordPressHTML::attributeMatch($refs); |
| 931 |
return $tag['before_attribute'].$tag['after_attribute']; |
| 932 |
} |
| 933 |
|
| 934 |
function sanitize_content ($content, $obj) { |
| 935 |
# This kind of sucks. I intend to replace it with |
| 936 |
# lib_filter sometime soon. |
| 937 |
foreach ($obj->strip_attrs as $pair): |
| 938 |
list($tag,$attr) = $pair; |
| 939 |
$pattern = FeedWordPressHTML::attributeRegex($tag, $attr); |
| 940 |
|
| 941 |
$content = preg_replace_callback ( |
| 942 |
$pattern, |
| 943 |
array(&$obj, 'strip_attribute_from_tag'), |
| 944 |
$content |
| 945 |
); |
| 946 |
endforeach; |
| 947 |
return $content; |
| 948 |
} /* SyndicatedPost::sanitize() */ |
| 949 |
|
| 950 |
##################### |
| 951 |
#### POST STATUS #### |
| 952 |
##################### |
| 953 |
|
| 954 |
/** |
| 955 |
* SyndicatedPost::filtered: check whether or not this post has been |
| 956 |
* screened out by a registered filter. |
| 957 |
* |
| 958 |
* @return bool TRUE iff post has been filtered out by a previous filter |
| 959 |
*/ |
| 960 |
function filtered () { |
| 961 |
return is_null($this->post); |
| 962 |
} /* SyndicatedPost::filtered() */ |
| 963 |
|
| 964 |
/** |
| 965 |
* SyndicatedPost::freshness: check whether post is a new post to be |
| 966 |
* inserted, a previously syndicated post that needs to be updated to |
| 967 |
* match the latest revision, or a previously syndicated post that is |
| 968 |
* still up-to-date. |
| 969 |
* |
| 970 |
* @return int A status code representing the freshness of the post |
| 971 |
* 0 = post already syndicated; no update needed |
| 972 |
* 1 = post already syndicated, but needs to be updated to latest |
| 973 |
* 2 = post has not yet been syndicated; needs to be created |
| 974 |
*/ |
| 975 |
function freshness () { |
| 976 |
global $wpdb; |
| 977 |
|
| 978 |
if ($this->filtered()) : // This should never happen. |
| 979 |
FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__); |
| 980 |
endif; |
| 981 |
|
| 982 |
if (is_null($this->_freshness)) : |
| 983 |
$guid = $wpdb->escape($this->guid()); |
| 984 |
|
| 985 |
$result = $wpdb->get_row(" |
| 986 |
SELECT id, guid, post_modified_gmt |
| 987 |
FROM $wpdb->posts WHERE guid='$guid' |
| 988 |
"); |
| 989 |
|
| 990 |
if (!$result) : |
| 991 |
$this->_freshness = 2; // New content |
| 992 |
else: |
| 993 |
$stored_update_hashes = get_post_custom_values('syndication_item_hash', $result->id); |
| 994 |
if (count($stored_update_hashes) > 0) : |
| 995 |
$stored_update_hash = $stored_update_hashes[0]; |
| 996 |
$update_hash_changed = ($stored_update_hash != $this->update_hash()); |
| 997 |
else : |
| 998 |
$update_hash_changed = true; // Can't find syndication meta-data |
| 999 |
endif; |
| 1000 |
|
| 1001 |
preg_match('/([0-9]+)-([0-9]+)-([0-9]+) ([0-9]+):([0-9]+):([0-9]+)/', $result->post_modified_gmt, $backref); |
| 1002 |
|
| 1003 |
$last_rev_ts = gmmktime($backref[4], $backref[5], $backref[6], $backref[2], $backref[3], $backref[1]); |
| 1004 |
$updated_ts = $this->updated(/*fallback=*/ true, /*default=*/ NULL); |
| 1005 |
|
| 1006 |
$frozen_values = get_post_custom_values('_syndication_freeze_updates', $result->id); |
| 1007 |
$frozen_post = (count($frozen_values) > 0 and 'yes' == $frozen_values[0]); |
| 1008 |
$frozen_feed = ('yes' == $this->link->setting('freeze updates', 'freeze_updates', NULL)); |
| 1009 |
|
| 1010 |
// Check timestamps... |
| 1011 |
$updated = ( |
| 1012 |
!is_null($updated_ts) |
| 1013 |
and ($updated_ts > $last_rev_ts) |
| 1014 |
); |
| 1015 |
|
| 1016 |
|
| 1017 |
// Or the hash... |
| 1018 |
$updated = ($updated or $update_hash_changed); |
| 1019 |
|
| 1020 |
// But only if the post is not frozen. |
| 1021 |
$updated = ( |
| 1022 |
$updated |
| 1023 |
and !$frozen_post |
| 1024 |
and !$frozen_feed |
| 1025 |
); |
| 1026 |
|
| 1027 |
if ($updated) : |
| 1028 |
$this->_freshness = 1; // Updated content |
| 1029 |
$this->_wp_id = $result->id; |
| 1030 |
else : |
| 1031 |
$this->_freshness = 0; // Same old, same old |
| 1032 |
$this->_wp_id = $result->id; |
| 1033 |
endif; |
| 1034 |
endif; |
| 1035 |
endif; |
| 1036 |
return $this->_freshness; |
| 1037 |
} |
| 1038 |
|
| 1039 |
################################################# |
| 1040 |
#### INTERNAL STORAGE AND MANAGEMENT METHODS #### |
| 1041 |
################################################# |
| 1042 |
|
| 1043 |
function wp_id () { |
| 1044 |
if ($this->filtered()) : // This should never happen. |
| 1045 |
FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__); |
| 1046 |
endif; |
| 1047 |
|
| 1048 |
if (is_null($this->_wp_id) and is_null($this->_freshness)) : |
| 1049 |
$fresh = $this->freshness(); // sets WP DB id in the process |
| 1050 |
endif; |
| 1051 |
return $this->_wp_id; |
| 1052 |
} |
| 1053 |
|
| 1054 |
function store () { |
| 1055 |
global $wpdb; |
| 1056 |
|
| 1057 |
if ($this->filtered()) : // This should never happen. |
| 1058 |
FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__); |
| 1059 |
endif; |
| 1060 |
|
| 1061 |
$freshness = $this->freshness(); |
| 1062 |
if ($freshness > 0) : |
| 1063 |
# -- Look up, or create, numeric ID for author |
| 1064 |
$this->post['post_author'] = $this->author_id ( |
| 1065 |
FeedWordPress::on_unfamiliar('author', $this->post['named']['unfamiliar']['author']) |
| 1066 |
); |
| 1067 |
|
| 1068 |
if (is_null($this->post['post_author'])) : |
| 1069 |
$this->post = NULL; |
| 1070 |
endif; |
| 1071 |
endif; |
| 1072 |
|
| 1073 |
if (!$this->filtered() and $freshness > 0) : |
| 1074 |
# -- Look up, or create, numeric ID for categories |
| 1075 |
list($pcats, $ptags) = $this->category_ids ( |
| 1076 |
$this->post['named']['category'], |
| 1077 |
FeedWordPress::on_unfamiliar('category', $this->post['named']['unfamiliar']['category']), |
| 1078 |
/*tags_too=*/ true |
| 1079 |
); |
| 1080 |
|
| 1081 |
$this->post['post_category'] = $pcats; |
| 1082 |
$this->post['tags_input'] = array_merge($this->post['tags_input'], $ptags); |
| 1083 |
|
| 1084 |
if (is_null($this->post['post_category'])) : |
| 1085 |
// filter mode on, no matching categories; drop the post |
| 1086 |
$this->post = NULL; |
| 1087 |
else : |
| 1088 |
// filter mode off or at least one match; now add on the feed and global presets |
| 1089 |
$this->post['post_category'] = array_merge ( |
| 1090 |
$this->post['post_category'], |
| 1091 |
$this->category_ids ( |
| 1092 |
$this->post['named']['preset/category'], |
| 1093 |
'default' |
| 1094 |
) |
| 1095 |
); |
| 1096 |
|
| 1097 |
if (count($this->post['post_category']) < 1) : |
| 1098 |
$this->post['post_category'][] = 1; // Default to category 1 ("Uncategorized" / "General") if nothing else |
| 1099 |
endif; |
| 1100 |
endif; |
| 1101 |
endif; |
| 1102 |
|
| 1103 |
if (!$this->filtered() and $freshness > 0) : |
| 1104 |
unset($this->post['named']); |
| 1105 |
$this->post = apply_filters('syndicated_post', $this->post, $this); |
| 1106 |
|
| 1107 |
// Allow for feed-specific syndicated_post filters. |
| 1108 |
$this->post = apply_filters( |
| 1109 |
"syndicated_post_".$this->link->uri(), |
| 1110 |
$this->post, |
| 1111 |
$this |
| 1112 |
); |
| 1113 |
endif; |
| 1114 |
|
| 1115 |
// Hook in early to make sure these get inserted if at all possible |
| 1116 |
add_action( |
| 1117 |
/*hook=*/ 'transition_post_status', |
| 1118 |
/*callback=*/ array(&$this, 'add_rss_meta'), |
| 1119 |
/*priority=*/ -10000, /* very early */ |
| 1120 |
/*arguments=*/ 3 |
| 1121 |
); |
| 1122 |
|
| 1123 |
if (!$this->filtered() and $freshness == 2) : |
| 1124 |
// The item has not yet been added. So let's add it. |
| 1125 |
FeedWordPress::diagnostic('syndicated_posts', 'Inserting new post "'.$this->post['post_title'].'"'); |
| 1126 |
|
| 1127 |
$this->insert_new(); |
| 1128 |
do_action('post_syndicated_item', $this->wp_id(), $this); |
| 1129 |
|
| 1130 |
$ret = 'new'; |
| 1131 |
elseif (!$this->filtered() and $freshness == 1) : |
| 1132 |
FeedWordPress::diagnostic('syndicated_posts', 'Updating existing post # '.$this->wp_id().', "'.$this->post['post_title'].'"'); |
| 1133 |
|
| 1134 |
$this->post['ID'] = $this->wp_id(); |
| 1135 |
$this->update_existing(); |
| 1136 |
do_action('update_syndicated_item', $this->wp_id(), $this); |
| 1137 |
|
| 1138 |
$ret = 'updated'; |
| 1139 |
else : |
| 1140 |
$ret = false; |
| 1141 |
endif; |
| 1142 |
|
| 1143 |
// Remove add_rss_meta hook |
| 1144 |
remove_action( |
| 1145 |
/*hook=*/ 'transition_post_status', |
| 1146 |
/*callback=*/ array(&$this, 'add_rss_meta'), |
| 1147 |
/*priority=*/ -10000, /* very early */ |
| 1148 |
/*arguments=*/ 3 |
| 1149 |
); |
| 1150 |
|
| 1151 |
return $ret; |
| 1152 |
} /* function SyndicatedPost::store () */ |
| 1153 |
|
| 1154 |
function insert_new () { |
| 1155 |
global $wpdb, $wp_db_version; |
| 1156 |
|
| 1157 |
$dbpost = $this->normalize_post(/*new=*/ true); |
| 1158 |
if (!is_null($dbpost)) : |
| 1159 |
if ($this->use_api('wp_insert_post')) : |
| 1160 |
$dbpost['post_pingback'] = false; // Tell WP 2.1 and 2.2 not to process for pingbacks |
| 1161 |
|
| 1162 |
// This is a ridiculous fucking kludge necessitated by WordPress 2.6 munging authorship meta-data |
| 1163 |
add_action('_wp_put_post_revision', array($this, 'fix_revision_meta')); |
| 1164 |
|
| 1165 |
// Kludge to prevent kses filters from stripping the |
| 1166 |
// content of posts when updating without a logged in |
| 1167 |
// user who has `unfiltered_html` capability. |
| 1168 |
add_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11); |
| 1169 |
|
| 1170 |
$this->_wp_id = wp_insert_post($dbpost); |
| 1171 |
|
| 1172 |
// Turn off ridiculous fucking kludges #1 and #2 |
| 1173 |
remove_action('_wp_put_post_revision', array($this, 'fix_revision_meta')); |
| 1174 |
remove_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11); |
| 1175 |
|
| 1176 |
$this->validate_post_id($dbpost, array(__CLASS__, __FUNCTION__)); |
| 1177 |
|
| 1178 |
// Unfortunately, as of WordPress 2.3, wp_insert_post() |
| 1179 |
// *still* offers no way to use a guid of your choice, |
| 1180 |
// and munges your post modified timestamp, too. |
| 1181 |
$result = $wpdb->query(" |
| 1182 |
UPDATE $wpdb->posts |
| 1183 |
SET |
| 1184 |
guid='{$dbpost['guid']}', |
| 1185 |
post_modified='{$dbpost['post_modified']}', |
| 1186 |
post_modified_gmt='{$dbpost['post_modified_gmt']}' |
| 1187 |
WHERE ID='{$this->_wp_id}' |
| 1188 |
"); |
| 1189 |
else : |
| 1190 |
# The right way to do this is the above. But, alas, |
| 1191 |
# in earlier versions of WordPress, wp_insert_post has |
| 1192 |
# too much behavior (mainly related to pings) that can't |
| 1193 |
# be overridden. In WordPress 1.5, it's enough of a |
| 1194 |
# resource hog to make PHP segfault after inserting |
| 1195 |
# 50-100 posts. This can get pretty annoying, especially |
| 1196 |
# if you are trying to update your feeds for the first |
| 1197 |
# time. |
| 1198 |
|
| 1199 |
$result = $wpdb->query(" |
| 1200 |
INSERT INTO $wpdb->posts |
| 1201 |
SET |
| 1202 |
guid = '{$dbpost['guid']}', |
| 1203 |
post_author = '{$dbpost['post_author']}', |
| 1204 |
post_date = '{$dbpost['post_date']}', |
| 1205 |
post_date_gmt = '{$dbpost['post_date_gmt']}', |
| 1206 |
post_content = '{$dbpost['post_content']}'," |
| 1207 |
.(isset($dbpost['post_excerpt']) ? "post_excerpt = '{$dbpost['post_excerpt']}'," : "")." |
| 1208 |
post_title = '{$dbpost['post_title']}', |
| 1209 |
post_name = '{$dbpost['post_name']}', |
| 1210 |
post_modified = '{$dbpost['post_modified']}', |
| 1211 |
post_modified_gmt = '{$dbpost['post_modified_gmt']}', |
| 1212 |
comment_status = '{$dbpost['comment_status']}', |
| 1213 |
ping_status = '{$dbpost['ping_status']}', |
| 1214 |
post_status = '{$dbpost['post_status']}' |
| 1215 |
"); |
| 1216 |
$this->_wp_id = $wpdb->insert_id; |
| 1217 |
|
| 1218 |
$this->validate_post_id($dbpost, array(__CLASS__, __FUNCTION__)); |
| 1219 |
|
| 1220 |
// WordPress 1.5.x - 2.0.x |
| 1221 |
wp_set_post_cats('1', $this->wp_id(), $this->post['post_category']); |
| 1222 |
|
| 1223 |
// Since we are not going through official channels, we need to |
| 1224 |
// manually tell WordPress that we've published a new post. |
| 1225 |
// We need to make sure to do this in order for FeedWordPress |
| 1226 |
// to play well with the staticize-reloaded plugin (something |
| 1227 |
// that a large aggregator website is going to *want* to be |
| 1228 |
// able to use). |
| 1229 |
do_action('publish_post', $this->_wp_id); |
| 1230 |
endif; |
| 1231 |
endif; |
| 1232 |
} /* SyndicatedPost::insert_new() */ |
| 1233 |
|
| 1234 |
function update_existing () { |
| 1235 |
global $wpdb; |
| 1236 |
|
| 1237 |
// Why the fuck doesn't wp_insert_post already do this? |
| 1238 |
$dbpost = $this->normalize_post(/*new=*/ false); |
| 1239 |
if (!is_null($dbpost)) : |
| 1240 |
if ($this->use_api('wp_insert_post')) : |
| 1241 |
$dbpost['post_pingback'] = false; // Tell WP 2.1 and 2.2 not to process for pingbacks |
| 1242 |
|
| 1243 |
// This is a ridiculous fucking kludge necessitated by WordPress 2.6 munging authorship meta-data |
| 1244 |
add_action('_wp_put_post_revision', array($this, 'fix_revision_meta')); |
| 1245 |
|
| 1246 |
// Kludge to prevent kses filters from stripping the |
| 1247 |
// content of posts when updating without a logged in |
| 1248 |
// user who has `unfiltered_html` capability. |
| 1249 |
add_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11); |
| 1250 |
|
| 1251 |
// Don't munge status fields that the user may have reset manually |
| 1252 |
if (function_exists('get_post_field')) : |
| 1253 |
$doNotMunge = array('post_status', 'comment_status', 'ping_status'); |
| 1254 |
foreach ($doNotMunge as $field) : |
| 1255 |
$dbpost[$field] = get_post_field($field, $this->wp_id()); |
| 1256 |
endforeach; |
| 1257 |
endif; |
| 1258 |
|
| 1259 |
$this->_wp_id = wp_insert_post($dbpost); |
| 1260 |
|
| 1261 |
// Turn off ridiculous fucking kludges #1 and #2 |
| 1262 |
remove_action('_wp_put_post_revision', array($this, 'fix_revision_meta')); |
| 1263 |
remove_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11); |
| 1264 |
|
| 1265 |
$this->validate_post_id($dbpost, array(__CLASS__, __FUNCTION__)); |
| 1266 |
|
| 1267 |
// Unfortunately, as of WordPress 2.3, wp_insert_post() |
| 1268 |
// munges your post modified timestamp. |
| 1269 |
$result = $wpdb->query(" |
| 1270 |
UPDATE $wpdb->posts |
| 1271 |
SET |
| 1272 |
post_modified='{$dbpost['post_modified']}', |
| 1273 |
post_modified_gmt='{$dbpost['post_modified_gmt']}' |
| 1274 |
WHERE ID='{$this->_wp_id}' |
| 1275 |
"); |
| 1276 |
else : |
| 1277 |
|
| 1278 |
$result = $wpdb->query(" |
| 1279 |
UPDATE $wpdb->posts |
| 1280 |
SET |
| 1281 |
post_author = '{$dbpost['post_author']}', |
| 1282 |
post_content = '{$dbpost['post_content']}'," |
| 1283 |
.(isset($dbpost['post_excerpt']) ? "post_excerpt = '{$dbpost['post_excerpt']}'," : "")." |
| 1284 |
post_title = '{$dbpost['post_title']}', |
| 1285 |
post_name = '{$dbpost['post_name']}', |
| 1286 |
post_modified = '{$dbpost['post_modified']}', |
| 1287 |
post_modified_gmt = '{$dbpost['post_modified_gmt']}' |
| 1288 |
WHERE guid='{$dbpost['guid']}' |
| 1289 |
"); |
| 1290 |
|
| 1291 |
// WordPress 2.1.x and up |
| 1292 |
if (function_exists('wp_set_post_categories')) : |
| 1293 |
wp_set_post_categories($this->wp_id(), $this->post['post_category']); |
| 1294 |
// WordPress 1.5.x - 2.0.x |
| 1295 |
elseif (function_exists('wp_set_post_cats')) : |
| 1296 |
wp_set_post_cats('1', $this->wp_id(), $this->post['post_category']); |
| 1297 |
// This should never happen. |
| 1298 |
else : |
| 1299 |
FeedWordPress::critical_bug(__CLASS__.'::'.__FUNCTION.'(): no post categorizing function', array("dbpost" => $dbpost, "this" => $this), __LINE__); |
| 1300 |
endif; |
| 1301 |
|
| 1302 |
// Since we are not going through official channels, we need to |
| 1303 |
// manually tell WordPress that we've published a new post. |
| 1304 |
// We need to make sure to do this in order for FeedWordPress |
| 1305 |
// to play well with the staticize-reloaded plugin (something |
| 1306 |
// that a large aggregator website is going to *want* to be |
| 1307 |
// able to use). |
| 1308 |
do_action('edit_post', $this->post['ID']); |
| 1309 |
endif; |
| 1310 |
endif; |
| 1311 |
} /* SyndicatedPost::update_existing() */ |
| 1312 |
|
| 1313 |
/** |
| 1314 |
* SyndicatedPost::normalize_post() |
| 1315 |
* |
| 1316 |
* @param bool $new If true, this post is to be inserted anew. If false, it is an update of an existing post. |
| 1317 |
* @return array A normalized representation of the post ready to be inserted into the database or sent to the WordPress API functions |
| 1318 |
*/ |
| 1319 |
function normalize_post ($new = true) { |
| 1320 |
global $wpdb; |
| 1321 |
|
| 1322 |
$out = array(); |
| 1323 |
|
| 1324 |
// Why the fuck doesn't wp_insert_post already do this? |
| 1325 |
foreach ($this->post as $key => $value) : |
| 1326 |
if (is_string($value)) : |
| 1327 |
$out[$key] = $wpdb->escape($value); |
| 1328 |
else : |
| 1329 |
$out[$key] = $value; |
| 1330 |
endif; |
| 1331 |
endforeach; |
| 1332 |
|
| 1333 |
if (strlen($out['post_title'].$out['post_content'].$out['post_excerpt']) == 0) : |
| 1334 |
// FIXME: Option for filtering out empty posts |
| 1335 |
endif; |
| 1336 |
if (strlen($out['post_title'])==0) : |
| 1337 |
$offset = (int) get_option('gmt_offset') * 60 * 60; |
| 1338 |
$out['post_title'] = |
| 1339 |
$this->post['meta']['syndication_source'] |
| 1340 |
.' '.gmdate('Y-m-d H:i:s', $this->published() + $offset); |
| 1341 |
// FIXME: Option for what to fill a blank title with... |
| 1342 |
endif; |
| 1343 |
|
| 1344 |
return $out; |
| 1345 |
} |
| 1346 |
|
| 1347 |
/** |
| 1348 |
* SyndicatedPost::validate_post_id() |
| 1349 |
* |
| 1350 |
* @param array $dbpost An array representing the post we attempted to insert or update |
| 1351 |
* @param mixed $ns A string or array representing the namespace (class, method) whence this method was called. |
| 1352 |
*/ |
| 1353 |
function validate_post_id ($dbpost, $ns) { |
| 1354 |
if (is_array($ns)) : $ns = implode('::', $ns); |
| 1355 |
else : $ns = (string) $ns; endif; |
| 1356 |
|
| 1357 |
// This should never happen. |
| 1358 |
if (!is_numeric($this->_wp_id) or ($this->_wp_id == 0)) : |
| 1359 |
FeedWordPress::critical_bug( |
| 1360 |
/*name=*/ $ns.'::_wp_id', |
| 1361 |
/*var =*/ array( |
| 1362 |
"\$this->_wp_id" => $this->_wp_id, |
| 1363 |
"\$dbpost" => $dbpost, |
| 1364 |
"\$this" => $this |
| 1365 |
), |
| 1366 |
/*line # =*/ __LINE__ |
| 1367 |
); |
| 1368 |
endif; |
| 1369 |
} /* SyndicatedPost::validate_post_id() */ |
| 1370 |
|
| 1371 |
/** |
| 1372 |
* SyndicatedPost::fix_revision_meta() - Fixes the way WP 2.6+ fucks up |
| 1373 |
* meta-data (authorship, etc.) when storing revisions of an updated |
| 1374 |
* syndicated post. |
| 1375 |
* |
| 1376 |
* In their infinite wisdom, the WordPress coders have made it completely |
| 1377 |
* impossible for a plugin that uses wp_insert_post() to set certain |
| 1378 |
* meta-data (such as the author) when you store an old revision of an |
| 1379 |
* updated post. Instead, it uses the WordPress defaults (= currently |
| 1380 |
* active user ID if the process is running with a user logged in, or |
| 1381 |
* = #0 if there is no user logged in). This results in bogus authorship |
| 1382 |
* data for revisions that are syndicated from off the feed, unless we |
| 1383 |
* use a ridiculous kludge like this to end-run the munging of meta-data |
| 1384 |
* by _wp_put_post_revision. |
| 1385 |
* |
| 1386 |
* @param int $revision_id The revision ID to fix up meta-data |
| 1387 |
*/ |
| 1388 |
function fix_revision_meta ($revision_id) { |
| 1389 |
global $wpdb; |
| 1390 |
|
| 1391 |
$post_author = (int) $this->post['post_author']; |
| 1392 |
|
| 1393 |
$revision_id = (int) $revision_id; |
| 1394 |
$wpdb->query(" |
| 1395 |
UPDATE $wpdb->posts |
| 1396 |
SET post_author={$this->post['post_author']} |
| 1397 |
WHERE post_type = 'revision' AND ID='$revision_id' |
| 1398 |
"); |
| 1399 |
} /* SyndicatedPost::fix_revision_meta () */ |
| 1400 |
|
| 1401 |
/** |
| 1402 |
* SyndicatedPost::avoid_kses_munge() -- If FeedWordPress is processing |
| 1403 |
* an automatic update, that generally means that wp_insert_post() is |
| 1404 |
* being called under the user credentials of whoever is viewing the |
| 1405 |
* blog at the time -- usually meaning no user at all. But if WordPress |
| 1406 |
* gets a wp_insert_post() when current_user_can('unfiltered_html') is |
| 1407 |
* false, it will run the content of the post through a kses function |
| 1408 |
* that strips out lots of HTML tags -- notably <object> and some others. |
| 1409 |
* This causes problems for syndicating (for example) feeds that contain |
| 1410 |
* YouTube videos. It also produces an unexpected asymmetry between |
| 1411 |
* automatically-initiated updates and updates initiated manually from |
| 1412 |
* the WordPress Dashboard (which are usually initiated under the |
| 1413 |
* credentials of a logged-in admin, and so don't get run through the |
| 1414 |
* kses function). So, to avoid the whole mess, what we do here is |
| 1415 |
* just forcibly disable the kses munging for a single syndicated post, |
| 1416 |
* by restoring the contents of the `post_content` field. |
| 1417 |
* |
| 1418 |
* @param string $content The content of the post, after other filters have gotten to it |
| 1419 |
* @return string The original content of the post, before other filters had a chance to munge it. |
| 1420 |
*/ |
| 1421 |
function avoid_kses_munge ($content) { |
| 1422 |
global $wpdb; |
| 1423 |
return $wpdb->escape($this->post['post_content']); |
| 1424 |
} |
| 1425 |
|
| 1426 |
// SyndicatedPost::add_rss_meta: adds interesting meta-data to each entry |
| 1427 |
// using the space for custom keys. The set of keys and values to add is |
| 1428 |
// specified by the keys and values of $post['meta']. This is used to |
| 1429 |
// store anything that the WordPress user might want to access from a |
| 1430 |
// template concerning the post's original source that isn't provided |
| 1431 |
// for by standard WP meta-data (i.e., any interesting data about the |
| 1432 |
// syndicated post other than author, title, timestamp, categories, and |
| 1433 |
// guid). It's also used to hook into WordPress's support for |
| 1434 |
// enclosures. |
| 1435 |
function add_rss_meta ($new_status, $old_status, $post) { |
| 1436 |
FeedWordPress::diagnostic('syndicated_posts:meta_data', 'Adding post meta-data: {'.implode(", ", array_keys($this->post['meta'])).'}'); |
| 1437 |
|
| 1438 |
global $wpdb; |
| 1439 |
if ( is_array($this->post) and isset($this->post['meta']) and is_array($this->post['meta']) ) : |
| 1440 |
$postId = $post->ID; |
| 1441 |
|
| 1442 |
// Aggregated posts should NOT send out pingbacks. |
| 1443 |
// WordPress 2.1-2.2 claim you can tell them not to |
| 1444 |
// using $post_pingback, but they don't listen, so we |
| 1445 |
// make sure here. |
| 1446 |
$result = $wpdb->query(" |
| 1447 |
DELETE FROM $wpdb->postmeta |
| 1448 |
WHERE post_id='$postId' AND meta_key='_pingme' |
| 1449 |
"); |
| 1450 |
|
| 1451 |
foreach ( $this->post['meta'] as $key => $values ) : |
| 1452 |
$eKey = $wpdb->escape($key); |
| 1453 |
|
| 1454 |
// If this is an update, clear out the old |
| 1455 |
// values to avoid duplication. |
| 1456 |
$result = $wpdb->query(" |
| 1457 |
DELETE FROM $wpdb->postmeta |
| 1458 |
WHERE post_id='$postId' AND meta_key='$eKey' |
| 1459 |
"); |
| 1460 |
|
| 1461 |
// Allow for either a single value or an array |
| 1462 |
if (!is_array($values)) $values = array($values); |
| 1463 |
foreach ( $values as $value ) : |
| 1464 |
FeedWordPress::diagnostic('syndicated_posts:meta_data', "Adding post meta-datum to post [$postId]: [$key] = ".FeedWordPress::val($value, /*no newlines=*/ true)); |
| 1465 |
add_post_meta($postId, $key, $value, /*unique=*/ false); |
| 1466 |
endforeach; |
| 1467 |
endforeach; |
| 1468 |
endif; |
| 1469 |
} /* SyndicatedPost::add_rss_meta () */ |
| 1470 |
|
| 1471 |
// SyndicatedPost::author_id (): get the ID for an author name from |
| 1472 |
// the feed. Create the author if necessary. |
| 1473 |
function author_id ($unfamiliar_author = 'create') { |
| 1474 |
global $wpdb; |
| 1475 |
|
| 1476 |
$a = $this->author(); |
| 1477 |
$author = $a['name']; |
| 1478 |
$email = (isset($a['email']) ? $a['email'] : NULL); |
| 1479 |
$url = (isset($a['uri']) ? $a['uri'] : NULL); |
| 1480 |
|
| 1481 |
$match_author_by_email = !('yes' == get_option("feedwordpress_do_not_match_author_by_email")); |
| 1482 |
if ($match_author_by_email and !FeedWordPress::is_null_email($email)) : |
| 1483 |
$test_email = $email; |
| 1484 |
else : |
| 1485 |
$test_email = NULL; |
| 1486 |
endif; |
| 1487 |
|
| 1488 |
// Never can be too careful... |
| 1489 |
$login = sanitize_user($author, /*strict=*/ true); |
| 1490 |
$login = apply_filters('pre_user_login', $login); |
| 1491 |
|
| 1492 |
$nice_author = sanitize_title($author); |
| 1493 |
$nice_author = apply_filters('pre_user_nicename', $nice_author); |
| 1494 |
|
| 1495 |
$reg_author = $wpdb->escape(preg_quote($author)); |
| 1496 |
$author = $wpdb->escape($author); |
| 1497 |
$email = $wpdb->escape($email); |
| 1498 |
$test_email = $wpdb->escape($test_email); |
| 1499 |
$url = $wpdb->escape($url); |
| 1500 |
|
| 1501 |
// Check for an existing author rule.... |
| 1502 |
if (isset($this->link->settings['map authors']['name'][strtolower(trim($author))])) : |
| 1503 |
$author_rule = $this->link->settings['map authors']['name'][strtolower(trim($author))]; |
| 1504 |
else : |
| 1505 |
$author_rule = NULL; |
| 1506 |
endif; |
| 1507 |
|
| 1508 |
// User name is mapped to a particular author. If that author ID exists, use it. |
| 1509 |
if (is_numeric($author_rule) and get_userdata((int) $author_rule)) : |
| 1510 |
$id = (int) $author_rule; |
| 1511 |
|
| 1512 |
// User name is filtered out |
| 1513 |
elseif ('filter' == $author_rule) : |
| 1514 |
$id = NULL; |
| 1515 |
|
| 1516 |
else : |
| 1517 |
// Check the database for an existing author record that might fit |
| 1518 |
|
| 1519 |
// First try the user core data table. |
| 1520 |
$id = $wpdb->get_var( |
| 1521 |
"SELECT ID FROM $wpdb->users |
| 1522 |
WHERE |
| 1523 |
TRIM(LCASE(user_login)) = TRIM(LCASE('$login')) |
| 1524 |
OR ( |
| 1525 |
LENGTH(TRIM(LCASE(user_email))) > 0 |
| 1526 |
AND TRIM(LCASE(user_email)) = TRIM(LCASE('$test_email')) |
| 1527 |
) |
| 1528 |
OR TRIM(LCASE(user_nicename)) = TRIM(LCASE('$nice_author')) |
| 1529 |
"); |
| 1530 |
|
| 1531 |
// If that fails, look for aliases in the user meta data table |
| 1532 |
if (is_null($id)) : |
| 1533 |
$id = $wpdb->get_var( |
| 1534 |
"SELECT user_id FROM $wpdb->usermeta |
| 1535 |
WHERE |
| 1536 |
(meta_key = 'description' AND TRIM(LCASE(meta_value)) = TRIM(LCASE('$author'))) |
| 1537 |
OR ( |
| 1538 |
meta_key = 'description' |
| 1539 |
AND TRIM(LCASE(meta_value)) |
| 1540 |
RLIKE CONCAT( |
| 1541 |
'(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*', |
| 1542 |
TRIM(LCASE('$reg_author')), |
| 1543 |
'( |\\t|\\r)*(\\n|\$)' |
| 1544 |
) |
| 1545 |
) |
| 1546 |
"); |
| 1547 |
endif; |
| 1548 |
|
| 1549 |
// ... if you don't find one, then do what you need to do |
| 1550 |
if (is_null($id)) : |
| 1551 |
if ($unfamiliar_author === 'create') : |
| 1552 |
$userdata = array(); |
| 1553 |
|
| 1554 |
// WordPress 3 is going to pitch a fit if we attempt to register |
| 1555 |
// more than one user account with an empty e-mail address, so we |
| 1556 |
// need *something* here. Ugh. |
| 1557 |
if (strlen($email) == 0 or FeedWordPress::is_null_email($email)) : |
| 1558 |
$url = parse_url($this->feed->channel['link']); |
| 1559 |
$email = $nice_author.'@'.$url['host']; |
| 1560 |
endif; |
| 1561 |
|
| 1562 |
#-- user table data |
| 1563 |
$userdata['ID'] = NULL; // new user |
| 1564 |
$userdata['user_login'] = $login; |
| 1565 |
$userdata['user_nicename'] = $nice_author; |
| 1566 |
$userdata['user_pass'] = substr(md5(uniqid(microtime())), 0, 6); // just something random to lock it up |
| 1567 |
$userdata['user_email'] = $email; |
| 1568 |
$userdata['user_url'] = $url; |
| 1569 |
$userdata['display_name'] = $author; |
| 1570 |
|
| 1571 |
$id = wp_insert_user($userdata); |
| 1572 |
elseif (is_numeric($unfamiliar_author) and get_userdata((int) $unfamiliar_author)) : |
| 1573 |
$id = (int) $unfamiliar_author; |
| 1574 |
elseif ($unfamiliar_author === 'default') : |
| 1575 |
$id = 1; |
| 1576 |
endif; |
| 1577 |
endif; |
| 1578 |
endif; |
| 1579 |
|
| 1580 |
if ($id) : |
| 1581 |
$this->link->settings['map authors']['name'][strtolower(trim($author))] = $id; |
| 1582 |
endif; |
| 1583 |
return $id; |
| 1584 |
} // function SyndicatedPost::author_id () |
| 1585 |
|
| 1586 |
// look up (and create) category ids from a list of categories |
| 1587 |
function category_ids ($cats, $unfamiliar_category = 'create', $tags_too = false) { |
| 1588 |
global $wpdb; |
| 1589 |
|
| 1590 |
// We need to normalize whitespace because (1) trailing |
| 1591 |
// whitespace can cause PHP and MySQL not to see eye to eye on |
| 1592 |
// VARCHAR comparisons for some versions of MySQL (cf. |
| 1593 |
// <http://dev.mysql.com/doc/mysql/en/char.html>), and (2) |
| 1594 |
// because I doubt most people want to make a semantic |
| 1595 |
// distinction between 'Computers' and 'Computers ' |
| 1596 |
$cats = array_map('trim', $cats); |
| 1597 |
|
| 1598 |
$tags = array(); |
| 1599 |
|
| 1600 |
$cat_ids = array (); |
| 1601 |
foreach ($cats as $cat_name) : |
| 1602 |
if (preg_match('/^{#([0-9]+)}$/', $cat_name, $backref)) : |
| 1603 |
$cat_id = (int) $backref[1]; |
| 1604 |
if (function_exists('is_term') and is_term($cat_id, 'category')) : |
| 1605 |
$cat_ids[] = $cat_id; |
| 1606 |
elseif (get_category($cat_id)) : |
| 1607 |
$cat_ids[] = $cat_id; |
| 1608 |
endif; |
| 1609 |
elseif (strlen($cat_name) > 0) : |
| 1610 |
$esc = $wpdb->escape($cat_name); |
| 1611 |
$resc = $wpdb->escape(preg_quote($cat_name)); |
| 1612 |
|
| 1613 |
// WordPress 2.3+ |
| 1614 |
if (function_exists('is_term')) : |
| 1615 |
$cat_id = is_term($cat_name, 'category'); |
| 1616 |
if ($cat_id) : |
| 1617 |
$cat_ids[] = $cat_id['term_id']; |
| 1618 |
// There must be a better way to do this... |
| 1619 |
elseif ($results = $wpdb->get_results( |
| 1620 |
"SELECT term_id |
| 1621 |
FROM $wpdb->term_taxonomy |
| 1622 |
WHERE |
| 1623 |
LOWER(description) RLIKE |
| 1624 |
CONCAT('(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*', LOWER('{$resc}'), '( |\\t|\\r)*(\\n|\$)')" |
| 1625 |
)) : |
| 1626 |
foreach ($results AS $term) : |
| 1627 |
$cat_ids[] = (int) $term->term_id; |
| 1628 |
endforeach; |
| 1629 |
elseif ('tag'==$unfamiliar_category) : |
| 1630 |
$tags[] = $cat_name; |
| 1631 |
elseif ('create'===$unfamiliar_category) : |
| 1632 |
$term = wp_insert_term($cat_name, 'category'); |
| 1633 |
if (is_wp_error($term)) : |
| 1634 |
FeedWordPress::noncritical_bug('term insertion problem', array('cat_name' => $cat_name, 'term' => $term, 'this' => $this), __LINE__); |
| 1635 |
else : |
| 1636 |
$cat_ids[] = $term['term_id']; |
| 1637 |
endif; |
| 1638 |
endif; |
| 1639 |
|
| 1640 |
// WordPress 1.5.x - 2.2.x |
| 1641 |
else : |
| 1642 |
$results = $wpdb->get_results( |
| 1643 |
"SELECT cat_ID |
| 1644 |
FROM $wpdb->categories |
| 1645 |
WHERE |
| 1646 |
(LOWER(cat_name) = LOWER('$esc')) |
| 1647 |
OR (LOWER(category_description) |
| 1648 |
RLIKE CONCAT('(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*', LOWER('{$resc}'), '( |\\t|\\r)*(\\n|\$)')) |
| 1649 |
"); |
| 1650 |
if ($results) : |
| 1651 |
foreach ($results as $term) : |
| 1652 |
$cat_ids[] = (int) $term->cat_ID; |
| 1653 |
endforeach; |
| 1654 |
elseif ('create'===$unfamiliar_category) : |
| 1655 |
if (function_exists('wp_insert_category')) : |
| 1656 |
$cat_id = wp_insert_category(array('cat_name' => $esc)); |
| 1657 |
// And into the database we go. |
| 1658 |
else : |
| 1659 |
$nice_kitty = sanitize_title($cat_name); |
| 1660 |
$wpdb->query(sprintf(" |
| 1661 |
INSERT INTO $wpdb->categories |
| 1662 |
SET |
| 1663 |
cat_name='%s', |
| 1664 |
category_nicename='%s' |
| 1665 |
", $esc, $nice_kitty |
| 1666 |
)); |
| 1667 |
$cat_id = $wpdb->insert_id; |
| 1668 |
endif; |
| 1669 |
$cat_ids[] = $cat_id; |
| 1670 |
endif; |
| 1671 |
endif; |
| 1672 |
endif; |
| 1673 |
endforeach; |
| 1674 |
|
| 1675 |
if ((count($cat_ids) == 0) and ($unfamiliar_category === 'filter')) : |
| 1676 |
$cat_ids = NULL; // Drop the post |
| 1677 |
else : |
| 1678 |
$cat_ids = array_unique($cat_ids); |
| 1679 |
endif; |
| 1680 |
|
| 1681 |
if ($tags_too) : $ret = array($cat_ids, $tags); |
| 1682 |
else : $ret = $cat_ids; |
| 1683 |
endif; |
| 1684 |
|
| 1685 |
return $ret; |
| 1686 |
} // function SyndicatedPost::category_ids () |
| 1687 |
|
| 1688 |
function use_api ($tag) { |
| 1689 |
global $wp_db_version; |
| 1690 |
switch ($tag) : |
| 1691 |
case 'wp_insert_post': |
| 1692 |
// Before 2.2, wp_insert_post does too much of the wrong stuff to use it |
| 1693 |
// In 1.5 it was such a resource hog it would make PHP segfault on big updates |
| 1694 |
$ret = (isset($wp_db_version) and $wp_db_version > FWP_SCHEMA_21); |
| 1695 |
break; |
| 1696 |
case 'post_status_pending': |
| 1697 |
$ret = (isset($wp_db_version) and $wp_db_version > FWP_SCHEMA_23); |
| 1698 |
break; |
| 1699 |
endswitch; |
| 1700 |
return $ret; |
| 1701 |
} // function SyndicatedPost::use_api () |
| 1702 |
|
| 1703 |
} /* class SyndicatedPost */ |
| 1704 |
|
| 1705 |
|
