| 1 |
<?php |
| 2 |
require_once(dirname(__FILE__).'/feedtime.class.php'); |
| 3 |
require_once(dirname(__FILE__).'/syndicatedpostterm.class.php'); |
| 4 |
require_once(dirname(__FILE__).'/syndicatedpostxpathquery.class.php'); |
| 5 |
|
| 6 |
/** |
| 7 |
* class SyndicatedPost: FeedWordPress uses to manage the conversion of |
| 8 |
* incoming items from the feed parser into posts for the WordPress |
| 9 |
* database. It contains several internal management methods primarily |
| 10 |
* of interest to someone working on the FeedWordPress source, as well |
| 11 |
* as some utility methods for extracting useful data from many |
| 12 |
* different feed formats, which may be useful to FeedWordPress users |
| 13 |
* who make use of feed data in PHP add-ons and filters. |
| 14 |
* |
| 15 |
* @version 2013.0525 |
| 16 |
*/ |
| 17 |
class SyndicatedPost { |
| 18 |
var $item = null; // MagpieRSS representation |
| 19 |
var $entry = null; // SimplePie_Item representation |
| 20 |
|
| 21 |
var $link = null; |
| 22 |
var $feed = null; |
| 23 |
var $feedmeta = null; |
| 24 |
|
| 25 |
var $xmlns = array (); |
| 26 |
|
| 27 |
var $post = array (); |
| 28 |
|
| 29 |
var $named = array (); |
| 30 |
var $preset_terms = array (); |
| 31 |
var $feed_terms = array (); |
| 32 |
|
| 33 |
var $_freshness = null; |
| 34 |
var $_wp_id = null; |
| 35 |
var $_wp_post = null; |
| 36 |
|
| 37 |
/** |
| 38 |
* SyndicatedPost constructor: Given a feed item and the source from |
| 39 |
* which it was taken, prepare a post that can be inserted into the |
| 40 |
* WordPress database on request, or updated in place if it has already |
| 41 |
* been syndicated. |
| 42 |
* |
| 43 |
* @param array $item The item syndicated from the feed. |
| 44 |
* @param SyndicatedLink $source The feed it was syndicated from. |
| 45 |
*/ |
| 46 |
function __construct ($item, &$source) { |
| 47 |
global $wpdb; |
| 48 |
|
| 49 |
if ( empty($item) && empty($source) ) |
| 50 |
return; |
| 51 |
|
| 52 |
if (is_array($item) |
| 53 |
and isset($item['simplepie']) |
| 54 |
and isset($item['magpie'])) : |
| 55 |
$this->entry = $item['simplepie']; |
| 56 |
$this->item = $item['magpie']; |
| 57 |
$item = $item['magpie']; |
| 58 |
elseif (is_a($item, 'SimplePie_Item')) : |
| 59 |
$this->entry = $item; |
| 60 |
|
| 61 |
// convert to Magpie for compat purposes |
| 62 |
$mp = new MagpieFromSimplePie($source->simplepie, $this->entry); |
| 63 |
$this->item = $mp->get_item(); |
| 64 |
|
| 65 |
// done with conversion object |
| 66 |
$mp = NULL; unset($mp); |
| 67 |
else : |
| 68 |
$this->item = $item; |
| 69 |
endif; |
| 70 |
|
| 71 |
$this->link =& $source; |
| 72 |
$this->feed = $source->magpie; |
| 73 |
$this->feedmeta = $source->settings; |
| 74 |
|
| 75 |
FeedWordPress::diagnostic('feed_items', 'Considering item ['.$this->guid().'] "'.$this->entry->get_title().'"'); |
| 76 |
|
| 77 |
# Dealing with namespaces can get so fucking fucked. |
| 78 |
$this->xmlns['forward'] = $source->magpie->_XMLNS_FAMILIAR; |
| 79 |
$this->xmlns['reverse'] = array(); |
| 80 |
foreach ($this->xmlns['forward'] as $url => $ns) : |
| 81 |
if (!isset($this->xmlns['reverse'][$ns])) : |
| 82 |
$this->xmlns['reverse'][$ns] = array(); |
| 83 |
endif; |
| 84 |
$this->xmlns['reverse'][$ns][] = $url; |
| 85 |
endforeach; |
| 86 |
|
| 87 |
// Fucking SimplePie. |
| 88 |
$this->xmlns['reverse']['rss'][] = ''; |
| 89 |
|
| 90 |
// Trigger global syndicated_item filter. |
| 91 |
$changed = apply_filters('syndicated_item', $this->item, $this); |
| 92 |
$this->item = $changed; |
| 93 |
|
| 94 |
// Allow for feed-specific syndicated_item filters. |
| 95 |
$changed = apply_filters( |
| 96 |
"syndicated_item_".$source->uri(), |
| 97 |
$this->item, |
| 98 |
$this |
| 99 |
); |
| 100 |
$this->item = $changed; |
| 101 |
|
| 102 |
# Filters can halt further processing by returning NULL |
| 103 |
if (is_null($this->item)) : |
| 104 |
$this->post = NULL; |
| 105 |
else : |
| 106 |
|
| 107 |
# Note that nothing is run through esc_sql() here. |
| 108 |
# That's deliberate. The escaping is done at the point |
| 109 |
# of insertion, not here, to avoid double-escaping and |
| 110 |
# to avoid screwing with syndicated_post filters |
| 111 |
|
| 112 |
$this->post['post_title'] = apply_filters( |
| 113 |
'syndicated_item_title', |
| 114 |
$this->entry->get_title(), $this |
| 115 |
); |
| 116 |
|
| 117 |
|
| 118 |
$this->named['author'] = apply_filters( |
| 119 |
'syndicated_item_author', |
| 120 |
$this->author(), $this |
| 121 |
); |
| 122 |
// This just gives us an alphanumeric name for the author. |
| 123 |
// We look up (or create) the numeric ID for the author |
| 124 |
// in SyndicatedPost::add(). |
| 125 |
|
| 126 |
$this->post['post_content'] = apply_filters( |
| 127 |
'syndicated_item_content', |
| 128 |
$this->content(), $this |
| 129 |
); |
| 130 |
|
| 131 |
$excerpt = apply_filters('syndicated_item_excerpt', $this->excerpt(), $this); |
| 132 |
if (!empty($excerpt)): |
| 133 |
$this->post['post_excerpt'] = $excerpt; |
| 134 |
endif; |
| 135 |
|
| 136 |
// Dealing with timestamps in WordPress is so fucking fucked. |
| 137 |
$offset = (int) get_option('gmt_offset') * 60 * 60; |
| 138 |
$post_date_gmt = $this->published(array('default' => -1)); |
| 139 |
$post_modified_gmt = $this->updated(array('default' => -1)); |
| 140 |
|
| 141 |
$this->post['post_date_gmt'] = gmdate('Y-m-d H:i:s', $post_date_gmt); |
| 142 |
$this->post['post_date'] = gmdate('Y-m-d H:i:s', $post_date_gmt + $offset); |
| 143 |
$this->post['post_modified_gmt'] = gmdate('Y-m-d H:i:s', $post_modified_gmt); |
| 144 |
$this->post['post_modified'] = gmdate('Y-m-d H:i:s', $post_modified_gmt + $offset); |
| 145 |
|
| 146 |
// Use feed-level preferences or the global default. |
| 147 |
$this->post['post_status'] = $this->link->syndicated_status('post', 'publish'); |
| 148 |
$this->post['comment_status'] = $this->link->syndicated_status('comment', 'closed'); |
| 149 |
$this->post['ping_status'] = $this->link->syndicated_status('ping', 'closed'); |
| 150 |
|
| 151 |
// Unique ID (hopefully a unique tag: URI); failing that, the permalink |
| 152 |
$this->post['guid'] = apply_filters('syndicated_item_guid', $this->guid(), $this); |
| 153 |
|
| 154 |
// User-supplied custom settings to apply to each post. |
| 155 |
// Do first so that FWP-generated custom settings will |
| 156 |
// overwrite if necessary; thus preventing any munging. |
| 157 |
$postMetaIn = $this->link->postmeta(array("parsed" => true)); |
| 158 |
$postMetaOut = array(); |
| 159 |
|
| 160 |
foreach ($postMetaIn as $key => $meta) : |
| 161 |
$postMetaOut[$key] = $meta->do_substitutions($this); |
| 162 |
endforeach; |
| 163 |
|
| 164 |
foreach ($postMetaOut as $key => $values) : |
| 165 |
$this->post['meta'][$key] = array(); |
| 166 |
foreach ($values as $value) : |
| 167 |
$this->post['meta'][$key][] = apply_filters("syndicated_post_meta_{$key}", $value, $this); |
| 168 |
endforeach; |
| 169 |
endforeach; |
| 170 |
|
| 171 |
// RSS 2.0 / Atom 1.0 enclosure support |
| 172 |
$enclosures = $this->entry->get_enclosures(); |
| 173 |
if (is_array($enclosures)) : foreach ($enclosures as $enclosure) : |
| 174 |
$this->post['meta']['enclosure'][] = |
| 175 |
apply_filters('syndicated_item_enclosure_url', $enclosure->get_link(), $this)."\n". |
| 176 |
apply_filters('syndicated_item_enclosure_length', $enclosure->get_length(), $this)."\n". |
| 177 |
apply_filters('syndicated_item_enclosure_type', $enclosure->get_type(), $this); |
| 178 |
endforeach; endif; |
| 179 |
|
| 180 |
// In case you want to point back to the blog this was |
| 181 |
// syndicated from. |
| 182 |
|
| 183 |
$sourcemeta['syndication_source'] = apply_filters( |
| 184 |
'syndicated_item_source_title', |
| 185 |
$this->link->name(), |
| 186 |
$this |
| 187 |
); |
| 188 |
$sourcemeta['syndication_source_uri'] = apply_filters( |
| 189 |
'syndicated_item_source_link', |
| 190 |
$this->link->homepage(), |
| 191 |
$this |
| 192 |
); |
| 193 |
$sourcemeta['syndication_source_id'] = apply_filters( |
| 194 |
'syndicated_item_source_id', |
| 195 |
$this->link->guid(), |
| 196 |
$this |
| 197 |
); |
| 198 |
|
| 199 |
// Make use of atom:source data, if present in an aggregated feed |
| 200 |
$entry_source = $this->source(); |
| 201 |
if (!is_null($entry_source)) : |
| 202 |
foreach ($entry_source as $what => $value) : |
| 203 |
if (!is_null($value)) : |
| 204 |
if ($what=='title') : $key = 'syndication_source'; |
| 205 |
elseif ($what=='feed') : $key = 'syndication_feed'; |
| 206 |
else : $key = "syndication_source_${what}"; |
| 207 |
endif; |
| 208 |
|
| 209 |
$sourcemeta["${key}_original"] = apply_filters( |
| 210 |
'syndicated_item_original_source_'.$what, |
| 211 |
$value, |
| 212 |
$this |
| 213 |
); |
| 214 |
endif; |
| 215 |
endforeach; |
| 216 |
endif; |
| 217 |
|
| 218 |
foreach ($sourcemeta as $meta_key => $value) : |
| 219 |
if (!is_null($value)) : |
| 220 |
$this->post['meta'][$meta_key] = $value; |
| 221 |
endif; |
| 222 |
endforeach; |
| 223 |
|
| 224 |
// Store information on human-readable and machine-readable comment URIs |
| 225 |
|
| 226 |
// Human-readable comment URI |
| 227 |
$commentLink = apply_filters('syndicated_item_comments', $this->comment_link(), $this); |
| 228 |
if (!is_null($commentLink)) : $this->post['meta']['rss:comments'] = $commentLink; endif; |
| 229 |
|
| 230 |
// Machine-readable content feed URI |
| 231 |
$commentFeed = apply_filters('syndicated_item_commentrss', $this->comment_feed(), $this); |
| 232 |
if (!is_null($commentFeed)) : $this->post['meta']['wfw:commentRSS'] = $commentFeed; endif; |
| 233 |
// Yeah, yeah, now I know that it's supposed to be |
| 234 |
// wfw:commentRss. Oh well. Path dependence, sucka. |
| 235 |
|
| 236 |
// Store information to identify the feed that this came from |
| 237 |
if (isset($this->feedmeta['link/uri'])) : |
| 238 |
$this->post['meta']['syndication_feed'] = $this->feedmeta['link/uri']; |
| 239 |
endif; |
| 240 |
if (isset($this->feedmeta['link/id'])) : |
| 241 |
$this->post['meta']['syndication_feed_id'] = $this->feedmeta['link/id']; |
| 242 |
endif; |
| 243 |
|
| 244 |
if (isset($this->item['source_link_self'])) : |
| 245 |
$this->post['meta']['syndication_feed_original'] = $this->item['source_link_self']; |
| 246 |
endif; |
| 247 |
|
| 248 |
// In case you want to know the external permalink... |
| 249 |
$this->post['meta']['syndication_permalink'] = apply_filters('syndicated_item_link', $this->permalink()); |
| 250 |
|
| 251 |
// Store a hash of the post content for checking whether something needs to be updated |
| 252 |
$this->post['meta']['syndication_item_hash'] = $this->update_hash(); |
| 253 |
|
| 254 |
// Categories, Tags, and other Terms: from settings assignments (global settings, subscription settings), |
| 255 |
// and from feed assignments (item metadata, post content) |
| 256 |
$this->preset_terms = apply_filters('syndicated_item_preset_terms', $this->get_terms_from_settings(), $this); |
| 257 |
$this->feed_terms = apply_filters('syndicated_item_feed_terms', $this->get_terms_from_feeds(), $this); |
| 258 |
|
| 259 |
$this->post['post_type'] = apply_filters('syndicated_post_type', $this->link->setting('syndicated post type', 'syndicated_post_type', 'post'), $this); |
| 260 |
endif; |
| 261 |
|
| 262 |
} /* SyndicatedPost::__construct() */ |
| 263 |
|
| 264 |
##################################### |
| 265 |
#### EXTRACT DATA FROM FEED ITEM #### |
| 266 |
##################################### |
| 267 |
|
| 268 |
function substitution_function ($name) { |
| 269 |
$ret = NULL; |
| 270 |
|
| 271 |
switch ($name) : |
| 272 |
// Allowed PHP string functions |
| 273 |
case 'trim': |
| 274 |
case 'ltrim': |
| 275 |
case 'rtrim': |
| 276 |
case 'strtoupper': |
| 277 |
case 'strtolower': |
| 278 |
case 'urlencode': |
| 279 |
case 'urldecode': |
| 280 |
$ret = $name; |
| 281 |
endswitch; |
| 282 |
return $ret; |
| 283 |
} |
| 284 |
|
| 285 |
/** |
| 286 |
* SyndicatedPost::query uses an XPath-like syntax to query arbitrary |
| 287 |
* elements within the syndicated item. |
| 288 |
* |
| 289 |
* @param string $path |
| 290 |
* @returns array of string values representing contents of matching |
| 291 |
* elements or attributes |
| 292 |
*/ |
| 293 |
function query ($path) { |
| 294 |
$xq = new SyndicatedPostXPathQuery(array("path" => $path)); |
| 295 |
|
| 296 |
$feedChannel = array_merge( |
| 297 |
$this->get_feed_root_element(), |
| 298 |
$this->get_feed_channel_elements() |
| 299 |
); |
| 300 |
|
| 301 |
$matches = $xq->match(array( |
| 302 |
"type" => $this->link->simplepie->get_type(), |
| 303 |
"xmlns" => $this->xmlns, |
| 304 |
"map" => array( |
| 305 |
"/" => array($this->entry->data), |
| 306 |
"item" => array($this->entry->data), |
| 307 |
"feed" => $feedChannel, |
| 308 |
"channel" => $feedChannel |
| 309 |
), |
| 310 |
"context" => $this->entry->data, |
| 311 |
"parent" => $feedChannel, |
| 312 |
)); |
| 313 |
|
| 314 |
return $matches; |
| 315 |
} /* SyndicatedPost::query() */ |
| 316 |
|
| 317 |
function get_feed_root_element () { |
| 318 |
$matches = array(); |
| 319 |
foreach ($this->link->simplepie->data['child'] as $ns => $root) : |
| 320 |
foreach ($root as $element => $data) : |
| 321 |
$matches = array_merge($matches, $data); |
| 322 |
endforeach; |
| 323 |
endforeach; |
| 324 |
return $matches; |
| 325 |
} /* SyndicatedPost::get_feed_root_element() */ |
| 326 |
|
| 327 |
function get_feed_channel_elements () { |
| 328 |
$rss = array( |
| 329 |
SIMPLEPIE_NAMESPACE_RSS_090, |
| 330 |
SIMPLEPIE_NAMESPACE_RSS_10, |
| 331 |
'http://backend.userland.com/RSS2', |
| 332 |
SIMPLEPIE_NAMESPACE_RSS_20, |
| 333 |
); |
| 334 |
|
| 335 |
$matches = array(); |
| 336 |
foreach ($rss as $ns) : |
| 337 |
$data = $this->link->simplepie->get_feed_tags($ns, 'channel'); |
| 338 |
if (!is_null($data)) : |
| 339 |
$matches = array_merge($matches, $data); |
| 340 |
endif; |
| 341 |
endforeach; |
| 342 |
return $matches; |
| 343 |
} /* SyndicatedPost::get_feed_channel_elements() */ |
| 344 |
|
| 345 |
function get_categories ($params = array()) { |
| 346 |
return $this->entry->get_categories(); |
| 347 |
} |
| 348 |
|
| 349 |
function title ($params = array()) { |
| 350 |
return $this->entry->get_title(); |
| 351 |
} /* SyndicatedPost::title () */ |
| 352 |
|
| 353 |
function content ($params = array()) { |
| 354 |
|
| 355 |
$params = wp_parse_args($params, array( |
| 356 |
"full only" => false, |
| 357 |
)); |
| 358 |
|
| 359 |
$content = NULL; |
| 360 |
|
| 361 |
// FIXME: This is one of the main places in the code still using |
| 362 |
// the outmoded SimplePie - to - Magpie construction. We could |
| 363 |
// replace using SimplePie_Item::get_tags() here. (Or if really |
| 364 |
// ambitious we could attempt to just use |
| 365 |
// SimplePie_Item::get_content() with content-only set to TRUE |
| 366 |
// and some sanitization in effect. -CJ 1jul14 |
| 367 |
|
| 368 |
// atom:content, standard method of providing full HTML content |
| 369 |
// in Atom feeds. |
| 370 |
if (isset($this->item['atom_content'])) : |
| 371 |
$content = $this->item['atom_content']; |
| 372 |
elseif (isset($this->item['atom']['atom_content'])) : |
| 373 |
$content = $this->item['atom']['atom_content']; |
| 374 |
|
| 375 |
// Some exotics: back in the day, before widespread convergence |
| 376 |
// on content:encoding, some RSS feeds took advantage of XML |
| 377 |
// namespaces to use an inline xhtml:body or xhtml:div element |
| 378 |
// for full-content HTML. (E.g. Sam Ruby's feed, IIRC.) |
| 379 |
elseif (isset($this->item['xhtml']['body'])) : |
| 380 |
$content = $this->item['xhtml']['body']; |
| 381 |
elseif (isset($this->item['xhtml']['div'])) : |
| 382 |
$content = $this->item['xhtml']['div']; |
| 383 |
|
| 384 |
// content:encoded, most common method of providing full HTML in |
| 385 |
// RSS 2.0 feeds. |
| 386 |
elseif (isset($this->item['content']['encoded']) and $this->item['content']['encoded']): |
| 387 |
$content = $this->item['content']['encoded']; |
| 388 |
|
| 389 |
// Fall back on elements that sometimes may contain full HTML |
| 390 |
// but sometimes not. |
| 391 |
elseif (!$params['full only']) : |
| 392 |
|
| 393 |
// description element is sometimes used for full HTML |
| 394 |
// sometimes for summary text in RSS. (By the letter of |
| 395 |
// the standard, it should just be for summary text.) |
| 396 |
if (isset($this->item['description'])) : |
| 397 |
$content = $this->item['description']; |
| 398 |
endif; |
| 399 |
|
| 400 |
endif; |
| 401 |
|
| 402 |
return $content; |
| 403 |
} /* SyndicatedPost::content() */ |
| 404 |
|
| 405 |
function excerpt () { |
| 406 |
# Identify and sanitize excerpt: atom:summary, or rss:description |
| 407 |
$excerpt = $this->entry->get_description(); |
| 408 |
|
| 409 |
# Many RSS feeds use rss:description, inadvisably, to |
| 410 |
# carry the entire post (typically with escaped HTML). |
| 411 |
# If that's what happened, we don't want the full |
| 412 |
# content for the excerpt. |
| 413 |
$content = $this->content(); |
| 414 |
|
| 415 |
// Ignore whitespace, case, and tag cruft. |
| 416 |
$theExcerpt = preg_replace('/\s+/', '', strtolower(strip_tags($excerpt))); |
| 417 |
$theContent = preg_replace('/\s+/', '', strtolower(strip_tags($content))); |
| 418 |
|
| 419 |
if ( empty($excerpt) or $theExcerpt == $theContent ) : |
| 420 |
# If content is available, generate an excerpt. |
| 421 |
if ( strlen(trim($content)) > 0 ) : |
| 422 |
$excerpt = strip_tags($content); |
| 423 |
if (strlen($excerpt) > 255) : |
| 424 |
$excerpt = substr($excerpt,0,252).'...'; |
| 425 |
endif; |
| 426 |
endif; |
| 427 |
endif; |
| 428 |
return $excerpt; |
| 429 |
} /* SyndicatedPost::excerpt() */ |
| 430 |
|
| 431 |
function permalink () { |
| 432 |
// Handles explicit <link> elements and also RSS 2.0 cases with |
| 433 |
// <guid isPermaLink="true">, etc. Hooray! |
| 434 |
$permalink = $this->entry->get_link(); |
| 435 |
return $permalink; |
| 436 |
} |
| 437 |
|
| 438 |
function created ($params = array()) { |
| 439 |
$unfiltered = false; $default = NULL; |
| 440 |
extract($params); |
| 441 |
|
| 442 |
$date = ''; |
| 443 |
if (isset($this->item['dc']['created'])) : |
| 444 |
$date = $this->item['dc']['created']; |
| 445 |
elseif (isset($this->item['dcterms']['created'])) : |
| 446 |
$date = $this->item['dcterms']['created']; |
| 447 |
elseif (isset($this->item['created'])): // Atom 0.3 |
| 448 |
$date = $this->item['created']; |
| 449 |
endif; |
| 450 |
|
| 451 |
$time = new FeedTime($date); |
| 452 |
$ts = $time->timestamp(); |
| 453 |
if (!$unfiltered) : |
| 454 |
apply_filters('syndicated_item_created', $ts, $this); |
| 455 |
endif; |
| 456 |
return $ts; |
| 457 |
} /* SyndicatedPost::created() */ |
| 458 |
|
| 459 |
function published ($params = array(), $default = NULL) { |
| 460 |
$fallback = true; $unfiltered = false; |
| 461 |
if (!is_array($params)) : // Old style |
| 462 |
$fallback = $params; |
| 463 |
else : // New style |
| 464 |
extract($params); |
| 465 |
endif; |
| 466 |
|
| 467 |
$date = ''; |
| 468 |
$ts = null; |
| 469 |
|
| 470 |
# RSS is a fucking mess. Figure out whether we have a date in |
| 471 |
# <dc:date>, <issued>, <pubDate>, etc., and get it into Unix |
| 472 |
# epoch format for reformatting. If we can't find anything, |
| 473 |
# we'll use the last-updated time. |
| 474 |
if (isset($this->item['dc']['date'])): // Dublin Core |
| 475 |
$date = $this->item['dc']['date']; |
| 476 |
elseif (isset($this->item['dcterms']['issued'])) : // Dublin Core extensions |
| 477 |
$date = $this->item['dcterms']['issued']; |
| 478 |
elseif (isset($this->item['published'])) : // Atom 1.0 |
| 479 |
$date = $this->item['published']; |
| 480 |
elseif (isset($this->item['issued'])): // Atom 0.3 |
| 481 |
$date = $this->item['issued']; |
| 482 |
elseif (isset($this->item['pubdate'])): // RSS 2.0 |
| 483 |
$date = $this->item['pubdate']; |
| 484 |
endif; |
| 485 |
|
| 486 |
if (strlen($date) > 0) : |
| 487 |
$time = new FeedTime($date); |
| 488 |
$ts = $time->timestamp(); |
| 489 |
elseif ($fallback) : // Fall back to <updated> / <modified> if present |
| 490 |
$ts = $this->updated(/*fallback=*/ false, /*default=*/ $default); |
| 491 |
endif; |
| 492 |
|
| 493 |
# If everything failed, then default to the current time. |
| 494 |
if (is_null($ts)) : |
| 495 |
if (-1 == $default) : |
| 496 |
$ts = time(); |
| 497 |
else : |
| 498 |
$ts = $default; |
| 499 |
endif; |
| 500 |
endif; |
| 501 |
|
| 502 |
if (!$unfiltered) : |
| 503 |
$ts = apply_filters('syndicated_item_published', $ts, $this); |
| 504 |
endif; |
| 505 |
return $ts; |
| 506 |
} /* SyndicatedPost::published() */ |
| 507 |
|
| 508 |
function updated ($params = array(), $default = -1) { |
| 509 |
$fallback = true; $unfiltered = false; |
| 510 |
if (!is_array($params)) : // Old style |
| 511 |
$fallback = $params; |
| 512 |
else : // New style |
| 513 |
extract($params); |
| 514 |
endif; |
| 515 |
|
| 516 |
$date = ''; |
| 517 |
$ts = null; |
| 518 |
|
| 519 |
# As far as I know, only dcterms and Atom have reliable ways to |
| 520 |
# specify when something was *modified* last. If neither is |
| 521 |
# available, then we'll try to get the time of publication. |
| 522 |
if (isset($this->item['dc']['modified'])) : // Not really correct |
| 523 |
$date = $this->item['dc']['modified']; |
| 524 |
elseif (isset($this->item['dcterms']['modified'])) : // Dublin Core extensions |
| 525 |
$date = $this->item['dcterms']['modified']; |
| 526 |
elseif (isset($this->item['modified'])): // Atom 0.3 |
| 527 |
$date = $this->item['modified']; |
| 528 |
elseif (isset($this->item['updated'])): // Atom 1.0 |
| 529 |
$date = $this->item['updated']; |
| 530 |
endif; |
| 531 |
|
| 532 |
if (strlen($date) > 0) : |
| 533 |
$time = new FeedTime($date); |
| 534 |
$ts = $time->timestamp(); |
| 535 |
elseif ($fallback) : // Fall back to issued / dc:date |
| 536 |
$ts = $this->published(/*fallback=*/ false, /*default=*/ $default); |
| 537 |
endif; |
| 538 |
|
| 539 |
# If everything failed, then default to the current time. |
| 540 |
if (is_null($ts)) : |
| 541 |
if (-1 == $default) : |
| 542 |
$ts = time(); |
| 543 |
else : |
| 544 |
$ts = $default; |
| 545 |
endif; |
| 546 |
endif; |
| 547 |
|
| 548 |
if (!$unfiltered) : |
| 549 |
apply_filters('syndicated_item_updated', $ts, $this); |
| 550 |
endif; |
| 551 |
return $ts; |
| 552 |
} /* SyndicatedPost::updated() */ |
| 553 |
|
| 554 |
var $_hashes = array(); |
| 555 |
function stored_hashes ($id = NULL) { |
| 556 |
if (is_null($id)) : |
| 557 |
$id = $this->wp_id(); |
| 558 |
endif; |
| 559 |
|
| 560 |
if (!isset($this->_hashes[$id])) : |
| 561 |
$this->_hashes[$id] = get_post_custom_values( |
| 562 |
'syndication_item_hash', $id |
| 563 |
); |
| 564 |
if (is_null($this->_hashes[$id])) : |
| 565 |
$this->_hashes[$id] = array(); |
| 566 |
endif; |
| 567 |
endif; |
| 568 |
return $this->_hashes[$id]; |
| 569 |
} |
| 570 |
|
| 571 |
function update_hash ($hashed = true) { |
| 572 |
// Basis for tracking possible changes to item. |
| 573 |
$hash = array( |
| 574 |
"title" => $this->entry->get_title(), |
| 575 |
"link" => $this->permalink(), |
| 576 |
"content" => $this->content(), |
| 577 |
"excerpt" => $this->excerpt(), |
| 578 |
); |
| 579 |
|
| 580 |
if ($hashed) : |
| 581 |
$hash = md5(serialize($hash)); |
| 582 |
endif; |
| 583 |
|
| 584 |
return $hash; |
| 585 |
} /* SyndicatedPost::update_hash() */ |
| 586 |
|
| 587 |
static function normalize_guid_prefix () { |
| 588 |
return trailingslashit(get_bloginfo('url')).'?guid='; |
| 589 |
} |
| 590 |
|
| 591 |
static function normalize_guid ($guid) { |
| 592 |
$guid = trim($guid); |
| 593 |
if (preg_match('/^[0-9a-z]{32}$/i', $guid)) : // MD5 |
| 594 |
$guid = SyndicatedPost::normalize_guid_prefix().strtolower($guid); |
| 595 |
elseif ((strlen(esc_url($guid)) == 0) or (esc_url($guid) != $guid)) : |
| 596 |
$guid = SyndicatedPost::normalize_guid_prefix().md5($guid); |
| 597 |
endif; |
| 598 |
$guid = trim($guid); |
| 599 |
return $guid; |
| 600 |
} /* SyndicatedPost::normalize_guid() */ |
| 601 |
|
| 602 |
function guid () { |
| 603 |
$guid = null; |
| 604 |
if (isset($this->item['id'])): // Atom 0.3 / 1.0 |
| 605 |
$guid = $this->item['id']; |
| 606 |
elseif (isset($this->item['atom']['id'])) : // Namespaced Atom |
| 607 |
$guid = $this->item['atom']['id']; |
| 608 |
elseif (isset($this->item['guid'])) : // RSS 2.0 |
| 609 |
$guid = $this->item['guid']; |
| 610 |
elseif (isset($this->item['dc']['identifier'])) : // yeah, right |
| 611 |
$guid = $this->item['dc']['identifier']; |
| 612 |
endif; |
| 613 |
|
| 614 |
// Un-set or too long to use as-is. Generate a tag: URI. |
| 615 |
if (is_null($guid) or strlen($guid) > 250) : |
| 616 |
// In case we need to check this again |
| 617 |
$original_guid = $guid; |
| 618 |
|
| 619 |
// The feed does not seem to have provided us with a |
| 620 |
// usable unique identifier, so we'll have to cobble |
| 621 |
// together a tag: URI that might work for us. The base |
| 622 |
// of the URI will be the host name of the feed source ... |
| 623 |
$bits = parse_url($this->link->uri()); |
| 624 |
$guid = 'tag:'.$bits['host']; |
| 625 |
|
| 626 |
// Some ill-mannered feeds (for example, certain feeds |
| 627 |
// coming from Google Calendar) have extraordinarily long |
| 628 |
// guids -- so long that they exceed the 255 character |
| 629 |
// width of the WordPress guid field. But if the string |
| 630 |
// gets clipped by MySQL, uniqueness tests will fail |
| 631 |
// forever after and the post will be endlessly |
| 632 |
// reduplicated. So, instead, Guids Of A Certain Length |
| 633 |
// are hashed down into a nice, manageable tag: URI. |
| 634 |
if (!is_null($original_guid)) : |
| 635 |
$guid .= ',2010-12-03:id.'.md5($original_guid); |
| 636 |
|
| 637 |
// If we have a date of creation, then we can use that |
| 638 |
// to uniquely identify the item. (On the other hand, if |
| 639 |
// the feed producer was consicentious enough to |
| 640 |
// generate dates of creation, she probably also was |
| 641 |
// conscientious enough to generate unique identifiers.) |
| 642 |
elseif (!is_null($this->created())) : |
| 643 |
$guid .= '://post.'.date('YmdHis', $this->created()); |
| 644 |
|
| 645 |
// Otherwise, use both the URI of the item, *and* the |
| 646 |
// item's title. We have to use both because titles are |
| 647 |
// often not unique, and sometimes links aren't unique |
| 648 |
// either (e.g. Bitch (S)HITLIST, Mozilla Dot Org news, |
| 649 |
// some podcasts). But it's rare to have *both* the same |
| 650 |
// title *and* the same link for two different items. So |
| 651 |
// this is about the best we can do. |
| 652 |
else : |
| 653 |
$link = $this->permalink(); |
| 654 |
if (is_null($link)) : $link = $this->link->uri(); endif; |
| 655 |
$guid .= '://'.md5($link.'/'.$this->item['title']); |
| 656 |
endif; |
| 657 |
endif; |
| 658 |
return $guid; |
| 659 |
} /* SyndicatedPost::guid() */ |
| 660 |
|
| 661 |
function author () { |
| 662 |
$author = array (); |
| 663 |
|
| 664 |
$aa = $this->entry->get_authors(); |
| 665 |
if (count($aa) > 0) : |
| 666 |
$a = reset($aa); |
| 667 |
|
| 668 |
$author = array( |
| 669 |
'name' => $a->get_name(), |
| 670 |
'email' => $a->get_email(), |
| 671 |
'uri' => $a->get_link(), |
| 672 |
); |
| 673 |
endif; |
| 674 |
|
| 675 |
if (FEEDWORDPRESS_COMPATIBILITY) : |
| 676 |
// Search through the MagpieRSS elements: Atom, Dublin Core, RSS |
| 677 |
if (isset($this->item['author_name'])): |
| 678 |
$author['name'] = $this->item['author_name']; |
| 679 |
elseif (isset($this->item['dc']['creator'])): |
| 680 |
$author['name'] = $this->item['dc']['creator']; |
| 681 |
elseif (isset($this->item['dc']['contributor'])): |
| 682 |
$author['name'] = $this->item['dc']['contributor']; |
| 683 |
elseif (isset($this->feed->channel['dc']['creator'])) : |
| 684 |
$author['name'] = $this->feed->channel['dc']['creator']; |
| 685 |
elseif (isset($this->feed->channel['dc']['contributor'])) : |
| 686 |
$author['name'] = $this->feed->channel['dc']['contributor']; |
| 687 |
elseif (isset($this->feed->channel['author_name'])) : |
| 688 |
$author['name'] = $this->feed->channel['author_name']; |
| 689 |
elseif ($this->feed->is_rss() and isset($this->item['author'])) : |
| 690 |
// The author element in RSS is allegedly an |
| 691 |
// e-mail address, but lots of people don't use |
| 692 |
// it that way. So let's make of it what we can. |
| 693 |
$author = parse_email_with_realname($this->item['author']); |
| 694 |
|
| 695 |
if (!isset($author['name'])) : |
| 696 |
if (isset($author['email'])) : |
| 697 |
$author['name'] = $author['email']; |
| 698 |
else : |
| 699 |
$author['name'] = $this->feed->channel['title']; |
| 700 |
endif; |
| 701 |
endif; |
| 702 |
endif; |
| 703 |
endif; |
| 704 |
|
| 705 |
if (!isset($author['name']) or is_null($author['name'])) : |
| 706 |
// Nothing found. Try some crappy defaults. |
| 707 |
if ($this->link->name()) : |
| 708 |
$author['name'] = $this->link->name(); |
| 709 |
else : |
| 710 |
$url = parse_url($this->link->uri()); |
| 711 |
$author['name'] = $url['host']; |
| 712 |
endif; |
| 713 |
endif; |
| 714 |
|
| 715 |
if (FEEDWORDPRESS_COMPATIBILITY) : |
| 716 |
if (isset($this->item['author_email'])): |
| 717 |
$author['email'] = $this->item['author_email']; |
| 718 |
elseif (isset($this->feed->channel['author_email'])) : |
| 719 |
$author['email'] = $this->feed->channel['author_email']; |
| 720 |
endif; |
| 721 |
|
| 722 |
if (isset($this->item['author_uri'])): |
| 723 |
$author['uri'] = $this->item['author_uri']; |
| 724 |
elseif (isset($this->item['author_url'])): |
| 725 |
$author['uri'] = $this->item['author_url']; |
| 726 |
elseif (isset($this->feed->channel['author_uri'])) : |
| 727 |
$author['uri'] = $this->item['author_uri']; |
| 728 |
elseif (isset($this->feed->channel['author_url'])) : |
| 729 |
$author['uri'] = $this->item['author_url']; |
| 730 |
elseif (isset($this->feed->channel['link'])) : |
| 731 |
$author['uri'] = $this->feed->channel['link']; |
| 732 |
endif; |
| 733 |
endif; |
| 734 |
|
| 735 |
return $author; |
| 736 |
} /* SyndicatedPost::author() */ |
| 737 |
|
| 738 |
/** |
| 739 |
* SyndicatedPost::get_terms_from_settings(): Return an array of terms to associate with the incoming |
| 740 |
* post based on the Categories, Tags, and other terms associated with each new post by the user's |
| 741 |
* settings (global and feed-specific). |
| 742 |
* |
| 743 |
* @since 2016.0331 |
| 744 |
* @return array of lists, each element has the taxonomy for a key ('category', 'post_tag', etc.), |
| 745 |
* and a list of term codes (either alphanumeric names, or ID numbers encoded in a format that |
| 746 |
* SyndicatedLink::category_ids() can understand) within that taxonomy |
| 747 |
* |
| 748 |
*/ |
| 749 |
public function get_terms_from_settings () { |
| 750 |
// Categories: start with default categories, if any. |
| 751 |
$cats = array(); |
| 752 |
if ('no' != $this->link->setting('add/category', NULL, 'yes')) : |
| 753 |
$fc = get_option("feedwordpress_syndication_cats"); |
| 754 |
if ($fc) : |
| 755 |
$cats = array_merge($cats, explode("\n", $fc)); |
| 756 |
endif; |
| 757 |
endif; |
| 758 |
|
| 759 |
$fc = $this->link->setting('cats',NULL, array()); |
| 760 |
if (is_array($fc)) : |
| 761 |
$cats = array_merge($cats, $fc); |
| 762 |
endif; |
| 763 |
$preset_terms['category'] = $cats; |
| 764 |
|
| 765 |
// Tags: start with default tags, if any |
| 766 |
$tags = array(); |
| 767 |
if ('no' != $this->link->setting('add/post_tag', NULL, 'yes')) : |
| 768 |
$ft = get_option("feedwordpress_syndication_tags", NULL); |
| 769 |
$tags = (is_null($ft) ? array() : explode(FEEDWORDPRESS_CAT_SEPARATOR, $ft)); |
| 770 |
endif; |
| 771 |
|
| 772 |
$ft = $this->link->setting('tags', NULL, array()); |
| 773 |
if (is_array($ft)) : |
| 774 |
$tags = array_merge($tags, $ft); |
| 775 |
endif; |
| 776 |
$preset_terms['post_tag'] = $tags; |
| 777 |
|
| 778 |
$taxonomies = $this->link->taxonomies(); |
| 779 |
$feedTerms = $this->link->setting('terms', NULL, array()); |
| 780 |
$globalTerms = get_option('feedwordpress_syndication_terms', array()); |
| 781 |
$specials = array('category' => 'cats', 'post_tag' => 'tags'); |
| 782 |
|
| 783 |
foreach ($taxonomies as $tax) : |
| 784 |
// category and tag settings have already previously been handled |
| 785 |
// but if this is from another taxonomy, then... |
| 786 |
if (!isset($specials[$tax])) : |
| 787 |
$terms = array(); |
| 788 |
|
| 789 |
// See if we should get the globals |
| 790 |
if ('no' != $this->link->setting("add/$tax", NULL, 'yes')) : |
| 791 |
if (isset($globalTerms[$tax])) : |
| 792 |
$terms = $globalTerms[$tax]; |
| 793 |
endif; |
| 794 |
endif; |
| 795 |
|
| 796 |
// Now merge in the locals |
| 797 |
if (isset($feedTerms[$tax])) : |
| 798 |
$terms = array_merge($terms, $feedTerms[$tax]); |
| 799 |
endif; |
| 800 |
|
| 801 |
// That's all, folks. |
| 802 |
$preset_terms[$tax] = $terms; |
| 803 |
endif; |
| 804 |
endforeach; |
| 805 |
|
| 806 |
return $preset_terms; |
| 807 |
} /* SyndicatedPost::get_terms_from_settings () */ |
| 808 |
|
| 809 |
/** |
| 810 |
* SyndicatedPost::get_terms_from_feeds(): Return an array of terms to associate with the incoming |
| 811 |
* post based on the contents of the subscribed feed (atom:category and rss:category elements, dc:subject |
| 812 |
* elements, tags embedded using microformats in the post content, etc.) |
| 813 |
* |
| 814 |
* @since 2016.0331 |
| 815 |
* @return array of lists, each element has the taxonomy for a key ('category', 'post_tag', etc.), |
| 816 |
* and a list of alphanumeric term names |
| 817 |
*/ |
| 818 |
public function get_terms_from_feeds () { |
| 819 |
// Now add categories from the post, if we have 'em |
| 820 |
$cats = array(); |
| 821 |
$post_cats = $this->entry->get_categories(); |
| 822 |
if (is_array($post_cats)) : foreach ($post_cats as $cat) : |
| 823 |
$cat_name = $cat->get_term(); |
| 824 |
if (!$cat_name) : $cat_name = $cat->get_label(); endif; |
| 825 |
|
| 826 |
if ($this->link->setting('cat_split', NULL, NULL)) : |
| 827 |
$pcre = "\007".$this->feedmeta['cat_split']."\007"; |
| 828 |
$cats = array_merge( |
| 829 |
$cats, |
| 830 |
preg_split( |
| 831 |
$pcre, |
| 832 |
$cat_name, |
| 833 |
-1 /*=no limit*/, |
| 834 |
PREG_SPLIT_NO_EMPTY |
| 835 |
) |
| 836 |
); |
| 837 |
else : |
| 838 |
$cats[] = $cat_name; |
| 839 |
endif; |
| 840 |
endforeach; endif; |
| 841 |
|
| 842 |
$feed_terms['category'] = apply_filters('syndicated_item_categories', $cats, $this); |
| 843 |
|
| 844 |
// Scan post for /a[@rel='tag'] and use as tags if present |
| 845 |
$tags = $this->inline_tags(); |
| 846 |
$feed_terms['post_tag'] = apply_filters('syndicated_item_tags', $tags, $this); |
| 847 |
|
| 848 |
return $feed_terms; |
| 849 |
} /* SyndicatedPost::get_terms_from_feeds () */ |
| 850 |
|
| 851 |
/** |
| 852 |
* SyndicatedPost::inline_tags: Return a list of all the tags embedded |
| 853 |
* in post content using the a[@rel="tag"] microformat. |
| 854 |
* |
| 855 |
* @since 2010.0630 |
| 856 |
* @return array of string values containing the name of each tag |
| 857 |
*/ |
| 858 |
function inline_tags () { |
| 859 |
$tags = array(); |
| 860 |
$content = $this->content(); |
| 861 |
$pattern = FeedWordPressHTML::tagWithAttributeRegex('a', 'rel', 'tag'); |
| 862 |
preg_match_all($pattern, $content, $refs, PREG_SET_ORDER); |
| 863 |
if (count($refs) > 0) : |
| 864 |
foreach ($refs as $ref) : |
| 865 |
$tag = FeedWordPressHTML::tagWithAttributeMatch($ref); |
| 866 |
$tags[] = $tag['content']; |
| 867 |
endforeach; |
| 868 |
endif; |
| 869 |
return $tags; |
| 870 |
} |
| 871 |
|
| 872 |
/** |
| 873 |
* SyndicatedPost::isTaggedAs: Test whether a feed item is |
| 874 |
* tagged / categorized with a given string. Case and leading and |
| 875 |
* trailing whitespace are ignored. |
| 876 |
* |
| 877 |
* @param string $tag Tag to check for |
| 878 |
* |
| 879 |
* @return bool Whether or not at least one of the categories / tags on |
| 880 |
* $this->item is set to $tag (modulo case and leading and trailing |
| 881 |
* whitespace) |
| 882 |
*/ |
| 883 |
function isTaggedAs ($tag) { |
| 884 |
$desiredTag = strtolower(trim($tag)); // Normalize case and whitespace |
| 885 |
|
| 886 |
// Check to see if this is tagged with $tag |
| 887 |
$currentCategory = 'category'; |
| 888 |
$currentCategoryNumber = 1; |
| 889 |
|
| 890 |
// If we have the new MagpieRSS, the number of category elements |
| 891 |
// on this item is stored under index "category#". |
| 892 |
if (isset($this->item['category#'])) : |
| 893 |
$numberOfCategories = (int) $this->item['category#']; |
| 894 |
|
| 895 |
// We REALLY shouldn't have the old and busted MagpieRSS, but in |
| 896 |
// case we do, it doesn't support multiple categories, but there |
| 897 |
// might still be a single value under the "category" index. |
| 898 |
elseif (isset($this->item['category'])) : |
| 899 |
$numberOfCategories = 1; |
| 900 |
|
| 901 |
// No standard category or tag elements on this feed item. |
| 902 |
else : |
| 903 |
$numberOfCategories = 0; |
| 904 |
|
| 905 |
endif; |
| 906 |
|
| 907 |
$isSoTagged = false; // Innocent until proven guilty |
| 908 |
|
| 909 |
// Loop through category elements; if there are multiple |
| 910 |
// elements, they are indexed as category, category#2, |
| 911 |
// category#3, ... category#N |
| 912 |
while ($currentCategoryNumber <= $numberOfCategories) : |
| 913 |
if ($desiredTag == strtolower(trim($this->item[$currentCategory]))) : |
| 914 |
$isSoTagged = true; // Got it! |
| 915 |
break; |
| 916 |
endif; |
| 917 |
|
| 918 |
$currentCategoryNumber += 1; |
| 919 |
$currentCategory = 'category#'.$currentCategoryNumber; |
| 920 |
endwhile; |
| 921 |
|
| 922 |
return $isSoTagged; |
| 923 |
} /* SyndicatedPost::isTaggedAs() */ |
| 924 |
|
| 925 |
/** |
| 926 |
* SyndicatedPost::enclosures: returns an array with any enclosures |
| 927 |
* that may be attached to this syndicated item. |
| 928 |
* |
| 929 |
* @param string $type If you only want enclosures that match a certain |
| 930 |
* MIME type or group of MIME types, you can limit the enclosures |
| 931 |
* that will be returned to only those with a MIME type which |
| 932 |
* matches this regular expression. |
| 933 |
* @return array |
| 934 |
*/ |
| 935 |
function enclosures ($type = '/.*/') { |
| 936 |
$enclosures = array(); |
| 937 |
|
| 938 |
if (isset($this->item['enclosure#'])) : |
| 939 |
// Loop through enclosure, enclosure#2, enclosure#3, .... |
| 940 |
for ($i = 1; $i <= $this->item['enclosure#']; $i++) : |
| 941 |
$eid = (($i > 1) ? "#{$id}" : ""); |
| 942 |
|
| 943 |
// Does it match the type we want? |
| 944 |
if (preg_match($type, $this->item["enclosure{$eid}@type"])) : |
| 945 |
$enclosures[] = array( |
| 946 |
"url" => $this->item["enclosure{$eid}@url"], |
| 947 |
"type" => $this->item["enclosure{$eid}@type"], |
| 948 |
"length" => $this->item["enclosure{$eid}@length"], |
| 949 |
); |
| 950 |
endif; |
| 951 |
endfor; |
| 952 |
endif; |
| 953 |
return $enclosures; |
| 954 |
} /* SyndicatedPost::enclosures() */ |
| 955 |
|
| 956 |
function source ($what = NULL) { |
| 957 |
$ret = NULL; |
| 958 |
$source = $this->entry->get_source(); |
| 959 |
if ($source) : |
| 960 |
$ret = array(); |
| 961 |
$ret['title'] = $source->get_title(); |
| 962 |
$ret['uri'] = $source->get_link(); |
| 963 |
$ret['feed'] = $source->get_link(0, 'self'); |
| 964 |
|
| 965 |
if ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'id')) : |
| 966 |
$ret['id'] = $id_tags[0]['data']; |
| 967 |
elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_ATOM_03, 'id')) : |
| 968 |
$ret['id'] = $id_tags[0]['data']; |
| 969 |
elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_RSS_20, 'guid')) : |
| 970 |
$ret['id'] = $id_tags[0]['data']; |
| 971 |
elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_RSS_10, 'guid')) : |
| 972 |
$ret['id'] = $id_tags[0]['data']; |
| 973 |
elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_RSS_090, 'guid')) : |
| 974 |
$ret['id'] = $id_tags[0]['data']; |
| 975 |
endif; |
| 976 |
endif; |
| 977 |
|
| 978 |
if (!is_null($what) and is_scalar($what)) : |
| 979 |
$ret = $ret[$what]; |
| 980 |
endif; |
| 981 |
return $ret; |
| 982 |
} |
| 983 |
|
| 984 |
function comment_link () { |
| 985 |
$url = null; |
| 986 |
|
| 987 |
// RSS 2.0 has a standard <comments> element: |
| 988 |
// "<comments> is an optional sub-element of <item>. If present, |
| 989 |
// it is the url of the comments page for the item." |
| 990 |
// <http://cyber.law.harvard.edu/rss/rss.html#ltcommentsgtSubelementOfLtitemgt> |
| 991 |
if (isset($this->item['comments'])) : |
| 992 |
$url = $this->item['comments']; |
| 993 |
endif; |
| 994 |
|
| 995 |
// The convention in Atom feeds is to use a standard <link> |
| 996 |
// element with @rel="replies" and @type="text/html". |
| 997 |
// Unfortunately, SimplePie_Item::get_links() allows us to filter |
| 998 |
// by the value of @rel, but not by the value of @type. *sigh* |
| 999 |
|
| 1000 |
// Try Atom 1.0 first |
| 1001 |
$linkElements = $this->entry->get_item_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'link'); |
| 1002 |
|
| 1003 |
// Fall back and try Atom 0.3 |
| 1004 |
if (is_null($linkElements)) : $linkElements = $this->entry->get_item_tags(SIMPLEPIE_NAMESPACE_ATOM_03, 'link'); endif; |
| 1005 |
|
| 1006 |
// Now loop through the elements, screening by @rel and @type |
| 1007 |
if (is_array($linkElements)) : foreach ($linkElements as $link) : |
| 1008 |
$rel = (isset($link['attribs']['']['rel']) ? $link['attribs']['']['rel'] : 'alternate'); |
| 1009 |
$type = (isset($link['attribs']['']['type']) ? $link['attribs']['']['type'] : NULL); |
| 1010 |
$href = (isset($link['attribs']['']['href']) ? $link['attribs']['']['href'] : NULL); |
| 1011 |
|
| 1012 |
if (strtolower($rel)=='replies' and $type=='text/html' and !is_null($href)) : |
| 1013 |
$url = $href; |
| 1014 |
endif; |
| 1015 |
endforeach; endif; |
| 1016 |
|
| 1017 |
return $url; |
| 1018 |
} |
| 1019 |
|
| 1020 |
function comment_feed () { |
| 1021 |
$feed = null; |
| 1022 |
|
| 1023 |
// Well Formed Web comment feeds extension for RSS 2.0 |
| 1024 |
// <http://www.sellsbrothers.com/spout/default.aspx?content=archive.htm#exposingRssComments> |
| 1025 |
// |
| 1026 |
// N.B.: Correct capitalization is wfw:commentRss, but |
| 1027 |
// wfw:commentRSS is common in the wild (partly due to a typo in |
| 1028 |
// the original spec). In any case, our item array is normalized |
| 1029 |
// to all lowercase anyways. |
| 1030 |
if (isset($this->item['wfw']['commentrss'])) : |
| 1031 |
$feed = $this->item['wfw']['commentrss']; |
| 1032 |
endif; |
| 1033 |
|
| 1034 |
// In Atom 1.0, the convention is to use a standard link element |
| 1035 |
// with @rel="replies". Sometimes this is also used to pass a |
| 1036 |
// link to the human-readable comments page, so we also need to |
| 1037 |
// check link/@type for a feed MIME type. |
| 1038 |
// |
| 1039 |
// Which is why I'm not using the SimplePie_Item::get_links() |
| 1040 |
// method here, incidentally: it doesn't allow you to filter by |
| 1041 |
// @type. *sigh* |
| 1042 |
if (isset($this->item['link_replies'])) : |
| 1043 |
// There may be multiple <link rel="replies"> elements; feeds have a feed MIME type |
| 1044 |
$N = isset($this->item['link_replies#']) ? $this->item['link_replies#'] : 1; |
| 1045 |
for ($i = 1; $i <= $N; $i++) : |
| 1046 |
$currentElement = 'link_replies'.(($i > 1) ? '#'.$i : ''); |
| 1047 |
if (isset($this->item[$currentElement.'@type']) |
| 1048 |
and preg_match("\007application/(atom|rss|rdf)\+xml\007i", $this->item[$currentElement.'@type'])) : |
| 1049 |
$feed = $this->item[$currentElement]; |
| 1050 |
endif; |
| 1051 |
endfor; |
| 1052 |
endif; |
| 1053 |
return $feed; |
| 1054 |
} /* SyndicatedPost::comment_feed() */ |
| 1055 |
|
| 1056 |
################################## |
| 1057 |
#### BUILT-IN CONTENT FILTERS #### |
| 1058 |
################################## |
| 1059 |
|
| 1060 |
var $uri_attrs = array ( |
| 1061 |
array('a', 'href'), |
| 1062 |
array('applet', 'codebase'), |
| 1063 |
array('area', 'href'), |
| 1064 |
array('blockquote', 'cite'), |
| 1065 |
array('body', 'background'), |
| 1066 |
array('del', 'cite'), |
| 1067 |
array('form', 'action'), |
| 1068 |
array('frame', 'longdesc'), |
| 1069 |
array('frame', 'src'), |
| 1070 |
array('iframe', 'longdesc'), |
| 1071 |
array('iframe', 'src'), |
| 1072 |
array('head', 'profile'), |
| 1073 |
array('img', 'longdesc'), |
| 1074 |
array('img', 'src'), |
| 1075 |
array('img', 'usemap'), |
| 1076 |
array('input', 'src'), |
| 1077 |
array('input', 'usemap'), |
| 1078 |
array('ins', 'cite'), |
| 1079 |
array('link', 'href'), |
| 1080 |
array('object', 'classid'), |
| 1081 |
array('object', 'codebase'), |
| 1082 |
array('object', 'data'), |
| 1083 |
array('object', 'usemap'), |
| 1084 |
array('q', 'cite'), |
| 1085 |
array('script', 'src') |
| 1086 |
); /* var SyndicatedPost::$uri_attrs */ |
| 1087 |
|
| 1088 |
var $_base = null; |
| 1089 |
|
| 1090 |
function resolve_single_relative_uri ($refs) { |
| 1091 |
$tag = FeedWordPressHTML::attributeMatch($refs); |
| 1092 |
$url = SimplePie_Misc::absolutize_url($tag['value'], $this->_base); |
| 1093 |
|
| 1094 |
return $tag['prefix'] . $url . $tag['suffix']; |
| 1095 |
} /* function SyndicatedPost::resolve_single_relative_uri() */ |
| 1096 |
|
| 1097 |
static function resolve_relative_uris ($content, $obj) { |
| 1098 |
$set = $obj->link->setting('resolve relative', 'resolve_relative', 'yes'); |
| 1099 |
if ($set and $set != 'no') : |
| 1100 |
// Fallback: if we don't have anything better, use the |
| 1101 |
// item link from the feed |
| 1102 |
$obj->_base = $obj->permalink(); // Reset the base for resolving relative URIs |
| 1103 |
|
| 1104 |
// What we should do here, properly, is to use |
| 1105 |
// SimplePie_Item::get_base() -- but that method is |
| 1106 |
// currently broken. Or getting down and dirty in the |
| 1107 |
// SimplePie representation of the content tags and |
| 1108 |
// grabbing the xml_base member for the content element. |
| 1109 |
// Maybe someday... |
| 1110 |
|
| 1111 |
foreach ($obj->uri_attrs as $pair) : |
| 1112 |
list($tag, $attr) = $pair; |
| 1113 |
$pattern = FeedWordPressHTML::attributeRegex($tag, $attr); |
| 1114 |
|
| 1115 |
// FIXME: Encountered issue while testing an extremely long (= 88827 characters) item |
| 1116 |
// Relying on preg_replace_callback() here can cause a PHP seg fault on my development |
| 1117 |
// server. preg_match_all() causes a similar problem. Apparently this is a PCRE issue |
| 1118 |
// Cf. discussion of similar issue <https://bugs.php.net/bug.php?id=65009> |
| 1119 |
$content = preg_replace_callback ( |
| 1120 |
$pattern, |
| 1121 |
array($obj, 'resolve_single_relative_uri'), |
| 1122 |
$content |
| 1123 |
); |
| 1124 |
|
| 1125 |
endforeach; |
| 1126 |
endif; |
| 1127 |
|
| 1128 |
return $content; |
| 1129 |
} /* function SyndicatedPost::resolve_relative_uris () */ |
| 1130 |
|
| 1131 |
var $strip_attrs = array ( |
| 1132 |
array('[a-z]+', 'target'), |
| 1133 |
// array('[a-z]+', 'style'), |
| 1134 |
// array('[a-z]+', 'on[a-z]+'), |
| 1135 |
); |
| 1136 |
|
| 1137 |
function strip_attribute_from_tag ($refs) { |
| 1138 |
$tag = FeedWordPressHTML::attributeMatch($refs); |
| 1139 |
return $tag['before_attribute'].$tag['after_attribute']; |
| 1140 |
} |
| 1141 |
|
| 1142 |
static function sanitize_content ($content, $obj) { |
| 1143 |
# This kind of sucks. I intend to replace it with |
| 1144 |
# lib_filter sometime soon. |
| 1145 |
foreach ($obj->strip_attrs as $pair): |
| 1146 |
list($tag,$attr) = $pair; |
| 1147 |
$pattern = FeedWordPressHTML::attributeRegex($tag, $attr); |
| 1148 |
|
| 1149 |
$content = preg_replace_callback ( |
| 1150 |
$pattern, |
| 1151 |
array($obj, 'strip_attribute_from_tag'), |
| 1152 |
$content |
| 1153 |
); |
| 1154 |
endforeach; |
| 1155 |
return $content; |
| 1156 |
} /* SyndicatedPost::sanitize() */ |
| 1157 |
|
| 1158 |
##################### |
| 1159 |
#### POST STATUS #### |
| 1160 |
##################### |
| 1161 |
|
| 1162 |
/** |
| 1163 |
* SyndicatedPost::filtered: check whether or not this post has been |
| 1164 |
* screened out by a registered filter. |
| 1165 |
* |
| 1166 |
* @return bool TRUE iff post has been filtered out by a previous filter |
| 1167 |
*/ |
| 1168 |
function filtered () { |
| 1169 |
return is_null($this->post); |
| 1170 |
} /* SyndicatedPost::filtered() */ |
| 1171 |
|
| 1172 |
/** |
| 1173 |
* SyndicatedPost::freshness: check whether post is a new post to be |
| 1174 |
* inserted, a previously syndicated post that needs to be updated to |
| 1175 |
* match the latest revision, or a previously syndicated post that is |
| 1176 |
* still up-to-date. |
| 1177 |
* |
| 1178 |
* @return int A status code representing the freshness of the post |
| 1179 |
* -1 = post already syndicated; has a revision that needs to be stored, but not updated to |
| 1180 |
* 0 = post already syndicated; no update needed |
| 1181 |
* 1 = post already syndicated, but needs to be updated to latest |
| 1182 |
* 2 = post has not yet been syndicated; needs to be created |
| 1183 |
*/ |
| 1184 |
function freshness ($format = 'number') { |
| 1185 |
global $wpdb; |
| 1186 |
|
| 1187 |
if ($this->filtered()) : // This should never happen. |
| 1188 |
FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__, __FILE__); |
| 1189 |
endif; |
| 1190 |
|
| 1191 |
if (is_null($this->_freshness)) : // Not yet checked and cached. |
| 1192 |
$guid = $this->post['guid']; |
| 1193 |
$eguid = esc_sql($this->post['guid']); |
| 1194 |
|
| 1195 |
$q = new WP_Query(array( |
| 1196 |
'fields' => '_synfresh', // id, guid, post_modified_gmt |
| 1197 |
'ignore_sticky_posts' => true, |
| 1198 |
'guid' => $guid, |
| 1199 |
)); |
| 1200 |
|
| 1201 |
$old_post = NULL; |
| 1202 |
if ($q->have_posts()) : |
| 1203 |
while ($q->have_posts()) : $q->the_post(); |
| 1204 |
$old_post = $q->post; |
| 1205 |
endwhile; |
| 1206 |
endif; |
| 1207 |
|
| 1208 |
if (is_null($old_post)) : // No post with this guid |
| 1209 |
FeedWordPress::diagnostic('feed_items:freshness', 'Item ['.$guid.'] "'.$this->entry->get_title().'" is a NEW POST.'); |
| 1210 |
$this->_wp_id = NULL; |
| 1211 |
$this->_freshness = 2; // New content |
| 1212 |
else : |
| 1213 |
// Presume there is nothing new until we find |
| 1214 |
// something new. |
| 1215 |
$updated = false; |
| 1216 |
$live = false; |
| 1217 |
|
| 1218 |
// Pull the list of existing revisions to get |
| 1219 |
// timestamps. |
| 1220 |
$revisions = wp_get_post_revisions($old_post->ID); |
| 1221 |
foreach ($revisions as $rev) : |
| 1222 |
$revisions_ts[] = mysql2date('G', $rev->post_modified_gmt); |
| 1223 |
endforeach; |
| 1224 |
|
| 1225 |
$revisions_ts[] = mysql2date('G', $old_post->post_modified_gmt); |
| 1226 |
$last_rev_ts = end($revisions_ts); |
| 1227 |
$updated_ts = $this->updated(/*fallback=*/ true, /*default=*/ NULL); |
| 1228 |
|
| 1229 |
// If we have an explicit updated timestamp, |
| 1230 |
// check that against existing stamps. |
| 1231 |
if (!is_null($updated_ts)) : |
| 1232 |
$updated = !in_array($updated_ts, $revisions_ts); |
| 1233 |
|
| 1234 |
// If this a newer revision, make it go |
| 1235 |
// live. If an older one, just record |
| 1236 |
// the contents. |
| 1237 |
$live = ($updated and ($updated_ts > $last_rev_ts)); |
| 1238 |
endif; |
| 1239 |
|
| 1240 |
// This is a revision we haven't seen before, judging by the date. |
| 1241 |
|
| 1242 |
$updatedReason = NULL; |
| 1243 |
if ($updated) : |
| 1244 |
$updatedReason = preg_replace( |
| 1245 |
"/\s+/", " ", |
| 1246 |
'has been marked with a new timestamp (' |
| 1247 |
.date('Y-m-d H:i:s', $updated_ts) |
| 1248 |
." > " |
| 1249 |
.date('Y-m-d H:i:s', $last_rev_ts) |
| 1250 |
.')' |
| 1251 |
); |
| 1252 |
|
| 1253 |
// The date does not indicate a new revision, so |
| 1254 |
// let's check the hash. |
| 1255 |
else : |
| 1256 |
// Or the hash... |
| 1257 |
$hash = $this->update_hash(); |
| 1258 |
$seen = $this->stored_hashes($old_post->ID); |
| 1259 |
if (count($seen) > 0) : |
| 1260 |
$updated = !in_array($hash, $seen); // Not seen yet? |
| 1261 |
else : |
| 1262 |
$updated = true; // Can't find syndication meta-data |
| 1263 |
endif; |
| 1264 |
|
| 1265 |
if ($updated and FeedWordPress::diagnostic_on('feed_items:freshness:reasons')) : |
| 1266 |
// In the absence of definitive |
| 1267 |
// timestamp information, we |
| 1268 |
// just have to assume that a |
| 1269 |
// hash we haven't seen before |
| 1270 |
// is a newer version. |
| 1271 |
$live = true; |
| 1272 |
|
| 1273 |
$updatedReason = ' has a not-yet-seen update hash: ' |
| 1274 |
.MyPHP::val($hash) |
| 1275 |
.' not in {' |
| 1276 |
.implode(", ", array_map(array('FeedWordPress', 'val'), $seen)) |
| 1277 |
.'}. Basis: ' |
| 1278 |
.MyPHP::val(array_keys($this->update_hash(false))); |
| 1279 |
endif; |
| 1280 |
endif; |
| 1281 |
|
| 1282 |
$frozen = false; |
| 1283 |
if ($updated) : // Ignore if the post is frozen |
| 1284 |
$frozen = ('yes' == $this->link->setting('freeze updates', 'freeze_updates', NULL)); |
| 1285 |
if (!$frozen) : |
| 1286 |
$frozen_values = get_post_custom_values('_syndication_freeze_updates', $old_post->ID); |
| 1287 |
$frozen = (count($frozen_values) > 0 and 'yes' == $frozen_values[0]); |
| 1288 |
|
| 1289 |
if ($frozen) : |
| 1290 |
$updatedReason = ' IS BLOCKED FROM BEING UPDATED BY A UPDATE LOCK ON THIS POST, EVEN THOUGH IT '.$updatedReason; |
| 1291 |
endif; |
| 1292 |
else : |
| 1293 |
$updatedReason = ' IS BLOCKED FROM BEING UPDATED BY A FEEDWORDPRESS UPDATE LOCK, EVEN THOUGH IT '.$updatedReason; |
| 1294 |
endif; |
| 1295 |
endif; |
| 1296 |
$live = ($live and !$frozen); |
| 1297 |
|
| 1298 |
if ($updated) : |
| 1299 |
FeedWordPress::diagnostic('feed_items:freshness', 'Item ['.$guid.'] "'.$this->entry->get_title().'" is an update of an existing post.'); |
| 1300 |
if (!is_null($updatedReason)) : |
| 1301 |
$updatedReason = preg_replace('/\s+/', ' ', $updatedReason); |
| 1302 |
FeedWordPress::diagnostic('feed_items:freshness:reasons', 'Item ['.$guid.'] "'.$this->entry->get_title().'" '.$updatedReason); |
| 1303 |
endif; |
| 1304 |
|
| 1305 |
$this->_freshness = apply_filters('syndicated_item_freshness', ($live ? 1 : -1), $updated, $frozen, $updated_ts, $last_rev_ts, $this); |
| 1306 |
|
| 1307 |
$this->_wp_id = $old_post->ID; |
| 1308 |
$this->_wp_post = $old_post; |
| 1309 |
|
| 1310 |
// We want this to keep a running list of all the |
| 1311 |
// processed update hashes. |
| 1312 |
$this->post['meta']['syndication_item_hash'] = array_merge( |
| 1313 |
$this->stored_hashes(), |
| 1314 |
array($this->update_hash()) |
| 1315 |
); |
| 1316 |
else : |
| 1317 |
FeedWordPress::diagnostic('feed_items:freshness', 'Item ['.$guid.'] "'.$this->entry->get_title().'" is a duplicate of an existing post.'); |
| 1318 |
$this->_freshness = 0; // Same old, same old |
| 1319 |
$this->_wp_id = $old_post->ID; |
| 1320 |
endif; |
| 1321 |
endif; |
| 1322 |
endif; |
| 1323 |
|
| 1324 |
switch ($format) : |
| 1325 |
case 'status' : |
| 1326 |
switch ($this->_freshness) : |
| 1327 |
case -1: |
| 1328 |
$ret = 'stored'; |
| 1329 |
break; |
| 1330 |
case 0: |
| 1331 |
$ret = NULL; |
| 1332 |
break; |
| 1333 |
case 1: |
| 1334 |
$ret = 'updated'; |
| 1335 |
break; |
| 1336 |
case 2: |
| 1337 |
default: |
| 1338 |
$ret = 'new'; |
| 1339 |
break; |
| 1340 |
endswitch; |
| 1341 |
break; |
| 1342 |
case 'number' : |
| 1343 |
default : |
| 1344 |
$ret = $this->_freshness; |
| 1345 |
endswitch; |
| 1346 |
|
| 1347 |
|
| 1348 |
return $ret; |
| 1349 |
} /* SyndicatedPost::freshness () */ |
| 1350 |
|
| 1351 |
function has_fresh_content () { |
| 1352 |
return ( ! $this->filtered() and $this->freshness() != 0 ); |
| 1353 |
} /* SyndicatedPost::has_fresh_content () */ |
| 1354 |
|
| 1355 |
function this_revision_needs_original_post ($freshness = NULL) { |
| 1356 |
if (is_null($freshness)) : |
| 1357 |
$freshness = $this->freshness(); |
| 1358 |
endif; |
| 1359 |
return ( $freshness >= 2 ); |
| 1360 |
} |
| 1361 |
|
| 1362 |
function this_revision_is_current ($freshness = NULL) { |
| 1363 |
if (is_null($freshness)) : |
| 1364 |
$freshness = $this->freshness(); |
| 1365 |
endif; |
| 1366 |
return ( $freshness >= 1 ); |
| 1367 |
} /* SyndicatedPost::this_revision_is_current () */ |
| 1368 |
|
| 1369 |
function fresh_content_is_update () { |
| 1370 |
return ($this->freshness() < 2); |
| 1371 |
} /* SyndicatedPost::fresh_content_is_update () */ |
| 1372 |
|
| 1373 |
function fresh_storage_diagnostic () { |
| 1374 |
$ret = NULL; |
| 1375 |
switch ($this->freshness()) : |
| 1376 |
case -1 : |
| 1377 |
$ret = 'Storing alternate revision of existing post # '.$this->wp_id().', "'.$this->post['post_title'].'"'; |
| 1378 |
break; |
| 1379 |
case 1 : |
| 1380 |
$ret = 'Updating existing post # '.$this->wp_id().', "'.$this->post['post_title'].'"'; |
| 1381 |
break; |
| 1382 |
case 2 : |
| 1383 |
default : |
| 1384 |
$ret = 'Inserting new post "'.$this->post['post_title'].'"'; |
| 1385 |
break; |
| 1386 |
endswitch; |
| 1387 |
return $ret; |
| 1388 |
} /* SyndicatedPost::fresh_storage_diagnostic() */ |
| 1389 |
|
| 1390 |
function fresh_storage_hook () { |
| 1391 |
$ret = NULL; |
| 1392 |
switch ($this->freshness()) : |
| 1393 |
case -1 : |
| 1394 |
case 1 : |
| 1395 |
$ret = 'update_syndicated_item'; |
| 1396 |
break; |
| 1397 |
case 2 : |
| 1398 |
default : |
| 1399 |
$ret = 'post_syndicated_item'; |
| 1400 |
break; |
| 1401 |
endswitch; |
| 1402 |
return $ret; |
| 1403 |
} /* SyndicatedPost::fresh_storage_hook () */ |
| 1404 |
|
| 1405 |
################################################# |
| 1406 |
#### INTERNAL STORAGE AND MANAGEMENT METHODS #### |
| 1407 |
################################################# |
| 1408 |
|
| 1409 |
function wp_id () { |
| 1410 |
if ($this->filtered()) : // This should never happen. |
| 1411 |
FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__, __FILE__); |
| 1412 |
endif; |
| 1413 |
|
| 1414 |
if (is_null($this->_wp_id) and is_null($this->_freshness)) : |
| 1415 |
$fresh = $this->freshness(); // sets WP DB id in the process |
| 1416 |
endif; |
| 1417 |
return $this->_wp_id; |
| 1418 |
} |
| 1419 |
|
| 1420 |
function store () { |
| 1421 |
global $wpdb; |
| 1422 |
|
| 1423 |
if ($this->filtered()) : // This should never happen. |
| 1424 |
FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__, __FILE__); |
| 1425 |
endif; |
| 1426 |
|
| 1427 |
$freshness = $this->freshness(); |
| 1428 |
if ($this->has_fresh_content()) : |
| 1429 |
# -- Look up, or create, numeric ID for author |
| 1430 |
$this->post['post_author'] = $this->author_id ( |
| 1431 |
$this->link->setting('unfamiliar author', 'unfamiliar_author', 'create') |
| 1432 |
); |
| 1433 |
|
| 1434 |
if (is_null($this->post['post_author'])) : |
| 1435 |
FeedWordPress::diagnostic('feed_items:rejected', 'Filtered out item ['.$this->guid().'] without syndication: no author available'); |
| 1436 |
$this->post = NULL; |
| 1437 |
endif; |
| 1438 |
endif; |
| 1439 |
|
| 1440 |
// We have to check again in case post has been filtered during |
| 1441 |
// the author_id lookup |
| 1442 |
if ($this->has_fresh_content()) : |
| 1443 |
$mapping = apply_filters('syndicated_post_terms_mapping', array( |
| 1444 |
'category' => array('abbr' => 'cats', 'unfamiliar' => 'category', 'domain' => array('category', 'post_tag')), |
| 1445 |
'post_tag' => array('abbr' => 'tags', 'unfamiliar' => 'post_tag', 'domain' => array('post_tag')), |
| 1446 |
), $this); |
| 1447 |
|
| 1448 |
$termSet = array(); $valid = null; |
| 1449 |
foreach ($this->feed_terms as $what => $anTerms) : |
| 1450 |
// Default to using the inclusive procedures (for cats) rather than exclusive (for inline tags) |
| 1451 |
$taxes = (isset($mapping[$what]) ? $mapping[$what] : $mapping['category']); |
| 1452 |
$unfamiliar = $taxes['unfamiliar']; |
| 1453 |
|
| 1454 |
if (!is_null($this->post)) : // Not filtered out yet |
| 1455 |
# -- Look up, or create, numeric ID for categories |
| 1456 |
$taxonomies = $this->link->setting("match/".$taxes['abbr'], 'match_'.$taxes['abbr'], $taxes['domain']); |
| 1457 |
|
| 1458 |
// Eliminate dummy variables |
| 1459 |
$taxonomies = array_filter($taxonomies, 'remove_dummy_zero'); |
| 1460 |
|
| 1461 |
// Allow FWP add-on filters to control the taxonomies we use to search for a term |
| 1462 |
$taxonomies = apply_filters("syndicated_post_terms_match", $taxonomies, $what, $this); |
| 1463 |
$taxonomies = apply_filters("syndicated_post_terms_match_${what}", $taxonomies, $this); |
| 1464 |
|
| 1465 |
// Allow FWP add-on filters to control with greater precision what happens on unmatched |
| 1466 |
$unmatched = apply_filters("syndicated_post_terms_unfamiliar", |
| 1467 |
$this->link->setting( |
| 1468 |
"unfamiliar {$unfamiliar}", |
| 1469 |
"unfamiliar_{$unfamiliar}", |
| 1470 |
'create:'.$unfamiliar |
| 1471 |
), |
| 1472 |
$what, |
| 1473 |
$this |
| 1474 |
); |
| 1475 |
|
| 1476 |
$terms = $this->category_ids ( |
| 1477 |
$anTerms, |
| 1478 |
$unmatched, |
| 1479 |
/*taxonomies=*/ $taxonomies, |
| 1480 |
array( |
| 1481 |
'singleton' => false, // I don't like surprises |
| 1482 |
'filters' => true, |
| 1483 |
) |
| 1484 |
); |
| 1485 |
|
| 1486 |
if (is_null($terms) or is_null($termSet)) : |
| 1487 |
// filtered out -- no matches |
| 1488 |
else : |
| 1489 |
$valid = true; |
| 1490 |
|
| 1491 |
// filter mode off, or at least one match |
| 1492 |
foreach ($terms as $tax => $term_ids) : |
| 1493 |
if (!isset($termSet[$tax])) : |
| 1494 |
$termSet[$tax] = array(); |
| 1495 |
endif; |
| 1496 |
$termSet[$tax] = array_merge($termSet[$tax], $term_ids); |
| 1497 |
endforeach; |
| 1498 |
endif; |
| 1499 |
endif; |
| 1500 |
endforeach; |
| 1501 |
|
| 1502 |
if (is_null($valid)) : // Plonked |
| 1503 |
$this->post = NULL; |
| 1504 |
else : // We can proceed |
| 1505 |
$this->post['tax_input'] = array(); |
| 1506 |
foreach ($termSet as $tax => $term_ids) : |
| 1507 |
if (!isset($this->post['tax_input'][$tax])) : |
| 1508 |
$this->post['tax_input'][$tax] = array(); |
| 1509 |
endif; |
| 1510 |
$this->post['tax_input'][$tax] = array_merge( |
| 1511 |
$this->post['tax_input'][$tax], |
| 1512 |
$term_ids |
| 1513 |
); |
| 1514 |
endforeach; |
| 1515 |
|
| 1516 |
// Now let's add on the feed and global presets |
| 1517 |
foreach ($this->preset_terms as $tax => $term_ids) : |
| 1518 |
if (!isset($this->post['tax_input'][$tax])) : |
| 1519 |
$this->post['tax_input'][$tax] = array(); |
| 1520 |
endif; |
| 1521 |
|
| 1522 |
$this->post['tax_input'][$tax] = array_merge ( |
| 1523 |
$this->post['tax_input'][$tax], |
| 1524 |
$this->category_ids ( |
| 1525 |
/*terms=*/ $term_ids, |
| 1526 |
/*unfamiliar=*/ 'create:'.$tax, // These are presets; for those added in a tagbox editor, the tag may not yet exist |
| 1527 |
/*taxonomies=*/ array($tax), |
| 1528 |
array( |
| 1529 |
'singleton' => true, |
| 1530 |
)) |
| 1531 |
); |
| 1532 |
endforeach; |
| 1533 |
endif; |
| 1534 |
endif; |
| 1535 |
|
| 1536 |
// We have to check again in case the post has been filtered |
| 1537 |
// during the category/tags/taxonomy terms lookup |
| 1538 |
if ($this->has_fresh_content()) : |
| 1539 |
// Filter some individual fields |
| 1540 |
|
| 1541 |
// If there already is a post slug (from syndication or by manual |
| 1542 |
// editing) don't cause WP to overwrite it by sending in a NULL |
| 1543 |
// post_name. Props Chris Fritz 2012-11-28. |
| 1544 |
$post_name = (is_null($this->_wp_post) ? NULL : $this->_wp_post->post_name); |
| 1545 |
|
| 1546 |
// Allow filters to set post slug. Props niska. |
| 1547 |
$post_name = apply_filters('syndicated_post_slug', $post_name, $this); |
| 1548 |
if (!empty($post_name)) : |
| 1549 |
$this->post['post_name'] = $post_name; |
| 1550 |
endif; |
| 1551 |
|
| 1552 |
$this->post = apply_filters('syndicated_post', $this->post, $this); |
| 1553 |
|
| 1554 |
// Allow for feed-specific syndicated_post filters. |
| 1555 |
$this->post = apply_filters( |
| 1556 |
"syndicated_post_".$this->link->uri(), |
| 1557 |
$this->post, |
| 1558 |
$this |
| 1559 |
); |
| 1560 |
endif; |
| 1561 |
|
| 1562 |
// Hook in early to make sure these get inserted if at all possible |
| 1563 |
add_action( |
| 1564 |
/*hook=*/ 'transition_post_status', |
| 1565 |
/*callback=*/ array($this, 'add_rss_meta'), |
| 1566 |
/*priority=*/ -10000, /* very early */ |
| 1567 |
/*arguments=*/ 3 |
| 1568 |
); |
| 1569 |
|
| 1570 |
$ret = false; |
| 1571 |
if ($this->has_fresh_content()) : |
| 1572 |
$diag = $this->fresh_storage_diagnostic(); |
| 1573 |
if (!is_null($diag)) : |
| 1574 |
FeedWordPress::diagnostic('syndicated_posts', $diag); |
| 1575 |
endif; |
| 1576 |
|
| 1577 |
$this->insert_post(/*update=*/ $this->fresh_content_is_update(), $this->freshness()); |
| 1578 |
|
| 1579 |
$hook = $this->fresh_storage_hook(); |
| 1580 |
if (!is_null($hook)) : |
| 1581 |
do_action($hook, $this->wp_id(), $this); |
| 1582 |
endif; |
| 1583 |
|
| 1584 |
$ret = $this->freshness('status'); |
| 1585 |
endif; |
| 1586 |
|
| 1587 |
// If this is a legit, non-filtered post, tag it as found on the |
| 1588 |
// feed regardless of fresh or stale status |
| 1589 |
if (!$this->filtered()) : |
| 1590 |
$key = '_feedwordpress_retire_me_' . $this->link->id; |
| 1591 |
delete_post_meta($this->wp_id(), $key); |
| 1592 |
|
| 1593 |
$status = get_post_field('post_status', $this->wp_id()); |
| 1594 |
if ('fwpretired'==$status and $this->link->is_non_incremental()) : |
| 1595 |
FeedWordPress::diagnostic('syndicated_posts', "Un-retiring previously retired post # ".$this->wp_id()." due to re-appearance on non-incremental feed."); |
| 1596 |
set_post_field('post_status', $this->post['post_status'], $this->wp_id()); |
| 1597 |
wp_transition_post_status($this->post['post_status'], $status, $this->post); |
| 1598 |
elseif ('fwpzapped'==$status) : |
| 1599 |
// Set this new revision up to be |
| 1600 |
// blanked on the next update. |
| 1601 |
add_post_meta($this->wp_id(), '_feedwordpress_zapped_blank_me', 2, /*single=*/ true); |
| 1602 |
endif; |
| 1603 |
endif; |
| 1604 |
|
| 1605 |
// Remove add_rss_meta hook |
| 1606 |
remove_action( |
| 1607 |
/*hook=*/ 'transition_post_status', |
| 1608 |
/*callback=*/ array($this, 'add_rss_meta'), |
| 1609 |
/*priority=*/ -10000, /* very early */ |
| 1610 |
/*arguments=*/ 3 |
| 1611 |
); |
| 1612 |
|
| 1613 |
return $ret; |
| 1614 |
} /* function SyndicatedPost::store () */ |
| 1615 |
|
| 1616 |
function insert_post ($update = false, $freshness = 2) { |
| 1617 |
global $wpdb; |
| 1618 |
|
| 1619 |
$dbpost = $this->normalize_post(/*new=*/ true); |
| 1620 |
|
| 1621 |
$ret = null; |
| 1622 |
|
| 1623 |
if (!is_null($dbpost)) : |
| 1624 |
$dbpost['post_pingback'] = false; // Tell WP 2.1 and 2.2 not to process for pingbacks |
| 1625 |
|
| 1626 |
// This is a ridiculous fucking kludge necessitated by WordPress 2.6 munging authorship meta-data |
| 1627 |
add_action('_wp_put_post_revision', array($this, 'fix_revision_meta')); |
| 1628 |
|
| 1629 |
// Kludge to prevent kses filters from stripping the |
| 1630 |
// content of posts when updating without a logged in |
| 1631 |
// user who has `unfiltered_html` capability. |
| 1632 |
$mungers = array('wp_filter_kses', 'wp_filter_post_kses'); |
| 1633 |
$removed = array(); |
| 1634 |
foreach ($mungers as $munger) : |
| 1635 |
if (has_filter('content_save_pre', $munger)) : |
| 1636 |
remove_filter('content_save_pre', $munger); |
| 1637 |
$removed[] = $munger; |
| 1638 |
endif; |
| 1639 |
endforeach; |
| 1640 |
|
| 1641 |
if ($update and function_exists('get_post_field')) : |
| 1642 |
// Don't munge status fields that the user may |
| 1643 |
// have reset manually |
| 1644 |
$doNotMunge = array('post_status', 'comment_status', 'ping_status'); |
| 1645 |
|
| 1646 |
foreach ($doNotMunge as $field) : |
| 1647 |
$dbpost[$field] = get_post_field($field, $this->wp_id()); |
| 1648 |
endforeach; |
| 1649 |
endif; |
| 1650 |
|
| 1651 |
// WP3's wp_insert_post scans current_user_can() for the |
| 1652 |
// tax_input, with no apparent way to override. Ugh. |
| 1653 |
add_action( |
| 1654 |
/*hook=*/ 'transition_post_status', |
| 1655 |
/*callback=*/ array($this, 'add_terms'), |
| 1656 |
/*priority=*/ -10001, /* very early */ |
| 1657 |
/*arguments=*/ 3 |
| 1658 |
); |
| 1659 |
|
| 1660 |
// WP3 appears to override whatever you give it for |
| 1661 |
// post_modified. Ugh. |
| 1662 |
add_action( |
| 1663 |
/*hook=*/ 'transition_post_status', |
| 1664 |
/*callback=*/ array($this, 'fix_post_modified_ts'), |
| 1665 |
/*priority=*/ -10000, /* very early */ |
| 1666 |
/*arguments=*/ 3 |
| 1667 |
); |
| 1668 |
|
| 1669 |
if ($update) : |
| 1670 |
$this->post['ID'] = $this->wp_id(); |
| 1671 |
$dbpost['ID'] = $this->post['ID']; |
| 1672 |
endif; |
| 1673 |
|
| 1674 |
// O.K., is this a new post? If so, we need to create |
| 1675 |
// the basic post record before we do anything else. |
| 1676 |
if ($this->this_revision_needs_original_post()) : |
| 1677 |
// *sigh*, for handling inconsistent slash expectations < 3.6 |
| 1678 |
$sdbpost = $this->db_sanitize_post($dbpost); |
| 1679 |
|
| 1680 |
// Go ahead and insert the first post record to |
| 1681 |
// anchor the revision history. |
| 1682 |
$this->_wp_id = wp_insert_post($sdbpost, /*return wp_error=*/ true); |
| 1683 |
|
| 1684 |
$dbpost['ID'] = $this->_wp_id; |
| 1685 |
endif; |
| 1686 |
|
| 1687 |
// Sanity check: if the attempt to insert post |
| 1688 |
// returned an error, then feeding that error |
| 1689 |
// object in to _wp_put_post_revision() would |
| 1690 |
// cause a fatal error. Better to break out. |
| 1691 |
if (!is_wp_error($this->_wp_id)) : |
| 1692 |
// Now that we've made sure the original exists, insert |
| 1693 |
// this version here as a revision. |
| 1694 |
$revision_id = _wp_put_post_revision($dbpost, /*autosave=*/ false); |
| 1695 |
|
| 1696 |
if (!$this->this_revision_needs_original_post()) : |
| 1697 |
|
| 1698 |
if ($this->this_revision_is_current()) : |
| 1699 |
|
| 1700 |
wp_restore_post_revision($revision_id); |
| 1701 |
|
| 1702 |
else : |
| 1703 |
|
| 1704 |
// If we do not activate this revision, then the |
| 1705 |
// add_rss_meta will not be called, which is |
| 1706 |
// more or less as it should be, but that means |
| 1707 |
// we have to actively record this revision's |
| 1708 |
// update hash from here. |
| 1709 |
$postId = $this->post['ID']; |
| 1710 |
$key = 'syndication_item_hash'; |
| 1711 |
$hash = $this->update_hash(); |
| 1712 |
FeedWordPress::diagnostic('syndicated_posts:meta_data', "Adding post meta-datum to post [$postId]: [$key] = ".FeedWordPress::val($hash, /*no newlines=*/ true)); |
| 1713 |
add_post_meta( $postId, $key, $hash, /*unique=*/ false ); |
| 1714 |
endif; |
| 1715 |
endif; |
| 1716 |
endif; |
| 1717 |
|
| 1718 |
remove_action( |
| 1719 |
/*hook=*/ 'transition_post_status', |
| 1720 |
/*callback=*/ array($this, 'add_terms'), |
| 1721 |
/*priority=*/ -10001, /* very early */ |
| 1722 |
/*arguments=*/ 3 |
| 1723 |
); |
| 1724 |
|
| 1725 |
remove_action( |
| 1726 |
/*hook=*/ 'transition_post_status', |
| 1727 |
/*callback=*/ array($this, 'fix_post_modified_ts'), |
| 1728 |
/*priority=*/ -10000, /* very early */ |
| 1729 |
/*arguments=*/ 3 |
| 1730 |
); |
| 1731 |
|
| 1732 |
// Turn off ridiculous fucking kludges #1 and #2 |
| 1733 |
remove_action('_wp_put_post_revision', array($this, 'fix_revision_meta')); |
| 1734 |
foreach ($removed as $filter) : |
| 1735 |
add_filter('content_save_pre', $filter); |
| 1736 |
endforeach; |
| 1737 |
|
| 1738 |
$this->validate_post_id($dbpost, $update, array(__CLASS__, __FUNCTION__)); |
| 1739 |
|
| 1740 |
$ret = $this->_wp_id; |
| 1741 |
endif; |
| 1742 |
return $ret; |
| 1743 |
} /* function SyndicatedPost::insert_post () */ |
| 1744 |
|
| 1745 |
function insert_new () { |
| 1746 |
$this->insert_post(/*update=*/ false, 1); |
| 1747 |
} /* SyndicatedPost::insert_new() */ |
| 1748 |
|
| 1749 |
function update_existing () { |
| 1750 |
$this->insert_post(/*update=*/ true, 2); |
| 1751 |
} /* SyndicatedPost::update_existing() */ |
| 1752 |
|
| 1753 |
/** |
| 1754 |
* SyndicatedPost::normalize_post() |
| 1755 |
* |
| 1756 |
* @param bool $new If true, this post is to be inserted anew. If false, it is an update of an existing post. |
| 1757 |
* @return array A normalized representation of the post ready to be inserted into the database or sent to the WordPress API functions |
| 1758 |
*/ |
| 1759 |
function normalize_post ($new = true) { |
| 1760 |
global $wpdb; |
| 1761 |
|
| 1762 |
$out = $this->post; |
| 1763 |
|
| 1764 |
$fullPost = $out['post_title'].$out['post_content']; |
| 1765 |
$fullPost .= (isset($out['post_excerpt']) ? $out['post_excerpt'] : ''); |
| 1766 |
if (strlen($fullPost) < 1) : |
| 1767 |
// FIXME: Option for filtering out empty posts |
| 1768 |
endif; |
| 1769 |
if (strlen($out['post_title'])==0) : |
| 1770 |
$offset = (int) get_option('gmt_offset') * 60 * 60; |
| 1771 |
if (isset($this->post['meta']['syndication_source'])) : |
| 1772 |
$source_title = $this->post['meta']['syndication_source']; |
| 1773 |
else : |
| 1774 |
$feed_url = parse_url($this->post['meta']['syndication_feed']); |
| 1775 |
$source_title = $feed_url['host']; |
| 1776 |
endif; |
| 1777 |
|
| 1778 |
$out['post_title'] = $source_title |
| 1779 |
.' '.gmdate('Y-m-d H:i:s', $this->published() + $offset); |
| 1780 |
// FIXME: Option for what to fill a blank title with... |
| 1781 |
endif; |
| 1782 |
|
| 1783 |
// Normalize the guid if necessary. |
| 1784 |
$out['guid'] = SyndicatedPost::normalize_guid($out['guid']); |
| 1785 |
|
| 1786 |
return $out; |
| 1787 |
} |
| 1788 |
|
| 1789 |
public function db_sanitize_post_check_encoding ($out) { |
| 1790 |
// Check encoding recursively: every string field needs to be checked |
| 1791 |
// for character encoding issues. This is a bit problematic because we |
| 1792 |
// *should* be using DB_CHARSET, but DB_CHARSET sometimes has values |
| 1793 |
// that work for MySQL but not for PHP mb_check_encoding. So instead |
| 1794 |
// we must rely on WordPress setting blog_charset and hope that the user |
| 1795 |
// has got their database encoding set up to roughly match |
| 1796 |
$charset = get_option('blog_charset', 'utf8'); |
| 1797 |
|
| 1798 |
foreach ($out as $key => $value) : |
| 1799 |
if (is_string($value)) : |
| 1800 |
|
| 1801 |
if (!function_exists('mb_check_encoding') or mb_check_encoding($value, $charset)) : |
| 1802 |
$out[$key] = $value; |
| 1803 |
else : |
| 1804 |
$fromCharset = mb_detect_encoding($value, mb_detect_order(), /*strict=*/ true); |
| 1805 |
$out[$key] = mb_convert_encoding($value, $charset, $fromCharset); |
| 1806 |
endif; |
| 1807 |
|
| 1808 |
elseif (is_array($value)) : |
| 1809 |
$out[$key] = $this->db_sanitize_post_check_encoding($value); |
| 1810 |
|
| 1811 |
else : |
| 1812 |
$out[$key] = $value; |
| 1813 |
endif; |
| 1814 |
|
| 1815 |
endforeach; |
| 1816 |
|
| 1817 |
return $out; |
| 1818 |
} /* SyndicatedPost::db_sanitize_post_check_encoding () */ |
| 1819 |
|
| 1820 |
function db_sanitize_post ($out) { |
| 1821 |
global $wp_db_version; |
| 1822 |
|
| 1823 |
$out = $this->db_sanitize_post_check_encoding($out); |
| 1824 |
|
| 1825 |
// < 3.6. Core API, including `wp_insert_post()`, expects |
| 1826 |
// properly slashed data. If `wp_slash()` exists, then |
| 1827 |
// this is after the big change-over in how data slashing |
| 1828 |
// was handled. |
| 1829 |
if (!function_exists('wp_slash')) : |
| 1830 |
|
| 1831 |
foreach ($out as $key => $value) : |
| 1832 |
if (is_string($value)) : |
| 1833 |
$out[$key] = esc_sql($value); |
| 1834 |
else : |
| 1835 |
$out[$key] = $value; |
| 1836 |
endif; |
| 1837 |
endforeach; |
| 1838 |
|
| 1839 |
// For revisions [@23416,@23554), core API expects |
| 1840 |
// unslashed data. Cf. <https://core.trac.wordpress.org/browser/trunk/wp-includes/post.php?rev=23416> |
| 1841 |
// NOOP for those revisions. |
| 1842 |
|
| 1843 |
// In revisions @23554 to present, `wp_insert_post()` |
| 1844 |
// expects slashed data once again. |
| 1845 |
// Cf. <https://core.trac.wordpress.org/changeset/23554/trunk/wp-includes/post.php?contextall=1> |
| 1846 |
// But at least now we can use the wp_slash API function to do that. |
| 1847 |
// Hooray. |
| 1848 |
|
| 1849 |
elseif ($wp_db_version >= 23524) : |
| 1850 |
|
| 1851 |
$out = wp_slash($out); |
| 1852 |
|
| 1853 |
endif; |
| 1854 |
|
| 1855 |
return $out; |
| 1856 |
} |
| 1857 |
|
| 1858 |
/** |
| 1859 |
* SyndicatedPost::validate_post_id() |
| 1860 |
* |
| 1861 |
* @param array $dbpost An array representing the post we attempted to insert or update |
| 1862 |
* @param mixed $ns A string or array representing the namespace (class, method) whence this method was called. |
| 1863 |
*/ |
| 1864 |
function validate_post_id ($dbpost, $is_update, $ns) { |
| 1865 |
if (is_array($ns)) : $ns = implode('::', $ns); |
| 1866 |
else : $ns = (string) $ns; endif; |
| 1867 |
|
| 1868 |
// This should never happen. |
| 1869 |
if (!is_numeric($this->_wp_id) or ($this->_wp_id == 0)) : |
| 1870 |
$verb = ($is_update ? 'update existing' : 'insert new'); |
| 1871 |
$guid = $this->guid(); |
| 1872 |
$url = $this->permalink(); |
| 1873 |
$feed = $this->link->uri(array('add_params' => true)); |
| 1874 |
|
| 1875 |
// wp_insert_post failed. Diagnostics, or barf up a critical bug |
| 1876 |
// notice if we are in debug mode. |
| 1877 |
$mesg = "Failed to $verb item [$guid]. WordPress API returned no valid post ID.\n" |
| 1878 |
."\t\tID = ".serialize($this->_wp_id)."\n" |
| 1879 |
."\t\tURL = ".MyPHP::val($url) |
| 1880 |
."\t\tFeed = ".MyPHP::val($feed); |
| 1881 |
|
| 1882 |
FeedWordPress::diagnostic('updated_feeds:errors', "WordPress API error: $mesg"); |
| 1883 |
FeedWordPress::diagnostic('feed_items:rejected', $mesg); |
| 1884 |
|
| 1885 |
$mesg = <<<EOM |
| 1886 |
The WordPress API returned an invalid post ID |
| 1887 |
when FeedWordPress tried to $verb item $guid |
| 1888 |
[URL: $url] |
| 1889 |
from the feed at $feed |
| 1890 |
|
| 1891 |
$ns::_wp_id |
| 1892 |
EOM; |
| 1893 |
FeedWordPress::noncritical_bug( |
| 1894 |
/*message=*/ $mesg, |
| 1895 |
/*var =*/ array( |
| 1896 |
"\$this->_wp_id" => $this->_wp_id, |
| 1897 |
"\$dbpost" => $dbpost, |
| 1898 |
), |
| 1899 |
/*line # =*/ __LINE__, /*filename=*/ __FILE__ |
| 1900 |
); |
| 1901 |
endif; |
| 1902 |
} /* SyndicatedPost::validate_post_id() */ |
| 1903 |
|
| 1904 |
/** |
| 1905 |
* SyndicatedPost::fix_revision_meta() - Ensures that we get the meta |
| 1906 |
* data (authorship, guid, etc.) that we want when storing revisions of |
| 1907 |
* a syndicated post. |
| 1908 |
* |
| 1909 |
* In their infinite wisdom, the WordPress coders seem to have made it |
| 1910 |
* completely impossible for a plugin that uses wp_insert_post() to set |
| 1911 |
* certain meta-data (such as the author) when you store an old revision |
| 1912 |
* of an updated post. Instead, it uses the WordPress defaults (= cur. |
| 1913 |
* active user ID if the process is running with a user logged in, or |
| 1914 |
* = #0 if there is no user logged in). This results in bogus authorship |
| 1915 |
* data for revisions that are syndicated from off the feed, unless we |
| 1916 |
* use a ridiculous kludge like this to end-run the munging of meta-data |
| 1917 |
* by _wp_put_post_revision. |
| 1918 |
* |
| 1919 |
* @param int $revision_id The revision ID to fix up meta-data |
| 1920 |
*/ |
| 1921 |
function fix_revision_meta ($revision_id) { |
| 1922 |
global $wpdb; |
| 1923 |
|
| 1924 |
$post_author = (int) $this->post['post_author']; |
| 1925 |
|
| 1926 |
$revision_id = (int) $revision_id; |
| 1927 |
|
| 1928 |
// Let's fix the author. |
| 1929 |
set_post_field('post_author', $this->post['post_author'], $revision_id); |
| 1930 |
|
| 1931 |
// Let's fix the GUID to a dummy URL with the update hash. |
| 1932 |
set_post_field('guid', 'http://feedwordpress.radgeek.com/?rev='.$this->update_hash(), $revision_id); |
| 1933 |
|
| 1934 |
// Let's fire an event for add-ons and filters |
| 1935 |
do_action('syndicated_post_fix_revision_meta', $revision_id, $this); |
| 1936 |
|
| 1937 |
} /* SyndicatedPost::fix_revision_meta () */ |
| 1938 |
|
| 1939 |
/** |
| 1940 |
* SyndicatedPost::add_terms() -- if FeedWordPress is processing an |
| 1941 |
* automatic update, that generally means that wp_insert_post() is being |
| 1942 |
* called under the user credentials of whoever is viewing the blog at |
| 1943 |
* the time -- which usually means no user at all. But wp_insert_post() |
| 1944 |
* checks current_user_can() before assigning any of the terms in a |
| 1945 |
* post's tax_input structure -- which is unfortunate, since |
| 1946 |
* current_user_can() always returns FALSE when there is no current user |
| 1947 |
* logged in. Meaning that automatic updates get no terms assigned. |
| 1948 |
* |
| 1949 |
* So, wp_insert_post() is not going to do the term assignments for us. |
| 1950 |
* If you want something done right.... |
| 1951 |
* |
| 1952 |
* @param string $new_status Unused action parameter. |
| 1953 |
* @param string $old_status Unused action parameter. |
| 1954 |
* @param object $post The database record for the post just inserted. |
| 1955 |
*/ |
| 1956 |
function add_terms ($new_status, $old_status, $post) { |
| 1957 |
|
| 1958 |
if ($new_status!='inherit') : // Bail if we are creating a revision. |
| 1959 |
if ( is_array($this->post) and isset($this->post['tax_input']) and is_array($this->post['tax_input']) ) : |
| 1960 |
foreach ($this->post['tax_input'] as $taxonomy => $terms) : |
| 1961 |
if (is_array($terms)) : |
| 1962 |
$terms = array_filter($terms); // strip out empties |
| 1963 |
endif; |
| 1964 |
|
| 1965 |
$res = wp_set_post_terms( |
| 1966 |
/*post_id=*/ $post->ID, |
| 1967 |
/*terms=*/ $terms, |
| 1968 |
/*taxonomy=*/ $taxonomy |
| 1969 |
); |
| 1970 |
|
| 1971 |
FeedWordPress::diagnostic( |
| 1972 |
'syndicated_posts:categories', |
| 1973 |
'Category: post('.json_encode($post->ID).') '.$taxonomy |
| 1974 |
.' := ' |
| 1975 |
.json_encode($terms) |
| 1976 |
.' / result: ' |
| 1977 |
.json_encode($res) |
| 1978 |
); |
| 1979 |
|
| 1980 |
endforeach; |
| 1981 |
endif; |
| 1982 |
endif; |
| 1983 |
|
| 1984 |
} /* SyndicatedPost::add_terms () */ |
| 1985 |
|
| 1986 |
/** |
| 1987 |
* SyndicatedPost::fix_post_modified_ts() -- We would like to set |
| 1988 |
* post_modified and post_modified_gmt to reflect the value of |
| 1989 |
* <atom:updated> or equivalent elements on the feed. Unfortunately, |
| 1990 |
* wp_insert_post() refuses to acknowledge explicitly-set post_modified |
| 1991 |
* fields and overwrites them, either with the post_date (if new) or the |
| 1992 |
* current timestamp (if updated). |
| 1993 |
* |
| 1994 |
* So, wp_insert_post() is not going to do the last-modified assignments |
| 1995 |
* for us. If you want something done right.... |
| 1996 |
* |
| 1997 |
* @param string $new_status Unused action parameter. |
| 1998 |
* @param string $old_status Unused action parameter. |
| 1999 |
* @param object $post The database record for the post just inserted. |
| 2000 |
*/ |
| 2001 |
function fix_post_modified_ts ($new_status, $old_status, $post) { |
| 2002 |
global $wpdb; |
| 2003 |
if ($new_status!='inherit') : // Bail if we are creating a revision. |
| 2004 |
$wpdb->update( $wpdb->posts, /*data=*/ array( |
| 2005 |
'post_modified' => $this->post['post_modified'], |
| 2006 |
'post_modified_gmt' => $this->post['post_modified_gmt'], |
| 2007 |
), /*where=*/ array('ID' => $post->ID) ); |
| 2008 |
endif; |
| 2009 |
} /* SyndicatedPost::fix_post_modified_ts () */ |
| 2010 |
|
| 2011 |
/** |
| 2012 |
* SyndicatedPost::add_rss_meta: adds interesting meta-data to each entry |
| 2013 |
* using the space for custom keys. The set of keys and values to add is |
| 2014 |
* specified by the keys and values of $post['meta']. This is used to |
| 2015 |
* store anything that the WordPress user might want to access from a |
| 2016 |
* template concerning the post's original source that isn't provided |
| 2017 |
* for by standard WP meta-data (i.e., any interesting data about the |
| 2018 |
* syndicated post other than author, title, timestamp, categories, and |
| 2019 |
* guid). It's also used to hook into WordPress's support for |
| 2020 |
* enclosures. |
| 2021 |
* |
| 2022 |
* @param string $new_status Unused action parameter. |
| 2023 |
* @param string $old_status Unused action parameter. |
| 2024 |
* @param object $post The database record for the post just inserted. |
| 2025 |
*/ |
| 2026 |
function add_rss_meta ($new_status, $old_status, $post) { |
| 2027 |
global $wpdb; |
| 2028 |
if ($new_status!='inherit') : // Bail if we are creating a revision. |
| 2029 |
FeedWordPress::diagnostic('syndicated_posts:meta_data', 'Adding post meta-data: {'.implode(", ", array_keys($this->post['meta'])).'}'); |
| 2030 |
|
| 2031 |
if ( is_array($this->post) and isset($this->post['meta']) and is_array($this->post['meta']) ) : |
| 2032 |
$postId = $post->ID; |
| 2033 |
|
| 2034 |
// Aggregated posts should NOT send out pingbacks. |
| 2035 |
// WordPress 2.1-2.2 claim you can tell them not to |
| 2036 |
// using $post_pingback, but they don't listen, so we |
| 2037 |
// make sure here. |
| 2038 |
$result = $wpdb->query(" |
| 2039 |
DELETE FROM $wpdb->postmeta |
| 2040 |
WHERE post_id='$postId' AND meta_key='_pingme' |
| 2041 |
"); |
| 2042 |
|
| 2043 |
foreach ( $this->post['meta'] as $key => $values ) : |
| 2044 |
$eKey = esc_sql($key); |
| 2045 |
|
| 2046 |
// If this is an update, clear out the old |
| 2047 |
// values to avoid duplication. |
| 2048 |
$result = $wpdb->query(" |
| 2049 |
DELETE FROM $wpdb->postmeta |
| 2050 |
WHERE post_id='$postId' AND meta_key='$eKey' |
| 2051 |
"); |
| 2052 |
|
| 2053 |
// Allow for either a single value or an array |
| 2054 |
if (!is_array($values)) $values = array($values); |
| 2055 |
foreach ( $values as $value ) : |
| 2056 |
FeedWordPress::diagnostic('syndicated_posts:meta_data', "Adding post meta-datum to post [$postId]: [$key] = ".MyPHP::val($value, /*no newlines=*/ true)); |
| 2057 |
add_post_meta($postId, $key, $value, /*unique=*/ false); |
| 2058 |
endforeach; |
| 2059 |
endforeach; |
| 2060 |
endif; |
| 2061 |
endif; |
| 2062 |
} /* SyndicatedPost::add_rss_meta () */ |
| 2063 |
|
| 2064 |
/** |
| 2065 |
* SyndicatedPost::author_id (): get the ID for an author name from |
| 2066 |
* the feed. Create the author if necessary. |
| 2067 |
* |
| 2068 |
* @param string $unfamiliar_author |
| 2069 |
* |
| 2070 |
* @return NULL|int The numeric ID of the author to attribute the post to |
| 2071 |
* NULL if the post should be filtered out. |
| 2072 |
*/ |
| 2073 |
function author_id ($unfamiliar_author = 'create') { |
| 2074 |
global $wpdb; |
| 2075 |
|
| 2076 |
$a = $this->named['author']; |
| 2077 |
|
| 2078 |
$source = $this->source(); |
| 2079 |
$forbidden = apply_filters('feedwordpress_forbidden_author_names', |
| 2080 |
array('admin', 'administrator', 'www', 'root')); |
| 2081 |
|
| 2082 |
// Prepare the list of candidates to try for author name: name from |
| 2083 |
// feed, original source title (if any), immediate source title live |
| 2084 |
// from feed, subscription title, prettied version of feed homepage URL, |
| 2085 |
// prettied version of feed URL, or, failing all, use "unknown author" |
| 2086 |
// as last resort |
| 2087 |
|
| 2088 |
$candidates = array(); |
| 2089 |
$candidates[] = $a['name']; |
| 2090 |
if (!is_null($source)) : $candidates[] = $source['title']; endif; |
| 2091 |
$candidates[] = $this->link->name(/*fromFeed=*/ true); |
| 2092 |
$candidates[] = $this->link->name(/*fromFeed=*/ false); |
| 2093 |
if (strlen($this->link->homepage()) > 0) : $candidates[] = feedwordpress_display_url($this->link->homepage()); endif; |
| 2094 |
$candidates[] = feedwordpress_display_url($this->link->uri()); |
| 2095 |
$candidates[] = 'unknown author'; |
| 2096 |
|
| 2097 |
// Pick the first one that works from the list, screening against empty |
| 2098 |
// or forbidden names. |
| 2099 |
|
| 2100 |
$author = NULL; |
| 2101 |
while (is_null($author) and ($candidate = each($candidates))) : |
| 2102 |
if (!is_null($candidate['value']) |
| 2103 |
and (strlen(trim($candidate['value'])) > 0) |
| 2104 |
and !in_array(strtolower(trim($candidate['value'])), $forbidden)) : |
| 2105 |
$author = $candidate['value']; |
| 2106 |
endif; |
| 2107 |
endwhile; |
| 2108 |
|
| 2109 |
$email = (isset($a['email']) ? $a['email'] : NULL); |
| 2110 |
$authorUrl = (isset($a['uri']) ? $a['uri'] : NULL); |
| 2111 |
|
| 2112 |
|
| 2113 |
$hostUrl = $this->link->homepage(); |
| 2114 |
if (is_null($hostUrl) or (strlen($hostUrl) < 0)) : |
| 2115 |
$hostUrl = $this->link->uri(); |
| 2116 |
endif; |
| 2117 |
|
| 2118 |
$match_author_by_email = !('yes' == get_option("feedwordpress_do_not_match_author_by_email")); |
| 2119 |
if ($match_author_by_email and !FeedWordPress::is_null_email($email)) : |
| 2120 |
$test_email = $email; |
| 2121 |
else : |
| 2122 |
$test_email = NULL; |
| 2123 |
endif; |
| 2124 |
|
| 2125 |
// Never can be too careful... |
| 2126 |
$login = sanitize_user($author, /*strict=*/ true); |
| 2127 |
|
| 2128 |
// Possible for, e.g., foreign script author names |
| 2129 |
if (strlen($login) < 1) : |
| 2130 |
// No usable characters in author name for a login. |
| 2131 |
// (Sometimes results from, e.g., foreign scripts.) |
| 2132 |
// |
| 2133 |
// We just need *something* in Western alphanumerics, |
| 2134 |
// so let's try the domain name. |
| 2135 |
// |
| 2136 |
// Uniqueness will be guaranteed below if necessary. |
| 2137 |
|
| 2138 |
$url = parse_url($hostUrl); |
| 2139 |
|
| 2140 |
$login = sanitize_user($url['host'], /*strict=*/ true); |
| 2141 |
if (strlen($login) < 1) : |
| 2142 |
// This isn't working. Frak it. |
| 2143 |
$login = 'syndicated'; |
| 2144 |
endif; |
| 2145 |
endif; |
| 2146 |
|
| 2147 |
$login = apply_filters('pre_user_login', $login); |
| 2148 |
|
| 2149 |
$nice_author = sanitize_title($author); |
| 2150 |
$nice_author = apply_filters('pre_user_nicename', $nice_author); |
| 2151 |
|
| 2152 |
$reg_author = esc_sql(preg_quote($author)); |
| 2153 |
$author = esc_sql($author); |
| 2154 |
$email = esc_sql($email); |
| 2155 |
$test_email = esc_sql($test_email); |
| 2156 |
$authorUrl = esc_sql($authorUrl); |
| 2157 |
|
| 2158 |
// Check for an existing author rule.... |
| 2159 |
if (isset($this->link->settings['map authors']['name']['*'])) : |
| 2160 |
$author_rule = $this->link->settings['map authors']['name']['*']; |
| 2161 |
elseif (isset($this->link->settings['map authors']['name'][strtolower(trim($author))])) : |
| 2162 |
$author_rule = $this->link->settings['map authors']['name'][strtolower(trim($author))]; |
| 2163 |
else : |
| 2164 |
$author_rule = NULL; |
| 2165 |
endif; |
| 2166 |
|
| 2167 |
// User name is mapped to a particular author. If that author ID exists, use it. |
| 2168 |
if (is_numeric($author_rule) and get_userdata((int) $author_rule)) : |
| 2169 |
$id = (int) $author_rule; |
| 2170 |
|
| 2171 |
// User name is filtered out |
| 2172 |
elseif ('filter' == $author_rule) : |
| 2173 |
$id = NULL; |
| 2174 |
|
| 2175 |
else : |
| 2176 |
// Check the database for an existing author record that might fit |
| 2177 |
|
| 2178 |
// First try the user core data table. |
| 2179 |
$id = $wpdb->get_var( |
| 2180 |
"SELECT ID FROM $wpdb->users |
| 2181 |
WHERE TRIM(LCASE(display_name)) = TRIM(LCASE('$author')) |
| 2182 |
OR TRIM(LCASE(user_login)) = TRIM(LCASE('$author')) |
| 2183 |
OR ( |
| 2184 |
LENGTH(TRIM(LCASE(user_email))) > 0 |
| 2185 |
AND TRIM(LCASE(user_email)) = TRIM(LCASE('$test_email')) |
| 2186 |
)"); |
| 2187 |
|
| 2188 |
// If that fails, look for aliases in the user meta data table |
| 2189 |
if (is_null($id)) : |
| 2190 |
$id = $wpdb->get_var( |
| 2191 |
"SELECT user_id FROM $wpdb->usermeta |
| 2192 |
WHERE |
| 2193 |
(meta_key = 'description' AND TRIM(LCASE(meta_value)) = TRIM(LCASE('$author'))) |
| 2194 |
OR ( |
| 2195 |
meta_key = 'description' |
| 2196 |
AND TRIM(LCASE(meta_value)) |
| 2197 |
RLIKE CONCAT( |
| 2198 |
'(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*', |
| 2199 |
TRIM(LCASE('$reg_author')), |
| 2200 |
'( |\\t|\\r)*(\\n|\$)' |
| 2201 |
) |
| 2202 |
) |
| 2203 |
"); |
| 2204 |
endif; |
| 2205 |
|
| 2206 |
// ... if you don't find one, then do what you need to do |
| 2207 |
if (is_null($id)) : |
| 2208 |
if ($unfamiliar_author === 'create') : |
| 2209 |
$userdata = array(); |
| 2210 |
|
| 2211 |
// WordPress 3 is going to pitch a fit if we attempt to register |
| 2212 |
// more than one user account with an empty e-mail address, so we |
| 2213 |
// need *something* here. Ugh. |
| 2214 |
if (strlen($email) == 0 or FeedWordPress::is_null_email($email)) : |
| 2215 |
$url = parse_url($hostUrl); |
| 2216 |
$email = $nice_author.'@'.$url['host']; |
| 2217 |
endif; |
| 2218 |
|
| 2219 |
#-- user table data |
| 2220 |
$userdata['ID'] = NULL; // new user |
| 2221 |
$userdata['user_login'] = $login; |
| 2222 |
$userdata['user_nicename'] = $nice_author; |
| 2223 |
$userdata['user_pass'] = substr(md5(uniqid(microtime())), 0, 6); // just something random to lock it up |
| 2224 |
$userdata['user_email'] = $email; |
| 2225 |
$userdata['user_url'] = $authorUrl; |
| 2226 |
$userdata['nickname'] = $author; |
| 2227 |
$parts = preg_split('/\s+/', trim($author), 2); |
| 2228 |
if (isset($parts[0])) : $userdata['first_name'] = $parts[0]; endif; |
| 2229 |
if (isset($parts[1])) : $userdata['last_name'] = $parts[1]; endif; |
| 2230 |
$userdata['display_name'] = $author; |
| 2231 |
$userdata['role'] = 'contributor'; |
| 2232 |
|
| 2233 |
do { // Keep trying until you get it right. Or until PHP crashes, I guess. |
| 2234 |
$id = wp_insert_user($userdata); |
| 2235 |
if (is_wp_error($id)) : |
| 2236 |
$codes = $id->get_error_code(); |
| 2237 |
switch ($codes) : |
| 2238 |
case 'empty_user_login' : |
| 2239 |
case 'existing_user_login' : |
| 2240 |
// Add a random disambiguator |
| 2241 |
$userdata['user_login'] .= substr(md5(uniqid(microtime())), 0, 6); |
| 2242 |
break; |
| 2243 |
case 'user_nicename_too_long' : |
| 2244 |
// Add a limited 50 caracters user_nicename based on user_login |
| 2245 |
$userdata['user_nicename'] = mb_substr( $userdata['user_login'], 0, 50 ); |
| 2246 |
break; |
| 2247 |
case 'existing_user_email' : |
| 2248 |
// No disassemble! |
| 2249 |
$parts = explode('@', $userdata['user_email'], 2); |
| 2250 |
|
| 2251 |
// Add a random disambiguator as a gmail-style username extension |
| 2252 |
$parts[0] .= '+'.substr(md5(uniqid(microtime())), 0, 6); |
| 2253 |
|
| 2254 |
// Reassemble |
| 2255 |
$userdata['user_email'] = $parts[0].'@'.$parts[1]; |
| 2256 |
break; |
| 2257 |
endswitch; |
| 2258 |
endif; |
| 2259 |
} while (is_wp_error($id)); |
| 2260 |
elseif (is_numeric($unfamiliar_author) and get_userdata((int) $unfamiliar_author)) : |
| 2261 |
$id = (int) $unfamiliar_author; |
| 2262 |
elseif ($unfamiliar_author === 'default') : |
| 2263 |
$id = 1; |
| 2264 |
endif; |
| 2265 |
endif; |
| 2266 |
endif; |
| 2267 |
|
| 2268 |
if ($id) : |
| 2269 |
$this->link->settings['map authors']['name'][strtolower(trim($author))] = $id; |
| 2270 |
|
| 2271 |
// Multisite: Check whether the author has been recorded |
| 2272 |
// on *this* blog before. If not, put her down as a |
| 2273 |
// Contributor for *this* blog. |
| 2274 |
$user = new WP_User((int) $id); |
| 2275 |
if (empty($user->roles)) : |
| 2276 |
$user->add_role('contributor'); |
| 2277 |
endif; |
| 2278 |
endif; |
| 2279 |
return $id; |
| 2280 |
} /* function SyndicatedPost::author_id () */ |
| 2281 |
|
| 2282 |
/** |
| 2283 |
* category_ids: look up (and create) category ids from a list of |
| 2284 |
* categories |
| 2285 |
* |
| 2286 |
* @param array $cats |
| 2287 |
* @param string $unfamiliar_category |
| 2288 |
* @param array|null $taxonomies |
| 2289 |
* @return array |
| 2290 |
*/ |
| 2291 |
function category_ids ($cats, $unfamiliar_category = 'create', $taxonomies = NULL, $params = array()) { |
| 2292 |
return $this->link->category_ids($this, $cats, $unfamiliar_category, $taxonomies, $params); |
| 2293 |
} /* SyndicatedPost::category_ids () */ |
| 2294 |
|
| 2295 |
} /* class SyndicatedPost */ |
| 2296 |
|
