← All changes | syndicatedpost.class.php +1602 -860 2010.0528 → 2022.0203 View file →

		@@ -1,6 +1,8 @@
1	1	<?php
2	2	require_once(dirname(__FILE__).'/feedtime.class.php');
	3	+require_once(dirname(__FILE__).'/syndicatedpostterm.class.php');
	4	+require_once(dirname(__FILE__).'/syndicatedpostxpathquery.class.php');
3	5
4	6	/**
5	7	* class SyndicatedPost: FeedWordPress uses to manage the conversion of
6	8	* incoming items from the feed parser into posts for the WordPress
		@@ -9,9 +11,9 @@
9	11	* as some utility methods for extracting useful data from many
10	12	* different feed formats, which may be useful to FeedWordPress users
11	13	* who make use of feed data in PHP add-ons and filters.
12	14	*
13		- * @version 2010.0528
	15	+ * @version 2017.1018
14	16	*/
15	17	class SyndicatedPost {
16	18	var $item = null; // MagpieRSS representation
17	19	var $entry = null; // SimplePie_Item representation
		@@ -18,15 +20,20 @@
18	20
19	21	var $link = null;
20	22	var $feed = null;
21	23	var $feedmeta = null;
22		-
	24	+
23	25	var $xmlns = array ();
24	26
25	27	var $post = array ();
26	28
	29	+ var $named = array ();
	30	+ var $preset_terms = array ();
	31	+ var $feed_terms = array ();
	32	+
27	33	var $_freshness = null;
28	34	var $_wp_id = null;
	35	+ var $_wp_post = null;
29	36
30	37	/**
31	38	* SyndicatedPost constructor: Given a feed item and the source from
32	39	* which it was taken, prepare a post that can be inserted into the
		@@ -35,11 +42,14 @@
35	42	*
36	43	* @param array $item The item syndicated from the feed.
37	44	* @param SyndicatedLink $source The feed it was syndicated from.
38	45	*/
39		- function SyndicatedPost ($item, $source) {
	46	+ public function __construct ($item, $source) {
40	47	global $wpdb;
41	48
	49	+ if ( empty($item) and empty($source) )
	50	+ return;
	51	+
42	52	if (is_array($item)
43	53	and isset($item['simplepie'])
44	54	and isset($item['magpie'])) :
45	55	$this->entry = $item['simplepie'];
		@@ -44,8 +54,17 @@
44	54	and isset($item['magpie'])) :
45	55	$this->entry = $item['simplepie'];
46	56	$this->item = $item['magpie'];
47	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);
48	67	else :
49	68	$this->item = $item;
50	69	endif;
51	70
		@@ -52,8 +71,10 @@
52	71	$this->link = $source;
53	72	$this->feed = $source->magpie;
54	73	$this->feedmeta = $source->settings;
55	74
	75	+ FeedWordPress::diagnostic('feed_items', 'Considering item ['.$this->guid().'] "'.$this->entry->get_title().'"');
	76	+
56	77	# Dealing with namespaces can get so fucking fucked.
57	78	$this->xmlns['forward'] = $source->magpie->_XMLNS_FAMILIAR;
58	79	$this->xmlns['reverse'] = array();
59	80	foreach ($this->xmlns['forward'] as $url => $ns) :
		@@ -59,44 +80,32 @@
59	80	foreach ($this->xmlns['forward'] as $url => $ns) :
60	81	if (!isset($this->xmlns['reverse'][$ns])) :
61	82	$this->xmlns['reverse'][$ns] = array();
62	83	endif;
63		- $this->xmlns['reverse'][$ns][] = $url;
	84	+ $this->xmlns['reverse'][$ns][] = $url;
64	85	endforeach;
65		-
	86	+
66	87	// Fucking SimplePie.
67	88	$this->xmlns['reverse']['rss'][] = '';
68	89
69		- # These globals were originally an ugly kludge around a bug in
70		- # apply_filters from WordPress 1.5. The bug was fixed in 1.5.1,
71		- # and I sure hope at this point that nobody writing filters for
72		- # FeedWordPress is still relying on them.
73		- #
74		- # Anyway, I hereby declare them DEPRECATED as of 8 February
75		- # 2010. I'll probably remove the globals within 1-2 releases in
76		- # the interests of code hygiene and memory usage. If you
77		- # currently use them in your filters, I advise you switch off to
78		- # accessing the public members SyndicatedPost::feed and
79		- # SyndicatedPost::feedmeta.
	90	+ // Trigger global syndicated_item filter.
	91	+ $changed = apply_filters('syndicated_item', $this->item, $this);
	92	+ $this->item = $changed;
80	93
81		- global $fwp_channel, $fwp_feedmeta;
82		- $fwp_channel = $this->feed; $fwp_feedmeta = $this->feedmeta;
83		-
84		- // Trigger global syndicated_item filter.
85		- $this->item = apply_filters('syndicated_item', $this->item, $this);
86		-
87	94	// Allow for feed-specific syndicated_item filters.
88		- $this->item = apply_filters(
	95	+ $changed = apply_filters(
89	96	"syndicated_item_".$source->uri(),
90	97	$this->item,
91	98	$this
92	99	);
	100	+ $this->item = $changed;
93	101
94	102	# Filters can halt further processing by returning NULL
95	103	if (is_null($this->item)) :
96	104	$this->post = NULL;
97	105	else :
98		- # Note that nothing is run through $wpdb->escape() here.
	106	+
	107	+ # Note that nothing is run through esc_sql() here.
99	108	# That's deliberate. The escaping is done at the point
100	109	# of insertion, not here, to avoid double-escaping and
101	110	# to avoid screwing with syndicated_post filters
102	111
		@@ -104,9 +113,10 @@
104	113	'syndicated_item_title',
105	114	$this->entry->get_title(), $this
106	115	);
107	116
108		- $this->post['named']['author'] = apply_filters(
	117	+
	118	+ $this->named['author'] = apply_filters(
109	119	'syndicated_item_author',
110	120	$this->author(), $this
111	121	);
112	122	// This just gives us an alphanumeric name for the author.
		@@ -118,23 +128,23 @@
118	128	$this->content(), $this
119	129	);
120	130
121	131	$excerpt = apply_filters('syndicated_item_excerpt', $this->excerpt(), $this);
122		- if (!is_null($excerpt)):
	132	+
	133	+ if (!empty($excerpt)):
123	134	$this->post['post_excerpt'] = $excerpt;
124	135	endif;
125		-
126		- $this->post['epoch']['issued'] = apply_filters('syndicated_item_published', $this->published(), $this);
127		- $this->post['epoch']['created'] = apply_filters('syndicated_item_created', $this->created(), $this);
128		- $this->post['epoch']['modified'] = apply_filters('syndicated_item_updated', $this->updated(), $this);
129	136
130	137	// Dealing with timestamps in WordPress is so fucking fucked.
131	138	$offset = (int) get_option('gmt_offset') * 60 * 60;
132		- $this->post['post_date'] = gmdate('Y-m-d H:i:s', $this->published(/*fallback=*/ true, /*default=*/ -1) + $offset);
133		- $this->post['post_modified'] = gmdate('Y-m-d H:i:s', $this->updated(/*fallback=*/ true, /*default=*/ -1) + $offset);
134		- $this->post['post_date_gmt'] = gmdate('Y-m-d H:i:s', $this->published(/*fallback=*/ true, /*default=*/ -1));
135		- $this->post['post_modified_gmt'] = gmdate('Y-m-d H:i:s', $this->updated(/*fallback=*/ true, /*default=*/ -1));
	139	+ $post_date_gmt = $this->published(array('default' => -1));
	140	+ $post_modified_gmt = $this->updated(array('default' => -1));
136	141
	142	+ $this->post['post_date_gmt'] = gmdate('Y-m-d H:i:s', $post_date_gmt);
	143	+ $this->post['post_date'] = gmdate('Y-m-d H:i:s', $post_date_gmt + $offset);
	144	+ $this->post['post_modified_gmt'] = gmdate('Y-m-d H:i:s', $post_modified_gmt);
	145	+ $this->post['post_modified'] = gmdate('Y-m-d H:i:s', $post_modified_gmt + $offset);
	146	+
137	147	// Use feed-level preferences or the global default.
138	148	$this->post['post_status'] = $this->link->syndicated_status('post', 'publish');
139	149	$this->post['comment_status'] = $this->link->syndicated_status('comment', 'closed');
140	150	$this->post['ping_status'] = $this->link->syndicated_status('ping', 'closed');
		@@ -141,57 +151,28 @@
141	151
142	152	// Unique ID (hopefully a unique tag: URI); failing that, the permalink
143	153	$this->post['guid'] = apply_filters('syndicated_item_guid', $this->guid(), $this);
144	154
145		- // User-supplied custom settings to apply to each post. Do first so that FWP-generated custom settings will overwrite if necessary; thus preventing any munging
146		- $default_custom_settings = get_option('feedwordpress_custom_settings');
147		- if ($default_custom_settings and !is_array($default_custom_settings)) :
148		- $default_custom_settings = unserialize($default_custom_settings);
149		- endif;
150		- if (!is_array($default_custom_settings)) :
151		- $default_custom_settings = array();
152		- endif;
153		-
154		- $custom_settings = (isset($this->link->settings['postmeta']) ? $this->link->settings['postmeta'] : null);
155		- if ($custom_settings and !is_array($custom_settings)) :
156		- $custom_settings = unserialize($custom_settings);
157		- endif;
158		- if (!is_array($custom_settings)) :
159		- $custom_settings = array();
160		- endif;
161		-
162		- $postMetaIn = array_merge($default_custom_settings, $custom_settings);
	155	+ // User-supplied custom settings to apply to each post.
	156	+ // Do first so that FWP-generated custom settings will
	157	+ // overwrite if necessary; thus preventing any munging.
	158	+ $postMetaIn = $this->link->postmeta(array("parsed" => true));
163	159	$postMetaOut = array();
164	160
165		- // Big ugly fuckin loop to do any element substitutions
166		- // that we may need.
167		- foreach ($postMetaIn as $key => $values) :
168		- if (is_string($values)) : $values = array($values); endif;
169		-
170		- $postMetaOut[$key] = array();
171		- foreach ($values as $value) :
172		- if (preg_match('/\$\( ([^)]+) \)/x', $value, $ref)) :
173		- $elements = $this->query($ref[1]);
174		- foreach ($elements as $element) :
175		- $postMetaOut[$key][] = str_replace(
176		- $ref[0],
177		- $element,
178		- $value
179		- );
180		- endforeach;
181		- else :
182		- $postMetaOut[$key][] = $value;
183		- endif;
184		- endforeach;
	161	+ foreach ($postMetaIn as $key => $meta) :
	162	+ $postMetaOut[$key] = $meta->do_substitutions($this);
185	163	endforeach;
186	164
187	165	foreach ($postMetaOut as $key => $values) :
	166	+ if (is_null($values)) { // have chosen to replace value with empty string
	167	+ $values = [''];
	168	+ }
188	169	$this->post['meta'][$key] = array();
189	170	foreach ($values as $value) :
190	171	$this->post['meta'][$key][] = apply_filters("syndicated_post_meta_{$key}", $value, $this);
191	172	endforeach;
192	173	endforeach;
193		-
	174	+
194	175	// RSS 2.0 / Atom 1.0 enclosure support
195	176	$enclosures = $this->entry->get_enclosures();
196	177	if (is_array($enclosures)) : foreach ($enclosures as $enclosure) :
197	178	$this->post['meta']['enclosure'][] =
		@@ -199,32 +180,54 @@
199	180	apply_filters('syndicated_item_enclosure_length', $enclosure->get_length(), $this)."\n".
200	181	apply_filters('syndicated_item_enclosure_type', $enclosure->get_type(), $this);
201	182	endforeach; endif;
202	183
203		- // In case you want to point back to the blog this was syndicated from
204		- if (isset($this->feed->channel['title'])) :
205		- $this->post['meta']['syndication_source'] = apply_filters('syndicated_item_source_title', $this->feed->channel['title'], $this);
206		- endif;
	184	+ // In case you want to point back to the blog this was
	185	+ // syndicated from.
207	186
208		- if (isset($this->feed->channel['link'])) :
209		- $this->post['meta']['syndication_source_uri'] = apply_filters('syndicated_item_source_link', $this->feed->channel['link'], $this);
210		- endif;
211		-
	187	+ $sourcemeta['syndication_source'] = apply_filters(
	188	+ 'syndicated_item_source_title',
	189	+ $this->link->name(),
	190	+ $this
	191	+ );
	192	+ $sourcemeta['syndication_source_uri'] = apply_filters(
	193	+ 'syndicated_item_source_link',
	194	+ $this->link->homepage(),
	195	+ $this
	196	+ );
	197	+ $sourcemeta['syndication_source_id'] = apply_filters(
	198	+ 'syndicated_item_source_id',
	199	+ $this->link->guid(),
	200	+ $this
	201	+ );
	202	+
212	203	// Make use of atom:source data, if present in an aggregated feed
213		- if (isset($this->item['source_title'])) :
214		- $this->post['meta']['syndication_source_original'] = $this->item['source_title'];
	204	+ $entry_source = $this->source();
	205	+ if (!is_null($entry_source)) :
	206	+ foreach ($entry_source as $what => $value) :
	207	+ if (!is_null($value)) :
	208	+ if ($what=='title') : $key = 'syndication_source';
	209	+ elseif ($what=='feed') : $key = 'syndication_feed';
	210	+ else : $key = "syndication_source_${what}";
	211	+ endif;
	212	+
	213	+ $sourcemeta["${key}_original"] = apply_filters(
	214	+ 'syndicated_item_original_source_'.$what,
	215	+ $value,
	216	+ $this
	217	+ );
	218	+ endif;
	219	+ endforeach;
215	220	endif;
216	221
217		- if (isset($this->item['source_link'])) :
218		- $this->post['meta']['syndication_source_uri_original'] = $this->item['source_link'];
219		- endif;
220		-
221		- if (isset($this->item['source_id'])) :
222		- $this->post['meta']['syndication_source_id_original'] = $this->item['source_id'];
223		- endif;
	222	+ foreach ($sourcemeta as $meta_key => $value) :
	223	+ if (!is_null($value)) :
	224	+ $this->post['meta'][$meta_key] = $value;
	225	+ endif;
	226	+ endforeach;
224	227
225	228	// Store information on human-readable and machine-readable comment URIs
226		-
	229	+
227	230	// Human-readable comment URI
228	231	$commentLink = apply_filters('syndicated_item_comments', $this->comment_link(), $this);
229	232	if (!is_null($commentLink)) : $this->post['meta']['rss:comments'] = $commentLink; endif;
230	233
		@@ -251,60 +254,38 @@
251	254
252	255	// Store a hash of the post content for checking whether something needs to be updated
253	256	$this->post['meta']['syndication_item_hash'] = $this->update_hash();
254	257
255		- // Feed-by-feed options for author and category creation
256		- $this->post['named']['unfamiliar']['author'] = (isset($this->feedmeta['unfamiliar author']) ? $this->feedmeta['unfamiliar author'] : null);
257		- $this->post['named']['unfamiliar']['category'] = (isset($this->feedmeta['unfamiliar category']) ? $this->feedmeta['unfamiliar category'] : null);
	258	+ // Categories, Tags, and other Terms: from settings assignments (global settings, subscription settings),
	259	+ // and from feed assignments (item metadata, post content)
	260	+ $this->preset_terms = apply_filters('syndicated_item_preset_terms', $this->get_terms_from_settings(), $this);
	261	+ $this->feed_terms = apply_filters('syndicated_item_feed_terms', $this->get_terms_from_feeds(), $this);
258	262
259		- // Categories: start with default categories, if any
260		- $fc = get_option("feedwordpress_syndication_cats");
261		- if ($fc) :
262		- $this->post['named']['preset/category'] = explode("\n", $fc);
263		- else :
264		- $this->post['named']['preset/category'] = array();
265		- endif;
266		-
267		- if (isset($this->feedmeta['cats']) and is_array($this->feedmeta['cats'])) :
268		- $this->post['named']['preset/category'] = array_merge($this->post['named']['preset/category'], $this->feedmeta['cats']);
269		- endif;
270		-
271		- // Now add categories from the post, if we have 'em
272		- $this->post['named']['category'] = array();
273		- if ( isset($this->item['category#']) ) :
274		- for ($i = 1; $i <= $this->item['category#']; $i++) :
275		- $cat_idx = (($i > 1) ? "#{$i}" : "");
276		- $cat = $this->item["category{$cat_idx}"];
277		-
278		- if ( isset($this->feedmeta['cat_split']) and strlen($this->feedmeta['cat_split']) > 0) :
279		- $pcre = "\007".$this->feedmeta['cat_split']."\007";
280		- $this->post['named']['category'] = array_merge($this->post['named']['category'], preg_split($pcre, $cat, -1 /*=no limit*/, PREG_SPLIT_NO_EMPTY));
281		- else :
282		- $this->post['named']['category'][] = $cat;
283		- endif;
284		- endfor;
285		- endif;
286		- $this->post['named']['category'] = apply_filters('syndicated_item_categories', $this->post['named']['category'], $this);
287		-
288		- // Tags: start with default tags, if any
289		- $ft = get_option("feedwordpress_syndication_tags");
290		- if ($ft) :
291		- $this->post['tags_input'] = explode(FEEDWORDPRESS_CAT_SEPARATOR, $ft);
292		- else :
293		- $this->post['tags_input'] = array();
294		- endif;
295		-
296		- if (isset($this->feedmeta['tags']) and is_array($this->feedmeta['tags'])) :
297		- $this->post['tags_input'] = array_merge($this->post['tags_input'], $this->feedmeta['tags']);
298		- endif;
299		- $this->post['tags_input'] = apply_filters('syndicated_item_tags', $this->post['tags_input'], $this);
	263	+ $this->post['post_type'] = apply_filters('syndicated_post_type', $this->link->setting('syndicated post type', 'syndicated_post_type', 'post'), $this);
300	264	endif;
301		- } /* SyndicatedPost::SyndicatedPost() */
	265	+ } /* SyndicatedPost::__construct() */
302	266
303	267	#####################################
304	268	#### EXTRACT DATA FROM FEED ITEM ####
305	269	#####################################
306		-
	270	+
	271	+ function substitution_function ($name) {
	272	+ $ret = NULL;
	273	+
	274	+ switch ($name) :
	275	+ // Allowed PHP string functions
	276	+ case 'trim':
	277	+ case 'ltrim':
	278	+ case 'rtrim':
	279	+ case 'strtoupper':
	280	+ case 'strtolower':
	281	+ case 'urlencode':
	282	+ case 'urldecode':
	283	+ $ret = $name;
	284	+ endswitch;
	285	+ return $ret;
	286	+ }
	287	+
307	288	/**
308	289	* SyndicatedPost::query uses an XPath-like syntax to query arbitrary
309	290	* elements within the syndicated item.
310	291	*
		@@ -311,211 +292,168 @@
311	292	* @param string $path
312	293	* @returns array of string values representing contents of matching
313	294	* elements or attributes
314	295	*/
315		- function query ($path) {
316		- $urlHash = array();
	296	+ public function query ($path) {
	297	+ $xq = new SyndicatedPostXPathQuery(array("path" => $path));
317	298
318		- // Allow {url} notation for namespaces. URLs will contain : and /, so...
319		- preg_match_all('/{([^}]+)}/', $path, $match, PREG_SET_ORDER);
320		- foreach ($match as $ref) :
321		- $urlHash[md5($ref[1])] = $ref[1];
322		- endforeach;
323		-
324		- foreach ($urlHash as $hash => $url) :
325		- $path = str_replace('{'.$url.'}', '{#'.$hash.'}', $path);
326		- endforeach;
	299	+ $feedChannel = array_merge(
	300	+ $this->get_feed_root_element(),
	301	+ $this->get_feed_channel_elements()
	302	+ );
327	303
328		- $path = explode('/', $path);
329		- foreach ($path as $index => $node) :
330		- if (preg_match('/{#([^}]+)}/', $node, $ref)) :
331		- if (isset($urlHash[$ref[1]])) :
332		- $path[$index] = str_replace(
333		- '{#'.$ref[1].'}',
334		- '{'.$urlHash[$ref[1]].'}',
335		- $node
336		- );
337		- endif;
338		- endif;
339		- endforeach;
	304	+ $matches = $xq->match(array(
	305	+ "type" => $this->link->simplepie->get_type(),
	306	+ "xmlns" => $this->xmlns,
	307	+ "map" => array(
	308	+ "/" => array($this->entry->data),
	309	+ "item" => array($this->entry->data),
	310	+ "feed" => $feedChannel,
	311	+ "channel" => $feedChannel
	312	+ ),
	313	+ "context" => $this->entry->data,
	314	+ "parent" => $feedChannel,
	315	+ ));
340	316
341		- // Start out with a get_item_tags query.
342		- $node = '';
343		- while (strlen($node)==0 and !is_null($node)) :
344		- $node = array_shift($path);
345		- endwhile;
346		-
347		- switch ($node) :
348		- case 'feed' :
349		- case 'channel' :
350		- $method = "get_${node}_tags";
351		- $node = array_shift($path);
352		- break;
353		- case 'item' :
354		- $node = array_shift($path);
355		- default :
356		- $method = NULL;
357		- endswitch;
	317	+ return $matches;
	318	+ } /* SyndicatedPost::query() */
358	319
359		- $data = array();
360		- if (!is_null($node)) :
361		- list($namespaces, $element) = $this->xpath_extended_name($node);
362		-
363		- $matches = array();
364		- foreach ($namespaces as $ns) :
365		- if (!is_null($method)) :
366		- $el = $this->link->simplepie->{$method}($ns, $element);
367		- else :
368		- $el = $this->entry->get_item_tags($ns, $element);
369		- endif;
370		-
371		- if (!is_null($el)) :
372		- $matches = array_merge($matches, $el);
373		- endif;
	320	+ function get_feed_root_element () {
	321	+ $matches = array();
	322	+ foreach ($this->link->simplepie->data['child'] as $ns => $root) :
	323	+ foreach ($root as $element => $data) :
	324	+ $matches = array_merge($matches, $data);
374	325	endforeach;
375		- $data = $matches;
376		-
377		- $node = array_shift($path);
378		- endif;
	326	+ endforeach;
	327	+ return $matches;
	328	+ } /* SyndicatedPost::get_feed_root_element() */
379	329
380		- while (!is_null($node)) :
381		- if (strlen($node) > 0) :
382		- $matches = array();
383		-
384		- list($ns, $element) = $this->xpath_extended_name($node);
385		-
386		- if (preg_match('/^@(.*)$/', $element, $ref)) :
387		- $element = $ref[1];
388		- $axis = 'attribs';
389		- else :
390		- $axis = 'child';
391		- endif;
	330	+ function get_feed_channel_elements () {
	331	+ $rss = array(
	332	+ SIMPLEPIE_NAMESPACE_RSS_090,
	333	+ SIMPLEPIE_NAMESPACE_RSS_10,
	334	+ 'http://backend.userland.com/RSS2',
	335	+ SIMPLEPIE_NAMESPACE_RSS_20,
	336	+ );
392	337
393		- foreach ($data as $datum) :
394		- foreach ($namespaces as $ns) :
395		- if (!is_string($datum)
396		- and isset($datum[$axis][$ns][$element])) :
397		- if (is_string($datum[$axis][$ns][$element])) :
398		- $matches[] = $datum[$axis][$ns][$element];
399		- else :
400		- $matches = array_merge($matches, $datum[$axis][$ns][$element]);
401		- endif;
402		- endif;
403		- endforeach;
404		- endforeach;
405		-
406		- $data = $matches;
407		- endif;
408		- $node = array_shift($path);
409		- endwhile;
410		-
411	338	$matches = array();
412		- foreach ($data as $datum) :
413		- if (is_string($datum)) :
414		- $matches[] = $datum;
415		- elseif (isset($datum['data'])) :
416		- $matches[] = $datum['data'];
	339	+ foreach ($rss as $ns) :
	340	+ $data = $this->link->simplepie->get_feed_tags($ns, 'channel');
	341	+ if (!is_null($data)) :
	342	+ $matches = array_merge($matches, $data);
417	343	endif;
418	344	endforeach;
419	345	return $matches;
420		- } /* SyndicatedPost::query() */
	346	+ } /* SyndicatedPost::get_feed_channel_elements() */
421	347
422		- function xpath_default_namespace () {
423		- // Get the default namespace.
424		- $type = $this->link->simplepie->get_type();
425		- if ($type & SIMPLEPIE_TYPE_ATOM_10) :
426		- $defaultNS = SIMPLEPIE_NAMESPACE_ATOM_10;
427		- elseif ($type & SIMPLEPIE_TYPE_ATOM_03) :
428		- $defaultNS = SIMPLEPIE_NAMESPACE_ATOM_03;
429		- elseif ($type & SIMPLEPIE_TYPE_RSS_090) :
430		- $defaultNS = SIMPLEPIE_NAMESPACE_RSS_090;
431		- elseif ($type & SIMPLEPIE_TYPE_RSS_10) :
432		- $defaultNS = SIMPLEPIE_NAMESPACE_RSS_10;
433		- elseif ($type & SIMPLEPIE_TYPE_RSS_20) :
434		- $defaultNS = SIMPLEPIE_NAMESPACE_RSS_20;
435		- else :
436		- $defaultNS = SIMPLEPIE_NAMESPACE_RSS_20;
437		- endif;
438		- return $defaultNS;
439		- } /* SyndicatedPost::xpath_default_namespace() */
	348	+ public function get_categories ($params = array()) {
	349	+ return $this->entry->get_categories();
	350	+ }
440	351
441		- function xpath_extended_name ($node) {
442		- $ns = NULL; $element = NULL;
	352	+ public function title ($params = array()) {
	353	+ return $this->entry->get_title();
	354	+ } /* SyndicatedPost::title () */
	355	+
	356	+ public function content ($params = array()) {
	357	+
	358	+ $params = wp_parse_args($params, array(
	359	+ "full only" => false,
	360	+ ));
443	361
444		- if (substr($node, 0, 1)=='@') :
445		- $attr = '@'; $node = substr($node, 1);
446		- else :
447		- $attr = '';
448		- endif;
449		-
450		- if (preg_match('/^{([^}]*)}(.*)$/', $node, $ref)) :
451		- $ns = array($ref[1]); $element = $ref[2];
452		- elseif (strpos($node, ':') !== FALSE) :
453		- list($xmlns, $element) = explode(':', $node, 2);
454		- if (isset($this->xmlns['reverse'][$xmlns])) :
455		- $ns = $this->xmlns['reverse'][$xmlns];
456		- else :
457		- $ns = array($xmlns);
458		- endif;
459		-
460		- // Fucking SimplePie. For attributes in default xmlns.
461		- if ($xmlns==$this->xmlns['forward'][$defaultNS[0]]) :
462		- $ns[] = '';
463		- endif;
464		- else :
465		- // Often in SimplePie, the default namespace gets stored
466		- // as an empty string rather than a URL.
467		- $ns = array($this->xpath_default_namespace(), '');
468		- $element = $node;
469		- endif;
470		- return array(array_unique($ns), $attr.$element);
471		- } /* SyndicatedPost::xpath_extended_name () */
472		-
473		- function content () {
474	362	$content = NULL;
	363	+
	364	+ // FIXME: This is one of the main places in the code still using
	365	+ // the outmoded SimplePie - to - Magpie construction. We could
	366	+ // replace using SimplePie_Item::get_tags() here. (Or if really
	367	+ // ambitious we could attempt to just use
	368	+ // SimplePie_Item::get_content() with content-only set to TRUE
	369	+ // and some sanitization in effect. -CJ 1jul14
	370	+
	371	+ // atom:content, standard method of providing full HTML content
	372	+ // in Atom feeds.
475	373	if (isset($this->item['atom_content'])) :
476	374	$content = $this->item['atom_content'];
	375	+ elseif (isset($this->item['atom']['atom_content'])) :
	376	+ $content = $this->item['atom']['atom_content'];
	377	+
	378	+ // Some exotics: back in the day, before widespread convergence
	379	+ // on content:encoding, some RSS feeds took advantage of XML
	380	+ // namespaces to use an inline xhtml:body or xhtml:div element
	381	+ // for full-content HTML. (E.g. Sam Ruby's feed, IIRC.)
477	382	elseif (isset($this->item['xhtml']['body'])) :
478	383	$content = $this->item['xhtml']['body'];
479	384	elseif (isset($this->item['xhtml']['div'])) :
480	385	$content = $this->item['xhtml']['div'];
	386	+
	387	+ // content:encoded, most common method of providing full HTML in
	388	+ // RSS 2.0 feeds.
481	389	elseif (isset($this->item['content']['encoded']) and $this->item['content']['encoded']):
482	390	$content = $this->item['content']['encoded'];
483		- else:
484		- $content = $this->item['description'];
	391	+
	392	+ // Fall back on elements that sometimes may contain full HTML
	393	+ // but sometimes not.
	394	+ elseif (!$params['full only']) :
	395	+
	396	+ // description element is sometimes used for full HTML
	397	+ // sometimes for summary text in RSS. (By the letter of
	398	+ // the standard, it should just be for summary text.)
	399	+ if (isset($this->item['description'])) :
	400	+ $content = $this->item['description'];
	401	+ endif;
	402	+
485	403	endif;
	404	+
486	405	return $content;
487	406	} /* SyndicatedPost::content() */
488	407
489		- function excerpt () {
	408	+ public function excerpt () {
490	409	# Identify and sanitize excerpt: atom:summary, or rss:description
491	410	$excerpt = $this->entry->get_description();
492		-
	411	+
493	412	# Many RSS feeds use rss:description, inadvisably, to
494	413	# carry the entire post (typically with escaped HTML).
495	414	# If that's what happened, we don't want the full
496	415	# content for the excerpt.
497	416	$content = $this->content();
498		- if ( is_null($excerpt) or $excerpt == $content ) :
	417	+
	418	+ // Ignore whitespace, case, and tag cruft.
	419	+ $theExcerpt = preg_replace('/\s+/', '', strtolower(strip_tags(html_entity_decode($excerpt))));
	420	+ $theContent = preg_replace('/\s+/', '', strtolower(strip_tags(html_entity_decode($content))));
	421	+ if ( empty($excerpt) or $theExcerpt == $theContent ) :
499	422	# If content is available, generate an excerpt.
500	423	if ( strlen(trim($content)) > 0 ) :
501	424	$excerpt = strip_tags($content);
502	425	if (strlen($excerpt) > 255) :
503		- $excerpt = substr($excerpt,0,252).'...';
	426	+ if (is_object($this->link) and is_object($this->link->simplepie)) :
	427	+ $encoding = $this->link->simplepie->get_encoding();
	428	+ else :
	429	+ $encoding = get_option('blog_charset', 'utf8');
	430	+ endif;
	431	+ $excerpt = mb_substr($excerpt,0,252,$encoding).'...';
504	432	endif;
505	433	endif;
506	434	endif;
	435	+
507	436	return $excerpt;
508	437	} /* SyndicatedPost::excerpt() */
509	438
510		- function permalink () {
	439	+ /**
	440	+ * SyndicatedPost::permalink: returns the permalink for the post, as provided by the
	441	+ * source feed.
	442	+ *
	443	+ * @return string The URL of the original permalink for this syndicated post
	444	+ */
	445	+ public function permalink () {
511	446	// Handles explicit <link> elements and also RSS 2.0 cases with
512	447	// <guid isPermaLink="true">, etc. Hooray!
513	448	$permalink = $this->entry->get_link();
514	449	return $permalink;
515		- }
	450	+ } /* SyndicatedPost::permalink () */
516	451
517		- function created () {
	452	+ public function created ($params = array()) {
	453	+ $unfiltered = false; $default = NULL;
	454	+ extract($params);
	455	+
518	456	$date = '';
519	457	if (isset($this->item['dc']['created'])) :
520	458	$date = $this->item['dc']['created'];
521	459	elseif (isset($this->item['dcterms']['created'])) :
		@@ -523,14 +461,26 @@
523	461	elseif (isset($this->item['created'])): // Atom 0.3
524	462	$date = $this->item['created'];
525	463	endif;
526	464
527		- $epoch = new FeedTime($date);
528		- return $epoch->timestamp();
	465	+ $time = new FeedTime($date);
	466	+ $ts = $time->timestamp();
	467	+ if (!$unfiltered) :
	468	+ apply_filters('syndicated_item_created', $ts, $this);
	469	+ endif;
	470	+ return $ts;
529	471	} /* SyndicatedPost::created() */
530	472
531		- function published ($fallback = true, $default = NULL) {
	473	+ public function published ($params = array(), $default = NULL) {
	474	+ $fallback = true; $unfiltered = false;
	475	+ if (!is_array($params)) : // Old style
	476	+ $fallback = $params;
	477	+ else : // New style
	478	+ extract($params);
	479	+ endif;
	480	+
532	481	$date = '';
	482	+ $ts = null;
533	483
534	484	# RSS is a fucking mess. Figure out whether we have a date in
535	485	# <dc:date>, <issued>, <pubDate>, etc., and get it into Unix
536	486	# epoch format for reformatting. If we can't find anything,
		@@ -545,30 +495,41 @@
545	495	$date = $this->item['issued'];
546	496	elseif (isset($this->item['pubdate'])): // RSS 2.0
547	497	$date = $this->item['pubdate'];
548	498	endif;
549		-
	499	+
550	500	if (strlen($date) > 0) :
551	501	$time = new FeedTime($date);
552		- $epoch = $time->timestamp();
	502	+ $ts = $time->timestamp();
553	503	elseif ($fallback) : // Fall back to <updated> / <modified> if present
554		- $epoch = $this->updated(/*fallback=*/ false, /*default=*/ $default);
	504	+ $ts = $this->updated(/*fallback=*/ false, /*default=*/ $default);
555	505	endif;
556		-
	506	+
557	507	# If everything failed, then default to the current time.
558		- if (is_null($epoch)) :
	508	+ if (is_null($ts)) :
559	509	if (-1 == $default) :
560		- $epoch = time();
	510	+ $ts = time();
561	511	else :
562		- $epoch = $default;
	512	+ $ts = $default;
563	513	endif;
564	514	endif;
565		-
566		- return $epoch;
	515	+
	516	+ if (!$unfiltered) :
	517	+ $ts = apply_filters('syndicated_item_published', $ts, $this);
	518	+ endif;
	519	+ return $ts;
567	520	} /* SyndicatedPost::published() */
568	521
569		- function updated ($fallback = true, $default = -1) {
	522	+ public function updated ($params = array(), $default = -1) {
	523	+ $fallback = true; $unfiltered = false;
	524	+ if (!is_array($params)) : // Old style
	525	+ $fallback = $params;
	526	+ else : // New style
	527	+ extract($params);
	528	+ endif;
	529	+
570	530	$date = '';
	531	+ $ts = null;
571	532
572	533	# As far as I know, only dcterms and Atom have reliable ways to
573	534	# specify when something was *modified* last. If neither is
574	535	# available, then we'll try to get the time of publication.
		@@ -577,61 +538,157 @@
577	538	elseif (isset($this->item['dcterms']['modified'])) : // Dublin Core extensions
578	539	$date = $this->item['dcterms']['modified'];
579	540	elseif (isset($this->item['modified'])): // Atom 0.3
580	541	$date = $this->item['modified'];
581		- elseif (isset($this->item['updated'])): // Atom 1.0
	542	+ elseif (isset($this->item['updated'])): // Atom 1.0
582	543	$date = $this->item['updated'];
583	544	endif;
584		-
	545	+
585	546	if (strlen($date) > 0) :
586	547	$time = new FeedTime($date);
587		- $epoch = $time->timestamp();
	548	+ $ts = $time->timestamp();
588	549	elseif ($fallback) : // Fall back to issued / dc:date
589		- $epoch = $this->published(/*fallback=*/ false, /*default=*/ $default);
	550	+ $ts = $this->published(/*fallback=*/ false, /*default=*/ $default);
590	551	endif;
591		-
	552	+
592	553	# If everything failed, then default to the current time.
593		- if (is_null($epoch)) :
	554	+ if (is_null($ts)) :
594	555	if (-1 == $default) :
595		- $epoch = time();
	556	+ $ts = time();
596	557	else :
597		- $epoch = $default;
	558	+ $ts = $default;
598	559	endif;
599	560	endif;
600	561
601		- return $epoch;
	562	+ if (!$unfiltered) :
	563	+ $ts = apply_filters('syndicated_item_updated', $ts, $this);
	564	+ endif;
	565	+ return $ts;
602	566	} /* SyndicatedPost::updated() */
603	567
604		- function update_hash () {
605		- return md5(serialize($this->item));
	568	+ var $_hashes = array();
	569	+ function stored_hashes ($id = NULL) {
	570	+ if (is_null($id)) :
	571	+ $id = $this->wp_id();
	572	+ endif;
	573	+
	574	+ if (!isset($this->_hashes[$id])) :
	575	+ $this->_hashes[$id] = get_post_custom_values(
	576	+ 'syndication_item_hash', $id
	577	+ );
	578	+ if (is_null($this->_hashes[$id])) :
	579	+ $this->_hashes[$id] = array();
	580	+ endif;
	581	+ endif;
	582	+ return $this->_hashes[$id];
	583	+ }
	584	+
	585	+ function update_hash ($hashed = true) {
	586	+ // Basis for tracking possible changes to item.
	587	+ $hash = array(
	588	+ "title" => $this->entry->get_title(),
	589	+ "link" => $this->permalink(),
	590	+ "content" => $this->content(),
	591	+ "excerpt" => $this->excerpt(),
	592	+ );
	593	+
	594	+ if ($hashed) :
	595	+ $hash = md5(serialize($hash));
	596	+ endif;
	597	+
	598	+ return $hash;
606	599	} /* SyndicatedPost::update_hash() */
607	600
608		- function guid () {
	601	+ /**
	602	+ * SyndicatedPost::normalize_guid_prefix(): generates a normalized URL
	603	+ * prefix (including scheme, authority, full path, and the beginning of
	604	+ * a query string) for creating guids that conform to WordPress's
	605	+ * internal constraints on the URL space for valid guids. To create a
	606	+ * normalized guid, just concatenate a valid URL query parameter value
	607	+ * to the returned URL.
	608	+ *
	609	+ * @return string The URL prefix generated.
	610	+ *
	611	+ * @uses trailingslashit()
	612	+ * @uses home_url()
	613	+ * @uses apply_filters()
	614	+ */
	615	+ static function normalize_guid_prefix () {
	616	+ $url = trailingslashit(home_url(/*path=*/ '', /*scheme=*/ 'http'));
	617	+ return apply_filters('syndicated_item_guid_normalized_prefix', $url . '?guid=');
	618	+ } /* SyndicatedPost::normalize_guid_prefix() */
	619	+
	620	+ static function normalize_guid ($guid) {
	621	+ $guid = trim($guid);
	622	+ if (preg_match('/^[0-9a-z]{32}$/i', $guid)) : // MD5
	623	+ $guid = SyndicatedPost::normalize_guid_prefix().strtolower($guid);
	624	+ elseif ((strlen(esc_url($guid)) == 0) or (esc_url($guid) != $guid)) :
	625	+ $guid = SyndicatedPost::normalize_guid_prefix().md5($guid);
	626	+ endif;
	627	+ $guid = trim($guid);
	628	+
	629	+ return $guid;
	630	+ } /* SyndicatedPost::normalize_guid() */
	631	+
	632	+ static function alternative_guid_prefix () {
	633	+ $url = trailingslashit(home_url(/*path=*/ '', /*scheme=*/ 'https'));
	634	+ return apply_filters('syndicated_item_guid_normalized_prefix', $url . '?guid=');
	635	+ }
	636	+ static function alternative_guid ($guid) {
	637	+ $guid = trim($guid);
	638	+ if (preg_match('/^[0-9a-z]{32}$/i', $guid)) : // MD5
	639	+ $guid = SyndicatedPost::alternative_guid_prefix().strtolower($guid);
	640	+ elseif ((strlen(esc_url($guid)) == 0) or (esc_url($guid) != $guid)) :
	641	+ $guid = SyndicatedPost::alternative_guid_prefix().md5($guid);
	642	+ endif;
	643	+ $guid = trim($guid);
	644	+
	645	+ return $guid;
	646	+ } /* SyndicatedPost::normalize_guid() */
	647	+
	648	+ public function guid () {
609	649	$guid = null;
610		- if (isset($this->item['id'])): // Atom 0.3 / 1.0
	650	+ if (isset($this->item['id'])): // Atom 0.3 / 1.0
611	651	$guid = $this->item['id'];
612		- elseif (isset($this->item['atom']['id'])) : // Namespaced Atom
	652	+ elseif (isset($this->item['atom']['id'])) : // Namespaced Atom
613	653	$guid = $this->item['atom']['id'];
614		- elseif (isset($this->item['guid'])) : // RSS 2.0
	654	+ elseif (isset($this->item['guid'])) : // RSS 2.0
615	655	$guid = $this->item['guid'];
616		- elseif (isset($this->item['dc']['identifier'])) :// yeah, right
	656	+ elseif (isset($this->item['dc']['identifier'])) : // yeah, right
617	657	$guid = $this->item['dc']['identifier'];
618		- else :
	658	+ endif;
	659	+
	660	+ // Un-set or too long to use as-is. Generate a tag: URI.
	661	+ if (is_null($guid) or strlen($guid) > 250) :
	662	+ // In case we need to check this again
	663	+ $original_guid = $guid;
	664	+
619	665	// The feed does not seem to have provided us with a
620		- // unique identifier, so we'll have to cobble together
621		- // a tag: URI that might work for us. The base of the
622		- // URI will be the host name of the feed source ...
623		- $bits = parse_url($this->feedmeta['link/uri']);
	666	+ // usable unique identifier, so we'll have to cobble
	667	+ // together a tag: URI that might work for us. The base
	668	+ // of the URI will be the host name of the feed source ...
	669	+ $bits = parse_url($this->link->uri());
624	670	$guid = 'tag:'.$bits['host'];
625	671
	672	+ // Some ill-mannered feeds (for example, certain feeds
	673	+ // coming from Google Calendar) have extraordinarily long
	674	+ // guids -- so long that they exceed the 255 character
	675	+ // width of the WordPress guid field. But if the string
	676	+ // gets clipped by MySQL, uniqueness tests will fail
	677	+ // forever after and the post will be endlessly
	678	+ // reduplicated. So, instead, Guids Of A Certain Length
	679	+ // are hashed down into a nice, manageable tag: URI.
	680	+ if (!is_null($original_guid)) :
	681	+ $guid .= ',2010-12-03:id.'.md5($original_guid);
	682	+
626	683	// If we have a date of creation, then we can use that
627	684	// to uniquely identify the item. (On the other hand, if
628	685	// the feed producer was consicentious enough to
629	686	// generate dates of creation, she probably also was
630	687	// conscientious enough to generate unique identifiers.)
631		- if (!is_null($this->created())) :
	688	+ elseif (!is_null($this->created())) :
632	689	$guid .= '://post.'.date('YmdHis', $this->created());
633		-
	690	+
634	691	// Otherwise, use both the URI of the item, *and* the
635	692	// item's title. We have to use both because titles are
636	693	// often not unique, and sometimes links aren't unique
637	694	// either (e.g. Bitch (S)HITLIST, Mozilla Dot Org news,
		@@ -638,58 +695,88 @@
638	695	// some podcasts). But it's rare to have *both* the same
639	696	// title *and* the same link for two different items. So
640	697	// this is about the best we can do.
641	698	else :
642		- $guid .= '://'.md5($this->item['link'].'/'.$this->item['title']);
	699	+ $link = $this->permalink();
	700	+ if (is_null($link)) : $link = $this->link->uri(); endif;
	701	+ $guid .= '://'.md5($link.'/'.$this->title());
643	702	endif;
644	703	endif;
645	704	return $guid;
646	705	} /* SyndicatedPost::guid() */
647		-
648		- function author () {
	706	+
	707	+ public function author () {
649	708	$author = array ();
650		-
651		- if (isset($this->item['author_name'])):
652		- $author['name'] = $this->item['author_name'];
653		- elseif (isset($this->item['dc']['creator'])):
654		- $author['name'] = $this->item['dc']['creator'];
655		- elseif (isset($this->item['dc']['contributor'])):
656		- $author['name'] = $this->item['dc']['contributor'];
657		- elseif (isset($this->feed->channel['dc']['creator'])) :
658		- $author['name'] = $this->feed->channel['dc']['creator'];
659		- elseif (isset($this->feed->channel['dc']['contributor'])) :
660		- $author['name'] = $this->feed->channel['dc']['contributor'];
661		- elseif (isset($this->feed->channel['author_name'])) :
662		- $author['name'] = $this->feed->channel['author_name'];
663		- elseif ($this->feed->is_rss() and isset($this->item['author'])) :
664		- // The author element in RSS is allegedly an
665		- // e-mail address, but lots of people don't use
666		- // it that way. So let's make of it what we can.
667		- $author = parse_email_with_realname($this->item['author']);
668		-
669		- if (!isset($author['name'])) :
670		- if (isset($author['email'])) :
671		- $author['name'] = $author['email'];
672		- else :
673		- $author['name'] = $this->feed->channel['title'];
	709	+
	710	+ $aa = $this->entry->get_authors();
	711	+ if (is_countable($aa) and count($aa) > 0) :
	712	+ $a = reset($aa);
	713	+
	714	+ $author = array(
	715	+ 'name' => $a->get_name(),
	716	+ 'email' => $a->get_email(),
	717	+ 'uri' => $a->get_link(),
	718	+ );
	719	+ endif;
	720	+
	721	+ if (FEEDWORDPRESS_COMPATIBILITY) :
	722	+ // Search through the MagpieRSS elements: Atom, Dublin Core, RSS
	723	+ if (isset($this->item['author_name'])):
	724	+ $author['name'] = $this->item['author_name'];
	725	+ elseif (isset($this->item['dc']['creator'])):
	726	+ $author['name'] = $this->item['dc']['creator'];
	727	+ elseif (isset($this->item['dc']['contributor'])):
	728	+ $author['name'] = $this->item['dc']['contributor'];
	729	+ elseif (isset($this->feed->channel['dc']['creator'])) :
	730	+ $author['name'] = $this->feed->channel['dc']['creator'];
	731	+ elseif (isset($this->feed->channel['dc']['contributor'])) :
	732	+ $author['name'] = $this->feed->channel['dc']['contributor'];
	733	+ elseif (isset($this->feed->channel['author_name'])) :
	734	+ $author['name'] = $this->feed->channel['author_name'];
	735	+ elseif ($this->feed->is_rss() and isset($this->item['author'])) :
	736	+ // The author element in RSS is allegedly an
	737	+ // e-mail address, but lots of people don't use
	738	+ // it that way. So let's make of it what we can.
	739	+ $author = parse_email_with_realname($this->item['author']);
	740	+
	741	+ if (!isset($author['name'])) :
	742	+ if (isset($author['email'])) :
	743	+ $author['name'] = $author['email'];
	744	+ else :
	745	+ $author['name'] = $this->feed->channel['title'];
	746	+ endif;
674	747	endif;
675	748	endif;
676		- else :
677		- $author['name'] = $this->feed->channel['title'];
678	749	endif;
679		-
680		- if (isset($this->item['author_email'])):
681		- $author['email'] = $this->item['author_email'];
682		- elseif (isset($this->feed->channel['author_email'])) :
683		- $author['email'] = $this->feed->channel['author_email'];
	750	+
	751	+ if (!isset($author['name']) or is_null($author['name'])) :
	752	+ // Nothing found. Try some crappy defaults.
	753	+ if ($this->link->name()) :
	754	+ $author['name'] = $this->link->name();
	755	+ else :
	756	+ $url = parse_url($this->link->uri());
	757	+ $author['name'] = $url['host'];
	758	+ endif;
684	759	endif;
685		-
686		- if (isset($this->item['author_url'])):
687		- $author['uri'] = $this->item['author_url'];
688		- elseif (isset($this->feed->channel['author_url'])) :
689		- $author['uri'] = $this->item['author_url'];
690		- else:
691		- $author['uri'] = $this->feed->channel['link'];
	760	+
	761	+ if (FEEDWORDPRESS_COMPATIBILITY) :
	762	+ if (isset($this->item['author_email'])):
	763	+ $author['email'] = $this->item['author_email'];
	764	+ elseif (isset($this->feed->channel['author_email'])) :
	765	+ $author['email'] = $this->feed->channel['author_email'];
	766	+ endif;
	767	+
	768	+ if (isset($this->item['author_uri'])):
	769	+ $author['uri'] = $this->item['author_uri'];
	770	+ elseif (isset($this->item['author_url'])):
	771	+ $author['uri'] = $this->item['author_url'];
	772	+ elseif (isset($this->feed->channel['author_uri'])) :
	773	+ $author['uri'] = $this->item['author_uri'];
	774	+ elseif (isset($this->feed->channel['author_url'])) :
	775	+ $author['uri'] = $this->item['author_url'];
	776	+ elseif (isset($this->feed->channel['link'])) :
	777	+ $author['uri'] = $this->feed->channel['link'];
	778	+ endif;
692	779	endif;
693	780
694	781	return $author;
695	782	} /* SyndicatedPost::author() */
		@@ -694,8 +781,142 @@
694	781	return $author;
695	782	} /* SyndicatedPost::author() */
696	783
697	784	/**
	785	+ * SyndicatedPost::get_terms_from_settings(): Return an array of terms to associate with the incoming
	786	+ * post based on the Categories, Tags, and other terms associated with each new post by the user's
	787	+ * settings (global and feed-specific).
	788	+ *
	789	+ * @since 2016.0331
	790	+ * @return array of lists, each element has the taxonomy for a key ('category', 'post_tag', etc.),
	791	+ * and a list of term codes (either alphanumeric names, or ID numbers encoded in a format that
	792	+ * SyndicatedLink::category_ids() can understand) within that taxonomy
	793	+ *
	794	+ */
	795	+ public function get_terms_from_settings () {
	796	+ // Categories: start with default categories, if any.
	797	+ $cats = array();
	798	+ if ('no' != $this->link->setting('add/category', NULL, 'yes')) :
	799	+ $fc = get_option("feedwordpress_syndication_cats");
	800	+ if ($fc) :
	801	+ $cats = array_merge($cats, explode("\n", $fc));
	802	+ endif;
	803	+ endif;
	804	+
	805	+ $fc = $this->link->setting('cats',NULL, array());
	806	+ if (is_array($fc)) :
	807	+ $cats = array_merge($cats, $fc);
	808	+ endif;
	809	+ $preset_terms['category'] = $cats;
	810	+
	811	+ // Tags: start with default tags, if any
	812	+ $tags = array();
	813	+ if ('no' != $this->link->setting('add/post_tag', NULL, 'yes')) :
	814	+ $ft = get_option("feedwordpress_syndication_tags", NULL);
	815	+ $tags = (is_null($ft) ? array() : explode(FEEDWORDPRESS_CAT_SEPARATOR, $ft));
	816	+ endif;
	817	+
	818	+ $ft = $this->link->setting('tags', NULL, array());
	819	+ if (is_array($ft)) :
	820	+ $tags = array_merge($tags, $ft);
	821	+ endif;
	822	+ $preset_terms['post_tag'] = $tags;
	823	+
	824	+ $taxonomies = $this->link->taxonomies();
	825	+ $feedTerms = $this->link->setting('terms', NULL, array());
	826	+ $globalTerms = get_option('feedwordpress_syndication_terms', array());
	827	+ $specials = array('category' => 'cats', 'post_tag' => 'tags');
	828	+
	829	+ foreach ($taxonomies as $tax) :
	830	+ // category and tag settings have already previously been handled
	831	+ // but if this is from another taxonomy, then...
	832	+ if (!isset($specials[$tax])) :
	833	+ $terms = array();
	834	+
	835	+ // See if we should get the globals
	836	+ if ('no' != $this->link->setting("add/$tax", NULL, 'yes')) :
	837	+ if (isset($globalTerms[$tax])) :
	838	+ $terms = $globalTerms[$tax];
	839	+ endif;
	840	+ endif;
	841	+
	842	+ // Now merge in the locals
	843	+ if (isset($feedTerms[$tax])) :
	844	+ $terms = array_merge($terms, $feedTerms[$tax]);
	845	+ endif;
	846	+
	847	+ // That's all, folks.
	848	+ $preset_terms[$tax] = $terms;
	849	+ endif;
	850	+ endforeach;
	851	+
	852	+ return $preset_terms;
	853	+ } /* SyndicatedPost::get_terms_from_settings () */
	854	+
	855	+ /**
	856	+ * SyndicatedPost::get_terms_from_feeds(): Return an array of terms to associate with the incoming
	857	+ * post based on the contents of the subscribed feed (atom:category and rss:category elements, dc:subject
	858	+ * elements, tags embedded using microformats in the post content, etc.)
	859	+ *
	860	+ * @since 2016.0331
	861	+ * @return array of lists, each element has the taxonomy for a key ('category', 'post_tag', etc.),
	862	+ * and a list of alphanumeric term names
	863	+ */
	864	+ public function get_terms_from_feeds () {
	865	+ // Now add categories from the post, if we have 'em
	866	+ $cats = array();
	867	+ $post_cats = $this->entry->get_categories();
	868	+ if (is_array($post_cats)) : foreach ($post_cats as $cat) :
	869	+ $cat_name = $cat->get_term();
	870	+ if (!$cat_name) : $cat_name = $cat->get_label(); endif;
	871	+
	872	+ if ($this->link->setting('cat_split', NULL, NULL)) :
	873	+ $pcre = "\007".$this->feedmeta['cat_split']."\007";
	874	+ $cats = array_merge(
	875	+ $cats,
	876	+ preg_split(
	877	+ $pcre,
	878	+ $cat_name,
	879	+ -1 /*=no limit*/,
	880	+ PREG_SPLIT_NO_EMPTY
	881	+ )
	882	+ );
	883	+ else :
	884	+ $cats[] = $cat_name;
	885	+ endif;
	886	+ endforeach; endif;
	887	+
	888	+ $feed_terms['category'] = apply_filters('syndicated_item_categories', $cats, $this);
	889	+
	890	+ // Scan post for /a[@rel='tag'] and use as tags if present
	891	+ $tags = $this->inline_tags();
	892	+ $feed_terms['post_tag'] = apply_filters('syndicated_item_tags', $tags, $this);
	893	+
	894	+ return $feed_terms;
	895	+ } /* SyndicatedPost::get_terms_from_feeds () */
	896	+
	897	+ /**
	898	+ * SyndicatedPost::inline_tags: Return a list of all the tags embedded
	899	+ * in post content using the a[@rel="tag"] microformat.
	900	+ *
	901	+ * @since 2010.0630
	902	+ * @return array of string values containing the name of each tag
	903	+ */
	904	+ function inline_tags () {
	905	+ $tags = array();
	906	+ $content = $this->content();
	907	+ $pattern = FeedWordPressHTML::tagWithAttributeRegex('a', 'rel', 'tag');
	908	+ preg_match_all($pattern, $content, $refs, PREG_SET_ORDER);
	909	+ if (is_countable($refs) and count($refs) > 0) :
	910	+ foreach ($refs as $ref) :
	911	+ $tag = FeedWordPressHTML::tagWithAttributeMatch($ref);
	912	+ $tags[] = $tag['content'];
	913	+ endforeach;
	914	+ endif;
	915	+ return $tags;
	916	+ }
	917	+
	918	+ /**
698	919	* SyndicatedPost::isTaggedAs: Test whether a feed item is
699	920	* tagged / categorized with a given string. Case and leading and
700	921	* trailing whitespace are ignored.
701	922	*
		@@ -700,9 +921,9 @@
700	921	* trailing whitespace are ignored.
701	922	*
702	923	* @param string $tag Tag to check for
703	924	*
704		- * @return bool Whether or not at least one of the categories / tags on
	925	+ * @return bool Whether or not at least one of the categories / tags on
705	926	* $this->item is set to $tag (modulo case and leading and trailing
706	927	* whitespace)
707	928	*/
708	929	function isTaggedAs ($tag) {
		@@ -715,9 +936,9 @@
715	936	// If we have the new MagpieRSS, the number of category elements
716	937	// on this item is stored under index "category#".
717	938	if (isset($this->item['category#'])) :
718	939	$numberOfCategories = (int) $this->item['category#'];
719		-
	940	+
720	941	// We REALLY shouldn't have the old and busted MagpieRSS, but in
721	942	// case we do, it doesn't support multiple categories, but there
722	943	// might still be a single value under the "category" index.
723	944	elseif (isset($this->item['category'])) :
		@@ -774,14 +995,42 @@
774	995	);
775	996	endif;
776	997	endfor;
777	998	endif;
778		- return $enclosures;
	999	+ return $enclosures;
779	1000	} /* SyndicatedPost::enclosures() */
780	1001
	1002	+ function source ($what = NULL) {
	1003	+ $ret = NULL;
	1004	+ $source = $this->entry->get_source();
	1005	+ if ($source) :
	1006	+ $ret = array();
	1007	+ $ret['title'] = $source->get_title();
	1008	+ $ret['uri'] = $source->get_link();
	1009	+ $ret['feed'] = $source->get_link(0, 'self');
	1010	+
	1011	+ if ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'id')) :
	1012	+ $ret['id'] = $id_tags[0]['data'];
	1013	+ elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_ATOM_03, 'id')) :
	1014	+ $ret['id'] = $id_tags[0]['data'];
	1015	+ elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_RSS_20, 'guid')) :
	1016	+ $ret['id'] = $id_tags[0]['data'];
	1017	+ elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_RSS_10, 'guid')) :
	1018	+ $ret['id'] = $id_tags[0]['data'];
	1019	+ elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_RSS_090, 'guid')) :
	1020	+ $ret['id'] = $id_tags[0]['data'];
	1021	+ endif;
	1022	+ endif;
	1023	+
	1024	+ if (!is_null($what) and is_scalar($what)) :
	1025	+ $ret = $ret[$what];
	1026	+ endif;
	1027	+ return $ret;
	1028	+ }
	1029	+
781	1030	function comment_link () {
782	1031	$url = null;
783		-
	1032	+
784	1033	// RSS 2.0 has a standard <comments> element:
785	1034	// "<comments> is an optional sub-element of <item>. If present,
786	1035	// it is the url of the comments page for the item."
787	1036	// <http://cyber.law.harvard.edu/rss/rss.html#ltcommentsgtSubelementOfLtitemgt>
		@@ -792,15 +1041,15 @@
792	1041	// The convention in Atom feeds is to use a standard <link>
793	1042	// element with @rel="replies" and @type="text/html".
794	1043	// Unfortunately, SimplePie_Item::get_links() allows us to filter
795	1044	// by the value of @rel, but not by the value of @type. *sigh*
796		-
	1045	+
797	1046	// Try Atom 1.0 first
798	1047	$linkElements = $this->entry->get_item_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'link');
799		-
	1048	+
800	1049	// Fall back and try Atom 0.3
801	1050	if (is_null($linkElements)) : $linkElements = $this->entry->get_item_tags(SIMPLEPIE_NAMESPACE_ATOM_03, 'link'); endif;
802		-
	1051	+
803	1052	// Now loop through the elements, screening by @rel and @type
804	1053	if (is_array($linkElements)) : foreach ($linkElements as $link) :
805	1054	$rel = (isset($link['attribs']['']['rel']) ? $link['attribs']['']['rel'] : 'alternate');
806	1055	$type = (isset($link['attribs']['']['type']) ? $link['attribs']['']['type'] : NULL);
		@@ -885,18 +1134,19 @@
885	1134	var $_base = null;
886	1135
887	1136	function resolve_single_relative_uri ($refs) {
888	1137	$tag = FeedWordPressHTML::attributeMatch($refs);
889		- $url = Relative_URI::resolve($tag['value'], $this->_base);
	1138	+ $url = SimplePie_Misc::absolutize_url($tag['value'], $this->_base);
	1139	+
890	1140	return $tag['prefix'] . $url . $tag['suffix'];
891	1141	} /* function SyndicatedPost::resolve_single_relative_uri() */
892	1142
893		- function resolve_relative_uris ($content, $obj) {
	1143	+ static function resolve_relative_uris ($content, $obj) {
894	1144	$set = $obj->link->setting('resolve relative', 'resolve_relative', 'yes');
895		- if ($set and $set != 'no') :
	1145	+ if ($set and $set != 'no') :
896	1146	// Fallback: if we don't have anything better, use the
897	1147	// item link from the feed
898		- $obj->_base = $obj->item['link']; // Reset the base for resolving relative URIs
	1148	+ $obj->_base = $obj->permalink(); // Reset the base for resolving relative URIs
899	1149
900	1150	// What we should do here, properly, is to use
901	1151	// SimplePie_Item::get_base() -- but that method is
902	1152	// currently broken. Or getting down and dirty in the
		@@ -906,16 +1156,22 @@
906	1156
907	1157	foreach ($obj->uri_attrs as $pair) :
908	1158	list($tag, $attr) = $pair;
909	1159	$pattern = FeedWordPressHTML::attributeRegex($tag, $attr);
	1160	+
	1161	+ // FIXME: Encountered issue while testing an extremely long (= 88827 characters) item
	1162	+ // Relying on preg_replace_callback() here can cause a PHP seg fault on my development
	1163	+ // server. preg_match_all() causes a similar problem. Apparently this is a PCRE issue
	1164	+ // Cf. discussion of similar issue <https://bugs.php.net/bug.php?id=65009>
910	1165	$content = preg_replace_callback (
911	1166	$pattern,
912		- array(&$obj, 'resolve_single_relative_uri'),
	1167	+ array($obj, 'resolve_single_relative_uri'),
913	1168	$content
914	1169	);
	1170	+
915	1171	endforeach;
916	1172	endif;
917		-
	1173	+
918	1174	return $content;
919	1175	} /* function SyndicatedPost::resolve_relative_uris () */
920	1176
921	1177	var $strip_attrs = array (
		@@ -928,9 +1184,9 @@
928	1184	$tag = FeedWordPressHTML::attributeMatch($refs);
929	1185	return $tag['before_attribute'].$tag['after_attribute'];
930	1186	}
931	1187
932		- function sanitize_content ($content, $obj) {
	1188	+ static function sanitize_content ($content, $obj) {
933	1189	# This kind of sucks. I intend to replace it with
934	1190	# lib_filter sometime soon.
935	1191	foreach ($obj->strip_attrs as $pair):
936	1192	list($tag,$attr) = $pair;
		@@ -937,9 +1193,9 @@
937	1193	$pattern = FeedWordPressHTML::attributeRegex($tag, $attr);
938	1194
939	1195	$content = preg_replace_callback (
940	1196	$pattern,
941		- array(&$obj, 'strip_attribute_from_tag'),
	1197	+ array($obj, 'strip_attribute_from_tag'),
942	1198	$content
943	1199	);
944	1200	endforeach;
945	1201	return $content;
		@@ -965,76 +1221,236 @@
965	1221	* match the latest revision, or a previously syndicated post that is
966	1222	* still up-to-date.
967	1223	*
968	1224	* @return int A status code representing the freshness of the post
	1225	+ * -1 = post already syndicated; has a revision that needs to be stored, but not updated to
969	1226	* 0 = post already syndicated; no update needed
970	1227	* 1 = post already syndicated, but needs to be updated to latest
971	1228	* 2 = post has not yet been syndicated; needs to be created
972	1229	*/
973		- function freshness () {
	1230	+ function freshness ($format = 'number') {
974	1231	global $wpdb;
975	1232
976	1233	if ($this->filtered()) : // This should never happen.
977		- FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__);
	1234	+ FeedWordPressDiagnostic::critical_bug('SyndicatedPost', $this, __LINE__, __FILE__);
978	1235	endif;
979		-
980		- if (is_null($this->_freshness)) :
981		- $guid = $wpdb->escape($this->guid());
982	1236
983		- $result = $wpdb->get_row("
984		- SELECT id, guid, post_modified_gmt
985		- FROM $wpdb->posts WHERE guid='$guid'
986		- ");
	1237	+ if (is_null($this->_freshness)) : // Not yet checked and cached.
	1238	+ $guid = $this->post['guid'];
	1239	+ $eguid = esc_sql($this->post['guid']);
987	1240
988		- if (!$result) :
	1241	+ $q = new WP_Query(array(
	1242	+ 'fields' => '_synfresh', // id, guid, post_modified_gmt
	1243	+ 'ignore_sticky_posts' => true,
	1244	+ 'guid' => $guid,
	1245	+ ));
	1246	+
	1247	+ $old_post = NULL;
	1248	+ if ($q->have_posts()) :
	1249	+ while ($q->have_posts()) : $q->the_post();
	1250	+ if (get_post_type($q->post->ID) == $this->post['post_type']):
	1251	+ $old_post = $q->post;
	1252	+ endif;
	1253	+ endwhile;
	1254	+ endif;
	1255	+
	1256	+ if (is_null($old_post)) : // No post with this guid
	1257	+ FeedWordPress::diagnostic('feed_items:freshness', 'Item ['.$guid.'] "'.$this->entry->get_title().'" is a NEW POST.');
	1258	+ $this->_wp_id = NULL;
989	1259	$this->_freshness = 2; // New content
990		- else:
991		- $stored_update_hashes = get_post_custom_values('syndication_item_hash', $result->id);
992		- if (count($stored_update_hashes) > 0) :
993		- $stored_update_hash = $stored_update_hashes[0];
994		- $update_hash_changed = ($stored_update_hash != $this->update_hash());
	1260	+ else :
	1261	+ // Presume there is nothing new until we find
	1262	+ // something new.
	1263	+ $updated = false;
	1264	+ $live = false;
	1265	+
	1266	+ // Pull the list of existing revisions to get
	1267	+ // timestamps.
	1268	+ $revisions = wp_get_post_revisions($old_post->ID);
	1269	+ foreach ($revisions as $rev) :
	1270	+ $revisions_ts[] = mysql2date('G', $rev->post_modified_gmt);
	1271	+ endforeach;
	1272	+
	1273	+ $revisions_ts[] = mysql2date('G', $old_post->post_modified_gmt);
	1274	+ $last_rev_ts = end($revisions_ts);
	1275	+ $updated_ts = $this->updated(/*fallback=*/ true, /*default=*/ NULL);
	1276	+
	1277	+ // If we have an explicit updated timestamp,
	1278	+ // check that against existing stamps.
	1279	+ if (!is_null($updated_ts)) :
	1280	+ $updated = !in_array($updated_ts, $revisions_ts);
	1281	+
	1282	+ // If this a newer revision, make it go
	1283	+ // live. If an older one, just record
	1284	+ // the contents.
	1285	+ $live = ($updated and ($updated_ts > $last_rev_ts));
	1286	+ endif;
	1287	+
	1288	+ // This is a revision we haven't seen before, judging by the date.
	1289	+
	1290	+ $updatedReason = NULL;
	1291	+ if ($updated) :
	1292	+ $updatedReason = preg_replace(
	1293	+ "/\s+/", " ",
	1294	+ 'has been marked with a new timestamp ('
	1295	+ .date('Y-m-d H:i:s', $updated_ts)
	1296	+ ." > "
	1297	+ .date('Y-m-d H:i:s', $last_rev_ts)
	1298	+ .')'
	1299	+ );
	1300	+
	1301	+ // The date does not indicate a new revision, so
	1302	+ // let's check the hash.
995	1303	else :
996		- $update_hash_changed = false;
	1304	+ // Or the hash...
	1305	+ $hash = $this->update_hash();
	1306	+ $seen = $this->stored_hashes($old_post->ID);
	1307	+ if (is_countable($seen) and count($seen) > 0) :
	1308	+ $updated = !in_array($hash, $seen); // Not seen yet?
	1309	+ else :
	1310	+ $updated = true; // Can't find syndication meta-data
	1311	+ endif;
	1312	+
	1313	+ if ($updated and FeedWordPressDiagnostic::is_on('feed_items:freshness:reasons')) :
	1314	+ // In the absence of definitive
	1315	+ // timestamp information, we
	1316	+ // just have to assume that a
	1317	+ // hash we haven't seen before
	1318	+ // is a newer version.
	1319	+ $live = true;
	1320	+
	1321	+ $updatedReason = ' has a not-yet-seen update hash: '
	1322	+ .MyPHP::val($hash)
	1323	+ .' not in {'
	1324	+ .implode(", ", array_map(array('FeedWordPress', 'val'), $seen))
	1325	+ .'}. Basis: '
	1326	+ .MyPHP::val(array_keys($this->update_hash(false)));
	1327	+ endif;
997	1328	endif;
998	1329
999		- preg_match('/([0-9]+)-([0-9]+)-([0-9]+) ([0-9]+):([0-9]+):([0-9]+)/', $result->post_modified_gmt, $backref);
	1330	+ $frozen = false;
	1331	+ if ($updated) : // Ignore if the post is frozen
	1332	+ $frozen = ('yes' == $this->link->setting('freeze updates', 'freeze_updates', NULL));
	1333	+ if (!$frozen) :
	1334	+ $frozen_value = get_post_meta($old_post->ID, '_syndication_freeze_updates', /*single=*/ true);
	1335	+ $frozen = (!is_null($frozen_value) and ('yes' == $frozen_value));
1000	1336
1001		- $last_rev_ts = gmmktime($backref[4], $backref[5], $backref[6], $backref[2], $backref[3], $backref[1]);
1002		- $updated_ts = $this->updated(/*fallback=*/ true, /*default=*/ NULL);
1003		-
1004		- $frozen_values = get_post_custom_values('_syndication_freeze_updates', $result->id);
1005		- $frozen_post = (count($frozen_values) > 0 and 'yes' == $frozen_values[0]);
1006		- $frozen_feed = ('yes' == $this->link->setting('freeze updates', 'freeze_updates', NULL));
	1337	+ if ($frozen) :
	1338	+ $updatedReason = ' IS BLOCKED FROM BEING UPDATED BY A UPDATE LOCK ON THIS POST, EVEN THOUGH IT '.$updatedReason;
	1339	+ endif;
	1340	+ else :
	1341	+ $updatedReason = ' IS BLOCKED FROM BEING UPDATED BY A FEEDWORDPRESS UPDATE LOCK, EVEN THOUGH IT '.$updatedReason;
	1342	+ endif;
	1343	+ endif;
	1344	+ $live = ($live and !$frozen);
1007	1345
1008		- // Check timestamps...
1009		- $updated = (
1010		- !is_null($updated_ts)
1011		- and ($updated_ts > $last_rev_ts)
1012		- );
1013		-
1014		-
1015		- // Or the hash...
1016		- $updated = ($updated or $update_hash_changed);
1017		-
1018		- // But only if the post is not frozen.
1019		- $updated = (
1020		- $updated
1021		- and !$frozen_post
1022		- and !$frozen_feed
1023		- );
1024		-
1025	1346	if ($updated) :
1026		- $this->_freshness = 1; // Updated content
1027		- $this->_wp_id = $result->id;
	1347	+ FeedWordPress::diagnostic('feed_items:freshness', 'Item ['.$guid.'] "'.$this->entry->get_title().'" is an update of an existing post.');
	1348	+ if (!is_null($updatedReason)) :
	1349	+ $updatedReason = preg_replace('/\s+/', ' ', $updatedReason);
	1350	+ FeedWordPress::diagnostic('feed_items:freshness:reasons', 'Item ['.$guid.'] "'.$this->entry->get_title().'" '.$updatedReason);
	1351	+ endif;
	1352	+
	1353	+ $this->_freshness = apply_filters('syndicated_item_freshness', ($live ? 1 : -1), $updated, $frozen, $updated_ts, $last_rev_ts, $this);
	1354	+
	1355	+ $this->_wp_id = $old_post->ID;
	1356	+ $this->_wp_post = $old_post;
	1357	+
	1358	+ // We want this to keep a running list of all the
	1359	+ // processed update hashes.
	1360	+ $this->post['meta']['syndication_item_hash'] = array_merge(
	1361	+ $this->stored_hashes(),
	1362	+ array($this->update_hash())
	1363	+ );
1028	1364	else :
	1365	+ FeedWordPress::diagnostic('feed_items:freshness', 'Item ['.$guid.'] "'.$this->entry->get_title().'" is a duplicate of an existing post.');
1029	1366	$this->_freshness = 0; // Same old, same old
1030		- $this->_wp_id = $result->id;
	1367	+ $this->_wp_id = $old_post->ID;
1031	1368	endif;
1032	1369	endif;
1033	1370	endif;
1034		- return $this->_freshness;
	1371	+
	1372	+ switch ($format) :
	1373	+ case 'status' :
	1374	+ switch ($this->_freshness) :
	1375	+ case -1:
	1376	+ $ret = 'stored';
	1377	+ break;
	1378	+ case 0:
	1379	+ $ret = NULL;
	1380	+ break;
	1381	+ case 1:
	1382	+ $ret = 'updated';
	1383	+ break;
	1384	+ case 2:
	1385	+ default:
	1386	+ $ret = 'new';
	1387	+ break;
	1388	+ endswitch;
	1389	+ break;
	1390	+ case 'number' :
	1391	+ default :
	1392	+ $ret = $this->_freshness;
	1393	+ endswitch;
	1394	+
	1395	+
	1396	+ return $ret;
	1397	+ } /* SyndicatedPost::freshness () */
	1398	+
	1399	+ function has_fresh_content () {
	1400	+ return ( ! $this->filtered() and $this->freshness() != 0 );
	1401	+ } /* SyndicatedPost::has_fresh_content () */
	1402	+
	1403	+ function this_revision_needs_original_post ($freshness = NULL) {
	1404	+ if (is_null($freshness)) :
	1405	+ $freshness = $this->freshness();
	1406	+ endif;
	1407	+ return ( $freshness >= 2 );
1035	1408	}
1036		-
	1409	+
	1410	+ function this_revision_is_current ($freshness = NULL) {
	1411	+ if (is_null($freshness)) :
	1412	+ $freshness = $this->freshness();
	1413	+ endif;
	1414	+ return ( $freshness >= 1 );
	1415	+ } /* SyndicatedPost::this_revision_is_current () */
	1416	+
	1417	+ function fresh_content_is_update () {
	1418	+ return ($this->freshness() < 2);
	1419	+ } /* SyndicatedPost::fresh_content_is_update () */
	1420	+
	1421	+ function fresh_storage_diagnostic () {
	1422	+ $ret = NULL;
	1423	+ switch ($this->freshness()) :
	1424	+ case -1 :
	1425	+ $ret = 'Storing alternate revision of existing post # '.$this->wp_id().', "'.$this->post['post_title'].'"';
	1426	+ break;
	1427	+ case 1 :
	1428	+ $ret = 'Updating existing post # '.$this->wp_id().', "'.$this->post['post_title'].'"';
	1429	+ break;
	1430	+ case 2 :
	1431	+ default :
	1432	+ $ret = 'Inserting new post "'.$this->post['post_title'].'"';
	1433	+ break;
	1434	+ endswitch;
	1435	+ return $ret;
	1436	+ } /* SyndicatedPost::fresh_storage_diagnostic() */
	1437	+
	1438	+ function fresh_storage_hook () {
	1439	+ $ret = NULL;
	1440	+ switch ($this->freshness()) :
	1441	+ case -1 :
	1442	+ case 1 :
	1443	+ $ret = 'update_syndicated_item';
	1444	+ break;
	1445	+ case 2 :
	1446	+ default :
	1447	+ $ret = 'post_syndicated_item';
	1448	+ break;
	1449	+ endswitch;
	1450	+ return $ret;
	1451	+ } /* SyndicatedPost::fresh_storage_hook () */
	1452	+
1037	1453	#################################################
1038	1454	#### INTERNAL STORAGE AND MANAGEMENT METHODS ####
1039	1455	#################################################
1040	1456
		@@ -1039,11 +1455,11 @@
1039	1455	#################################################
1040	1456
1041	1457	function wp_id () {
1042	1458	if ($this->filtered()) : // This should never happen.
1043		- FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__);
	1459	+ FeedWordPressDiagnostic::critical_bug('SyndicatedPost', $this, __LINE__, __FILE__);
1044	1460	endif;
1045		-
	1461	+
1046	1462	if (is_null($this->_wp_id) and is_null($this->_freshness)) :
1047	1463	$fresh = $this->freshness(); // sets WP DB id in the process
1048	1464	endif;
1049	1465	return $this->_wp_id;
		@@ -1048,59 +1464,183 @@
1048	1464	endif;
1049	1465	return $this->_wp_id;
1050	1466	}
1051	1467
1052		- function store () {
	1468	+ /**
	1469	+ * SyndicatedPost::secure_author_id(). Look up, or create, a numeric ID
	1470	+ * for the author of the incoming post.
	1471	+ *
	1472	+ * side effect: int|NULL stored in $this->post['post_author']
	1473	+ * side effect: IF no valid author is found, NULL stored in $this->post
	1474	+ * side effect: diagnostic output in case item is rejected with NULL author
	1475	+ *
	1476	+ * @used-by SyndicatedPost::store
	1477	+ *
	1478	+ * @uses SyndicatedPost::post
	1479	+ * @uses SyndicatedPost::author_id
	1480	+ * @uses SyndicatedLink::setting
	1481	+ * @uses FeedWordPress::diagnostic
	1482	+ */
	1483	+ protected function secure_author_id () {
	1484	+ # -- Look up, or create, numeric ID for author
	1485	+ $this->post['post_author'] = $this->author_id (
	1486	+ $this->link->setting('unfamiliar author', 'unfamiliar_author', 'create')
	1487	+ );
	1488	+
	1489	+ if (is_null($this->post['post_author'])) :
	1490	+ FeedWordPress::diagnostic('feed_items:rejected', 'Filtered out item ['.$this->guid().'] without syndication: no author available');
	1491	+ $this->post = NULL;
	1492	+ endif;
	1493	+ } /* SyndicatedPost::secure_author_id() */
	1494	+
	1495	+ /**
	1496	+ * SyndicatedPost::secure_term_ids(). Look up, or create, numeric IDs
	1497	+ * for the terms (categories, tags, etc.) assigned to the incoming post,
	1498	+ * whether by global settings, feed settings, or by the tags on the feed.
	1499	+ *
	1500	+ * side effect: array of term ids stored in $this->post['tax_input']
	1501	+ * side effect: IF settings or filters determine post should be filtered out,
	1502	+ * NULL stored in $this->post
	1503	+ *
	1504	+ * @used-by SyndicatedPost::store
	1505	+ *
	1506	+ * @uses apply_filters
	1507	+ * @uses SyndicatedLink::setting
	1508	+ * @uses SyndicatedPost::category_ids
	1509	+ * @uses SyndicatedPost::preset_terms
	1510	+ * @uses SyndicatedPost::post
	1511	+ */
	1512	+ protected function secure_term_ids () {
	1513	+ $mapping = apply_filters('syndicated_post_terms_mapping', array(
	1514	+ 'category' => array('abbr' => 'cats', 'unfamiliar' => 'category', 'domain' => array('category', 'post_tag')),
	1515	+ 'post_tag' => array('abbr' => 'tags', 'unfamiliar' => 'post_tag', 'domain' => array('post_tag')),
	1516	+ ), $this);
	1517	+
	1518	+ $termSet = array(); $valid = null;
	1519	+ foreach ($this->feed_terms as $what => $anTerms) :
	1520	+ // Default to using the inclusive procedures (for cats) rather than exclusive (for inline tags)
	1521	+ $taxes = (isset($mapping[$what]) ? $mapping[$what] : $mapping['category']);
	1522	+ $unfamiliar = $taxes['unfamiliar'];
	1523	+
	1524	+ if (!is_null($this->post)) : // Not filtered out yet
	1525	+ # -- Look up, or create, numeric ID for categories
	1526	+ $taxonomies = $this->link->setting("match/".$taxes['abbr'], 'match_'.$taxes['abbr'], $taxes['domain']);
	1527	+
	1528	+ // Eliminate dummy variables
	1529	+ $taxonomies = array_filter($taxonomies, 'remove_dummy_zero');
	1530	+
	1531	+ // Allow FWP add-on filters to control the taxonomies we use to search for a term
	1532	+ $taxonomies = apply_filters("syndicated_post_terms_match", $taxonomies, $what, $this);
	1533	+ $taxonomies = apply_filters("syndicated_post_terms_match_${what}", $taxonomies, $this);
	1534	+
	1535	+ // Allow FWP add-on filters to control with greater precision what happens on unmatched
	1536	+ $unmatched = apply_filters("syndicated_post_terms_unfamiliar",
	1537	+ $this->link->setting(
	1538	+ "unfamiliar {$unfamiliar}",
	1539	+ "unfamiliar_{$unfamiliar}",
	1540	+ 'create:'.$unfamiliar
	1541	+ ),
	1542	+ $what,
	1543	+ $this
	1544	+ );
	1545	+
	1546	+ $terms = $this->category_ids (
	1547	+ $anTerms,
	1548	+ $unmatched,
	1549	+ /*taxonomies=*/ $taxonomies,
	1550	+ array(
	1551	+ 'singleton' => false, // I don't like surprises
	1552	+ 'filters' => true,
	1553	+ )
	1554	+ );
	1555	+
	1556	+ if (is_null($terms) or is_null($termSet)) :
	1557	+ // filtered out -- no matches
	1558	+ else :
	1559	+ $valid = true;
	1560	+
	1561	+ // filter mode off, or at least one match
	1562	+ foreach ($terms as $tax => $term_ids) :
	1563	+ if (!isset($termSet[$tax])) :
	1564	+ $termSet[$tax] = array();
	1565	+ endif;
	1566	+ $termSet[$tax] = array_merge($termSet[$tax], $term_ids);
	1567	+ endforeach;
	1568	+ endif;
	1569	+ endif;
	1570	+ endforeach;
	1571	+
	1572	+ if (is_null($valid)) : // Plonked
	1573	+ $this->post = NULL;
	1574	+ else : // We can proceed
	1575	+ $this->post['tax_input'] = array();
	1576	+ foreach ($termSet as $tax => $term_ids) :
	1577	+ if (!isset($this->post['tax_input'][$tax])) :
	1578	+ $this->post['tax_input'][$tax] = array();
	1579	+ endif;
	1580	+ $this->post['tax_input'][$tax] = array_merge(
	1581	+ $this->post['tax_input'][$tax],
	1582	+ $term_ids
	1583	+ );
	1584	+ endforeach;
	1585	+
	1586	+ // Now let's add on the feed and global presets
	1587	+ foreach ($this->preset_terms as $tax => $term_ids) :
	1588	+ if (!isset($this->post['tax_input'][$tax])) :
	1589	+ $this->post['tax_input'][$tax] = array();
	1590	+ endif;
	1591	+
	1592	+ $this->post['tax_input'][$tax] = array_merge (
	1593	+ $this->post['tax_input'][$tax],
	1594	+ $this->category_ids (
	1595	+ /*terms=*/ $term_ids,
	1596	+ /*unfamiliar=*/ 'create:'.$tax, // These are presets; for those added in a tagbox editor, the tag may not yet exist
	1597	+ /*taxonomies=*/ array($tax),
	1598	+ array(
	1599	+ 'singleton' => true,
	1600	+ ))
	1601	+ );
	1602	+ endforeach;
	1603	+ endif;
	1604	+ } /* SyndicatedPost::secure_term_ids() */
	1605	+
	1606	+ /**
	1607	+ * SyndicatedPost::store
	1608	+ *
	1609	+ * @uses SyndicatedPost::secure_author_id
	1610	+ */
	1611	+ public function store () {
1053	1612	global $wpdb;
1054	1613
1055	1614	if ($this->filtered()) : // This should never happen.
1056		- FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__);
	1615	+ FeedWordPressDiagnostic::critical_bug('SyndicatedPost', $this, __LINE__, __FILE__);
1057	1616	endif;
1058		-
	1617	+
1059	1618	$freshness = $this->freshness();
1060		- if ($freshness > 0) :
1061		- # -- Look up, or create, numeric ID for author
1062		- $this->post['post_author'] = $this->author_id (
1063		- FeedWordPress::on_unfamiliar('author', $this->post['named']['unfamiliar']['author'])
1064		- );
	1619	+ if ($this->has_fresh_content()) :
	1620	+ $this->secure_author_id();
	1621	+ endif;
1065	1622
1066		- if (is_null($this->post['post_author'])) :
1067		- $this->post = NULL;
1068		- endif;
	1623	+ if ($this->has_fresh_content()) : // Was this filtered during author_id lookup?
	1624	+ $this->secure_term_ids();
1069	1625	endif;
1070		-
1071		- if (!$this->filtered() and $freshness > 0) :
1072		- # -- Look up, or create, numeric ID for categories
1073		- list($pcats, $ptags) = $this->category_ids (
1074		- $this->post['named']['category'],
1075		- FeedWordPress::on_unfamiliar('category', $this->post['named']['unfamiliar']['category']),
1076		- /*tags_too=*/ true
1077		- );
1078	1626
1079		- $this->post['post_category'] = $pcats;
1080		- $this->post['tags_input'] = array_merge($this->post['tags_input'], $ptags);
	1627	+ // We have to check again in case the post has been filtered
	1628	+ // during the category/tags/taxonomy terms lookup
	1629	+ if ($this->has_fresh_content()) :
	1630	+ // Filter some individual fields
1081	1631
1082		- if (is_null($this->post['post_category'])) :
1083		- // filter mode on, no matching categories; drop the post
1084		- $this->post = NULL;
1085		- else :
1086		- // filter mode off or at least one match; now add on the feed and global presets
1087		- $this->post['post_category'] = array_merge (
1088		- $this->post['post_category'],
1089		- $this->category_ids (
1090		- $this->post['named']['preset/category'],
1091		- 'default'
1092		- )
1093		- );
	1632	+ // If there already is a post slug (from syndication or by manual
	1633	+ // editing) don't cause WP to overwrite it by sending in a NULL
	1634	+ // post_name. Props Chris Fritz 2012-11-28.
	1635	+ $post_name = (is_null($this->_wp_post) ? NULL : $this->_wp_post->post_name);
1094	1636
1095		- if (count($this->post['post_category']) < 1) :
1096		- $this->post['post_category'][] = 1; // Default to category 1 ("Uncategorized" / "General") if nothing else
1097		- endif;
	1637	+ // Allow filters to set post slug. Props niska.
	1638	+ $post_name = apply_filters('syndicated_post_slug', $post_name, $this);
	1639	+ if (!empty($post_name)) :
	1640	+ $this->post['post_name'] = $post_name;
1098	1641	endif;
1099		- endif;
1100		-
1101		- if (!$this->filtered() and $freshness > 0) :
1102		- unset($this->post['named']);
	1642	+
1103	1643	$this->post = apply_filters('syndicated_post', $this->post, $this);
1104	1644
1105	1645	// Allow for feed-specific syndicated_post filters.
1106	1646	$this->post = apply_filters(
		@@ -1108,9 +1648,9 @@
1108	1648	$this->post,
1109	1649	$this
1110	1650	);
1111	1651	endif;
1112		-
	1652	+
1113	1653	// Hook in early to make sure these get inserted if at all possible
1114	1654	add_action(
1115	1655	/*hook=*/ 'transition_post_status',
1116	1656	/*callback=*/ array($this, 'add_rss_meta'),
		@@ -1116,191 +1656,203 @@
1116	1656	/*callback=*/ array($this, 'add_rss_meta'),
1117	1657	/*priority=*/ -10000, /* very early */
1118	1658	/*arguments=*/ 3
1119	1659	);
	1660	+
	1661	+ $ret = false;
	1662	+ if ($this->has_fresh_content()) :
	1663	+ $diag = $this->fresh_storage_diagnostic();
	1664	+ if (!is_null($diag)) :
	1665	+ FeedWordPress::diagnostic('syndicated_posts', $diag);
	1666	+ endif;
	1667	+
	1668	+ $this->insert_post(/*update=*/ $this->fresh_content_is_update(), $this->freshness());
1120	1669
1121		- if (!$this->filtered() and $freshness == 2) :
1122		- // The item has not yet been added. So let's add it.
1123		- $this->insert_new();
1124		- do_action('post_syndicated_item', $this->wp_id(), $this);
	1670	+ $hook = $this->fresh_storage_hook();
	1671	+ if (!is_null($hook)) :
	1672	+ do_action($hook, $this->wp_id(), $this);
	1673	+ endif;
	1674	+
	1675	+ $ret = $this->freshness('status');
	1676	+ endif;
1125	1677
1126		- $ret = 'new';
1127		- elseif (!$this->filtered() and $freshness == 1) :
1128		- $this->post['ID'] = $this->wp_id();
1129		- $this->update_existing();
1130		- do_action('update_syndicated_item', $this->wp_id(), $this);
	1678	+ // If this is a legit, non-filtered post, tag it as found on the
	1679	+ // feed regardless of fresh or stale status
	1680	+ if (!$this->filtered()) :
	1681	+ $key = '_feedwordpress_retire_me_' . $this->link->id;
	1682	+ delete_post_meta($this->wp_id(), $key);
1131	1683
1132		- $ret = 'updated';
1133		- else :
1134		- $ret = false;
	1684	+ $status = get_post_field('post_status', $this->wp_id());
	1685	+ if ('fwpretired'==$status and $this->link->is_non_incremental()) :
	1686	+ FeedWordPress::diagnostic('syndicated_posts', "Un-retiring previously retired post # ".$this->wp_id()." due to re-appearance on non-incremental feed.");
	1687	+ set_post_field('post_status', $this->post['post_status'], $this->wp_id());
	1688	+ wp_transition_post_status($this->post['post_status'], $status, $this->post);
	1689	+ elseif ('fwpzapped'==$status) :
	1690	+ // Set this new revision up to be
	1691	+ // blanked on the next update.
	1692	+ add_post_meta($this->wp_id(), '_feedwordpress_zapped_blank_me', 2, /*single=*/ true);
	1693	+ endif;
1135	1694	endif;
1136	1695
1137	1696	// Remove add_rss_meta hook
1138	1697	remove_action(
1139	1698	/*hook=*/ 'transition_post_status',
1140		- /*callback=*/ array($this, 'add_rss_meta')
	1699	+ /*callback=*/ array($this, 'add_rss_meta'),
	1700	+ /*priority=*/ -10000, /* very early */
	1701	+ /*arguments=*/ 3
1141	1702	);
1142	1703
1143	1704	return $ret;
1144	1705	} /* function SyndicatedPost::store () */
1145		-
1146		- function insert_new () {
1147		- global $wpdb, $wp_db_version;
1148	1706
	1707	+ function insert_post ($update = false, $freshness = 2) {
	1708	+ global $wpdb;
	1709	+
1149	1710	$dbpost = $this->normalize_post(/*new=*/ true);
	1711	+
	1712	+ $ret = null;
	1713	+
1150	1714	if (!is_null($dbpost)) :
1151		- if ($this->use_api('wp_insert_post')) :
1152		- $dbpost['post_pingback'] = false; // Tell WP 2.1 and 2.2 not to process for pingbacks
1153		-
1154		- // This is a ridiculous fucking kludge necessitated by WordPress 2.6 munging authorship meta-data
1155		- add_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
	1715	+ $dbpost['post_pingback'] = false; // Tell WP 2.1 and 2.2 not to process for pingbacks
	1716	+
	1717	+ // This is a ridiculous fucking kludge necessitated by WordPress 2.6 munging authorship meta-data
	1718	+ add_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
	1719	+
	1720	+ // Kludge to prevent kses filters from stripping the
	1721	+ // content of posts when updating without a logged in
	1722	+ // user who has `unfiltered_html` capability.
	1723	+ $mungers = array('wp_filter_kses', 'wp_filter_post_kses');
	1724	+ $removed = array();
	1725	+ foreach ($mungers as $munger) :
	1726	+ if (has_filter('content_save_pre', $munger)) :
	1727	+ remove_filter('content_save_pre', $munger);
	1728	+ $removed[] = $munger;
	1729	+ endif;
	1730	+ endforeach;
	1731	+
	1732	+ if ($update and function_exists('get_post_field')) :
	1733	+ // Don't munge status fields that the user may
	1734	+ // have reset manually
	1735	+ $doNotMunge = array('post_status', 'comment_status', 'ping_status');
	1736	+
	1737	+ foreach ($doNotMunge as $field) :
	1738	+ $dbpost[$field] = get_post_field($field, $this->wp_id());
	1739	+ endforeach;
	1740	+ endif;
	1741	+
	1742	+ // WP3's wp_insert_post scans current_user_can() for the
	1743	+ // tax_input, with no apparent way to override. Ugh.
	1744	+ add_action(
	1745	+ /*hook=*/ 'transition_post_status',
	1746	+ /*callback=*/ array($this, 'add_terms'),
	1747	+ /*priority=*/ -10001, /* very early */
	1748	+ /*arguments=*/ 3
	1749	+ );
	1750	+
	1751	+ // WP3 appears to override whatever you give it for
	1752	+ // post_modified. Ugh.
	1753	+ add_action(
	1754	+ /*hook=*/ 'transition_post_status',
	1755	+ /*callback=*/ array($this, 'fix_post_modified_ts'),
	1756	+ /*priority=*/ -10000, /* very early */
	1757	+ /*arguments=*/ 3
	1758	+ );
	1759	+
	1760	+ if ($update) :
	1761	+ $this->post['ID'] = $this->wp_id();
	1762	+ $dbpost['ID'] = $this->post['ID'];
	1763	+ endif;
	1764	+
	1765	+ // O.K., is this a new post? If so, we need to create
	1766	+ // the basic post record before we do anything else.
	1767	+ if ($this->this_revision_needs_original_post()) :
	1768	+ // *sigh*, for handling inconsistent slash expectations < 3.6
	1769	+ $sdbpost = $this->db_sanitize_post($dbpost);
1156	1770
1157		- // Kludge to prevent kses filters from stripping the
1158		- // content of posts when updating without a logged in
1159		- // user who has `unfiltered_html` capability.
1160		- add_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11);
	1771	+ // Go ahead and insert the first post record to
	1772	+ // anchor the revision history.
	1773	+
	1774	+ $this->_wp_id = wp_insert_post($sdbpost, /*return wp_error=*/ true);
1161	1775
1162		- $this->_wp_id = wp_insert_post($dbpost);
1163		-
1164		- // Turn off ridiculous fucking kludges #1 and #2
1165		- remove_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
1166		- remove_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11);
1167		-
1168		- $this->validate_post_id($dbpost, array(__CLASS__, __FUNCTION__));
1169		-
1170		- // Unfortunately, as of WordPress 2.3, wp_insert_post()
1171		- // *still* offers no way to use a guid of your choice,
1172		- // and munges your post modified timestamp, too.
1173		- $result = $wpdb->query("
1174		- UPDATE $wpdb->posts
1175		- SET
1176		- guid='{$dbpost['guid']}',
1177		- post_modified='{$dbpost['post_modified']}',
1178		- post_modified_gmt='{$dbpost['post_modified_gmt']}'
1179		- WHERE ID='{$this->_wp_id}'
1180		- ");
1181		- else :
1182		- # The right way to do this is the above. But, alas,
1183		- # in earlier versions of WordPress, wp_insert_post has
1184		- # too much behavior (mainly related to pings) that can't
1185		- # be overridden. In WordPress 1.5, it's enough of a
1186		- # resource hog to make PHP segfault after inserting
1187		- # 50-100 posts. This can get pretty annoying, especially
1188		- # if you are trying to update your feeds for the first
1189		- # time.
1190		-
1191		- $result = $wpdb->query("
1192		- INSERT INTO $wpdb->posts
1193		- SET
1194		- guid = '{$dbpost['guid']}',
1195		- post_author = '{$dbpost['post_author']}',
1196		- post_date = '{$dbpost['post_date']}',
1197		- post_date_gmt = '{$dbpost['post_date_gmt']}',
1198		- post_content = '{$dbpost['post_content']}',"
1199		- .(isset($dbpost['post_excerpt']) ? "post_excerpt = '{$dbpost['post_excerpt']}'," : "")."
1200		- post_title = '{$dbpost['post_title']}',
1201		- post_name = '{$dbpost['post_name']}',
1202		- post_modified = '{$dbpost['post_modified']}',
1203		- post_modified_gmt = '{$dbpost['post_modified_gmt']}',
1204		- comment_status = '{$dbpost['comment_status']}',
1205		- ping_status = '{$dbpost['ping_status']}',
1206		- post_status = '{$dbpost['post_status']}'
1207		- ");
1208		- $this->_wp_id = $wpdb->insert_id;
1209		-
1210		- $this->validate_post_id($dbpost, array(__CLASS__, __FUNCTION__));
1211		-
1212		- // WordPress 1.5.x - 2.0.x
1213		- wp_set_post_cats('1', $this->wp_id(), $this->post['post_category']);
1214		-
1215		- // Since we are not going through official channels, we need to
1216		- // manually tell WordPress that we've published a new post.
1217		- // We need to make sure to do this in order for FeedWordPress
1218		- // to play well with the staticize-reloaded plugin (something
1219		- // that a large aggregator website is going to *want* to be
1220		- // able to use).
1221		- do_action('publish_post', $this->_wp_id);
	1776	+ $dbpost['ID'] = $this->_wp_id;
1222	1777	endif;
1223		- endif;
1224		- } /* SyndicatedPost::insert_new() */
	1778	+
	1779	+ // Sanity check: if the attempt to insert post
	1780	+ // returned an error, then feeding that error
	1781	+ // object in to _wp_put_post_revision() would
	1782	+ // cause a fatal error. Better to break out.
	1783	+ if (!is_wp_error($this->_wp_id)) :
	1784	+ // Now that we've made sure the original exists, insert
	1785	+ // this version here as a revision.
	1786	+ $revision_id = _wp_put_post_revision($dbpost, /*autosave=*/ false);
1225	1787
1226		- function update_existing () {
1227		- global $wpdb;
	1788	+ if (!$this->this_revision_needs_original_post()) :
	1789	+
	1790	+ if ($this->this_revision_is_current()) :
1228	1791
1229		- // Why the fuck doesn't wp_insert_post already do this?
1230		- $dbpost = $this->normalize_post(/*new=*/ false);
1231		- if (!is_null($dbpost)) :
1232		- if ($this->use_api('wp_insert_post')) :
1233		- $dbpost['post_pingback'] = false; // Tell WP 2.1 and 2.2 not to process for pingbacks
1234		-
1235		- // This is a ridiculous fucking kludge necessitated by WordPress 2.6 munging authorship meta-data
1236		- add_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
1237		-
1238		- // Kludge to prevent kses filters from stripping the
1239		- // content of posts when updating without a logged in
1240		- // user who has `unfiltered_html` capability.
1241		- add_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11);
	1792	+ wp_restore_post_revision($revision_id);
1242	1793
1243		- // Don't munge status fields that the user may have reset manually
1244		- if (function_exists('get_post_field')) :
1245		- $doNotMunge = array('post_status', 'comment_status', 'ping_status');
1246		- foreach ($doNotMunge as $field) :
1247		- $dbpost[$field] = get_post_field($field, $this->wp_id());
1248		- endforeach;
1249		- endif;
	1794	+ else :
1250	1795
1251		- $this->_wp_id = wp_insert_post($dbpost);
1252		-
1253		- // Turn off ridiculous fucking kludges #1 and #2
1254		- remove_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
1255		- remove_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11);
1256		-
1257		- $this->validate_post_id($dbpost, array(__CLASS__, __FUNCTION__));
1258		-
1259		- // Unfortunately, as of WordPress 2.3, wp_insert_post()
1260		- // munges your post modified timestamp.
1261		- $result = $wpdb->query("
1262		- UPDATE $wpdb->posts
1263		- SET
1264		- post_modified='{$dbpost['post_modified']}',
1265		- post_modified_gmt='{$dbpost['post_modified_gmt']}'
1266		- WHERE ID='{$this->_wp_id}'
1267		- ");
1268		- else :
1269		-
1270		- $result = $wpdb->query("
1271		- UPDATE $wpdb->posts
1272		- SET
1273		- post_author = '{$dbpost['post_author']}',
1274		- post_content = '{$dbpost['post_content']}',"
1275		- .(isset($dbpost['post_excerpt']) ? "post_excerpt = '{$dbpost['post_excerpt']}'," : "")."
1276		- post_title = '{$dbpost['post_title']}',
1277		- post_name = '{$dbpost['post_name']}',
1278		- post_modified = '{$dbpost['post_modified']}',
1279		- post_modified_gmt = '{$dbpost['post_modified_gmt']}'
1280		- WHERE guid='{$dbpost['guid']}'
1281		- ");
1282		-
1283		- // WordPress 2.1.x and up
1284		- if (function_exists('wp_set_post_categories')) :
1285		- wp_set_post_categories($this->wp_id(), $this->post['post_category']);
1286		- // WordPress 1.5.x - 2.0.x
1287		- elseif (function_exists('wp_set_post_cats')) :
1288		- wp_set_post_cats('1', $this->wp_id(), $this->post['post_category']);
1289		- // This should never happen.
1290		- else :
1291		- FeedWordPress::critical_bug(__CLASS__.'::'.__FUNCTION.'(): no post categorizing function', array("dbpost" => $dbpost, "this" => $this), __LINE__);
	1796	+ // If we do not activate this revision, then the
	1797	+ // add_rss_meta will not be called, which is
	1798	+ // more or less as it should be, but that means
	1799	+ // we have to actively record this revision's
	1800	+ // update hash from here.
	1801	+ $postId = $this->post['ID'];
	1802	+ $key = 'syndication_item_hash';
	1803	+ $hash = $this->update_hash();
	1804	+ FeedWordPress::diagnostic('syndicated_posts:meta_data', "Adding post meta-datum to post [$postId]: [$key] = ".FeedWordPress::val($hash, /*no newlines=*/ true));
	1805	+ add_post_meta( $postId, $key, $hash, /*unique=*/ false );
	1806	+ endif;
1292	1807	endif;
1293		-
1294		- // Since we are not going through official channels, we need to
1295		- // manually tell WordPress that we've published a new post.
1296		- // We need to make sure to do this in order for FeedWordPress
1297		- // to play well with the staticize-reloaded plugin (something
1298		- // that a large aggregator website is going to *want* to be
1299		- // able to use).
1300		- do_action('edit_post', $this->post['ID']);
1301	1808	endif;
	1809	+
	1810	+ remove_action(
	1811	+ /*hook=*/ 'transition_post_status',
	1812	+ /*callback=*/ array($this, 'add_terms'),
	1813	+ /*priority=*/ -10001, /* very early */
	1814	+ /*arguments=*/ 3
	1815	+ );
	1816	+
	1817	+ remove_action(
	1818	+ /*hook=*/ 'transition_post_status',
	1819	+ /*callback=*/ array($this, 'fix_post_modified_ts'),
	1820	+ /*priority=*/ -10000, /* very early */
	1821	+ /*arguments=*/ 3
	1822	+ );
	1823	+
	1824	+ // Turn off ridiculous fucking kludges #1 and #2
	1825	+ remove_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
	1826	+ foreach ($removed as $filter) :
	1827	+ add_filter('content_save_pre', $filter);
	1828	+ endforeach;
	1829	+
	1830	+ $this->validate_post_id($dbpost, $update, array(__CLASS__, __FUNCTION__));
	1831	+
	1832	+ $ret = $this->_wp_id;
1302	1833	endif;
	1834	+ return $ret;
	1835	+ } /* function SyndicatedPost::insert_post () */
	1836	+
	1837	+ /**
	1838	+ * SyndicatedPost::insert_new(). Uses the data collected in this post object to insert
	1839	+ * a new post into the wp_posts table.
	1840	+ *
	1841	+ * @uses SyndicatedPost::insert_post
	1842	+ */
	1843	+ function insert_new () {
	1844	+ $this->insert_post(/*update=*/ false, 1);
	1845	+ } /* SyndicatedPost::insert_new() */
	1846	+
	1847	+ /**
	1848	+ * SyndicatedPost::insert_new(). Uses the data collected in this post object to update
	1849	+ * an existing post in the wp_posts table.
	1850	+ *
	1851	+ * @uses SyndicatedPost::insert_post
	1852	+ */
	1853	+ function update_existing () {
	1854	+ $this->insert_post(/*update=*/ true, 2);
1303	1855	} /* SyndicatedPost::update_existing() */
1304	1856
1305	1857	/**
1306	1858	* SyndicatedPost::normalize_post()
		@@ -1310,33 +1862,104 @@
1310	1862	*/
1311	1863	function normalize_post ($new = true) {
1312	1864	global $wpdb;
1313	1865
1314		- $out = array();
	1866	+ $out = $this->post;
1315	1867
1316		- // Why the fuck doesn't wp_insert_post already do this?
1317		- foreach ($this->post as $key => $value) :
1318		- if (is_string($value)) :
1319		- $out[$key] = $wpdb->escape($value);
1320		- else :
1321		- $out[$key] = $value;
1322		- endif;
1323		- endforeach;
1324		-
1325		- if (strlen($out['post_title'].$out['post_content'].$out['post_excerpt']) == 0) :
	1868	+ $fullPost = $out['post_title'].$out['post_content'];
	1869	+ $fullPost .= (isset($out['post_excerpt']) ? $out['post_excerpt'] : '');
	1870	+ if (strlen($fullPost) < 1) :
1326	1871	// FIXME: Option for filtering out empty posts
1327	1872	endif;
1328	1873	if (strlen($out['post_title'])==0) :
1329	1874	$offset = (int) get_option('gmt_offset') * 60 * 60;
1330		- $out['post_title'] =
1331		- $this->post['meta']['syndication_source']
	1875	+ if (isset($this->post['meta']['syndication_source'])) :
	1876	+ $source_title = $this->post['meta']['syndication_source'];
	1877	+ else :
	1878	+ $feed_url = parse_url($this->post['meta']['syndication_feed']);
	1879	+ $source_title = $feed_url['host'];
	1880	+ endif;
	1881	+
	1882	+ $out['post_title'] = $source_title
1332	1883	.' '.gmdate('Y-m-d H:i:s', $this->published() + $offset);
1333	1884	// FIXME: Option for what to fill a blank title with...
1334	1885	endif;
1335	1886
	1887	+ // Normalize the guid if necessary.
	1888	+ $out['guid'] = SyndicatedPost::normalize_guid($out['guid']);
	1889	+
1336	1890	return $out;
1337	1891	}
1338	1892
	1893	+ public function db_sanitize_post_check_encoding ($out) {
	1894	+ // Check encoding recursively: every string field needs to be checked
	1895	+ // for character encoding issues. This is a bit problematic because we
	1896	+ // *should* be using DB_CHARSET, but DB_CHARSET sometimes has values
	1897	+ // that work for MySQL but not for PHP mb_check_encoding. So instead
	1898	+ // we must rely on WordPress setting blog_charset and hope that the user
	1899	+ // has got their database encoding set up to roughly match
	1900	+ $charset = get_option('blog_charset', 'utf8');
	1901	+
	1902	+ foreach ($out as $key => $value) :
	1903	+ if (is_string($value)) :
	1904	+
	1905	+ if (!function_exists('mb_check_encoding') or mb_check_encoding($value, $charset)) :
	1906	+ $out[$key] = $value;
	1907	+ else :
	1908	+ $fromCharset = mb_detect_encoding($value, mb_detect_order(), /*strict=*/ true);
	1909	+ $out[$key] = mb_convert_encoding($value, $charset, $fromCharset);
	1910	+ endif;
	1911	+
	1912	+ elseif (is_array($value)) :
	1913	+ $out[$key] = $this->db_sanitize_post_check_encoding($value);
	1914	+
	1915	+ else :
	1916	+ $out[$key] = $value;
	1917	+ endif;
	1918	+
	1919	+ endforeach;
	1920	+
	1921	+ return $out;
	1922	+ } /* SyndicatedPost::db_sanitize_post_check_encoding () */
	1923	+
	1924	+ function db_sanitize_post ($out) {
	1925	+ global $wp_db_version;
	1926	+
	1927	+ $out = $this->db_sanitize_post_check_encoding($out);
	1928	+
	1929	+ // < 3.6. Core API, including `wp_insert_post()`, expects
	1930	+ // properly slashed data. If `wp_slash()` exists, then
	1931	+ // this is after the big change-over in how data slashing
	1932	+ // was handled.
	1933	+ if (!function_exists('wp_slash')) :
	1934	+
	1935	+ foreach ($out as $key => $value) :
	1936	+ if (is_string($value)) :
	1937	+ $out[$key] = esc_sql($value);
	1938	+ else :
	1939	+ $out[$key] = $value;
	1940	+ endif;
	1941	+ endforeach;
	1942	+
	1943	+ // For revisions [@23416,@23554), core API expects
	1944	+ // unslashed data. Cf. <https://core.trac.wordpress.org/browser/trunk/wp-includes/post.php?rev=23416>
	1945	+ // NOOP for those revisions.
	1946	+
	1947	+ // In revisions @23554 to present, `wp_insert_post()`
	1948	+ // expects slashed data once again.
	1949	+ // Cf. <https://core.trac.wordpress.org/changeset/23554/trunk/wp-includes/post.php?contextall=1>
	1950	+ // But at least now we can use the wp_slash API function to do that.
	1951	+ // Hooray.
	1952	+
	1953	+ elseif ($wp_db_version >= 23524) :
	1954	+
	1955	+ $out = wp_slash($out);
	1956	+
	1957	+ endif;
	1958	+
	1959	+ return $out;
	1960	+ }
	1961	+
1339	1962	/**
1340	1963	* SyndicatedPost::validate_post_id()
1341	1964	*
1342	1965	* @param array $dbpost An array representing the post we attempted to insert or update
		@@ -1341,35 +1964,57 @@
1341	1964	*
1342	1965	* @param array $dbpost An array representing the post we attempted to insert or update
1343	1966	* @param mixed $ns A string or array representing the namespace (class, method) whence this method was called.
1344	1967	*/
1345		- function validate_post_id ($dbpost, $ns) {
	1968	+ function validate_post_id ($dbpost, $is_update, $ns) {
1346	1969	if (is_array($ns)) : $ns = implode('::', $ns);
1347	1970	else : $ns = (string) $ns; endif;
1348		-
	1971	+
1349	1972	// This should never happen.
1350	1973	if (!is_numeric($this->_wp_id) or ($this->_wp_id == 0)) :
1351		- FeedWordPress::critical_bug(
1352		- /*name=*/ $ns.'::_wp_id',
	1974	+ $verb = ($is_update ? 'update existing' : 'insert new');
	1975	+ $guid = $this->guid();
	1976	+ $url = $this->permalink();
	1977	+ $feed = $this->link->uri(array('add_params' => true));
	1978	+
	1979	+ // wp_insert_post failed. Diagnostics, or barf up a critical bug
	1980	+ // notice if we are in debug mode.
	1981	+ $mesg = "Failed to $verb item [$guid]. WordPress API returned no valid post ID.\n"
	1982	+ ."\t\tID = ".serialize($this->_wp_id)."\n"
	1983	+ ."\t\tURL = ".MyPHP::val($url)
	1984	+ ."\t\tFeed = ".MyPHP::val($feed);
	1985	+
	1986	+ FeedWordPress::diagnostic('updated_feeds:errors', "WordPress API error: $mesg");
	1987	+ FeedWordPress::diagnostic('feed_items:rejected', $mesg);
	1988	+
	1989	+ $mesg = <<<EOM
	1990	+The WordPress API returned an invalid post ID
	1991	+ when FeedWordPress tried to $verb item $guid
	1992	+ [URL: $url]
	1993	+ from the feed at $feed
	1994	+
	1995	+$ns::_wp_id
	1996	+EOM;
	1997	+ FeedWordPressDiagnostic::noncritical_bug(
	1998	+ /*message=*/ $mesg,
1353	1999	/*var =*/ array(
1354	2000	"\$this->_wp_id" => $this->_wp_id,
1355	2001	"\$dbpost" => $dbpost,
1356		- "\$this" => $this
1357	2002	),
1358		- /*line # =*/ __LINE__
	2003	+ /*line # =*/ __LINE__, /*filename=*/ __FILE__
1359	2004	);
1360	2005	endif;
1361	2006	} /* SyndicatedPost::validate_post_id() */
1362		-
	2007	+
1363	2008	/**
1364		- * SyndicatedPost::fix_revision_meta() - Fixes the way WP 2.6+ fucks up
1365		- * meta-data (authorship, etc.) when storing revisions of an updated
1366		- * syndicated post.
	2009	+ * SyndicatedPost::fix_revision_meta() - Ensures that we get the meta
	2010	+ * data (authorship, guid, etc.) that we want when storing revisions of
	2011	+ * a syndicated post.
1367	2012	*
1368		- * In their infinite wisdom, the WordPress coders have made it completely
1369		- * impossible for a plugin that uses wp_insert_post() to set certain
1370		- * meta-data (such as the author) when you store an old revision of an
1371		- * updated post. Instead, it uses the WordPress defaults (= currently
	2013	+ * In their infinite wisdom, the WordPress coders seem to have made it
	2014	+ * completely impossible for a plugin that uses wp_insert_post() to set
	2015	+ * certain meta-data (such as the author) when you store an old revision
	2016	+ * of an updated post. Instead, it uses the WordPress defaults (= cur.
1372	2017	* active user ID if the process is running with a user logged in, or
1373	2018	* = #0 if there is no user logged in). This results in bogus authorship
1374	2019	* data for revisions that are syndicated from off the feed, unless we
1375	2020	* use a ridiculous kludge like this to end-run the munging of meta-data
		@@ -1378,97 +2023,204 @@
1378	2023	* @param int $revision_id The revision ID to fix up meta-data
1379	2024	*/
1380	2025	function fix_revision_meta ($revision_id) {
1381	2026	global $wpdb;
	2027	+
	2028	+ $post_author = (int) $this->post['post_author'];
	2029	+
	2030	+ $revision_id = (int) $revision_id;
1382	2031
1383		- $post_author = (int) $this->post['post_author'];
	2032	+ // Let's fix the author.
	2033	+ set_post_field('post_author', $this->post['post_author'], $revision_id);
1384	2034
1385		- $revision_id = (int) $revision_id;
1386		- $wpdb->query("
1387		- UPDATE $wpdb->posts
1388		- SET post_author={$this->post['post_author']}
1389		- WHERE post_type = 'revision' AND ID='$revision_id'
1390		- ");
	2035	+ // Let's fix the GUID to a dummy URL with the update hash.
	2036	+ set_post_field('guid', 'http://feedwordpress.radgeek.com/?rev='.$this->update_hash(), $revision_id);
	2037	+
	2038	+ // Let's fire an event for add-ons and filters
	2039	+ do_action('syndicated_post_fix_revision_meta', $revision_id, $this);
	2040	+
1391	2041	} /* SyndicatedPost::fix_revision_meta () */
1392	2042
1393	2043	/**
1394		- * SyndicatedPost::avoid_kses_munge() -- If FeedWordPress is processing
1395		- * an automatic update, that generally means that wp_insert_post() is
1396		- * being called under the user credentials of whoever is viewing the
1397		- * blog at the time -- usually meaning no user at all. But if WordPress
1398		- * gets a wp_insert_post() when current_user_can('unfiltered_html') is
1399		- * false, it will run the content of the post through a kses function
1400		- * that strips out lots of HTML tags -- notably <object> and some others.
1401		- * This causes problems for syndicating (for example) feeds that contain
1402		- * YouTube videos. It also produces an unexpected asymmetry between
1403		- * automatically-initiated updates and updates initiated manually from
1404		- * the WordPress Dashboard (which are usually initiated under the
1405		- * credentials of a logged-in admin, and so don't get run through the
1406		- * kses function). So, to avoid the whole mess, what we do here is
1407		- * just forcibly disable the kses munging for a single syndicated post,
1408		- * by restoring the contents of the `post_content` field.
	2044	+ * SyndicatedPost::add_terms() -- if FeedWordPress is processing an
	2045	+ * automatic update, that generally means that wp_insert_post() is being
	2046	+ * called under the user credentials of whoever is viewing the blog at
	2047	+ * the time -- which usually means no user at all. But wp_insert_post()
	2048	+ * checks current_user_can() before assigning any of the terms in a
	2049	+ * post's tax_input structure -- which is unfortunate, since
	2050	+ * current_user_can() always returns FALSE when there is no current user
	2051	+ * logged in. Meaning that automatic updates get no terms assigned.
1409	2052	*
1410		- * @param string $content The content of the post, after other filters have gotten to it
1411		- * @return string The original content of the post, before other filters had a chance to munge it.
	2053	+ * So, wp_insert_post() is not going to do the term assignments for us.
	2054	+ * If you want something done right....
	2055	+ *
	2056	+ * @param string $new_status Unused action parameter.
	2057	+ * @param string $old_status Unused action parameter.
	2058	+ * @param object $post The database record for the post just inserted.
1412	2059	*/
1413		- function avoid_kses_munge ($content) {
1414		- global $wpdb;
1415		- return $wpdb->escape($this->post['post_content']);
1416		- }
1417		-
1418		- // SyndicatedPost::add_rss_meta: adds interesting meta-data to each entry
1419		- // using the space for custom keys. The set of keys and values to add is
1420		- // specified by the keys and values of $post['meta']. This is used to
1421		- // store anything that the WordPress user might want to access from a
1422		- // template concerning the post's original source that isn't provided
1423		- // for by standard WP meta-data (i.e., any interesting data about the
1424		- // syndicated post other than author, title, timestamp, categories, and
1425		- // guid). It's also used to hook into WordPress's support for
1426		- // enclosures.
	2060	+ function add_terms ($new_status, $old_status, $post) {
	2061	+
	2062	+ if ($new_status!='inherit') : // Bail if we are creating a revision.
	2063	+ if ( is_array($this->post) and isset($this->post['tax_input']) and is_array($this->post['tax_input']) ) :
	2064	+ foreach ($this->post['tax_input'] as $taxonomy => $terms) :
	2065	+ if (is_array($terms)) :
	2066	+ $terms = array_filter($terms); // strip out empties
	2067	+ endif;
	2068	+
	2069	+ $res = wp_set_post_terms(
	2070	+ /*post_id=*/ $post->ID,
	2071	+ /*terms=*/ $terms,
	2072	+ /*taxonomy=*/ $taxonomy
	2073	+ );
	2074	+
	2075	+ FeedWordPress::diagnostic(
	2076	+ 'syndicated_posts:categories',
	2077	+ 'Category: post('.json_encode($post->ID).') '.$taxonomy
	2078	+ .' := '
	2079	+ .json_encode($terms)
	2080	+ .' / result: '
	2081	+ .json_encode($res)
	2082	+ );
	2083	+
	2084	+ endforeach;
	2085	+ endif;
	2086	+ endif;
	2087	+
	2088	+ } /* SyndicatedPost::add_terms () */
	2089	+
	2090	+ /**
	2091	+ * SyndicatedPost::fix_post_modified_ts() -- We would like to set
	2092	+ * post_modified and post_modified_gmt to reflect the value of
	2093	+ * <atom:updated> or equivalent elements on the feed. Unfortunately,
	2094	+ * wp_insert_post() refuses to acknowledge explicitly-set post_modified
	2095	+ * fields and overwrites them, either with the post_date (if new) or the
	2096	+ * current timestamp (if updated).
	2097	+ *
	2098	+ * So, wp_insert_post() is not going to do the last-modified assignments
	2099	+ * for us. If you want something done right....
	2100	+ *
	2101	+ * @param string $new_status Unused action parameter.
	2102	+ * @param string $old_status Unused action parameter.
	2103	+ * @param object $post The database record for the post just inserted.
	2104	+ */
	2105	+ function fix_post_modified_ts ($new_status, $old_status, $post) {
	2106	+ global $wpdb;
	2107	+ if ($new_status!='inherit') : // Bail if we are creating a revision.
	2108	+ $wpdb->update( $wpdb->posts, /*data=*/ array(
	2109	+ 'post_modified' => $this->post['post_modified'],
	2110	+ 'post_modified_gmt' => $this->post['post_modified_gmt'],
	2111	+ ), /*where=*/ array('ID' => $post->ID) );
	2112	+ endif;
	2113	+ } /* SyndicatedPost::fix_post_modified_ts () */
	2114	+
	2115	+ /**
	2116	+ * SyndicatedPost::add_rss_meta: adds interesting meta-data to each entry
	2117	+ * using the space for custom keys. The set of keys and values to add is
	2118	+ * specified by the keys and values of $post['meta']. This is used to
	2119	+ * store anything that the WordPress user might want to access from a
	2120	+ * template concerning the post's original source that isn't provided
	2121	+ * for by standard WP meta-data (i.e., any interesting data about the
	2122	+ * syndicated post other than author, title, timestamp, categories, and
	2123	+ * guid). It's also used to hook into WordPress's support for
	2124	+ * enclosures.
	2125	+ *
	2126	+ * @param string $new_status Unused action parameter.
	2127	+ * @param string $old_status Unused action parameter.
	2128	+ * @param object $post The database record for the post just inserted.
	2129	+ */
1427	2130	function add_rss_meta ($new_status, $old_status, $post) {
1428	2131	global $wpdb;
1429		- if ( is_array($this->post) and isset($this->post['meta']) and is_array($this->post['meta']) ) :
1430		- $postId = $this->wp_id();
1431		-
1432		- // Aggregated posts should NOT send out pingbacks.
1433		- // WordPress 2.1-2.2 claim you can tell them not to
1434		- // using $post_pingback, but they don't listen, so we
1435		- // make sure here.
1436		- $result = $wpdb->query("
1437		- DELETE FROM $wpdb->postmeta
1438		- WHERE post_id='$postId' AND meta_key='_pingme'
1439		- ");
	2132	+ if ($new_status!='inherit') : // Bail if we are creating a revision.
	2133	+ FeedWordPress::diagnostic('syndicated_posts:meta_data', 'Adding post meta-data: {'.implode(", ", array_keys($this->post['meta'])).'}');
1440	2134
1441		- foreach ( $this->post['meta'] as $key => $values ) :
	2135	+ if ( is_array($this->post) and isset($this->post['meta']) and is_array($this->post['meta']) ) :
	2136	+ $postId = $post->ID;
1442	2137
1443		- $eKey = $wpdb->escape($key);
1444		-
1445		- // If this is an update, clear out the old
1446		- // values to avoid duplication.
	2138	+ // Aggregated posts should NOT send out pingbacks.
	2139	+ // WordPress 2.1-2.2 claim you can tell them not to
	2140	+ // using $post_pingback, but they don't listen, so we
	2141	+ // make sure here.
1447	2142	$result = $wpdb->query("
1448	2143	DELETE FROM $wpdb->postmeta
1449		- WHERE post_id='$postId' AND meta_key='$eKey'
	2144	+ WHERE post_id='$postId' AND meta_key='_pingme'
1450	2145	");
1451	2146
1452		- // Allow for either a single value or an array
1453		- if (!is_array($values)) $values = array($values);
1454		- foreach ( $values as $value ) :
1455		- add_post_meta($postId, $key, $value, /*unique=*/ false);
	2147	+ foreach ( $this->post['meta'] as $key => $values ) :
	2148	+ $eKey = esc_sql($key);
	2149	+
	2150	+ // If this is an update, clear out the old
	2151	+ // values to avoid duplication.
	2152	+ $result = $wpdb->query("
	2153	+ DELETE FROM $wpdb->postmeta
	2154	+ WHERE post_id='$postId' AND meta_key='$eKey'
	2155	+ ");
	2156	+
	2157	+ // Allow for either a single value or an array
	2158	+ if (!is_array($values)) $values = array($values);
	2159	+ foreach ( $values as $value ) :
	2160	+ FeedWordPress::diagnostic('syndicated_posts:meta_data', "Adding post meta-datum to post [$postId]: [$key] = ".MyPHP::val($value, /*no newlines=*/ true));
	2161	+ add_post_meta($postId, $key, $value, /*unique=*/ false);
	2162	+ endforeach;
1456	2163	endforeach;
1457		- endforeach;
	2164	+ endif;
1458	2165	endif;
1459	2166	} /* SyndicatedPost::add_rss_meta () */
1460	2167
1461		- // SyndicatedPost::author_id (): get the ID for an author name from
1462		- // the feed. Create the author if necessary.
	2168	+ /**
	2169	+ * SyndicatedPost::author_id (): get the ID for an author name from
	2170	+ * the feed. Create the author if necessary.
	2171	+ *
	2172	+ * @param string $unfamiliar_author
	2173	+ *
	2174	+ * @return NULL|int The numeric ID of the author to attribute the post to
	2175	+ * NULL if the post should be filtered out.
	2176	+ */
1463	2177	function author_id ($unfamiliar_author = 'create') {
1464	2178	global $wpdb;
1465	2179
1466		- $a = $this->author();
1467		- $author = $a['name'];
	2180	+ $a = $this->named['author'];
	2181	+
	2182	+ $source = $this->source();
	2183	+ $forbidden = apply_filters('feedwordpress_forbidden_author_names',
	2184	+ array('admin', 'administrator', 'www', 'root'));
	2185	+
	2186	+ // Prepare the list of candidates to try for author name: name from
	2187	+ // feed, original source title (if any), immediate source title live
	2188	+ // from feed, subscription title, prettied version of feed homepage URL,
	2189	+ // prettied version of feed URL, or, failing all, use "unknown author"
	2190	+ // as last resort
	2191	+
	2192	+ $candidates = array();
	2193	+ $candidates[] = $a['name'];
	2194	+ if (!is_null($source)) : $candidates[] = $source['title']; endif;
	2195	+ $candidates[] = $this->link->name(/*fromFeed=*/ true);
	2196	+ $candidates[] = $this->link->name(/*fromFeed=*/ false);
	2197	+ if (strlen($this->link->homepage()) > 0) : $candidates[] = feedwordpress_display_url($this->link->homepage()); endif;
	2198	+ $candidates[] = feedwordpress_display_url($this->link->uri());
	2199	+ $candidates[] = 'unknown author';
	2200	+
	2201	+ // Pick the first one that works from the list, screening against empty
	2202	+ // or forbidden names.
	2203	+
	2204	+ $author = NULL;
	2205	+ foreach ($candidates as $candidate) {
	2206	+ if (!is_null($candidate)
	2207	+ and (strlen(trim($candidate)) > 0)
	2208	+ and !in_array(strtolower(trim($candidate)), $forbidden)) :
	2209	+ $author = $candidate;
	2210	+ break;
	2211	+ endif;
	2212	+ }
	2213	+
1468	2214	$email = (isset($a['email']) ? $a['email'] : NULL);
1469		- $url = (isset($a['uri']) ? $a['uri'] : NULL);
	2215	+ $authorUrl = (isset($a['uri']) ? $a['uri'] : NULL);
1470	2216
	2217	+
	2218	+ $hostUrl = $this->link->homepage();
	2219	+ if (is_null($hostUrl) or (strlen($hostUrl) < 0)) :
	2220	+ $hostUrl = $this->link->uri();
	2221	+ endif;
	2222	+
1471	2223	$match_author_by_email = !('yes' == get_option("feedwordpress_do_not_match_author_by_email"));
1472	2224	if ($match_author_by_email and !FeedWordPress::is_null_email($email)) :
1473	2225	$test_email = $email;
1474	2226	else :
		@@ -1476,21 +2228,43 @@
1476	2228	endif;
1477	2229
1478	2230	// Never can be too careful...
1479	2231	$login = sanitize_user($author, /*strict=*/ true);
	2232	+
	2233	+ // Possible for, e.g., foreign script author names
	2234	+ if (strlen($login) < 1) :
	2235	+ // No usable characters in author name for a login.
	2236	+ // (Sometimes results from, e.g., foreign scripts.)
	2237	+ //
	2238	+ // We just need *something* in Western alphanumerics,
	2239	+ // so let's try the domain name.
	2240	+ //
	2241	+ // Uniqueness will be guaranteed below if necessary.
	2242	+
	2243	+ $url = parse_url($hostUrl);
	2244	+
	2245	+ $login = sanitize_user($url['host'], /*strict=*/ true);
	2246	+ if (strlen($login) < 1) :
	2247	+ // This isn't working. Frak it.
	2248	+ $login = 'syndicated';
	2249	+ endif;
	2250	+ endif;
	2251	+
1480	2252	$login = apply_filters('pre_user_login', $login);
1481	2253
1482	2254	$nice_author = sanitize_title($author);
1483	2255	$nice_author = apply_filters('pre_user_nicename', $nice_author);
1484	2256
1485		- $reg_author = $wpdb->escape(preg_quote($author));
1486		- $author = $wpdb->escape($author);
1487		- $email = $wpdb->escape($email);
1488		- $test_email = $wpdb->escape($test_email);
1489		- $url = $wpdb->escape($url);
	2257	+ $reg_author = esc_sql(preg_quote($author));
	2258	+ $author = esc_sql($author);
	2259	+ $email = esc_sql($email);
	2260	+ $test_email = esc_sql($test_email);
	2261	+ $authorUrl = esc_sql($authorUrl);
1490	2262
1491	2263	// Check for an existing author rule....
1492		- if (isset($this->link->settings['map authors']['name'][strtolower(trim($author))])) :
	2264	+ if (isset($this->link->settings['map authors']['name']['*'])) :
	2265	+ $author_rule = $this->link->settings['map authors']['name']['*'];
	2266	+ elseif (isset($this->link->settings['map authors']['name'][strtolower(trim($author))])) :
1493	2267	$author_rule = $this->link->settings['map authors']['name'][strtolower(trim($author))];
1494	2268	else :
1495	2269	$author_rule = NULL;
1496	2270	endif;
		@@ -1501,9 +2275,9 @@
1501	2275
1502	2276	// User name is filtered out
1503	2277	elseif ('filter' == $author_rule) :
1504	2278	$id = NULL;
1505		-
	2279	+
1506	2280	else :
1507	2281	// Check the database for an existing author record that might fit
1508	2282
1509	2283	// First try the user core data table.
		@@ -1508,17 +2282,15 @@
1508	2282
1509	2283	// First try the user core data table.
1510	2284	$id = $wpdb->get_var(
1511	2285	"SELECT ID FROM $wpdb->users
1512		- WHERE
1513		- TRIM(LCASE(user_login)) = TRIM(LCASE('$login'))
1514		- OR (
1515		- LENGTH(TRIM(LCASE(user_email))) > 0
1516		- AND TRIM(LCASE(user_email)) = TRIM(LCASE('$test_email'))
1517		- )
1518		- OR TRIM(LCASE(user_nicename)) = TRIM(LCASE('$nice_author'))
1519		- ");
1520		-
	2286	+ WHERE TRIM(LCASE(display_name)) = TRIM(LCASE('$author'))
	2287	+ OR TRIM(LCASE(user_login)) = TRIM(LCASE('$author'))
	2288	+ OR (
	2289	+ LENGTH(TRIM(LCASE(user_email))) > 0
	2290	+ AND TRIM(LCASE(user_email)) = TRIM(LCASE('$test_email'))
	2291	+ )");
	2292	+
1521	2293	// If that fails, look for aliases in the user meta data table
1522	2294	if (is_null($id)) :
1523	2295	$id = $wpdb->get_var(
1524	2296	"SELECT user_id FROM $wpdb->usermeta
		@@ -1540,13 +2312,13 @@
1540	2312	if (is_null($id)) :
1541	2313	if ($unfamiliar_author === 'create') :
1542	2314	$userdata = array();
1543	2315
1544		- // WordPress 3 is going to pitch a fit if we attempt to register
1545		- // more than one user account with an empty e-mail address, so we
1546		- // need *something* here. Ugh.
	2316	+ #-- we need *something* for the email here or WordPress
	2317	+ #-- is liable to pitch a fit. So, make something up if
	2318	+ #-- necessary. (Ugh.)
1547	2319	if (strlen($email) == 0 or FeedWordPress::is_null_email($email)) :
1548		- $url = parse_url($this->feed->channel['link']);
	2320	+ $url = parse_url($hostUrl);
1549	2321	$email = $nice_author.'@'.$url['host'];
1550	2322	endif;
1551	2323
1552	2324	#-- user table data
		@@ -1554,12 +2326,79 @@
1554	2326	$userdata['user_login'] = $login;
1555	2327	$userdata['user_nicename'] = $nice_author;
1556	2328	$userdata['user_pass'] = substr(md5(uniqid(microtime())), 0, 6); // just something random to lock it up
1557	2329	$userdata['user_email'] = $email;
1558		- $userdata['user_url'] = $url;
	2330	+ $userdata['user_url'] = $authorUrl;
	2331	+ $userdata['nickname'] = $author;
	2332	+
	2333	+ $parts = preg_split('/\s+/', trim($author), 2);
	2334	+ if (isset($parts[0])) : $userdata['first_name'] = $parts[0]; endif;
	2335	+ if (isset($parts[1])) : $userdata['last_name'] = $parts[1]; endif;
	2336	+
1559	2337	$userdata['display_name'] = $author;
	2338	+ $userdata['role'] = 'contributor';
1560	2339
1561		- $id = wp_insert_user($userdata);
	2340	+ #-- loop. Keep trying to add the user until you get it
	2341	+ #-- right. Or until PHP crashes, I guess.
	2342	+ $insanity = 0;
	2343	+ do {
	2344	+ $id = wp_insert_user($userdata);
	2345	+ if (is_wp_error($id)) :
	2346	+ $codes = $id->get_error_code();
	2347	+ switch ($codes) :
	2348	+ case 'empty_user_login' :
	2349	+ case 'existing_user_login' :
	2350	+ case 'invalid_username' :
	2351	+ // Add a random disambiguator
	2352	+ $userdata['user_login'] .= substr(md5(uniqid(microtime())), 0, 6);
	2353	+ break;
	2354	+ case 'user_login_too_long' :
	2355	+ // Limit length to 53 characters; if we end up needing a random disambiguator,
	2356	+ // we should still have space to add it.
	2357	+ $userdata['user_login'] = mb_substr( $userdata['user_login'], 0, 53 );
	2358	+ break;
	2359	+ case 'user_nicename_too_long' :
	2360	+ // Add a limited 50 characters user_nicename based on user_login
	2361	+ $userdata['user_nicename'] = mb_substr( $userdata['user_login'], 0, 50 );
	2362	+ break;
	2363	+ case 'existing_user_email' :
	2364	+ // Disassemble email for username, host
	2365	+ $parts = explode('@', $userdata['user_email'], 2);
	2366	+
	2367	+ // Add a random disambiguator as a gmail-style username extension
	2368	+ $parts[0] .= '+'.substr(md5(uniqid(microtime())), 0, 6);
	2369	+
	2370	+ // Reassemble
	2371	+ $userdata['user_email'] = $parts[0].'@'.$parts[1];
	2372	+ break;
	2373	+ default :
	2374	+ if ( $insanity > 10 ) :
	2375	+ // Try some settings that are unlikely to cause complaint...
	2376	+ $url = parse_url($hostUrl);
	2377	+
	2378	+ $userdata['user_login'] = substr(md5(uniqid(microtime())), 0, 6);
	2379	+ $userdata['user_nicename'] = $userdata['user_login'];
	2380	+ $userdata['user_email'] = 'noreply@' . $url['host'];
	2381	+ elseif ( $insanity > 50 ) :
	2382	+ // Stop doing the same thing and expecting a different result
	2383	+ break;
	2384	+ endif;
	2385	+ endswitch;
	2386	+ endif;
	2387	+ $insanity = $insanity + 1;
	2388	+ } while (is_wp_error($id));
	2389	+
	2390	+ // $id should now contain the numeric ID of a newly minted
	2391	+ // user account. Let's mark them as having been generated
	2392	+ // by FeedWordPress in the usermeta table, as per the
	2393	+ // suggestion of @boonebgorges, in case we need to process,
	2394	+ // winnow, filter, or merge syndicated author accounts, &c.
	2395	+ if (!is_wp_error($id)) :
	2396	+ add_user_meta($id, 'feedwordpress_generated', 1);
	2397	+ else :
	2398	+ $id = null;
	2399	+ endif;
	2400	+
1562	2401	elseif (is_numeric($unfamiliar_author) and get_userdata((int) $unfamiliar_author)) :
1563	2402	$id = (int) $unfamiliar_author;
1564	2403	elseif ($unfamiliar_author === 'default') :
1565	2404	$id = 1;
		@@ -1568,127 +2407,30 @@
1568	2407	endif;
1569	2408
1570	2409	if ($id) :
1571	2410	$this->link->settings['map authors']['name'][strtolower(trim($author))] = $id;
1572		- endif;
1573		- return $id;
1574		- } // function SyndicatedPost::author_id ()
1575	2411
1576		- // look up (and create) category ids from a list of categories
1577		- function category_ids ($cats, $unfamiliar_category = 'create', $tags_too = false) {
1578		- global $wpdb;
1579		-
1580		- // We need to normalize whitespace because (1) trailing
1581		- // whitespace can cause PHP and MySQL not to see eye to eye on
1582		- // VARCHAR comparisons for some versions of MySQL (cf.
1583		- // <http://dev.mysql.com/doc/mysql/en/char.html>), and (2)
1584		- // because I doubt most people want to make a semantic
1585		- // distinction between 'Computers' and 'Computers '
1586		- $cats = array_map('trim', $cats);
1587		-
1588		- $tags = array();
1589		-
1590		- $cat_ids = array ();
1591		- foreach ($cats as $cat_name) :
1592		- if (preg_match('/^{#([0-9]+)}$/', $cat_name, $backref)) :
1593		- $cat_id = (int) $backref[1];
1594		- if (function_exists('is_term') and is_term($cat_id, 'category')) :
1595		- $cat_ids[] = $cat_id;
1596		- elseif (get_category($cat_id)) :
1597		- $cat_ids[] = $cat_id;
1598		- endif;
1599		- elseif (strlen($cat_name) > 0) :
1600		- $esc = $wpdb->escape($cat_name);
1601		- $resc = $wpdb->escape(preg_quote($cat_name));
1602		-
1603		- // WordPress 2.3+
1604		- if (function_exists('is_term')) :
1605		- $cat_id = is_term($cat_name, 'category');
1606		- if ($cat_id) :
1607		- $cat_ids[] = $cat_id['term_id'];
1608		- // There must be a better way to do this...
1609		- elseif ($results = $wpdb->get_results(
1610		- "SELECT term_id
1611		- FROM $wpdb->term_taxonomy
1612		- WHERE
1613		- LOWER(description) RLIKE
1614		- CONCAT('(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*', LOWER('{$resc}'), '( |\\t|\\r)*(\\n|\$)')"
1615		- )) :
1616		- foreach ($results AS $term) :
1617		- $cat_ids[] = (int) $term->term_id;
1618		- endforeach;
1619		- elseif ('tag'==$unfamiliar_category) :
1620		- $tags[] = $cat_name;
1621		- elseif ('create'===$unfamiliar_category) :
1622		- $term = wp_insert_term($cat_name, 'category');
1623		- if (is_wp_error($term)) :
1624		- FeedWordPress::noncritical_bug('term insertion problem', array('cat_name' => $cat_name, 'term' => $term, 'this' => $this), __LINE__);
1625		- else :
1626		- $cat_ids[] = $term['term_id'];
1627		- endif;
1628		- endif;
1629		-
1630		- // WordPress 1.5.x - 2.2.x
1631		- else :
1632		- $results = $wpdb->get_results(
1633		- "SELECT cat_ID
1634		- FROM $wpdb->categories
1635		- WHERE
1636		- (LOWER(cat_name) = LOWER('$esc'))
1637		- OR (LOWER(category_description)
1638		- RLIKE CONCAT('(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*', LOWER('{$resc}'), '( |\\t|\\r)*(\\n|\$)'))
1639		- ");
1640		- if ($results) :
1641		- foreach ($results as $term) :
1642		- $cat_ids[] = (int) $term->cat_ID;
1643		- endforeach;
1644		- elseif ('create'===$unfamiliar_category) :
1645		- if (function_exists('wp_insert_category')) :
1646		- $cat_id = wp_insert_category(array('cat_name' => $esc));
1647		- // And into the database we go.
1648		- else :
1649		- $nice_kitty = sanitize_title($cat_name);
1650		- $wpdb->query(sprintf("
1651		- INSERT INTO $wpdb->categories
1652		- SET
1653		- cat_name='%s',
1654		- category_nicename='%s'
1655		- ", $esc, $nice_kitty
1656		- ));
1657		- $cat_id = $wpdb->insert_id;
1658		- endif;
1659		- $cat_ids[] = $cat_id;
1660		- endif;
1661		- endif;
	2412	+ // Multisite: Check whether the author has been recorded
	2413	+ // on *this* blog before. If not, put her down as a
	2414	+ // Contributor for *this* blog.
	2415	+ $user = new WP_User((int) $id);
	2416	+ if (empty($user->roles)) :
	2417	+ $user->add_role('contributor');
1662	2418	endif;
1663		- endforeach;
1664		-
1665		- if ((count($cat_ids) == 0) and ($unfamiliar_category === 'filter')) :
1666		- $cat_ids = NULL; // Drop the post
1667		- else :
1668		- $cat_ids = array_unique($cat_ids);
1669	2419	endif;
1670		-
1671		- if ($tags_too) : $ret = array($cat_ids, $tags);
1672		- else : $ret = $cat_ids;
1673		- endif;
	2420	+ return $id;
	2421	+ } /* function SyndicatedPost::author_id () */
1674	2422
1675		- return $ret;
1676		- } // function SyndicatedPost::category_ids ()
	2423	+ /**
	2424	+ * category_ids: look up (and create) category ids from a list of
	2425	+ * categories
	2426	+ *
	2427	+ * @param array $cats
	2428	+ * @param string $unfamiliar_category
	2429	+ * @param array|null $taxonomies
	2430	+ * @return array
	2431	+ */
	2432	+ function category_ids ($cats, $unfamiliar_category = 'create', $taxonomies = NULL, $params = array()) {
	2433	+ return $this->link->category_ids($this, $cats, $unfamiliar_category, $taxonomies, $params);
	2434	+ } /* SyndicatedPost::category_ids () */
1677	2435
1678		- function use_api ($tag) {
1679		- global $wp_db_version;
1680		- switch ($tag) :
1681		- case 'wp_insert_post':
1682		- // Before 2.2, wp_insert_post does too much of the wrong stuff to use it
1683		- // In 1.5 it was such a resource hog it would make PHP segfault on big updates
1684		- $ret = (isset($wp_db_version) and $wp_db_version > FWP_SCHEMA_21);
1685		- break;
1686		- case 'post_status_pending':
1687		- $ret = (isset($wp_db_version) and $wp_db_version > FWP_SCHEMA_23);
1688		- break;
1689		- endswitch;
1690		- return $ret;
1691		- } // function SyndicatedPost::use_api ()
1692		-
1693	2436	} /* class SyndicatedPost */
1694		-