← All changes | syndicatedpost.class.php +1768 -813 2009.0707 → 2011.0721 View file →

		@@ -1,40 +1,114 @@
1	1	<?php
	2	+require_once(dirname(__FILE__).'/feedtime.class.php');
	3	+
	4	+/**
	5	+ * class SyndicatedPost: FeedWordPress uses to manage the conversion of
	6	+ * incoming items from the feed parser into posts for the WordPress
	7	+ * database. It contains several internal management methods primarily
	8	+ * of interest to someone working on the FeedWordPress source, as well
	9	+ * as some utility methods for extracting useful data from many
	10	+ * different feed formats, which may be useful to FeedWordPress users
	11	+ * who make use of feed data in PHP add-ons and filters.
	12	+ *
	13	+ * @version 2011.0601
	14	+ */
2	15	class SyndicatedPost {
3		- var $item = null;
4		-
	16	+ var $item = null; // MagpieRSS representation
	17	+ var $entry = null; // SimplePie_Item representation
	18	+
5	19	var $link = null;
6	20	var $feed = null;
7	21	var $feedmeta = null;
8	22
	23	+ var $xmlns = array ();
	24	+
9	25	var $post = array ();
10	26
	27	+ var $named = array ();
	28	+ var $preset_terms = array ();
	29	+ var $feed_terms = array ();
	30	+
11	31	var $_freshness = null;
12	32	var $_wp_id = null;
13	33
14		- function SyndicatedPost ($item, $link) {
	34	+ /**
	35	+ * SyndicatedPost constructor: Given a feed item and the source from
	36	+ * which it was taken, prepare a post that can be inserted into the
	37	+ * WordPress database on request, or updated in place if it has already
	38	+ * been syndicated.
	39	+ *
	40	+ * @param array $item The item syndicated from the feed.
	41	+ * @param SyndicatedLink $source The feed it was syndicated from.
	42	+ */
	43	+ function SyndicatedPost ($item, &$source) {
15	44	global $wpdb;
16	45
17		- $this->link = $link;
18		- $feedmeta = $link->settings;
19		- $feed = $link->magpie;
	46	+ if (is_array($item)
	47	+ and isset($item['simplepie'])
	48	+ and isset($item['magpie'])) :
	49	+ $this->entry = $item['simplepie'];
	50	+ $this->item = $item['magpie'];
	51	+ $item = $item['magpie'];
	52	+ elseif (is_a($item, 'SimplePie_Item')) :
	53	+ $this->entry = $item;
	54	+
	55	+ // convert to Magpie for compat purposes
	56	+ $mp = new MagpieFromSimplePie($source->simplepie, $this->entry);
	57	+ $this->item = $mp->get_item();
	58	+
	59	+ // done with conversion object
	60	+ $mp = NULL; unset($mp);
	61	+ else :
	62	+ $this->item = $item;
	63	+ endif;
20	64
21		- # This is ugly as all hell. I'd like to use apply_filters()'s
22		- # alleged support for a variable argument count, but this seems
23		- # to have been broken in WordPress 1.5. It'll be fixed somehow
24		- # in WP 1.5.1, but I'm aiming at WP 1.5 compatibility across
25		- # the board here.
	65	+ $this->link =& $source;
	66	+ $this->feed = $source->magpie;
	67	+ $this->feedmeta = $source->settings;
	68	+
	69	+ FeedWordPress::diagnostic('feed_items', 'Considering item ['.$this->guid().'] "'.$this->entry->get_title().'"');
	70	+
	71	+ # Dealing with namespaces can get so fucking fucked.
	72	+ $this->xmlns['forward'] = $source->magpie->_XMLNS_FAMILIAR;
	73	+ $this->xmlns['reverse'] = array();
	74	+ foreach ($this->xmlns['forward'] as $url => $ns) :
	75	+ if (!isset($this->xmlns['reverse'][$ns])) :
	76	+ $this->xmlns['reverse'][$ns] = array();
	77	+ endif;
	78	+ $this->xmlns['reverse'][$ns][] = $url;
	79	+ endforeach;
	80	+
	81	+ // Fucking SimplePie.
	82	+ $this->xmlns['reverse']['rss'][] = '';
	83	+
	84	+ # These globals were originally an ugly kludge around a bug in
	85	+ # apply_filters from WordPress 1.5. The bug was fixed in 1.5.1,
	86	+ # and I sure hope at this point that nobody writing filters for
	87	+ # FeedWordPress is still relying on them.
26	88	#
27		- # Cf.: <http://mosquito.wordpress.org/view.php?id=901>
	89	+ # Anyway, I hereby declare them DEPRECATED as of 8 February
	90	+ # 2010. I'll probably remove the globals within 1-2 releases in
	91	+ # the interests of code hygiene and memory usage. If you
	92	+ # currently use them in your filters, I advise you switch off to
	93	+ # accessing the public members SyndicatedPost::feed and
	94	+ # SyndicatedPost::feedmeta.
	95	+
28	96	global $fwp_channel, $fwp_feedmeta;
29		- $fwp_channel = $feed; $fwp_feedmeta = $feedmeta;
	97	+ $fwp_channel = $this->feed; $fwp_feedmeta = $this->feedmeta;
30	98
31		- $this->feed = $feed;
32		- $this->feedmeta = $feedmeta;
33		-
34		- $this->item = $item;
35		- $this->item = apply_filters('syndicated_item', $this->item, $this);
36		-
	99	+ // Trigger global syndicated_item filter.
	100	+ $changed = apply_filters('syndicated_item', $this->item, $this);
	101	+ $this->item = $changed;
	102	+
	103	+ // Allow for feed-specific syndicated_item filters.
	104	+ $changed = apply_filters(
	105	+ "syndicated_item_".$source->uri(),
	106	+ $this->item,
	107	+ $this
	108	+ );
	109	+ $this->item = $changed;
	110	+
37	111	# Filters can halt further processing by returning NULL
38	112	if (is_null($this->item)) :
39	113	$this->post = NULL;
40	114	else :
		@@ -42,62 +116,41 @@
42	116	# That's deliberate. The escaping is done at the point
43	117	# of insertion, not here, to avoid double-escaping and
44	118	# to avoid screwing with syndicated_post filters
45	119
46		- $this->post['post_title'] = apply_filters('syndicated_item_title', $this->item['title'], $this);
	120	+ $this->post['post_title'] = apply_filters(
	121	+ 'syndicated_item_title',
	122	+ $this->entry->get_title(), $this
	123	+ );
47	124
48		- // This just gives us an alphanumeric representation of
49		- // the author. We will look up (or create) the numeric
50		- // ID for the author in SyndicatedPost::add()
51		- $this->post['named']['author'] = apply_filters('syndicated_item_author', $this->author(), $this);
	125	+ $this->named['author'] = apply_filters(
	126	+ 'syndicated_item_author',
	127	+ $this->author(), $this
	128	+ );
	129	+ // This just gives us an alphanumeric name for the author.
	130	+ // We look up (or create) the numeric ID for the author
	131	+ // in SyndicatedPost::add().
52	132
53		- # Identify content and sanitize it.
54		- # ---------------------------------
55		- if (isset($this->item['atom_content'])) :
56		- $content = $this->item['atom_content'];
57		- elseif (isset($this->item['xhtml']['body'])) :
58		- $content = $this->item['xhtml']['body'];
59		- elseif (isset($this->item['xhtml']['div'])) :
60		- $content = $this->item['xhtml']['div'];
61		- elseif (isset($this->item['content']['encoded']) and $this->item['content']['encoded']):
62		- $content = $this->item['content']['encoded'];
63		- else:
64		- $content = $this->item['description'];
65		- endif;
66		- $this->post['post_content'] = apply_filters('syndicated_item_content', $content, $this);
67		-
68		- # Identify and sanitize excerpt
69		- $excerpt = NULL;
70		- if ( isset($this->item['description']) and $this->item['description'] ) :
71		- $excerpt = $this->item['description'];
72		- elseif ( isset($content) and $content ) :
73		- $excerpt = strip_tags($content);
74		- if (strlen($excerpt) > 255) :
75		- $excerpt = substr($excerpt,0,252).'...';
76		- endif;
77		- endif;
78		- $excerpt = apply_filters('syndicated_item_excerpt', $excerpt, $this);
79		-
80		- if (!is_null($excerpt)):
	133	+ $this->post['post_content'] = apply_filters(
	134	+ 'syndicated_item_content',
	135	+ $this->content(), $this
	136	+ );
	137	+
	138	+ $excerpt = apply_filters('syndicated_item_excerpt', $this->excerpt(), $this);
	139	+ if (!empty($excerpt)):
81	140	$this->post['post_excerpt'] = $excerpt;
82	141	endif;
83		-
84		- // This is unnecessary if we use wp_insert_post
85		- if (!$this->use_api('wp_insert_post')) :
86		- $this->post['post_name'] = sanitize_title($this->post['post_title']);
87		- endif;
88	142
89		- $this->post['epoch']['issued'] = apply_filters('syndicated_item_published', $this->published(), $this);
90		- $this->post['epoch']['created'] = apply_filters('syndicated_item_created', $this->created(), $this);
91		- $this->post['epoch']['modified'] = apply_filters('syndicated_item_updated', $this->updated(), $this);
92		-
93	143	// Dealing with timestamps in WordPress is so fucking fucked.
94	144	$offset = (int) get_option('gmt_offset') * 60 * 60;
95		- $this->post['post_date'] = gmdate('Y-m-d H:i:s', $this->published() + $offset);
96		- $this->post['post_modified'] = gmdate('Y-m-d H:i:s', $this->updated() + $offset);
97		- $this->post['post_date_gmt'] = gmdate('Y-m-d H:i:s', $this->published());
98		- $this->post['post_modified_gmt'] = gmdate('Y-m-d H:i:s', $this->updated());
	145	+ $post_date_gmt = $this->published(array('default' => -1));
	146	+ $post_modified_gmt = $this->updated(array('default' => -1));
99	147
	148	+ $this->post['post_date_gmt'] = gmdate('Y-m-d H:i:s', $post_date_gmt);
	149	+ $this->post['post_date'] = gmdate('Y-m-d H:i:s', $post_date_gmt + $offset);
	150	+ $this->post['post_modified_gmt'] = gmdate('Y-m-d H:i:s', $post_modified_gmt);
	151	+ $this->post['post_modified'] = gmdate('Y-m-d H:i:s', $post_modified_gmt + $offset);
	152	+
100	153	// Use feed-level preferences or the global default.
101	154	$this->post['post_status'] = $this->link->syndicated_status('post', 'publish');
102	155	$this->post['comment_status'] = $this->link->syndicated_status('comment', 'closed');
103	156	$this->post['ping_status'] = $this->link->syndicated_status('ping', 'closed');
		@@ -106,9 +159,9 @@
106	159	$this->post['guid'] = apply_filters('syndicated_item_guid', $this->guid(), $this);
107	160
108	161	// 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
109	162	$default_custom_settings = get_option('feedwordpress_custom_settings');
110		- if ($default_custom_settings) :
	163	+ if ($default_custom_settings and !is_array($default_custom_settings)) :
111	164	$default_custom_settings = unserialize($default_custom_settings);
112	165	endif;
113	166	if (!is_array($default_custom_settings)) :
114	167	$default_custom_settings = array();
		@@ -114,164 +167,1147 @@
114	167	$default_custom_settings = array();
115	168	endif;
116	169
117	170	$custom_settings = (isset($this->link->settings['postmeta']) ? $this->link->settings['postmeta'] : null);
118		- if ($custom_settings) :
	171	+ if ($custom_settings and !is_array($custom_settings)) :
119	172	$custom_settings = unserialize($custom_settings);
120	173	endif;
121	174	if (!is_array($custom_settings)) :
122	175	$custom_settings = array();
123	176	endif;
124		- $this->post['meta'] = array_merge($default_custom_settings, $custom_settings);
	177	+
	178	+ $postMetaIn = array_merge($default_custom_settings, $custom_settings);
	179	+ $postMetaOut = array();
125	180
	181	+ // Big ugly fuckin loop to do any element substitutions
	182	+ // that we may need.
	183	+ foreach ($postMetaIn as $key => $values) :
	184	+ if (is_string($values)) : $values = array($values); endif;
	185	+
	186	+ $postMetaOut[$key] = array();
	187	+ foreach ($values as $value) :
	188	+ if (preg_match('/\$\( ([^)]+) \)/x', $value, $ref)) :
	189	+ $elements = $this->query($ref[1]);
	190	+ foreach ($elements as $element) :
	191	+ $postMetaOut[$key][] = str_replace(
	192	+ $ref[0],
	193	+ $element,
	194	+ $value
	195	+ );
	196	+ endforeach;
	197	+ else :
	198	+ $postMetaOut[$key][] = $value;
	199	+ endif;
	200	+ endforeach;
	201	+ endforeach;
	202	+
	203	+ foreach ($postMetaOut as $key => $values) :
	204	+ $this->post['meta'][$key] = array();
	205	+ foreach ($values as $value) :
	206	+ $this->post['meta'][$key][] = apply_filters("syndicated_post_meta_{$key}", $value, $this);
	207	+ endforeach;
	208	+ endforeach;
	209	+
126	210	// RSS 2.0 / Atom 1.0 enclosure support
127		- if ( isset($this->item['enclosure#']) ) :
128		- for ($i = 1; $i <= $this->item['enclosure#']; $i++) :
129		- $eid = (($i > 1) ? "#{$id}" : "");
130		- $this->post['meta']['enclosure'][] =
131		- apply_filters('syndicated_item_enclosure_url', $this->item["enclosure{$eid}@url"], $this)."\n".
132		- apply_filters('syndicated_item_enclosure_length', $this->item["enclosure{$eid}@length"], $this)."\n".
133		- apply_filters('syndicated_item_enclosure_type', $this->item["enclosure{$eid}@type"], $this);
134		- endfor;
	211	+ $enclosures = $this->entry->get_enclosures();
	212	+ if (is_array($enclosures)) : foreach ($enclosures as $enclosure) :
	213	+ $this->post['meta']['enclosure'][] =
	214	+ apply_filters('syndicated_item_enclosure_url', $enclosure->get_link(), $this)."\n".
	215	+ apply_filters('syndicated_item_enclosure_length', $enclosure->get_length(), $this)."\n".
	216	+ apply_filters('syndicated_item_enclosure_type', $enclosure->get_type(), $this);
	217	+ endforeach; endif;
	218	+
	219	+ // In case you want to point back to the blog this was
	220	+ // syndicated from.
	221	+
	222	+ $sourcemeta['syndication_source'] = apply_filters(
	223	+ 'syndicated_item_source_title',
	224	+ $this->link->name(),
	225	+ $this
	226	+ );
	227	+ $sourcemeta['syndication_source_uri'] = apply_filters(
	228	+ 'syndicated_item_source_link',
	229	+ $this->link->homepage(),
	230	+ $this
	231	+ );
	232	+ $sourcemeta['syndication_source_id'] = apply_filters(
	233	+ 'syndicated_item_source_id',
	234	+ $this->link->guid(),
	235	+ $this
	236	+ );
	237	+
	238	+ // Make use of atom:source data, if present in an aggregated feed
	239	+ $entry_source = $this->source();
	240	+ if (!is_null($entry_source)) :
	241	+ foreach ($entry_source as $what => $value) :
	242	+ if (!is_null($value)) :
	243	+ if ($what=='title') : $key = 'syndication_source';
	244	+ elseif ($what=='feed') : $key = 'syndication_feed';
	245	+ else : $key = "syndication_source_${what}";
	246	+ endif;
	247	+
	248	+ $sourcemeta["${key}_original"] = apply_filters(
	249	+ 'syndicated_item_original_source_'.$what,
	250	+ $value,
	251	+ $this
	252	+ );
	253	+ endif;
	254	+ endforeach;
135	255	endif;
	256	+
	257	+ foreach ($sourcemeta as $meta_key => $value) :
	258	+ if (!is_null($value)) :
	259	+ $this->post['meta'][$meta_key] = $value;
	260	+ endif;
	261	+ endforeach;
136	262
137		- // In case you want to point back to the blog this was syndicated from
138		- if (isset($this->feed->channel['title'])) :
139		- $this->post['meta']['syndication_source'] = apply_filters('syndicated_item_source_title', $this->feed->channel['title'], $this);
	263	+ // Store information on human-readable and machine-readable comment URIs
	264	+
	265	+ // Human-readable comment URI
	266	+ $commentLink = apply_filters('syndicated_item_comments', $this->comment_link(), $this);
	267	+ if (!is_null($commentLink)) : $this->post['meta']['rss:comments'] = $commentLink; endif;
	268	+
	269	+ // Machine-readable content feed URI
	270	+ $commentFeed = apply_filters('syndicated_item_commentrss', $this->comment_feed(), $this);
	271	+ if (!is_null($commentFeed)) : $this->post['meta']['wfw:commentRSS'] = $commentFeed; endif;
	272	+ // Yeah, yeah, now I know that it's supposed to be
	273	+ // wfw:commentRss. Oh well. Path dependence, sucka.
	274	+
	275	+ // Store information to identify the feed that this came from
	276	+ if (isset($this->feedmeta['link/uri'])) :
	277	+ $this->post['meta']['syndication_feed'] = $this->feedmeta['link/uri'];
140	278	endif;
	279	+ if (isset($this->feedmeta['link/id'])) :
	280	+ $this->post['meta']['syndication_feed_id'] = $this->feedmeta['link/id'];
	281	+ endif;
141	282
142		- if (isset($this->feed->channel['link'])) :
143		- $this->post['meta']['syndication_source_uri'] = apply_filters('syndicated_item_source_link', $this->feed->channel['link'], $this);
	283	+ if (isset($this->item['source_link_self'])) :
	284	+ $this->post['meta']['syndication_feed_original'] = $this->item['source_link_self'];
144	285	endif;
	286	+
	287	+ // In case you want to know the external permalink...
	288	+ $this->post['meta']['syndication_permalink'] = apply_filters('syndicated_item_link', $this->permalink());
	289	+
	290	+ // Store a hash of the post content for checking whether something needs to be updated
	291	+ $this->post['meta']['syndication_item_hash'] = $this->update_hash();
	292	+
	293	+ // Categories: start with default categories, if any.
	294	+ $cats = array();
	295	+ if ('no' != $this->link->setting('add/category', NULL, 'yes')) :
	296	+ $fc = get_option("feedwordpress_syndication_cats");
	297	+ if ($fc) :
	298	+ $cats = array_merge($cats, explode("\n", $fc));
	299	+ endif;
	300	+ endif;
	301	+
	302	+ $fc = $this->link->setting('cats', NULL, array());
	303	+ if (is_array($fc)) :
	304	+ $cats = array_merge($cats, $fc);
	305	+ endif;
	306	+ $this->preset_terms['category'] = $cats;
145	307
146		- // Make use of atom:source data, if present in an aggregated feed
147		- if (isset($this->item['source_title'])) :
148		- $this->post['meta']['syndication_source_original'] = $this->item['source_title'];
	308	+ // Now add categories from the post, if we have 'em
	309	+ $cats = array();
	310	+ $post_cats = $this->entry->get_categories();
	311	+ if (is_array($post_cats)) : foreach ($post_cats as $cat) :
	312	+ $cat_name = $cat->get_term();
	313	+ if (!$cat_name) : $cat_name = $cat->get_label(); endif;
	314	+
	315	+ if ($this->link->setting('cat_split', NULL, NULL)) :
	316	+ $pcre = "\007".$this->feedmeta['cat_split']."\007";
	317	+ $cats = array_merge(
	318	+ $cats,
	319	+ preg_split(
	320	+ $pcre,
	321	+ $cat_name,
	322	+ -1 /*=no limit*/,
	323	+ PREG_SPLIT_NO_EMPTY
	324	+ )
	325	+ );
	326	+ else :
	327	+ $cats[] = $cat_name;
	328	+ endif;
	329	+ endforeach; endif;
	330	+
	331	+ $this->feed_terms['category'] = apply_filters('syndicated_item_categories', $cats, $this);
	332	+
	333	+ // Tags: start with default tags, if any
	334	+ $tags = array();
	335	+ if ('no' != $this->link->setting('add/post_tag', NULL, 'yes')) :
	336	+ $ft = get_option("feedwordpress_syndication_tags", NULL);
	337	+ $tags = (is_null($ft) ? array() : explode(FEEDWORDPRESS_CAT_SEPARATOR, $ft));
149	338	endif;
	339	+
	340	+ $ft = $this->link->setting('tags', NULL, array());
	341	+ if (is_array($ft)) :
	342	+ $tags = array_merge($tags, $ft);
	343	+ endif;
	344	+ $this->preset_terms['post_tag'] = $tags;
	345	+
	346	+ // Scan post for /a[@rel='tag'] and use as tags if present
	347	+ $tags = $this->inline_tags();
	348	+ $this->feed_terms['post_tag'] = apply_filters('syndicated_item_tags', $tags, $this);
	349	+
	350	+ $taxonomies = $this->link->taxonomies();
	351	+ $feedTerms = $this->link->setting('terms', NULL, array());
	352	+ $globalTerms = get_option('feedwordpress_syndication_terms', array());
150	353
151		- if (isset($this->item['source_link'])) :
152		- $this->post['meta']['syndication_source_uri_original'] = $this->item['source_link'];
	354	+ $specials = array('category' => 'cats', 'post_tag' => 'tags');
	355	+ foreach ($taxonomies as $tax) :
	356	+ if (!isset($specials[$tax])) :
	357	+ $terms = array();
	358	+
	359	+ // See if we should get the globals
	360	+ if ('no' != $this->link->setting("add/$tax", NULL, 'yes')) :
	361	+ if (isset($globalTerms[$tax])) :
	362	+ $terms = $globalTerms[$tax];
	363	+ endif;
	364	+ endif;
	365	+
	366	+ // Now merge in the locals
	367	+ if (isset($feedTerms[$tax])) :
	368	+ $terms = array_merge($terms, $feedTerms[$tax]);
	369	+ endif;
	370	+
	371	+ // That's all, folks.
	372	+ $this->preset_terms[$tax] = $terms;
	373	+ endif;
	374	+ endforeach;
	375	+
	376	+ $this->post['post_type'] = apply_filters('syndicated_post_type', $this->link->setting('syndicated post type', 'syndicated_post_type', 'post'), $this);
	377	+ endif;
	378	+ } /* SyndicatedPost::SyndicatedPost() */
	379	+
	380	+ #####################################
	381	+ #### EXTRACT DATA FROM FEED ITEM ####
	382	+ #####################################
	383	+
	384	+ /**
	385	+ * SyndicatedPost::query uses an XPath-like syntax to query arbitrary
	386	+ * elements within the syndicated item.
	387	+ *
	388	+ * @param string $path
	389	+ * @returns array of string values representing contents of matching
	390	+ * elements or attributes
	391	+ */
	392	+ function query ($path) {
	393	+ $urlHash = array();
	394	+
	395	+ // Allow {url} notation for namespaces. URLs will contain : and /, so...
	396	+ preg_match_all('/{([^}]+)}/', $path, $match, PREG_SET_ORDER);
	397	+ foreach ($match as $ref) :
	398	+ $urlHash[md5($ref[1])] = $ref[1];
	399	+ endforeach;
	400	+
	401	+ foreach ($urlHash as $hash => $url) :
	402	+ $path = str_replace('{'.$url.'}', '{#'.$hash.'}', $path);
	403	+ endforeach;
	404	+
	405	+ $path = explode('/', $path);
	406	+ foreach ($path as $index => $node) :
	407	+ if (preg_match('/{#([^}]+)}/', $node, $ref)) :
	408	+ if (isset($urlHash[$ref[1]])) :
	409	+ $path[$index] = str_replace(
	410	+ '{#'.$ref[1].'}',
	411	+ '{'.$urlHash[$ref[1]].'}',
	412	+ $node
	413	+ );
	414	+ endif;
153	415	endif;
	416	+ endforeach;
	417	+
	418	+ // Start out with a get_item_tags query.
	419	+ $node = '';
	420	+ while (strlen($node)==0 and !is_null($node)) :
	421	+ $node = array_shift($path);
	422	+ endwhile;
	423	+
	424	+ switch ($node) :
	425	+ case 'feed' :
	426	+ case 'channel' :
	427	+ $node = array_shift($path);
	428	+ $data = $this->get_feed_root_element();
	429	+ $data = array_merge($data, $this->get_feed_channel_elements());
	430	+ break;
	431	+ case 'item' :
	432	+ $node = array_shift($path);
	433	+ default :
	434	+ $data = array($this->entry->data);
	435	+ $method = NULL;
	436	+ endswitch;
	437	+
	438	+ while (!is_null($node)) :
	439	+ if (strlen($node) > 0) :
	440	+ $matches = array();
	441	+
	442	+ list($axis, $element) = $this->xpath_name_and_axis($node);
	443	+
	444	+ foreach ($data as $datum) :
	445	+ if (!is_string($datum) and isset($datum[$axis])) :
	446	+ foreach ($datum[$axis] as $ns => $elements) :
	447	+ if (isset($elements[$element])) :
	448	+ // Potential match.
	449	+ // Check namespace.
	450	+ if (is_string($elements[$element])) : // Attribute
	451	+ $addenda = array($elements[$element]);
	452	+ $contexts = array($datum);
	453	+ else : // Element
	454	+ $addenda = $elements[$element];
	455	+ $contexts = $elements[$element];
	456	+ endif;
	457	+
	458	+ foreach ($addenda as $index => $addendum) :
	459	+ $context = $contexts[$index];
	460	+
	461	+ $namespaces = $this->xpath_possible_namespaces($node, $context);
	462	+ if (in_array($ns, $namespaces)) :
	463	+ $matches[] = $addendum;
	464	+ endif;
	465	+ endforeach;
	466	+ endif;
	467	+ endforeach;
	468	+ endif;
	469	+ endforeach;
	470	+
	471	+ $data = $matches;
	472	+ endif;
	473	+ $node = array_shift($path);
	474	+ endwhile;
	475	+
	476	+ $matches = array();
	477	+ foreach ($data as $datum) :
	478	+ if (is_string($datum)) :
	479	+ $matches[] = $datum;
	480	+ elseif (isset($datum['data'])) :
	481	+ $matches[] = $datum['data'];
	482	+ endif;
	483	+ endforeach;
	484	+ return $matches;
	485	+ } /* SyndicatedPost::query() */
	486	+
	487	+ function get_feed_root_element () {
	488	+ $matches = array();
	489	+ foreach ($this->link->simplepie->data['child'] as $ns => $root) :
	490	+ foreach ($root as $element => $data) :
	491	+ $matches = array_merge($matches, $data);
	492	+ endforeach;
	493	+ endforeach;
	494	+ return $matches;
	495	+ } /* SyndicatedPost::get_feed_root_element() */
	496	+
	497	+ function get_feed_channel_elements () {
	498	+ $rss = array(
	499	+ SIMPLEPIE_NAMESPACE_RSS_090,
	500	+ SIMPLEPIE_NAMESPACE_RSS_10,
	501	+ 'http://backend.userland.com/RSS2',
	502	+ SIMPLEPIE_NAMESPACE_RSS_20,
	503	+ );
	504	+
	505	+ $matches = array();
	506	+ foreach ($rss as $ns) :
	507	+ $data = $this->link->simplepie->get_feed_tags($ns, 'channel');
	508	+ if (!is_null($data)) :
	509	+ $matches = array_merge($matches, $data);
	510	+ endif;
	511	+ endforeach;
	512	+ return $matches;
	513	+ } /* SyndicatedPost::get_feed_channel_elements() */
	514	+
	515	+ function xpath_default_namespace () {
	516	+ // Get the default namespace.
	517	+ $type = $this->link->simplepie->get_type();
	518	+ if ($type & SIMPLEPIE_TYPE_ATOM_10) :
	519	+ $defaultNS = SIMPLEPIE_NAMESPACE_ATOM_10;
	520	+ elseif ($type & SIMPLEPIE_TYPE_ATOM_03) :
	521	+ $defaultNS = SIMPLEPIE_NAMESPACE_ATOM_03;
	522	+ elseif ($type & SIMPLEPIE_TYPE_RSS_090) :
	523	+ $defaultNS = SIMPLEPIE_NAMESPACE_RSS_090;
	524	+ elseif ($type & SIMPLEPIE_TYPE_RSS_10) :
	525	+ $defaultNS = SIMPLEPIE_NAMESPACE_RSS_10;
	526	+ elseif ($type & SIMPLEPIE_TYPE_RSS_20) :
	527	+ $defaultNS = SIMPLEPIE_NAMESPACE_RSS_20;
	528	+ else :
	529	+ $defaultNS = SIMPLEPIE_NAMESPACE_RSS_20;
	530	+ endif;
	531	+ return $defaultNS;
	532	+ } /* SyndicatedPost::xpath_default_namespace() */
	533	+
	534	+ function xpath_name_and_axis ($node) {
	535	+ $ns = NULL; $element = NULL;
	536	+
	537	+ if (substr($node, 0, 1)=='@') :
	538	+ $axis = 'attribs'; $node = substr($node, 1);
	539	+ else :
	540	+ $axis = 'child';
	541	+ endif;
	542	+
	543	+ if (preg_match('/^{([^}]*)}(.*)$/', $node, $ref)) :
	544	+ $element = $ref[2];
	545	+ elseif (strpos($node, ':') !== FALSE) :
	546	+ list($xmlns, $element) = explode(':', $node, 2);
	547	+ else :
	548	+ $element = $node;
	549	+ endif;
	550	+ return array($axis, $element);
	551	+ } /* SyndicatedPost::xpath_local_name () */
	552	+
	553	+ function xpath_possible_namespaces ($node, $datum = array()) {
	554	+ $ns = NULL; $element = NULL;
	555	+
	556	+ if (substr($node, 0, 1)=='@') :
	557	+ $attr = '@'; $node = substr($node, 1);
	558	+ else :
	559	+ $attr = '';
	560	+ endif;
	561	+
	562	+ if (preg_match('/^{([^}]*)}(.*)$/', $node, $ref)) :
	563	+ $ns = array($ref[1]);
	564	+ elseif (strpos($node, ':') !== FALSE) :
	565	+ list($xmlns, $element) = explode(':', $node, 2);
154	566
155		- if (isset($this->item['source_id'])) :
156		- $this->post['meta']['syndication_source_id_original'] = $this->item['source_id'];
	567	+ if (isset($this->xmlns['reverse'][$xmlns])) :
	568	+ $ns = $this->xmlns['reverse'][$xmlns];
	569	+ else :
	570	+ $ns = array($xmlns);
157	571	endif;
158		-
159		- // Store information on human-readable and machine-readable comment URIs
160		- if (isset($this->item['comments'])) :
161		- $this->post['meta']['rss:comments'] = apply_filters('syndicated_item_comments', $this->item['comments']);
	572	+
	573	+ // Fucking SimplePie. For attributes in default xmlns.
	574	+ $defaultNS = $this->xpath_default_namespace();
	575	+ if (isset($this->xmlns['forward'][$defaultNS])
	576	+ and ($xmlns==$this->xmlns['forward'][$defaultNS])) :
	577	+ $ns[] = '';
162	578	endif;
163		- if (isset($this->item['wfw']['commentrss'])) :
164		- $this->post['meta']['wfw:commentRSS'] = apply_filters('syndicated_item_commentrss', $this->item['wfw']['commentrss']);
	579	+
	580	+ if (isset($datum['xmlns'])) :
	581	+ if (isset($datum['xmlns'][$xmlns])) :
	582	+ $ns[] = $datum['xmlns'][$xmlns];
	583	+ endif;
165	584	endif;
	585	+ else :
	586	+ // Often in SimplePie, the default namespace gets stored
	587	+ // as an empty string rather than a URL.
	588	+ $ns = array($this->xpath_default_namespace(), '');
	589	+ endif;
	590	+ return array_unique($ns);
	591	+ } /* SyndicatedPost::xpath_possible_namespaces() */
166	592
167		- // Store information to identify the feed that this came from
168		- $this->post['meta']['syndication_feed'] = $this->feedmeta['link/uri'];
169		- $this->post['meta']['syndication_feed_id'] = $this->feedmeta['link/id'];
	593	+ function content () {
	594	+ $content = NULL;
	595	+ if (isset($this->item['atom_content'])) :
	596	+ $content = $this->item['atom_content'];
	597	+ elseif (isset($this->item['xhtml']['body'])) :
	598	+ $content = $this->item['xhtml']['body'];
	599	+ elseif (isset($this->item['xhtml']['div'])) :
	600	+ $content = $this->item['xhtml']['div'];
	601	+ elseif (isset($this->item['content']['encoded']) and $this->item['content']['encoded']):
	602	+ $content = $this->item['content']['encoded'];
	603	+ elseif (isset($this->item['description'])) :
	604	+ $content = $this->item['description'];
	605	+ endif;
	606	+ return $content;
	607	+ } /* SyndicatedPost::content() */
170	608
171		- if (isset($this->item['source_link_self'])) :
172		- $this->post['meta']['syndication_feed_original'] = $this->item['source_link_self'];
	609	+ function excerpt () {
	610	+ # Identify and sanitize excerpt: atom:summary, or rss:description
	611	+ $excerpt = $this->entry->get_description();
	612	+
	613	+ # Many RSS feeds use rss:description, inadvisably, to
	614	+ # carry the entire post (typically with escaped HTML).
	615	+ # If that's what happened, we don't want the full
	616	+ # content for the excerpt.
	617	+ $content = $this->content();
	618	+
	619	+ // Ignore whitespace, case, and tag cruft.
	620	+ $theExcerpt = preg_replace('/\s+/', '', strtolower(strip_tags($excerpt)));
	621	+ $theContent = preg_replace('/\s+/', '', strtolower(strip_tags($content)));
	622	+
	623	+ if ( empty($excerpt) or $theExcerpt == $theContent ) :
	624	+ # If content is available, generate an excerpt.
	625	+ if ( strlen(trim($content)) > 0 ) :
	626	+ $excerpt = strip_tags($content);
	627	+ if (strlen($excerpt) > 255) :
	628	+ $excerpt = substr($excerpt,0,252).'...';
	629	+ endif;
173	630	endif;
	631	+ endif;
	632	+ return $excerpt;
	633	+ } /* SyndicatedPost::excerpt() */
174	634
175		- // In case you want to know the external permalink...
176		- $this->post['meta']['syndication_permalink'] = apply_filters('syndicated_item_link', $this->item['link']);
	635	+ function permalink () {
	636	+ // Handles explicit <link> elements and also RSS 2.0 cases with
	637	+ // <guid isPermaLink="true">, etc. Hooray!
	638	+ $permalink = $this->entry->get_link();
	639	+ return $permalink;
	640	+ }
177	641
178		- // Store a hash of the post content for checking whether something needs to be updated
179		- $this->post['meta']['syndication_item_hash'] = $this->update_hash();
	642	+ function created ($params = array()) {
	643	+ $unfiltered = false; $default = NULL;
	644	+ extract($params);
	645	+
	646	+ $date = '';
	647	+ if (isset($this->item['dc']['created'])) :
	648	+ $date = $this->item['dc']['created'];
	649	+ elseif (isset($this->item['dcterms']['created'])) :
	650	+ $date = $this->item['dcterms']['created'];
	651	+ elseif (isset($this->item['created'])): // Atom 0.3
	652	+ $date = $this->item['created'];
	653	+ endif;
180	654
181		- // Feed-by-feed options for author and category creation
182		- $this->post['named']['unfamiliar']['author'] = (isset($this->feedmeta['unfamiliar author']) ? $this->feedmeta['unfamiliar author'] : null);
183		- $this->post['named']['unfamiliar']['category'] = (isset($this->feedmeta['unfamiliar category']) ? $this->feedmeta['unfamiliar category'] : null);
	655	+ $time = new FeedTime($date);
	656	+ $ts = $time->timestamp();
	657	+ if (!$unfiltered) :
	658	+ apply_filters('syndicated_item_created', $ts, $this);
	659	+ endif;
	660	+ return $ts;
	661	+ } /* SyndicatedPost::created() */
184	662
185		- // Categories: start with default categories, if any
186		- $fc = get_option("feedwordpress_syndication_cats");
187		- if ($fc) :
188		- $this->post['named']['preset/category'] = explode("\n", $fc);
	663	+ function published ($params = array(), $default = NULL) {
	664	+ $fallback = true; $unfiltered = false;
	665	+ if (!is_array($params)) : // Old style
	666	+ $fallback = $params;
	667	+ else : // New style
	668	+ extract($params);
	669	+ endif;
	670	+
	671	+ $date = '';
	672	+ $ts = null;
	673	+
	674	+ # RSS is a fucking mess. Figure out whether we have a date in
	675	+ # <dc:date>, <issued>, <pubDate>, etc., and get it into Unix
	676	+ # epoch format for reformatting. If we can't find anything,
	677	+ # we'll use the last-updated time.
	678	+ if (isset($this->item['dc']['date'])): // Dublin Core
	679	+ $date = $this->item['dc']['date'];
	680	+ elseif (isset($this->item['dcterms']['issued'])) : // Dublin Core extensions
	681	+ $date = $this->item['dcterms']['issued'];
	682	+ elseif (isset($this->item['published'])) : // Atom 1.0
	683	+ $date = $this->item['published'];
	684	+ elseif (isset($this->item['issued'])): // Atom 0.3
	685	+ $date = $this->item['issued'];
	686	+ elseif (isset($this->item['pubdate'])): // RSS 2.0
	687	+ $date = $this->item['pubdate'];
	688	+ endif;
	689	+
	690	+ if (strlen($date) > 0) :
	691	+ $time = new FeedTime($date);
	692	+ $ts = $time->timestamp();
	693	+ elseif ($fallback) : // Fall back to <updated> / <modified> if present
	694	+ $ts = $this->updated(/*fallback=*/ false, /*default=*/ $default);
	695	+ endif;
	696	+
	697	+ # If everything failed, then default to the current time.
	698	+ if (is_null($ts)) :
	699	+ if (-1 == $default) :
	700	+ $ts = time();
189	701	else :
190		- $this->post['named']['preset/category'] = array();
	702	+ $ts = $default;
191	703	endif;
	704	+ endif;
	705	+
	706	+ if (!$unfiltered) :
	707	+ $ts = apply_filters('syndicated_item_published', $ts, $this);
	708	+ endif;
	709	+ return $ts;
	710	+ } /* SyndicatedPost::published() */
192	711
193		- if (isset($this->feedmeta['cats']) and is_array($this->feedmeta['cats'])) :
194		- $this->post['named']['preset/category'] = array_merge($this->post['named']['preset/category'], $this->feedmeta['cats']);
	712	+ function updated ($params = array(), $default = -1) {
	713	+ $fallback = true; $unfiltered = false;
	714	+ if (!is_array($params)) : // Old style
	715	+ $fallback = $params;
	716	+ else : // New style
	717	+ extract($params);
	718	+ endif;
	719	+
	720	+ $date = '';
	721	+ $ts = null;
	722	+
	723	+ # As far as I know, only dcterms and Atom have reliable ways to
	724	+ # specify when something was *modified* last. If neither is
	725	+ # available, then we'll try to get the time of publication.
	726	+ if (isset($this->item['dc']['modified'])) : // Not really correct
	727	+ $date = $this->item['dc']['modified'];
	728	+ elseif (isset($this->item['dcterms']['modified'])) : // Dublin Core extensions
	729	+ $date = $this->item['dcterms']['modified'];
	730	+ elseif (isset($this->item['modified'])): // Atom 0.3
	731	+ $date = $this->item['modified'];
	732	+ elseif (isset($this->item['updated'])): // Atom 1.0
	733	+ $date = $this->item['updated'];
	734	+ endif;
	735	+
	736	+ if (strlen($date) > 0) :
	737	+ $time = new FeedTime($date);
	738	+ $ts = $time->timestamp();
	739	+ elseif ($fallback) : // Fall back to issued / dc:date
	740	+ $ts = $this->published(/*fallback=*/ false, /*default=*/ $default);
	741	+ endif;
	742	+
	743	+ # If everything failed, then default to the current time.
	744	+ if (is_null($ts)) :
	745	+ if (-1 == $default) :
	746	+ $ts = time();
	747	+ else :
	748	+ $ts = $default;
195	749	endif;
	750	+ endif;
196	751
197		- // Now add categories from the post, if we have 'em
198		- $this->post['named']['category'] = array();
199		- if ( isset($this->item['category#']) ) :
200		- for ($i = 1; $i <= $this->item['category#']; $i++) :
201		- $cat_idx = (($i > 1) ? "#{$i}" : "");
202		- $cat = $this->item["category{$cat_idx}"];
	752	+ if (!$unfiltered) :
	753	+ apply_filters('syndicated_item_updated', $ts, $this);
	754	+ endif;
	755	+ return $ts;
	756	+ } /* SyndicatedPost::updated() */
203	757
204		- if ( isset($this->feedmeta['cat_split']) and strlen($this->feedmeta['cat_split']) > 0) :
205		- $pcre = "\007".$this->feedmeta['cat_split']."\007";
206		- $this->post['named']['category'] = array_merge($this->post['named']['category'], preg_split($pcre, $cat, -1 /*=no limit*/, PREG_SPLIT_NO_EMPTY));
207		- else :
208		- $this->post['named']['category'][] = $cat;
209		- endif;
210		- endfor;
	758	+ var $_hashes = array();
	759	+ function stored_hashes ($id = NULL) {
	760	+ if (is_null($id)) :
	761	+ $id = $this->wp_id();
	762	+ endif;
	763	+
	764	+ if (!isset($this->_hashes[$id])) :
	765	+ $this->_hashes[$id] = get_post_custom_values(
	766	+ 'syndication_item_hash', $id
	767	+ );
	768	+ if (is_null($this->_hashes[$id])) :
	769	+ $this->_hashes[$id] = array();
211	770	endif;
212		- $this->post['named']['category'] = apply_filters('syndicated_item_categories', $this->post['named']['category'], $this);
	771	+ endif;
	772	+ return $this->_hashes[$id];
	773	+ }
	774	+
	775	+ function update_hash () {
	776	+ return md5(serialize($this->item));
	777	+ } /* SyndicatedPost::update_hash() */
	778	+
	779	+ /*static*/ function normalize_guid_prefix () {
	780	+ return trailingslashit(get_bloginfo('url')).'?guid=';
	781	+ }
	782	+
	783	+ /*static*/ function normalize_guid ($guid) {
	784	+ $guid = trim($guid);
	785	+ if (preg_match('/^[0-9a-z]{32}$/i', $guid)) : // MD5
	786	+ $guid = SyndicatedPost::normalize_guid_prefix().strtolower($guid);
	787	+ elseif ((strlen(esc_url($guid)) == 0) or (esc_url($guid) != $guid)) :
	788	+ $guid = SyndicatedPost::normalize_guid_prefix().md5($guid);
	789	+ endif;
	790	+ $guid = trim($guid);
	791	+ return $guid;
	792	+ } /* SyndicatedPost::normalize_guid() */
	793	+
	794	+ function guid () {
	795	+ $guid = null;
	796	+ if (isset($this->item['id'])): // Atom 0.3 / 1.0
	797	+ $guid = $this->item['id'];
	798	+ elseif (isset($this->item['atom']['id'])) : // Namespaced Atom
	799	+ $guid = $this->item['atom']['id'];
	800	+ elseif (isset($this->item['guid'])) : // RSS 2.0
	801	+ $guid = $this->item['guid'];
	802	+ elseif (isset($this->item['dc']['identifier'])) : // yeah, right
	803	+ $guid = $this->item['dc']['identifier'];
	804	+ endif;
	805	+
	806	+ // Un-set or too long to use as-is. Generate a tag: URI.
	807	+ if (is_null($guid) or strlen($guid) > 250) :
	808	+ // In case we need to check this again
	809	+ $original_guid = $guid;
213	810
214		- // Tags: start with default tags, if any
215		- $ft = get_option("feedwordpress_syndication_tags");
216		- if ($ft) :
217		- $this->post['tags_input'] = explode(FEEDWORDPRESS_CAT_SEPARATOR, $ft);
	811	+ // The feed does not seem to have provided us with a
	812	+ // usable unique identifier, so we'll have to cobble
	813	+ // together a tag: URI that might work for us. The base
	814	+ // of the URI will be the host name of the feed source ...
	815	+ $bits = parse_url($this->link->uri());
	816	+ $guid = 'tag:'.$bits['host'];
	817	+
	818	+ // Some ill-mannered feeds (for example, certain feeds
	819	+ // coming from Google Calendar) have extraordinarily long
	820	+ // guids -- so long that they exceed the 255 character
	821	+ // width of the WordPress guid field. But if the string
	822	+ // gets clipped by MySQL, uniqueness tests will fail
	823	+ // forever after and the post will be endlessly
	824	+ // reduplicated. So, instead, Guids Of A Certain Length
	825	+ // are hashed down into a nice, manageable tag: URI.
	826	+ if (!is_null($original_guid)) :
	827	+ $guid .= ',2010-12-03:id.'.md5($original_guid);
	828	+
	829	+ // If we have a date of creation, then we can use that
	830	+ // to uniquely identify the item. (On the other hand, if
	831	+ // the feed producer was consicentious enough to
	832	+ // generate dates of creation, she probably also was
	833	+ // conscientious enough to generate unique identifiers.)
	834	+ elseif (!is_null($this->created())) :
	835	+ $guid .= '://post.'.date('YmdHis', $this->created());
	836	+
	837	+ // Otherwise, use both the URI of the item, *and* the
	838	+ // item's title. We have to use both because titles are
	839	+ // often not unique, and sometimes links aren't unique
	840	+ // either (e.g. Bitch (S)HITLIST, Mozilla Dot Org news,
	841	+ // some podcasts). But it's rare to have *both* the same
	842	+ // title *and* the same link for two different items. So
	843	+ // this is about the best we can do.
218	844	else :
219		- $this->post['tags_input'] = array();
	845	+ $link = $this->permalink();
	846	+ if (is_null($link)) : $link = $this->link->uri(); endif;
	847	+ $guid .= '://'.md5($link.'/'.$this->item['title']);
220	848	endif;
	849	+ endif;
	850	+ return $guid;
	851	+ } /* SyndicatedPost::guid() */
	852	+
	853	+ function author () {
	854	+ $author = array ();
	855	+
	856	+ if (isset($this->item['author_name'])):
	857	+ $author['name'] = $this->item['author_name'];
	858	+ elseif (isset($this->item['dc']['creator'])):
	859	+ $author['name'] = $this->item['dc']['creator'];
	860	+ elseif (isset($this->item['dc']['contributor'])):
	861	+ $author['name'] = $this->item['dc']['contributor'];
	862	+ elseif (isset($this->feed->channel['dc']['creator'])) :
	863	+ $author['name'] = $this->feed->channel['dc']['creator'];
	864	+ elseif (isset($this->feed->channel['dc']['contributor'])) :
	865	+ $author['name'] = $this->feed->channel['dc']['contributor'];
	866	+ elseif (isset($this->feed->channel['author_name'])) :
	867	+ $author['name'] = $this->feed->channel['author_name'];
	868	+ elseif ($this->feed->is_rss() and isset($this->item['author'])) :
	869	+ // The author element in RSS is allegedly an
	870	+ // e-mail address, but lots of people don't use
	871	+ // it that way. So let's make of it what we can.
	872	+ $author = parse_email_with_realname($this->item['author']);
221	873
222		- if (isset($this->feedmeta['tags']) and is_array($this->feedmeta['tags'])) :
223		- $this->post['tags_input'] = array_merge($this->post['tags_input'], $this->feedmeta['tags']);
	874	+ if (!isset($author['name'])) :
	875	+ if (isset($author['email'])) :
	876	+ $author['name'] = $author['email'];
	877	+ else :
	878	+ $author['name'] = $this->feed->channel['title'];
	879	+ endif;
224	880	endif;
	881	+ elseif ($this->link->name()) :
	882	+ $author['name'] = $this->link->name();
	883	+ else :
	884	+ $url = parse_url($this->link->uri());
	885	+ $author['name'] = $url['host'];
	886	+ endif;
	887	+
	888	+ if (isset($this->item['author_email'])):
	889	+ $author['email'] = $this->item['author_email'];
	890	+ elseif (isset($this->feed->channel['author_email'])) :
	891	+ $author['email'] = $this->feed->channel['author_email'];
	892	+ endif;
	893	+
	894	+ if (isset($this->item['author_uri'])):
	895	+ $author['uri'] = $this->item['author_uri'];
	896	+ elseif (isset($this->item['author_url'])):
	897	+ $author['uri'] = $this->item['author_url'];
	898	+ elseif (isset($this->feed->channel['author_uri'])) :
	899	+ $author['uri'] = $this->item['author_uri'];
	900	+ elseif (isset($this->feed->channel['author_url'])) :
	901	+ $author['uri'] = $this->item['author_url'];
	902	+ elseif (isset($this->feed->channel['link'])) :
	903	+ $author['uri'] = $this->feed->channel['link'];
	904	+ endif;
225	905
	906	+ return $author;
	907	+ } /* SyndicatedPost::author() */
	908	+
	909	+ /**
	910	+ * SyndicatedPost::inline_tags: Return a list of all the tags embedded
	911	+ * in post content using the a[@rel="tag"] microformat.
	912	+ *
	913	+ * @since 2010.0630
	914	+ * @return array of string values containing the name of each tag
	915	+ */
	916	+ function inline_tags () {
	917	+ $tags = array();
	918	+ $content = $this->content();
	919	+ $pattern = FeedWordPressHTML::tagWithAttributeRegex('a', 'rel', 'tag');
	920	+ preg_match_all($pattern, $content, $refs, PREG_SET_ORDER);
	921	+ if (count($refs) > 0) :
	922	+ foreach ($refs as $ref) :
	923	+ $tag = FeedWordPressHTML::tagWithAttributeMatch($ref);
	924	+ $tags[] = $tag['content'];
	925	+ endforeach;
226	926	endif;
227		- } // SyndicatedPost::SyndicatedPost()
	927	+ return $tags;
	928	+ }
	929	+
	930	+ /**
	931	+ * SyndicatedPost::isTaggedAs: Test whether a feed item is
	932	+ * tagged / categorized with a given string. Case and leading and
	933	+ * trailing whitespace are ignored.
	934	+ *
	935	+ * @param string $tag Tag to check for
	936	+ *
	937	+ * @return bool Whether or not at least one of the categories / tags on
	938	+ * $this->item is set to $tag (modulo case and leading and trailing
	939	+ * whitespace)
	940	+ */
	941	+ function isTaggedAs ($tag) {
	942	+ $desiredTag = strtolower(trim($tag)); // Normalize case and whitespace
228	943
	944	+ // Check to see if this is tagged with $tag
	945	+ $currentCategory = 'category';
	946	+ $currentCategoryNumber = 1;
	947	+
	948	+ // If we have the new MagpieRSS, the number of category elements
	949	+ // on this item is stored under index "category#".
	950	+ if (isset($this->item['category#'])) :
	951	+ $numberOfCategories = (int) $this->item['category#'];
	952	+
	953	+ // We REALLY shouldn't have the old and busted MagpieRSS, but in
	954	+ // case we do, it doesn't support multiple categories, but there
	955	+ // might still be a single value under the "category" index.
	956	+ elseif (isset($this->item['category'])) :
	957	+ $numberOfCategories = 1;
	958	+
	959	+ // No standard category or tag elements on this feed item.
	960	+ else :
	961	+ $numberOfCategories = 0;
	962	+
	963	+ endif;
	964	+
	965	+ $isSoTagged = false; // Innocent until proven guilty
	966	+
	967	+ // Loop through category elements; if there are multiple
	968	+ // elements, they are indexed as category, category#2,
	969	+ // category#3, ... category#N
	970	+ while ($currentCategoryNumber <= $numberOfCategories) :
	971	+ if ($desiredTag == strtolower(trim($this->item[$currentCategory]))) :
	972	+ $isSoTagged = true; // Got it!
	973	+ break;
	974	+ endif;
	975	+
	976	+ $currentCategoryNumber += 1;
	977	+ $currentCategory = 'category#'.$currentCategoryNumber;
	978	+ endwhile;
	979	+
	980	+ return $isSoTagged;
	981	+ } /* SyndicatedPost::isTaggedAs() */
	982	+
	983	+ /**
	984	+ * SyndicatedPost::enclosures: returns an array with any enclosures
	985	+ * that may be attached to this syndicated item.
	986	+ *
	987	+ * @param string $type If you only want enclosures that match a certain
	988	+ * MIME type or group of MIME types, you can limit the enclosures
	989	+ * that will be returned to only those with a MIME type which
	990	+ * matches this regular expression.
	991	+ * @return array
	992	+ */
	993	+ function enclosures ($type = '/.*/') {
	994	+ $enclosures = array();
	995	+
	996	+ if (isset($this->item['enclosure#'])) :
	997	+ // Loop through enclosure, enclosure#2, enclosure#3, ....
	998	+ for ($i = 1; $i <= $this->item['enclosure#']; $i++) :
	999	+ $eid = (($i > 1) ? "#{$id}" : "");
	1000	+
	1001	+ // Does it match the type we want?
	1002	+ if (preg_match($type, $this->item["enclosure{$eid}@type"])) :
	1003	+ $enclosures[] = array(
	1004	+ "url" => $this->item["enclosure{$eid}@url"],
	1005	+ "type" => $this->item["enclosure{$eid}@type"],
	1006	+ "length" => $this->item["enclosure{$eid}@length"],
	1007	+ );
	1008	+ endif;
	1009	+ endfor;
	1010	+ endif;
	1011	+ return $enclosures;
	1012	+ } /* SyndicatedPost::enclosures() */
	1013	+
	1014	+ function source ($what = NULL) {
	1015	+ $ret = NULL;
	1016	+ $source = $this->entry->get_source();
	1017	+ if ($source) :
	1018	+ $ret = array();
	1019	+ $ret['title'] = $source->get_title();
	1020	+ $ret['uri'] = $source->get_link();
	1021	+ $ret['feed'] = $source->get_link(0, 'self');
	1022	+
	1023	+ if ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'id')) :
	1024	+ $ret['id'] = $id_tags[0]['data'];
	1025	+ elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_ATOM_03, 'id')) :
	1026	+ $ret['id'] = $id_tags[0]['data'];
	1027	+ elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_RSS_20, 'guid')) :
	1028	+ $ret['id'] = $id_tags[0]['data'];
	1029	+ elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_RSS_10, 'guid')) :
	1030	+ $ret['id'] = $id_tags[0]['data'];
	1031	+ elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_RSS_090, 'guid')) :
	1032	+ $ret['id'] = $id_tags[0]['data'];
	1033	+ endif;
	1034	+ endif;
	1035	+
	1036	+ if (!is_null($what) and is_scalar($what)) :
	1037	+ $ret = $ret[$what];
	1038	+ endif;
	1039	+ return $ret;
	1040	+ }
	1041	+
	1042	+ function comment_link () {
	1043	+ $url = null;
	1044	+
	1045	+ // RSS 2.0 has a standard <comments> element:
	1046	+ // "<comments> is an optional sub-element of <item>. If present,
	1047	+ // it is the url of the comments page for the item."
	1048	+ // <http://cyber.law.harvard.edu/rss/rss.html#ltcommentsgtSubelementOfLtitemgt>
	1049	+ if (isset($this->item['comments'])) :
	1050	+ $url = $this->item['comments'];
	1051	+ endif;
	1052	+
	1053	+ // The convention in Atom feeds is to use a standard <link>
	1054	+ // element with @rel="replies" and @type="text/html".
	1055	+ // Unfortunately, SimplePie_Item::get_links() allows us to filter
	1056	+ // by the value of @rel, but not by the value of @type. *sigh*
	1057	+
	1058	+ // Try Atom 1.0 first
	1059	+ $linkElements = $this->entry->get_item_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'link');
	1060	+
	1061	+ // Fall back and try Atom 0.3
	1062	+ if (is_null($linkElements)) : $linkElements = $this->entry->get_item_tags(SIMPLEPIE_NAMESPACE_ATOM_03, 'link'); endif;
	1063	+
	1064	+ // Now loop through the elements, screening by @rel and @type
	1065	+ if (is_array($linkElements)) : foreach ($linkElements as $link) :
	1066	+ $rel = (isset($link['attribs']['']['rel']) ? $link['attribs']['']['rel'] : 'alternate');
	1067	+ $type = (isset($link['attribs']['']['type']) ? $link['attribs']['']['type'] : NULL);
	1068	+ $href = (isset($link['attribs']['']['href']) ? $link['attribs']['']['href'] : NULL);
	1069	+
	1070	+ if (strtolower($rel)=='replies' and $type=='text/html' and !is_null($href)) :
	1071	+ $url = $href;
	1072	+ endif;
	1073	+ endforeach; endif;
	1074	+
	1075	+ return $url;
	1076	+ }
	1077	+
	1078	+ function comment_feed () {
	1079	+ $feed = null;
	1080	+
	1081	+ // Well Formed Web comment feeds extension for RSS 2.0
	1082	+ // <http://www.sellsbrothers.com/spout/default.aspx?content=archive.htm#exposingRssComments>
	1083	+ //
	1084	+ // N.B.: Correct capitalization is wfw:commentRss, but
	1085	+ // wfw:commentRSS is common in the wild (partly due to a typo in
	1086	+ // the original spec). In any case, our item array is normalized
	1087	+ // to all lowercase anyways.
	1088	+ if (isset($this->item['wfw']['commentrss'])) :
	1089	+ $feed = $this->item['wfw']['commentrss'];
	1090	+ endif;
	1091	+
	1092	+ // In Atom 1.0, the convention is to use a standard link element
	1093	+ // with @rel="replies". Sometimes this is also used to pass a
	1094	+ // link to the human-readable comments page, so we also need to
	1095	+ // check link/@type for a feed MIME type.
	1096	+ //
	1097	+ // Which is why I'm not using the SimplePie_Item::get_links()
	1098	+ // method here, incidentally: it doesn't allow you to filter by
	1099	+ // @type. *sigh*
	1100	+ if (isset($this->item['link_replies'])) :
	1101	+ // There may be multiple <link rel="replies"> elements; feeds have a feed MIME type
	1102	+ $N = isset($this->item['link_replies#']) ? $this->item['link_replies#'] : 1;
	1103	+ for ($i = 1; $i <= $N; $i++) :
	1104	+ $currentElement = 'link_replies'.(($i > 1) ? '#'.$i : '');
	1105	+ if (isset($this->item[$currentElement.'@type'])
	1106	+ and preg_match("\007application/(atom|rss|rdf)\+xml\007i", $this->item[$currentElement.'@type'])) :
	1107	+ $feed = $this->item[$currentElement];
	1108	+ endif;
	1109	+ endfor;
	1110	+ endif;
	1111	+ return $feed;
	1112	+ } /* SyndicatedPost::comment_feed() */
	1113	+
	1114	+ ##################################
	1115	+ #### BUILT-IN CONTENT FILTERS ####
	1116	+ ##################################
	1117	+
	1118	+ var $uri_attrs = array (
	1119	+ array('a', 'href'),
	1120	+ array('applet', 'codebase'),
	1121	+ array('area', 'href'),
	1122	+ array('blockquote', 'cite'),
	1123	+ array('body', 'background'),
	1124	+ array('del', 'cite'),
	1125	+ array('form', 'action'),
	1126	+ array('frame', 'longdesc'),
	1127	+ array('frame', 'src'),
	1128	+ array('iframe', 'longdesc'),
	1129	+ array('iframe', 'src'),
	1130	+ array('head', 'profile'),
	1131	+ array('img', 'longdesc'),
	1132	+ array('img', 'src'),
	1133	+ array('img', 'usemap'),
	1134	+ array('input', 'src'),
	1135	+ array('input', 'usemap'),
	1136	+ array('ins', 'cite'),
	1137	+ array('link', 'href'),
	1138	+ array('object', 'classid'),
	1139	+ array('object', 'codebase'),
	1140	+ array('object', 'data'),
	1141	+ array('object', 'usemap'),
	1142	+ array('q', 'cite'),
	1143	+ array('script', 'src')
	1144	+ ); /* var SyndicatedPost::$uri_attrs */
	1145	+
	1146	+ var $_base = null;
	1147	+
	1148	+ function resolve_single_relative_uri ($refs) {
	1149	+ $tag = FeedWordPressHTML::attributeMatch($refs);
	1150	+ $url = SimplePie_Misc::absolutize_url($tag['value'], $this->_base);
	1151	+ return $tag['prefix'] . $url . $tag['suffix'];
	1152	+ } /* function SyndicatedPost::resolve_single_relative_uri() */
	1153	+
	1154	+ function resolve_relative_uris ($content, $obj) {
	1155	+ $set = $obj->link->setting('resolve relative', 'resolve_relative', 'yes');
	1156	+ if ($set and $set != 'no') :
	1157	+ // Fallback: if we don't have anything better, use the
	1158	+ // item link from the feed
	1159	+ $obj->_base = $obj->permalink(); // Reset the base for resolving relative URIs
	1160	+
	1161	+ // What we should do here, properly, is to use
	1162	+ // SimplePie_Item::get_base() -- but that method is
	1163	+ // currently broken. Or getting down and dirty in the
	1164	+ // SimplePie representation of the content tags and
	1165	+ // grabbing the xml_base member for the content element.
	1166	+ // Maybe someday...
	1167	+
	1168	+ foreach ($obj->uri_attrs as $pair) :
	1169	+ list($tag, $attr) = $pair;
	1170	+ $pattern = FeedWordPressHTML::attributeRegex($tag, $attr);
	1171	+ $content = preg_replace_callback (
	1172	+ $pattern,
	1173	+ array(&$obj, 'resolve_single_relative_uri'),
	1174	+ $content
	1175	+ );
	1176	+ endforeach;
	1177	+ endif;
	1178	+
	1179	+ return $content;
	1180	+ } /* function SyndicatedPost::resolve_relative_uris () */
	1181	+
	1182	+ var $strip_attrs = array (
	1183	+ array('[a-z]+', 'target'),
	1184	+// array('[a-z]+', 'style'),
	1185	+// array('[a-z]+', 'on[a-z]+'),
	1186	+ );
	1187	+
	1188	+ function strip_attribute_from_tag ($refs) {
	1189	+ $tag = FeedWordPressHTML::attributeMatch($refs);
	1190	+ return $tag['before_attribute'].$tag['after_attribute'];
	1191	+ }
	1192	+
	1193	+ function sanitize_content ($content, $obj) {
	1194	+ # This kind of sucks. I intend to replace it with
	1195	+ # lib_filter sometime soon.
	1196	+ foreach ($obj->strip_attrs as $pair):
	1197	+ list($tag,$attr) = $pair;
	1198	+ $pattern = FeedWordPressHTML::attributeRegex($tag, $attr);
	1199	+
	1200	+ $content = preg_replace_callback (
	1201	+ $pattern,
	1202	+ array(&$obj, 'strip_attribute_from_tag'),
	1203	+ $content
	1204	+ );
	1205	+ endforeach;
	1206	+ return $content;
	1207	+ } /* SyndicatedPost::sanitize() */
	1208	+
	1209	+ #####################
	1210	+ #### POST STATUS ####
	1211	+ #####################
	1212	+
	1213	+ /**
	1214	+ * SyndicatedPost::filtered: check whether or not this post has been
	1215	+ * screened out by a registered filter.
	1216	+ *
	1217	+ * @return bool TRUE iff post has been filtered out by a previous filter
	1218	+ */
229	1219	function filtered () {
230	1220	return is_null($this->post);
231		- }
	1221	+ } /* SyndicatedPost::filtered() */
232	1222
	1223	+ /**
	1224	+ * SyndicatedPost::freshness: check whether post is a new post to be
	1225	+ * inserted, a previously syndicated post that needs to be updated to
	1226	+ * match the latest revision, or a previously syndicated post that is
	1227	+ * still up-to-date.
	1228	+ *
	1229	+ * @return int A status code representing the freshness of the post
	1230	+ * 0 = post already syndicated; no update needed
	1231	+ * 1 = post already syndicated, but needs to be updated to latest
	1232	+ * 2 = post has not yet been syndicated; needs to be created
	1233	+ */
233	1234	function freshness () {
234	1235	global $wpdb;
235	1236
236	1237	if ($this->filtered()) : // This should never happen.
237		- FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__);
	1238	+ FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__, __FILE__);
238	1239	endif;
239	1240
240		- if (is_null($this->_freshness)) :
	1241	+ if (is_null($this->_freshness)) : // Not yet checked and cached.
241	1242	$guid = $wpdb->escape($this->guid());
242	1243
243		- $result = $wpdb->get_row("
244		- SELECT id, guid, post_modified_gmt
245		- FROM $wpdb->posts WHERE guid='$guid'
246		- ");
247		-
248		- if (!$result) :
	1244	+ $q = new WP_Query(array(
	1245	+ 'fields' => '_synfresh', // id, guid, post_modified_gmt
	1246	+ 'ignore_sticky_posts' => true,
	1247	+ 'guid' => $this->guid(),
	1248	+ ));
	1249	+
	1250	+ $old_post = NULL;
	1251	+ if ($q->have_posts()) :
	1252	+ while ($q->have_posts()) : $q->the_post();
	1253	+ $old_post = $q->post;
	1254	+ endwhile;
	1255	+ endif;
	1256	+
	1257	+ if (is_null($old_post)) : // No post with this guid
	1258	+ FeedWordPress::diagnostic('feed_items:freshness', 'Item ['.$this->guid().'] "'.$this->entry->get_title().'" is a NEW POST.');
	1259	+ $this->_wp_id = NULL;
249	1260	$this->_freshness = 2; // New content
250		- else:
251		- $stored_update_hashes = get_post_custom_values('syndication_item_hash', $result->id);
252		- if (count($stored_update_hashes) > 0) :
253		- $stored_update_hash = $stored_update_hashes[0];
254		- $update_hash_changed = ($stored_update_hash != $this->update_hash());
255		- else :
256		- $update_hash_changed = false;
257		- endif;
	1261	+ else :
	1262	+ preg_match('/([0-9]+)-([0-9]+)-([0-9]+) ([0-9]+):([0-9]+):([0-9]+)/', $old_post->post_modified_gmt, $backref);
258	1263
259		- preg_match('/([0-9]+)-([0-9]+)-([0-9]+) ([0-9]+):([0-9]+):([0-9]+)/', $result->post_modified_gmt, $backref);
260		-
261	1264	$last_rev_ts = gmmktime($backref[4], $backref[5], $backref[6], $backref[2], $backref[3], $backref[1]);
262	1265	$updated_ts = $this->updated(/*fallback=*/ true, /*default=*/ NULL);
263		- $updated = ((
	1266	+
	1267	+ // Check timestamps...
	1268	+ $updated = (
264	1269	!is_null($updated_ts)
265	1270	and ($updated_ts > $last_rev_ts)
266		- ) or $update_hash_changed);
	1271	+ );
	1272	+
	1273	+
	1274	+ if (!$updated) :
	1275	+ // Or the hash...
	1276	+ $hash = $this->update_hash();
	1277	+ $seen = $this->stored_hashes($old_post->ID);
	1278	+ if (count($seen) > 0) :
	1279	+ $updated = !in_array($hash, $seen); // Not seen yet?
	1280	+ else :
	1281	+ $updated = true; // Can't find syndication meta-data
	1282	+ endif;
	1283	+ endif;
	1284	+
	1285	+ $frozen = false;
	1286	+ if ($updated) : // Ignore if the post is frozen
	1287	+ $frozen = ('yes' == $this->link->setting('freeze updates', 'freeze_updates', NULL));
	1288	+ if (!$frozen) :
	1289	+ $frozen_values = get_post_custom_values('_syndication_freeze_updates', $old_post->ID);
	1290	+ $frozen = (count($frozen_values) > 0 and 'yes' == $frozen_values[0]);
	1291	+ endif;
	1292	+ endif;
	1293	+ $updated = ($updated and !$frozen);
267	1294
268	1295	if ($updated) :
	1296	+ FeedWordPress::diagnostic('feed_items:freshness', 'Item ['.$this->guid().'] "'.$this->entry->get_title().'" is an update of an existing post.');
269	1297	$this->_freshness = 1; // Updated content
270		- $this->_wp_id = $result->id;
	1298	+ $this->_wp_id = $old_post->ID;
	1299	+
	1300	+ // We want this to keep a running list of all the
	1301	+ // processed update hashes.
	1302	+ $this->post['meta']['syndication_item_hash'] = array_merge(
	1303	+ $this->stored_hashes(),
	1304	+ array($this->update_hash())
	1305	+ );
271	1306	else :
	1307	+ FeedWordPress::diagnostic('feed_items:freshness', 'Item ['.$this->guid().'] "'.$this->entry->get_title().'" is a duplicate of an existing post.');
272	1308	$this->_freshness = 0; // Same old, same old
273		- $this->_wp_id = $result->id;
	1309	+ $this->_wp_id = $old_post->ID;
274	1310	endif;
275	1311	endif;
276	1312	endif;
277	1313	return $this->_freshness;
		@@ -276,11 +1312,15 @@
276	1312	endif;
277	1313	return $this->_freshness;
278	1314	}
279	1315
	1316	+ #################################################
	1317	+ #### INTERNAL STORAGE AND MANAGEMENT METHODS ####
	1318	+ #################################################
	1319	+
280	1320	function wp_id () {
281	1321	if ($this->filtered()) : // This should never happen.
282		- FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__);
	1322	+ FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__, __FILE__);
283	1323	endif;
284	1324
285	1325	if (is_null($this->_wp_id) and is_null($this->_freshness)) :
286	1326	$fresh = $this->freshness(); // sets WP DB id in the process
		@@ -291,9 +1331,9 @@
291	1331	function store () {
292	1332	global $wpdb;
293	1333
294	1334	if ($this->filtered()) : // This should never happen.
295		- FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__);
	1335	+ FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__, __FILE__);
296	1336	endif;
297	1337
298	1338	$freshness = $this->freshness();
299	1339	if ($freshness > 0) :
		@@ -298,229 +1338,235 @@
298	1338	$freshness = $this->freshness();
299	1339	if ($freshness > 0) :
300	1340	# -- Look up, or create, numeric ID for author
301	1341	$this->post['post_author'] = $this->author_id (
302		- FeedWordPress::on_unfamiliar('author', $this->post['named']['unfamiliar']['author'])
	1342	+ $this->link->setting('unfamiliar author', 'unfamiliar_author', 'create')
303	1343	);
304	1344
305	1345	if (is_null($this->post['post_author'])) :
	1346	+ FeedWordPress::diagnostic('feed_items:rejected', 'Filtered out item ['.$this->guid().'] without syndication: no author available');
306	1347	$this->post = NULL;
307	1348	endif;
308	1349	endif;
309	1350
310	1351	if (!$this->filtered() and $freshness > 0) :
311		- # -- Look up, or create, numeric ID for categories
312		- list($pcats, $ptags) = $this->category_ids (
313		- $this->post['named']['category'],
314		- FeedWordPress::on_unfamiliar('category', $this->post['named']['unfamiliar']['category']),
315		- /*tags_too=*/ true
	1352	+ $consider = array(
	1353	+ 'category' => array('abbr' => 'cats', 'domain' => array('category', 'post_tag')),
	1354	+ 'post_tag' => array('abbr' => 'tags', 'domain' => array('post_tag')),
316	1355	);
317	1356
318		- $this->post['post_category'] = $pcats;
319		- $this->post['tags_input'] = array_merge($this->post['tags_input'], $ptags);
	1357	+ $termSet = array(); $valid = null;
	1358	+ foreach ($consider as $what => $taxes) :
	1359	+ if (!is_null($this->post)) : // Not filtered out yet
	1360	+ # -- Look up, or create, numeric ID for categories
	1361	+ $taxonomies = $this->link->setting("match/".$taxes['abbr'], 'match_'.$taxes['abbr'], $taxes['domain']);
	1362	+
	1363	+ // Eliminate dummy variables
	1364	+ $taxonomies = array_filter($taxonomies, 'remove_dummy_zero');
	1365	+
	1366	+ $terms = $this->category_ids (
	1367	+ $this->feed_terms[$what],
	1368	+ $this->link->setting("unfamiliar {$what}", "unfamiliar_{$what}", 'create:'.$what),
	1369	+ /*taxonomies=*/ $taxonomies,
	1370	+ array(
	1371	+ 'singleton' => false, // I don't like surprises
	1372	+ 'filters' => true,
	1373	+ )
	1374	+ );
	1375	+
	1376	+ if (is_null($terms) or is_null($termSet)) :
	1377	+ // filtered out -- no matches
	1378	+ else :
	1379	+ $valid = true;
320	1380
321		- if (is_null($this->post['post_category'])) :
322		- // filter mode on, no matching categories; drop the post
	1381	+ // filter mode off, or at least one match
	1382	+ foreach ($terms as $tax => $term_ids) :
	1383	+ if (!isset($termSet[$tax])) :
	1384	+ $termSet[$tax] = array();
	1385	+ endif;
	1386	+ $termSet[$tax] = array_merge($termSet[$tax], $term_ids);
	1387	+ endforeach;
	1388	+ endif;
	1389	+ endif;
	1390	+ endforeach;
	1391	+
	1392	+ if (is_null($valid)) : // Plonked
323	1393	$this->post = NULL;
324		- else :
325		- // filter mode off or at least one match; now add on the feed and global presets
326		- $this->post['post_category'] = array_merge (
327		- $this->post['post_category'],
328		- $this->category_ids (
329		- $this->post['named']['preset/category'],
330		- 'default'
331		- )
332		- );
	1394	+ else : // We can proceed
	1395	+ $this->post['tax_input'] = array();
	1396	+ foreach ($termSet as $tax => $term_ids) :
	1397	+ if (!isset($this->post['tax_input'][$tax])) :
	1398	+ $this->post['tax_input'][$tax] = array();
	1399	+ endif;
	1400	+ $this->post['tax_input'][$tax] = array_merge(
	1401	+ $this->post['tax_input'][$tax],
	1402	+ $term_ids
	1403	+ );
	1404	+ endforeach;
333	1405
334		- if (count($this->post['post_category']) < 1) :
335		- $this->post['post_category'][] = 1; // Default to category 1 ("Uncategorized" / "General") if nothing else
336		- endif;
	1406	+ // Now let's add on the feed and global presets
	1407	+ foreach ($this->preset_terms as $tax => $term_ids) :
	1408	+ if (!isset($this->post['tax_input'][$tax])) :
	1409	+ $this->post['tax_input'][$tax] = array();
	1410	+ endif;
	1411	+
	1412	+ $this->post['tax_input'][$tax] = array_merge (
	1413	+ $this->post['tax_input'][$tax],
	1414	+ $this->category_ids (
	1415	+ /*terms=*/ $term_ids,
	1416	+ /*unfamiliar=*/ 'create:'.$tax, // These are presets; for those added in a tagbox editor, the tag may not yet exist
	1417	+ /*taxonomies=*/ array($tax),
	1418	+ array(
	1419	+ 'singleton' => true,
	1420	+ ))
	1421	+ );
	1422	+ endforeach;
337	1423	endif;
338	1424	endif;
339	1425
340	1426	if (!$this->filtered() and $freshness > 0) :
341		- unset($this->post['named']);
	1427	+ // Filter some individual fields
	1428	+
	1429	+ // Allow filters to set post slug. Props niska.
	1430	+ $post_name = apply_filters('syndicated_post_slug', NULL, $this);
	1431	+ if (!empty($post_name)) :
	1432	+ $this->post['post_name'] = $post_name;
	1433	+ endif;
	1434	+
342	1435	$this->post = apply_filters('syndicated_post', $this->post, $this);
	1436	+
	1437	+ // Allow for feed-specific syndicated_post filters.
	1438	+ $this->post = apply_filters(
	1439	+ "syndicated_post_".$this->link->uri(),
	1440	+ $this->post,
	1441	+ $this
	1442	+ );
343	1443	endif;
344	1444
345		- if (!$this->filtered() and $freshness == 2) :
346		- // The item has not yet been added. So let's add it.
347		- $this->insert_new();
348		- $this->add_rss_meta();
349		- do_action('post_syndicated_item', $this->wp_id());
	1445	+ // Hook in early to make sure these get inserted if at all possible
	1446	+ add_action(
	1447	+ /*hook=*/ 'transition_post_status',
	1448	+ /*callback=*/ array(&$this, 'add_rss_meta'),
	1449	+ /*priority=*/ -10000, /* very early */
	1450	+ /*arguments=*/ 3
	1451	+ );
350	1452
351		- $ret = 'new';
352		- elseif (!$this->filtered() and $freshness == 1) :
353		- $this->post['ID'] = $this->wp_id();
354		- $this->update_existing();
355		- $this->add_rss_meta();
356		- do_action('update_syndicated_item', $this->wp_id());
	1453	+ $retval = array(1 => 'updated', 2 => 'new');
	1454	+
	1455	+ $ret = false;
	1456	+ if (!$this->filtered() and isset($retval[$freshness])) :
	1457	+ $diag = array(
	1458	+ 1 => 'Updating existing post # '.$this->wp_id().', "'.$this->post['post_title'].'"',
	1459	+ 2 => 'Inserting new post "'.$this->post['post_title'].'"',
	1460	+ );
	1461	+ FeedWordPress::diagnostic('syndicated_posts', $diag[$freshness]);
357	1462
358		- $ret = 'updated';
359		- else :
360		- $ret = false;
	1463	+ $this->insert_post(/*update=*/ ($freshness == 1));
	1464	+
	1465	+ $hook = array( 1 => 'update_syndicated_item', 2 => 'post_syndicated_item' );
	1466	+ do_action($hook[$freshness], $this->wp_id(), $this);
	1467	+
	1468	+ $ret = $retval[$freshness];
361	1469	endif;
362		-
	1470	+
	1471	+ // Remove add_rss_meta hook
	1472	+ remove_action(
	1473	+ /*hook=*/ 'transition_post_status',
	1474	+ /*callback=*/ array(&$this, 'add_rss_meta'),
	1475	+ /*priority=*/ -10000, /* very early */
	1476	+ /*arguments=*/ 3
	1477	+ );
	1478	+
363	1479	return $ret;
364		- } // function SyndicatedPost::store ()
	1480	+ } /* function SyndicatedPost::store () */
365	1481
366		- function insert_new () {
367		- global $wpdb, $wp_db_version;
	1482	+ function insert_post ($update = false) {
	1483	+ global $wpdb;
368	1484
369	1485	$dbpost = $this->normalize_post(/*new=*/ true);
370	1486	if (!is_null($dbpost)) :
371		- if ($this->use_api('wp_insert_post')) :
372		- $dbpost['post_pingback'] = false; // Tell WP 2.1 and 2.2 not to process for pingbacks
	1487	+ $dbpost['post_pingback'] = false; // Tell WP 2.1 and 2.2 not to process for pingbacks
373	1488
374		- // This is a ridiculous fucking kludge necessitated by WordPress 2.6 munging authorship meta-data
375		- add_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
	1489	+ // This is a ridiculous fucking kludge necessitated by WordPress 2.6 munging authorship meta-data
	1490	+ add_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
376	1491
377		- // Kludge to prevent kses filters from stripping the
378		- // content of posts when updating without a logged in
379		- // user who has `unfiltered_html` capability.
380		- add_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11);
	1492	+ // Kludge to prevent kses filters from stripping the
	1493	+ // content of posts when updating without a logged in
	1494	+ // user who has `unfiltered_html` capability.
	1495	+ $mungers = array('wp_filter_kses', 'wp_filter_post_kses');
	1496	+ $removed = array();
	1497	+ foreach ($mungers as $munger) :
	1498	+ if (has_filter('content_save_pre', $munger)) :
	1499	+ remove_filter('content_save_pre', $munger);
	1500	+ $removed[] = $munger;
	1501	+ endif;
	1502	+ endforeach;
	1503	+
	1504	+ if ($update and function_exists('get_post_field')) :
	1505	+ // Don't munge status fields that the user may
	1506	+ // have reset manually
	1507	+ $doNotMunge = array('post_status', 'comment_status', 'ping_status');
381	1508
382		- $this->_wp_id = wp_insert_post($dbpost);
383		-
384		- // Turn off ridiculous fucking kludges #1 and #2
385		- remove_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
386		- remove_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11);
387		-
388		- $this->validate_post_id($dbpost, array(__CLASS__, __FUNCTION__));
389		-
390		- // Unfortunately, as of WordPress 2.3, wp_insert_post()
391		- // *still* offers no way to use a guid of your choice,
392		- // and munges your post modified timestamp, too.
393		- $result = $wpdb->query("
394		- UPDATE $wpdb->posts
395		- SET
396		- guid='{$dbpost['guid']}',
397		- post_modified='{$dbpost['post_modified']}',
398		- post_modified_gmt='{$dbpost['post_modified_gmt']}'
399		- WHERE ID='{$this->_wp_id}'
400		- ");
401		- else :
402		- # The right way to do this is the above. But, alas,
403		- # in earlier versions of WordPress, wp_insert_post has
404		- # too much behavior (mainly related to pings) that can't
405		- # be overridden. In WordPress 1.5, it's enough of a
406		- # resource hog to make PHP segfault after inserting
407		- # 50-100 posts. This can get pretty annoying, especially
408		- # if you are trying to update your feeds for the first
409		- # time.
410		-
411		- $result = $wpdb->query("
412		- INSERT INTO $wpdb->posts
413		- SET
414		- guid = '{$dbpost['guid']}',
415		- post_author = '{$dbpost['post_author']}',
416		- post_date = '{$dbpost['post_date']}',
417		- post_date_gmt = '{$dbpost['post_date_gmt']}',
418		- post_content = '{$dbpost['post_content']}',"
419		- .(isset($dbpost['post_excerpt']) ? "post_excerpt = '{$dbpost['post_excerpt']}'," : "")."
420		- post_title = '{$dbpost['post_title']}',
421		- post_name = '{$dbpost['post_name']}',
422		- post_modified = '{$dbpost['post_modified']}',
423		- post_modified_gmt = '{$dbpost['post_modified_gmt']}',
424		- comment_status = '{$dbpost['comment_status']}',
425		- ping_status = '{$dbpost['ping_status']}',
426		- post_status = '{$dbpost['post_status']}'
427		- ");
428		- $this->_wp_id = $wpdb->insert_id;
429		-
430		- $this->validate_post_id($dbpost, array(__CLASS__, __FUNCTION__));
431		-
432		- // WordPress 1.5.x - 2.0.x
433		- wp_set_post_cats('1', $this->wp_id(), $this->post['post_category']);
434		-
435		- // Since we are not going through official channels, we need to
436		- // manually tell WordPress that we've published a new post.
437		- // We need to make sure to do this in order for FeedWordPress
438		- // to play well with the staticize-reloaded plugin (something
439		- // that a large aggregator website is going to *want* to be
440		- // able to use).
441		- do_action('publish_post', $this->_wp_id);
	1509	+ foreach ($doNotMunge as $field) :
	1510	+ $dbpost[$field] = get_post_field($field, $this->wp_id());
	1511	+ endforeach;
442	1512	endif;
443		- endif;
444		- } /* SyndicatedPost::insert_new() */
	1513	+
	1514	+ // WP3's wp_insert_post scans current_user_can() for the
	1515	+ // tax_input, with no apparent way to override. Ugh.
	1516	+ add_action(
	1517	+ /*hook=*/ 'transition_post_status',
	1518	+ /*callback=*/ array(&$this, 'add_terms'),
	1519	+ /*priority=*/ -10001, /* very early */
	1520	+ /*arguments=*/ 3
	1521	+ );
	1522	+
	1523	+ // WP3 appears to override whatever you give it for
	1524	+ // post_modified. Ugh.
	1525	+ add_action(
	1526	+ /*hook=*/ 'transition_post_status',
	1527	+ /*callback=*/ array(&$this, 'fix_post_modified_ts'),
	1528	+ /*priority=*/ -10000, /* very early */
	1529	+ /*arguments=*/ 3
	1530	+ );
445	1531
446		- function update_existing () {
447		- global $wpdb;
	1532	+ if ($update) :
	1533	+ $this->post['ID'] = $this->wp_id();
	1534	+ $dbpost['ID'] = $this->post['ID'];
	1535	+ endif;
	1536	+
	1537	+ $this->_wp_id = wp_insert_post($dbpost, /*return wp_error=*/ true);
448	1538
449		- // Why the fuck doesn't wp_insert_post already do this?
450		- $dbpost = $this->normalize_post(/*new=*/ false);
451		- if (!is_null($dbpost)) :
452		- if ($this->use_api('wp_insert_post')) :
453		- $dbpost['post_pingback'] = false; // Tell WP 2.1 and 2.2 not to process for pingbacks
454		-
455		- // This is a ridiculous fucking kludge necessitated by WordPress 2.6 munging authorship meta-data
456		- add_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
457		-
458		- // Kludge to prevent kses filters from stripping the
459		- // content of posts when updating without a logged in
460		- // user who has `unfiltered_html` capability.
461		- add_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11);
	1539	+ remove_action(
	1540	+ /*hook=*/ 'transition_post_status',
	1541	+ /*callback=*/ array(&$this, 'add_terms'),
	1542	+ /*priority=*/ -10001, /* very early */
	1543	+ /*arguments=*/ 3
	1544	+ );
462	1545
463		- // Don't munge status fields that the user may have reset manually
464		- if (function_exists('get_post_field')) :
465		- $doNotMunge = array('post_status', 'comment_status', 'ping_status');
466		- foreach ($doNotMunge as $field) :
467		- $dbpost[$field] = get_post_field($field, $this->wp_id());
468		- endforeach;
469		- endif;
	1546	+ remove_action(
	1547	+ /*hook=*/ 'transition_post_status',
	1548	+ /*callback=*/ array(&$this, 'fix_post_modified_ts'),
	1549	+ /*priority=*/ -10000, /* very early */
	1550	+ /*arguments=*/ 3
	1551	+ );
470	1552
471		- $this->_wp_id = wp_insert_post($dbpost);
	1553	+ // Turn off ridiculous fucking kludges #1 and #2
	1554	+ remove_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
	1555	+ foreach ($removed as $filter) :
	1556	+ add_filter('content_save_pre', $removed);
	1557	+ endforeach;
	1558	+
	1559	+ $this->validate_post_id($dbpost, $update, array(__CLASS__, __FUNCTION__));
	1560	+ endif;
	1561	+ } /* function SyndicatedPost::insert_post () */
472	1562
473		- // Turn off ridiculous fucking kludges #1 and #2
474		- remove_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
475		- remove_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11);
476		-
477		- $this->validate_post_id($dbpost, array(__CLASS__, __FUNCTION__));
478		-
479		- // Unfortunately, as of WordPress 2.3, wp_insert_post()
480		- // munges your post modified timestamp.
481		- $result = $wpdb->query("
482		- UPDATE $wpdb->posts
483		- SET
484		- post_modified='{$dbpost['post_modified']}',
485		- post_modified_gmt='{$dbpost['post_modified_gmt']}'
486		- WHERE ID='{$this->_wp_id}'
487		- ");
488		- else :
489		-
490		- $result = $wpdb->query("
491		- UPDATE $wpdb->posts
492		- SET
493		- post_author = '{$dbpost['post_author']}',
494		- post_content = '{$dbpost['post_content']}',"
495		- .(isset($dbpost['post_excerpt']) ? "post_excerpt = '{$dbpost['post_excerpt']}'," : "")."
496		- post_title = '{$dbpost['post_title']}',
497		- post_name = '{$dbpost['post_name']}',
498		- post_modified = '{$dbpost['post_modified']}',
499		- post_modified_gmt = '{$dbpost['post_modified_gmt']}'
500		- WHERE guid='{$dbpost['guid']}'
501		- ");
502		-
503		- // WordPress 2.1.x and up
504		- if (function_exists('wp_set_post_categories')) :
505		- wp_set_post_categories($this->wp_id(), $this->post['post_category']);
506		- // WordPress 1.5.x - 2.0.x
507		- elseif (function_exists('wp_set_post_cats')) :
508		- wp_set_post_cats('1', $this->wp_id(), $this->post['post_category']);
509		- // This should never happen.
510		- else :
511		- FeedWordPress::critical_bug(__CLASS__.'::'.__FUNCTION.'(): no post categorizing function', array("dbpost" => $dbpost, "this" => $this), __LINE__);
512		- endif;
513		-
514		- // Since we are not going through official channels, we need to
515		- // manually tell WordPress that we've published a new post.
516		- // We need to make sure to do this in order for FeedWordPress
517		- // to play well with the staticize-reloaded plugin (something
518		- // that a large aggregator website is going to *want* to be
519		- // able to use).
520		- do_action('edit_post', $this->post['ID']);
521		- endif;
522		- endif;
	1563	+ function insert_new () {
	1564	+ $this->insert_post(/*update=*/ false);
	1565	+ } /* SyndicatedPost::insert_new() */
	1566	+
	1567	+ function update_existing () {
	1568	+ $this->insert_post(/*update=*/ true);
523	1569	} /* SyndicatedPost::update_existing() */
524	1570
525	1571	/**
526	1572	* SyndicatedPost::normalize_post()
		@@ -530,12 +1576,34 @@
530	1576	*/
531	1577	function normalize_post ($new = true) {
532	1578	global $wpdb;
533	1579
534		- $out = array();
	1580	+ $out = $this->post;
535	1581
	1582	+ $fullPost = $out['post_title'].$out['post_content'];
	1583	+ $fullPost .= (isset($out['post_excerpt']) ? $out['post_excerpt'] : '');
	1584	+ if (strlen($fullPost) < 1) :
	1585	+ // FIXME: Option for filtering out empty posts
	1586	+ endif;
	1587	+ if (strlen($out['post_title'])==0) :
	1588	+ $offset = (int) get_option('gmt_offset') * 60 * 60;
	1589	+ if (isset($this->post['meta']['syndication_source'])) :
	1590	+ $source_title = $this->post['meta']['syndication_source'];
	1591	+ else :
	1592	+ $feed_url = parse_url($this->post['meta']['syndication_feed']);
	1593	+ $source_title = $feed_url['host'];
	1594	+ endif;
	1595	+
	1596	+ $out['post_title'] = $source_title
	1597	+ .' '.gmdate('Y-m-d H:i:s', $this->published() + $offset);
	1598	+ // FIXME: Option for what to fill a blank title with...
	1599	+ endif;
	1600	+
	1601	+ // Normalize the guid if necessary.
	1602	+ $out['guid'] = SyndicatedPost::normalize_guid($out['guid']);
	1603	+
536	1604	// Why the fuck doesn't wp_insert_post already do this?
537		- foreach ($this->post as $key => $value) :
	1605	+ foreach ($out as $key => $value) :
538	1606	if (is_string($value)) :
539	1607	$out[$key] = $wpdb->escape($value);
540	1608	else :
541	1609	$out[$key] = $value;
		@@ -541,19 +1609,8 @@
541	1609	$out[$key] = $value;
542	1610	endif;
543	1611	endforeach;
544	1612
545		- if (strlen($out['post_title'].$out['post_content'].$out['post_excerpt']) == 0) :
546		- // FIXME: Option for filtering out empty posts
547		- endif;
548		- if (strlen($out['post_title'])==0) :
549		- $offset = (int) get_option('gmt_offset') * 60 * 60;
550		- $out['post_title'] =
551		- $this->post['meta']['syndication_source']
552		- .' '.gmdate('Y-m-d H:i:s', $this->published() + $offset);
553		- // FIXME: Option for what to fill a blank title with...
554		- endif;
555		-
556	1613	return $out;
557	1614	}
558	1615
559	1616	/**
		@@ -561,22 +1618,43 @@
561	1618	*
562	1619	* @param array $dbpost An array representing the post we attempted to insert or update
563	1620	* @param mixed $ns A string or array representing the namespace (class, method) whence this method was called.
564	1621	*/
565		- function validate_post_id ($dbpost, $ns) {
	1622	+ function validate_post_id ($dbpost, $is_update, $ns) {
566	1623	if (is_array($ns)) : $ns = implode('::', $ns);
567	1624	else : $ns = (string) $ns; endif;
568		-
	1625	+
569	1626	// This should never happen.
570	1627	if (!is_numeric($this->_wp_id) or ($this->_wp_id == 0)) :
571		- FeedWordPress::critical_bug(
572		- /*name=*/ $ns.'::_wp_id',
	1628	+ $verb = ($is_update ? 'update existing' : 'insert new');
	1629	+ $guid = $this->guid();
	1630	+ $url = $this->permalink();
	1631	+ $feed = $this->link->uri(array('add_params' => true));
	1632	+
	1633	+ // wp_insert_post failed. Diagnostics, or barf up a critical bug
	1634	+ // notice if we are in debug mode.
	1635	+ $mesg = "Failed to $verb item [$guid]. WordPress API returned no valid post ID.\n"
	1636	+ ."\t\tID = ".serialize($this->_wp_id)."\n"
	1637	+ ."\t\tURL = ".FeedWordPress::val($url)
	1638	+ ."\t\tFeed = ".FeedWordPress::val($feed);
	1639	+ FeedWordPress::diagnostic('updated_feeds:errors', "WordPress API error: $mesg");
	1640	+ FeedWordPress::diagnostic('feed_items:rejected', $mesg);
	1641	+
	1642	+ $mesg = <<<EOM
	1643	+The WordPress API returned an invalid post ID
	1644	+ when FeedWordPress tried to $verb item $guid
	1645	+ [URL: $url]
	1646	+ from the feed at $feed
	1647	+
	1648	+$ns::_wp_id
	1649	+EOM;
	1650	+ FeedWordPress::noncritical_bug(
	1651	+ /*message=*/ $mesg,
573	1652	/*var =*/ array(
574	1653	"\$this->_wp_id" => $this->_wp_id,
575	1654	"\$dbpost" => $dbpost,
576		- "\$this" => $this
577	1655	),
578		- /*line # =*/ __LINE__
	1656	+ /*line # =*/ __LINE__, /*filename=*/ __FILE__
579	1657	);
580	1658	endif;
581	1659	} /* SyndicatedPost::validate_post_id() */
582	1660
		@@ -608,94 +1686,171 @@
608	1686	SET post_author={$this->post['post_author']}
609	1687	WHERE post_type = 'revision' AND ID='$revision_id'
610	1688	");
611	1689	} /* SyndicatedPost::fix_revision_meta () */
612		-
	1690	+
613	1691	/**
614		- * SyndicatedPost::avoid_kses_munge() -- If FeedWordPress is processing
615		- * an automatic update, that generally means that wp_insert_post() is
616		- * being called under the user credentials of whoever is viewing the
617		- * blog at the time -- usually meaning no user at all. But if WordPress
618		- * gets a wp_insert_post() when current_user_can('unfiltered_html') is
619		- * false, it will run the content of the post through a kses function
620		- * that strips out lots of HTML tags -- notably <object> and some others.
621		- * This causes problems for syndicating (for example) feeds that contain
622		- * YouTube videos. It also produces an unexpected asymmetry between
623		- * automatically-initiated updates and updates initiated manually from
624		- * the WordPress Dashboard (which are usually initiated under the
625		- * credentials of a logged-in admin, and so don't get run through the
626		- * kses function). So, to avoid the whole mess, what we do here is
627		- * just forcibly disable the kses munging for a single syndicated post,
628		- * by restoring the contents of the `post_content` field.
	1692	+ * SyndicatedPost::add_terms() -- if FeedWordPress is processing an
	1693	+ * automatic update, that generally means that wp_insert_post() is being
	1694	+ * called under the user credentials of whoever is viewing the blog at
	1695	+ * the time -- which usually means no user at all. But wp_insert_post()
	1696	+ * checks current_user_can() before assigning any of the terms in a
	1697	+ * post's tax_input structure -- which is unfortunate, since
	1698	+ * current_user_can() always returns FALSE when there is no current user
	1699	+ * logged in. Meaning that automatic updates get no terms assigned.
629	1700	*
630		- * @param string $content The content of the post, after other filters have gotten to it
631		- * @return string The original content of the post, before other filters had a chance to munge it.
	1701	+ * So, wp_insert_post() is not going to do the term assignments for us.
	1702	+ * If you want something done right....
	1703	+ *
	1704	+ * @param string $new_status Unused action parameter.
	1705	+ * @param string $old_status Unused action parameter.
	1706	+ * @param object $post The database record for the post just inserted.
632	1707	*/
633		- function avoid_kses_munge ($content) {
634		- global $wpdb;
635		- return $wpdb->escape($this->post['post_content']);
636		- }
637		-
638		- // SyndicatedPost::add_rss_meta: adds interesting meta-data to each entry
639		- // using the space for custom keys. The set of keys and values to add is
640		- // specified by the keys and values of $post['meta']. This is used to
641		- // store anything that the WordPress user might want to access from a
642		- // template concerning the post's original source that isn't provided
643		- // for by standard WP meta-data (i.e., any interesting data about the
644		- // syndicated post other than author, title, timestamp, categories, and
645		- // guid). It's also used to hook into WordPress's support for
646		- // enclosures.
647		- function add_rss_meta () {
	1708	+ function add_terms ($new_status, $old_status, $post) {
	1709	+ if ($new_status!='inherit') : // Bail if we are creating a revision.
	1710	+ if ( is_array($this->post) and isset($this->post['tax_input']) and is_array($this->post['tax_input']) ) :
	1711	+ foreach ($this->post['tax_input'] as $taxonomy => $terms) :
	1712	+ if (is_array($terms)) :
	1713	+ $terms = array_filter($terms); // strip out empties
	1714	+ endif;
	1715	+ wp_set_post_terms($post->ID, $terms, $taxonomy);
	1716	+ endforeach;
	1717	+ endif;
	1718	+ endif;
	1719	+ } /* SyndicatedPost::add_terms () */
	1720	+
	1721	+ /**
	1722	+ * SyndicatedPost::fix_post_modified_ts() -- We would like to set
	1723	+ * post_modified and post_modified_gmt to reflect the value of
	1724	+ * <atom:updated> or equivalent elements on the feed. Unfortunately,
	1725	+ * wp_insert_post() refuses to acknowledge explicitly-set post_modified
	1726	+ * fields and overwrites them, either with the post_date (if new) or the
	1727	+ * current timestamp (if updated).
	1728	+ *
	1729	+ * So, wp_insert_post() is not going to do the last-modified assignments
	1730	+ * for us. If you want something done right....
	1731	+ *
	1732	+ * @param string $new_status Unused action parameter.
	1733	+ * @param string $old_status Unused action parameter.
	1734	+ * @param object $post The database record for the post just inserted.
	1735	+ */
	1736	+ function fix_post_modified_ts ($new_status, $old_status, $post) {
648	1737	global $wpdb;
649		- if ( is_array($this->post) and isset($this->post['meta']) and is_array($this->post['meta']) ) :
650		- $postId = $this->wp_id();
651		-
652		- // Aggregated posts should NOT send out pingbacks.
653		- // WordPress 2.1-2.2 claim you can tell them not to
654		- // using $post_pingback, but they don't listen, so we
655		- // make sure here.
656		- $result = $wpdb->query("
657		- DELETE FROM $wpdb->postmeta
658		- WHERE post_id='$postId' AND meta_key='_pingme'
659		- ");
	1738	+ if ($new_status!='inherit') : // Bail if we are creating a revision.
	1739	+ $wpdb->update( $wpdb->posts, /*data=*/ array(
	1740	+ 'post_modified' => $this->post['post_modified'],
	1741	+ 'post_modified_gmt' => $this->post['post_modified_gmt'],
	1742	+ ), /*where=*/ array('ID' => $post->ID) );
	1743	+ endif;
	1744	+ } /* SyndicatedPost::fix_post_modified_ts () */
	1745	+
	1746	+ /**
	1747	+ * SyndicatedPost::add_rss_meta: adds interesting meta-data to each entry
	1748	+ * using the space for custom keys. The set of keys and values to add is
	1749	+ * specified by the keys and values of $post['meta']. This is used to
	1750	+ * store anything that the WordPress user might want to access from a
	1751	+ * template concerning the post's original source that isn't provided
	1752	+ * for by standard WP meta-data (i.e., any interesting data about the
	1753	+ * syndicated post other than author, title, timestamp, categories, and
	1754	+ * guid). It's also used to hook into WordPress's support for
	1755	+ * enclosures.
	1756	+ *
	1757	+ * @param string $new_status Unused action parameter.
	1758	+ * @param string $old_status Unused action parameter.
	1759	+ * @param object $post The database record for the post just inserted.
	1760	+ */
	1761	+ function add_rss_meta ($new_status, $old_status, $post) {
	1762	+ global $wpdb;
	1763	+ if ($new_status!='inherit') : // Bail if we are creating a revision.
	1764	+ FeedWordPress::diagnostic('syndicated_posts:meta_data', 'Adding post meta-data: {'.implode(", ", array_keys($this->post['meta'])).'}');
660	1765
661		- foreach ( $this->post['meta'] as $key => $values ) :
662		-
663		- $key = $wpdb->escape($key);
664		-
665		- // If this is an update, clear out the old
666		- // values to avoid duplication.
	1766	+ if ( is_array($this->post) and isset($this->post['meta']) and is_array($this->post['meta']) ) :
	1767	+ $postId = $post->ID;
	1768	+
	1769	+ // Aggregated posts should NOT send out pingbacks.
	1770	+ // WordPress 2.1-2.2 claim you can tell them not to
	1771	+ // using $post_pingback, but they don't listen, so we
	1772	+ // make sure here.
667	1773	$result = $wpdb->query("
668	1774	DELETE FROM $wpdb->postmeta
669		- WHERE post_id='$postId' AND meta_key='$key'
	1775	+ WHERE post_id='$postId' AND meta_key='_pingme'
670	1776	");
671		-
672		- // Allow for either a single value or an array
673		- if (!is_array($values)) $values = array($values);
674		- foreach ( $values as $value ) :
675		- $value = $wpdb->escape($value);
	1777	+
	1778	+ foreach ( $this->post['meta'] as $key => $values ) :
	1779	+ $eKey = $wpdb->escape($key);
	1780	+
	1781	+ // If this is an update, clear out the old
	1782	+ // values to avoid duplication.
676	1783	$result = $wpdb->query("
677		- INSERT INTO $wpdb->postmeta
678		- SET
679		- post_id='$postId',
680		- meta_key='$key',
681		- meta_value='$value'
	1784	+ DELETE FROM $wpdb->postmeta
	1785	+ WHERE post_id='$postId' AND meta_key='$eKey'
682	1786	");
	1787	+
	1788	+ // Allow for either a single value or an array
	1789	+ if (!is_array($values)) $values = array($values);
	1790	+ foreach ( $values as $value ) :
	1791	+ FeedWordPress::diagnostic('syndicated_posts:meta_data', "Adding post meta-datum to post [$postId]: [$key] = ".FeedWordPress::val($value, /*no newlines=*/ true));
	1792	+ add_post_meta($postId, $key, $value, /*unique=*/ false);
	1793	+ endforeach;
683	1794	endforeach;
684		- endforeach;
	1795	+ endif;
685	1796	endif;
686	1797	} /* SyndicatedPost::add_rss_meta () */
687	1798
688		- // SyndicatedPost::author_id (): get the ID for an author name from
689		- // the feed. Create the author if necessary.
	1799	+ /**
	1800	+ * SyndicatedPost::author_id (): get the ID for an author name from
	1801	+ * the feed. Create the author if necessary.
	1802	+ *
	1803	+ * @param string $unfamiliar_author
	1804	+ *
	1805	+ * @return NULL|int The numeric ID of the author to attribute the post to
	1806	+ * NULL if the post should be filtered out.
	1807	+ */
690	1808	function author_id ($unfamiliar_author = 'create') {
691	1809	global $wpdb;
692	1810
693		- $a = $this->author();
694		- $author = $a['name'];
695		- $email = $a['email'];
696		- $url = $a['uri'];
	1811	+ $a = $this->named['author'];
697	1812
	1813	+ $source = $this->source();
	1814	+ $forbidden = apply_filters('feedwordpress_forbidden_author_names',
	1815	+ array('admin', 'administrator', 'www', 'root'));
	1816	+
	1817	+ // Prepare the list of candidates to try for author name: name from
	1818	+ // feed, original source title (if any), immediate source title live
	1819	+ // from feed, subscription title, prettied version of feed homepage URL,
	1820	+ // prettied version of feed URL, or, failing all, use "unknown author"
	1821	+ // as last resort
	1822	+
	1823	+ $candidates = array();
	1824	+ $candidates[] = $a['name'];
	1825	+ if (!is_null($source)) : $candidates[] = $source['title']; endif;
	1826	+ $candidates[] = $this->link->name(/*fromFeed=*/ true);
	1827	+ $candidates[] = $this->link->name(/*fromFeed=*/ false);
	1828	+ if (strlen($this->link->homepage()) > 0) : $candidates[] = feedwordpress_display_url($this->link->homepage()); endif;
	1829	+ $candidates[] = feedwordpress_display_url($this->link->uri());
	1830	+ $candidates[] = 'unknown author';
	1831	+
	1832	+ // Pick the first one that works from the list, screening against empty
	1833	+ // or forbidden names.
	1834	+
	1835	+ $author = NULL;
	1836	+ while (is_null($author) and ($candidate = each($candidates))) :
	1837	+ if (!is_null($candidate['value'])
	1838	+ and (strlen(trim($candidate['value'])) > 0)
	1839	+ and !in_array(strtolower(trim($candidate['value'])), $forbidden)) :
	1840	+ $author = $candidate['value'];
	1841	+ endif;
	1842	+ endwhile;
	1843	+
	1844	+ $email = (isset($a['email']) ? $a['email'] : NULL);
	1845	+ $authorUrl = (isset($a['uri']) ? $a['uri'] : NULL);
	1846	+
	1847	+
	1848	+ $hostUrl = $this->link->homepage();
	1849	+ if (is_null($hostUrl) or (strlen($hostUrl) < 0)) :
	1850	+ $hostUrl = $this->link->uri();
	1851	+ endif;
	1852	+
698	1853	$match_author_by_email = !('yes' == get_option("feedwordpress_do_not_match_author_by_email"));
699	1854	if ($match_author_by_email and !FeedWordPress::is_null_email($email)) :
700	1855	$test_email = $email;
701	1856	else :
		@@ -703,8 +1858,28 @@
703	1858	endif;
704	1859
705	1860	// Never can be too careful...
706	1861	$login = sanitize_user($author, /*strict=*/ true);
	1862	+
	1863	+ // Possible for, e.g., foreign script author names
	1864	+ if (strlen($login) < 1) :
	1865	+ // No usable characters in author name for a login.
	1866	+ // (Sometimes results from, e.g., foreign scripts.)
	1867	+ //
	1868	+ // We just need *something* in Western alphanumerics,
	1869	+ // so let's try the domain name.
	1870	+ //
	1871	+ // Uniqueness will be guaranteed below if necessary.
	1872	+
	1873	+ $url = parse_url($hostUrl);
	1874	+
	1875	+ $login = sanitize_user($url['host'], /*strict=*/ true);
	1876	+ if (strlen($login) < 1) :
	1877	+ // This isn't working. Frak it.
	1878	+ $login = 'syndicated';
	1879	+ endif;
	1880	+ endif;
	1881	+
707	1882	$login = apply_filters('pre_user_login', $login);
708	1883
709	1884	$nice_author = sanitize_title($author);
710	1885	$nice_author = apply_filters('pre_user_nicename', $nice_author);
		@@ -712,12 +1887,14 @@
712	1887	$reg_author = $wpdb->escape(preg_quote($author));
713	1888	$author = $wpdb->escape($author);
714	1889	$email = $wpdb->escape($email);
715	1890	$test_email = $wpdb->escape($test_email);
716		- $url = $wpdb->escape($url);
	1891	+ $authorUrl = $wpdb->escape($authorUrl);
717	1892
718	1893	// Check for an existing author rule....
719		- if (isset($this->link->settings['map authors']['name'][strtolower(trim($author))])) :
	1894	+ if (isset($this->link->settings['map authors']['name']['*'])) :
	1895	+ $author_rule = $this->link->settings['map authors']['name']['*'];
	1896	+ elseif (isset($this->link->settings['map authors']['name'][strtolower(trim($author))])) :
720	1897	$author_rule = $this->link->settings['map authors']['name'][strtolower(trim($author))];
721	1898	else :
722	1899	$author_rule = NULL;
723	1900	endif;
		@@ -732,65 +1909,34 @@
732	1909
733	1910	else :
734	1911	// Check the database for an existing author record that might fit
735	1912
736		- #-- WordPress 2.0+
737		- if (fwp_test_wp_version(FWP_SCHEMA_HAS_USERMETA)) :
738		-
739		- // First try the user core data table.
	1913	+ // First try the user core data table.
	1914	+ $id = $wpdb->get_var(
	1915	+ "SELECT ID FROM $wpdb->users
	1916	+ WHERE TRIM(LCASE(display_name)) = TRIM(LCASE('$author'))
	1917	+ OR TRIM(LCASE(user_login)) = TRIM(LCASE('$author'))
	1918	+ OR (
	1919	+ LENGTH(TRIM(LCASE(user_email))) > 0
	1920	+ AND TRIM(LCASE(user_email)) = TRIM(LCASE('$test_email'))
	1921	+ )");
	1922	+
	1923	+ // If that fails, look for aliases in the user meta data table
	1924	+ if (is_null($id)) :
740	1925	$id = $wpdb->get_var(
741		- "SELECT ID FROM $wpdb->users
	1926	+ "SELECT user_id FROM $wpdb->usermeta
742	1927	WHERE
743		- TRIM(LCASE(user_login)) = TRIM(LCASE('$login'))
	1928	+ (meta_key = 'description' AND TRIM(LCASE(meta_value)) = TRIM(LCASE('$author')))
744	1929	OR (
745		- LENGTH(TRIM(LCASE(user_email))) > 0
746		- AND TRIM(LCASE(user_email)) = TRIM(LCASE('$test_email'))
747		- )
748		- OR TRIM(LCASE(user_nicename)) = TRIM(LCASE('$nice_author'))
749		- ");
750		-
751		- // If that fails, look for aliases in the user meta data table
752		- if (is_null($id)) :
753		- $id = $wpdb->get_var(
754		- "SELECT user_id FROM $wpdb->usermeta
755		- WHERE
756		- (meta_key = 'description' AND TRIM(LCASE(meta_value)) = TRIM(LCASE('$author')))
757		- OR (
758		- meta_key = 'description'
759		- AND TRIM(LCASE(meta_value))
760		- RLIKE CONCAT(
761		- '(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*',
762		- TRIM(LCASE('$reg_author')),
763		- '( |\\t|\\r)*(\\n|\$)'
764		- )
765		- )
766		- ");
767		- endif;
768		-
769		- #-- WordPress 1.5.x
770		- else :
771		- $id = $wpdb->get_var(
772		- "SELECT ID from $wpdb->users
773		- WHERE
774		- TRIM(LCASE(user_login)) = TRIM(LCASE('$login')) OR
775		- (
776		- LENGTH(TRIM(LCASE(user_email))) > 0
777		- AND TRIM(LCASE(user_email)) = TRIM(LCASE('$test_email'))
778		- ) OR
779		- TRIM(LCASE(user_firstname)) = TRIM(LCASE('$author')) OR
780		- TRIM(LCASE(user_nickname)) = TRIM(LCASE('$author')) OR
781		- TRIM(LCASE(user_nicename)) = TRIM(LCASE('$nice_author')) OR
782		- TRIM(LCASE(user_description)) = TRIM(LCASE('$author')) OR
783		- (
784		- LOWER(user_description)
	1930	+ meta_key = 'description'
	1931	+ AND TRIM(LCASE(meta_value))
785	1932	RLIKE CONCAT(
786	1933	'(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*',
787		- LCASE('$reg_author'),
	1934	+ TRIM(LCASE('$reg_author')),
788	1935	'( |\\t|\\r)*(\\n|\$)'
789	1936	)
790	1937	)
791	1938	");
792		-
793	1939	endif;
794	1940
795	1941	// ... if you don't find one, then do what you need to do
796	1942	if (is_null($id)) :
		@@ -796,8 +1942,16 @@
796	1942	if (is_null($id)) :
797	1943	if ($unfamiliar_author === 'create') :
798	1944	$userdata = array();
799	1945
	1946	+ // WordPress 3 is going to pitch a fit if we attempt to register
	1947	+ // more than one user account with an empty e-mail address, so we
	1948	+ // need *something* here. Ugh.
	1949	+ if (strlen($email) == 0 or FeedWordPress::is_null_email($email)) :
	1950	+ $url = parse_url($hostUrl);
	1951	+ $email = $nice_author.'@'.$url['host'];
	1952	+ endif;
	1953	+
800	1954	#-- user table data
801	1955	$userdata['ID'] = NULL; // new user
802	1956	$userdata['user_login'] = $login;
803	1957	$userdata['user_nicename'] = $nice_author;
		@@ -802,12 +1956,39 @@
802	1956	$userdata['user_login'] = $login;
803	1957	$userdata['user_nicename'] = $nice_author;
804	1958	$userdata['user_pass'] = substr(md5(uniqid(microtime())), 0, 6); // just something random to lock it up
805	1959	$userdata['user_email'] = $email;
806		- $userdata['user_url'] = $url;
	1960	+ $userdata['user_url'] = $authorUrl;
	1961	+ $userdata['nickname'] = $author;
	1962	+ $parts = preg_split('/\s+/', trim($author), 2);
	1963	+ if (isset($parts[0])) : $userdata['first_name'] = $parts[0]; endif;
	1964	+ if (isset($parts[1])) : $userdata['last_name'] = $parts[1]; endif;
807	1965	$userdata['display_name'] = $author;
808		-
809		- $id = wp_insert_user($userdata);
	1966	+ $userdata['role'] = 'contributor';
	1967	+
	1968	+ do { // Keep trying until you get it right. Or until PHP crashes, I guess.
	1969	+ $id = wp_insert_user($userdata);
	1970	+ if (is_wp_error($id)) :
	1971	+ $codes = $id->get_error_code();
	1972	+ switch ($codes) :
	1973	+ case 'empty_user_login' :
	1974	+ case 'existing_user_login' :
	1975	+ // Add a random disambiguator
	1976	+ $userdata['user_login'] .= substr(md5(uniqid(microtime())), 0, 6);
	1977	+ break;
	1978	+ case 'existing_user_email' :
	1979	+ // No disassemble!
	1980	+ $parts = explode('@', $userdata['user_email'], 2);
	1981	+
	1982	+ // Add a random disambiguator as a gmail-style username extension
	1983	+ $parts[0] .= '+'.substr(md5(uniqid(microtime())), 0, 6);
	1984	+
	1985	+ // Reassemble
	1986	+ $userdata['user_email'] = $parts[0].'@'.$parts[1];
	1987	+ break;
	1988	+ endswitch;
	1989	+ endif;
	1990	+ } while (is_wp_error($id));
810	1991	elseif (is_numeric($unfamiliar_author) and get_userdata((int) $unfamiliar_author)) :
811	1992	$id = (int) $unfamiliar_author;
812	1993	elseif ($unfamiliar_author === 'default') :
813	1994	$id = 1;
		@@ -816,16 +1997,38 @@
816	1997	endif;
817	1998
818	1999	if ($id) :
819	2000	$this->link->settings['map authors']['name'][strtolower(trim($author))] = $id;
	2001	+
	2002	+ // Multisite: Check whether the author has been recorded
	2003	+ // on *this* blog before. If not, put her down as a
	2004	+ // Contributor for *this* blog.
	2005	+ $user = new WP_User((int) $id);
	2006	+ if (empty($user->roles)) :
	2007	+ $user->add_role('contributor');
	2008	+ endif;
820	2009	endif;
821	2010	return $id;
822		- } // function SyndicatedPost::author_id ()
	2011	+ } /* function SyndicatedPost::author_id () */
823	2012
824		- // look up (and create) category ids from a list of categories
825		- function category_ids ($cats, $unfamiliar_category = 'create', $tags_too = false) {
826		- global $wpdb;
	2013	+ /**
	2014	+ * category_ids: look up (and create) category ids from a list of categories
	2015	+ *
	2016	+ * @param array $cats
	2017	+ * @param string $unfamiliar_category
	2018	+ * @param array|null $taxonomies
	2019	+ * @return array
	2020	+ */
	2021	+ function category_ids ($cats, $unfamiliar_category = 'create', $taxonomies = NULL, $params = array()) {
	2022	+ $singleton = (isset($params['singleton']) ? $params['singleton'] : true);
	2023	+ $allowFilters = (isset($params['filters']) ? $params['filters'] : false);
	2024	+
	2025	+ $catTax = 'category';
827	2026
	2027	+ if (is_null($taxonomies)) :
	2028	+ $taxonomies = array('category');
	2029	+ endif;
	2030	+
828	2031	// We need to normalize whitespace because (1) trailing
829	2032	// whitespace can cause PHP and MySQL not to see eye to eye on
830	2033	// VARCHAR comparisons for some versions of MySQL (cf.
831	2034	// <http://dev.mysql.com/doc/mysql/en/char.html>), and (2)
		@@ -832,96 +2035,100 @@
832	2035	// because I doubt most people want to make a semantic
833	2036	// distinction between 'Computers' and 'Computers '
834	2037	$cats = array_map('trim', $cats);
835	2038
836		- $tags = array();
	2039	+ $terms = array();
	2040	+ foreach ($taxonomies as $tax) :
	2041	+ $terms[$tax] = array();
	2042	+ endforeach;
	2043	+
	2044	+ foreach ($cats as $cat_name) :
	2045	+ if (preg_match('/^{([^#}]*)#([0-9]+)}$/', $cat_name, $backref)) :
	2046	+ $cat_id = (int) $backref[2];
	2047	+ $tax = $backref[1];
	2048	+ if (strlen($tax) < 1) :
	2049	+ $tax = $catTax;
	2050	+ endif;
837	2051
838		- $cat_ids = array ();
839		- foreach ($cats as $cat_name) :
840		- if (preg_match('/^{#([0-9]+)}$/', $cat_name, $backref)) :
841		- $cat_id = (int) $backref[1];
842		- if (function_exists('is_term') and is_term($cat_id, 'category')) :
843		- $cat_ids[] = $cat_id;
844		- elseif (get_category($cat_id)) :
845		- $cat_ids[] = $cat_id;
	2052	+ $term = term_exists($cat_id, $tax);
	2053	+ if (!is_wp_error($term) and !!$term) :
	2054	+ if (!isset($terms[$tax])) :
	2055	+ $terms[$tax] = array();
	2056	+ endif;
	2057	+ $terms[$tax][] = $cat_id;
846	2058	endif;
847	2059	elseif (strlen($cat_name) > 0) :
848		- $esc = $wpdb->escape($cat_name);
849		- $resc = $wpdb->escape(preg_quote($cat_name));
850		-
851		- // WordPress 2.3+
852		- if (function_exists('is_term')) :
853		- $cat_id = is_term($cat_name, 'category');
854		- if ($cat_id) :
855		- $cat_ids[] = $cat_id['term_id'];
856		- // There must be a better way to do this...
857		- elseif ($results = $wpdb->get_results(
858		- "SELECT term_id
859		- FROM $wpdb->term_taxonomy
860		- WHERE
861		- LOWER(description) RLIKE
862		- CONCAT('(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*', LOWER('{$resc}'), '( |\\t|\\r)*(\\n|\$)')"
863		- )) :
864		- foreach ($results AS $term) :
865		- $cat_ids[] = (int) $term->term_id;
866		- endforeach;
867		- elseif ('tag'==$unfamiliar_category) :
868		- $tags[] = $cat_name;
869		- elseif ('create'===$unfamiliar_category) :
870		- $term = wp_insert_term($cat_name, 'category');
871		- if (is_wp_error($term)) :
872		- FeedWordPress::noncritical_bug('term insertion problem', array('cat_name' => $cat_name, 'term' => $term, 'this' => $this), __LINE__);
873		- else :
874		- $cat_ids[] = $term['term_id'];
	2060	+ $familiar = false;
	2061	+ foreach ($taxonomies as $tax) :
	2062	+ if ($tax!='category' or strtolower($cat_name)!='uncategorized') :
	2063	+ $term = term_exists($cat_name, $tax);
	2064	+ if (!is_wp_error($term) and !!$term) :
	2065	+ $familiar = true;
	2066	+
	2067	+ if (is_array($term)) :
	2068	+ $term_id = (int) $term['term_id'];
	2069	+ else :
	2070	+ $term_id = (int) $term;
	2071	+ endif;
	2072	+
	2073	+ if (!isset($terms[$tax])) :
	2074	+ $terms[$tax] = array();
	2075	+ endif;
	2076	+ $terms[$tax][] = $term_id;
	2077	+ break; // We're done here.
875	2078	endif;
876	2079	endif;
	2080	+ endforeach;
877	2081
878		- // WordPress 1.5.x - 2.2.x
879		- else :
880		- $results = $wpdb->get_results(
881		- "SELECT cat_ID
882		- FROM $wpdb->categories
883		- WHERE
884		- (LOWER(cat_name) = LOWER('$esc'))
885		- OR (LOWER(category_description)
886		- RLIKE CONCAT('(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*', LOWER('{$resc}'), '( |\\t|\\r)*(\\n|\$)'))
887		- ");
888		- if ($results) :
889		- foreach ($results as $term) :
890		- $cat_ids[] = (int) $term->cat_ID;
891		- endforeach;
892		- elseif ('create'===$unfamiliar_category) :
893		- if (function_exists('wp_insert_category')) :
894		- $cat_id = wp_insert_category(array('cat_name' => $cat_name));
895		- // And into the database we go.
	2082	+ if (!$familiar) :
	2083	+ if ('tag'==$unfamiliar_category) :
	2084	+ $unfamiliar_category = 'create:post_tag';
	2085	+ endif;
	2086	+
	2087	+ if (preg_match('/^create(:(.*))?$/i', $unfamiliar_category, $ref)) :
	2088	+ $tax = $catTax; // Default
	2089	+ if (isset($ref[2]) and strlen($ref[2]) > 2) :
	2090	+ $tax = $ref[2];
	2091	+ endif;
	2092	+ $term = wp_insert_term($cat_name, $tax);
	2093	+ if (is_wp_error($term)) :
	2094	+ FeedWordPress::noncritical_bug('term insertion problem', array('cat_name' => $cat_name, 'term' => $term, 'this' => $this), __LINE__, __FILE__);
896	2095	else :
897		- $nice_kitty = sanitize_title($cat_name);
898		- $wpdb->query(sprintf("
899		- INSERT INTO $wpdb->categories
900		- SET
901		- cat_name='%s',
902		- category_nicename='%s'
903		- ", $wpdb->escape($cat_name), $nice_kitty
904		- ));
905		- $cat_id = $wpdb->insert_id;
	2096	+ if (!isset($terms[$tax])) :
	2097	+ $terms[$tax] = array();
	2098	+ endif;
	2099	+ $terms[$tax][] = (int) $term['term_id'];
906	2100	endif;
907		- $cat_ids[] = $cat_id;
908	2101	endif;
909	2102	endif;
910	2103	endif;
911	2104	endforeach;
912	2105
913		- if ((count($cat_ids) == 0) and ($unfamiliar_category === 'filter')) :
914		- $cat_ids = NULL; // Drop the post
915		- else :
916		- $cat_ids = array_unique($cat_ids);
	2106	+ $filtersOn = $allowFilters;
	2107	+ if ($allowFilters) :
	2108	+ $filters = array_filter(
	2109	+ $this->link->setting('match/filter', 'match_filter', array()),
	2110	+ 'remove_dummy_zero'
	2111	+ );
	2112	+ $filtersOn = ($filtersOn and is_array($filters) and (count($filters) > 0));
917	2113	endif;
918	2114
919		- if ($tags_too) : $ret = array($cat_ids, $tags);
920		- else : $ret = $cat_ids;
	2115	+ // Check for filter conditions
	2116	+ foreach ($terms as $tax => $term_ids) :
	2117	+ if ($filtersOn
	2118	+ and (count($term_ids)==0)
	2119	+ and in_array($tax, $filters)) :
	2120	+ $terms = NULL; // Drop the post
	2121	+ break;
	2122	+ else :
	2123	+ $terms[$tax] = array_unique($term_ids);
	2124	+ endif;
	2125	+ endforeach;
	2126	+
	2127	+ if ($singleton and count($terms)==1) : // If we only searched one, just return the term IDs
	2128	+ $terms = end($terms);
921	2129	endif;
922		-
923		- return $ret;
	2130	+ return $terms;
924	2131	} // function SyndicatedPost::category_ids ()
925	2132
926	2133	function use_api ($tag) {
927	2134	global $wp_db_version;
		@@ -937,258 +2144,6 @@
937	2144	endswitch;
938	2145	return $ret;
939	2146	} // function SyndicatedPost::use_api ()
940	2147
941		- #### EXTRACT DATA FROM FEED ITEM ####
942		-
943		- function created () {
944		- $epoch = null;
945		- if (isset($this->item['dc']['created'])) :
946		- $epoch = @parse_w3cdtf($this->item['dc']['created']);
947		- elseif (isset($this->item['dcterms']['created'])) :
948		- $epoch = @parse_w3cdtf($this->item['dcterms']['created']);
949		- elseif (isset($this->item['created'])): // Atom 0.3
950		- $epoch = @parse_w3cdtf($this->item['created']);
951		- endif;
952		- return $epoch;
953		- }
954		- function published ($fallback = true) {
955		- $epoch = null;
956		-
957		- # RSS is a fucking mess. Figure out whether we have a date in
958		- # <dc:date>, <issued>, <pubDate>, etc., and get it into Unix
959		- # epoch format for reformatting. If we can't find anything,
960		- # we'll use the last-updated time.
961		- if (isset($this->item['dc']['date'])): // Dublin Core
962		- $epoch = @parse_w3cdtf($this->item['dc']['date']);
963		- elseif (isset($this->item['dcterms']['issued'])) : // Dublin Core extensions
964		- $epoch = @parse_w3cdtf($this->item['dcterms']['issued']);
965		- elseif (isset($this->item['published'])) : // Atom 1.0
966		- $epoch = @parse_w3cdtf($this->item['published']);
967		- elseif (isset($this->item['issued'])): // Atom 0.3
968		- $epoch = @parse_w3cdtf($this->item['issued']);
969		- elseif (isset($this->item['pubdate'])): // RSS 2.0
970		- $epoch = strtotime($this->item['pubdate']);
971		- elseif ($fallback) : // Fall back to <updated> / <modified> if present
972		- $epoch = $this->updated(/*fallback=*/ false);
973		- endif;
974		-
975		- # If everything failed, then default to the current time.
976		- if (is_null($epoch)) :
977		- if (-1 == $default) :
978		- $epoch = time();
979		- else :
980		- $epoch = $default;
981		- endif;
982		- endif;
983		-
984		- return $epoch;
985		- }
986		- function updated ($fallback = true, $default = -1) {
987		- $epoch = null;
988		-
989		- # As far as I know, only dcterms and Atom have reliable ways to
990		- # specify when something was *modified* last. If neither is
991		- # available, then we'll try to get the time of publication.
992		- if (isset($this->item['dc']['modified'])) : // Not really correct
993		- $epoch = @parse_w3cdtf($this->item['dc']['modified']);
994		- elseif (isset($this->item['dcterms']['modified'])) : // Dublin Core extensions
995		- $epoch = @parse_w3cdtf($this->item['dcterms']['modified']);
996		- elseif (isset($this->item['modified'])): // Atom 0.3
997		- $epoch = @parse_w3cdtf($this->item['modified']);
998		- elseif (isset($this->item['updated'])): // Atom 1.0
999		- $epoch = @parse_w3cdtf($this->item['updated']);
1000		- elseif ($fallback) : // Fall back to issued / dc:date
1001		- $epoch = $this->published(/*fallback=*/ false, /*default=*/ $default);
1002		- endif;
1003		-
1004		- # If everything failed, then default to the current time.
1005		- if (is_null($epoch)) :
1006		- if (-1 == $default) :
1007		- $epoch = time();
1008		- else :
1009		- $epoch = $default;
1010		- endif;
1011		- endif;
1012		-
1013		- return $epoch;
1014		- }
1015		-
1016		- function update_hash () {
1017		- return md5(serialize($this->item));
1018		- }
1019		-
1020		- function guid () {
1021		- $guid = null;
1022		- if (isset($this->item['id'])): // Atom 0.3 / 1.0
1023		- $guid = $this->item['id'];
1024		- elseif (isset($this->item['atom']['id'])) : // Namespaced Atom
1025		- $guid = $this->item['atom']['id'];
1026		- elseif (isset($this->item['guid'])) : // RSS 2.0
1027		- $guid = $this->item['guid'];
1028		- elseif (isset($this->item['dc']['identifier'])) :// yeah, right
1029		- $guid = $this->item['dc']['identifier'];
1030		- else :
1031		- // The feed does not seem to have provided us with a
1032		- // unique identifier, so we'll have to cobble together
1033		- // a tag: URI that might work for us. The base of the
1034		- // URI will be the host name of the feed source ...
1035		- $bits = parse_url($this->feedmeta['link/uri']);
1036		- $guid = 'tag:'.$bits['host'];
1037		-
1038		- // If we have a date of creation, then we can use that
1039		- // to uniquely identify the item. (On the other hand, if
1040		- // the feed producer was consicentious enough to
1041		- // generate dates of creation, she probably also was
1042		- // conscientious enough to generate unique identifiers.)
1043		- if (!is_null($this->created())) :
1044		- $guid .= '://post.'.date('YmdHis', $this->created());
1045		-
1046		- // Otherwise, use both the URI of the item, *and* the
1047		- // item's title. We have to use both because titles are
1048		- // often not unique, and sometimes links aren't unique
1049		- // either (e.g. Bitch (S)HITLIST, Mozilla Dot Org news,
1050		- // some podcasts). But it's rare to have *both* the same
1051		- // title *and* the same link for two different items. So
1052		- // this is about the best we can do.
1053		- else :
1054		- $guid .= '://'.md5($this->item['link'].'/'.$this->item['title']);
1055		- endif;
1056		- endif;
1057		- return $guid;
1058		- }
1059		-
1060		- function author () {
1061		- $author = array ();
1062		-
1063		- if (isset($this->item['author_name'])):
1064		- $author['name'] = $this->item['author_name'];
1065		- elseif (isset($this->item['dc']['creator'])):
1066		- $author['name'] = $this->item['dc']['creator'];
1067		- elseif (isset($this->item['dc']['contributor'])):
1068		- $author['name'] = $this->item['dc']['contributor'];
1069		- elseif (isset($this->feed->channel['dc']['creator'])) :
1070		- $author['name'] = $this->feed->channel['dc']['creator'];
1071		- elseif (isset($this->feed->channel['dc']['contributor'])) :
1072		- $author['name'] = $this->feed->channel['dc']['contributor'];
1073		- elseif (isset($this->feed->channel['author_name'])) :
1074		- $author['name'] = $this->feed->channel['author_name'];
1075		- elseif ($this->feed->is_rss() and isset($this->item['author'])) :
1076		- // The author element in RSS is allegedly an
1077		- // e-mail address, but lots of people don't use
1078		- // it that way. So let's make of it what we can.
1079		- $author = parse_email_with_realname($this->item['author']);
1080		-
1081		- if (!isset($author['name'])) :
1082		- if (isset($author['email'])) :
1083		- $author['name'] = $author['email'];
1084		- else :
1085		- $author['name'] = $this->feed->channel['title'];
1086		- endif;
1087		- endif;
1088		- else :
1089		- $author['name'] = $this->feed->channel['title'];
1090		- endif;
1091		-
1092		- if (isset($this->item['author_email'])):
1093		- $author['email'] = $this->item['author_email'];
1094		- elseif (isset($this->feed->channel['author_email'])) :
1095		- $author['email'] = $this->feed->channel['author_email'];
1096		- endif;
1097		-
1098		- if (isset($this->item['author_url'])):
1099		- $author['uri'] = $this->item['author_url'];
1100		- elseif (isset($this->feed->channel['author_url'])) :
1101		- $author['uri'] = $this->item['author_url'];
1102		- else:
1103		- $author['uri'] = $this->feed->channel['link'];
1104		- endif;
1105		-
1106		- return $author;
1107		- } // SyndicatedPost::author()
1108		-
1109		- var $uri_attrs = array (
1110		- array('a', 'href'),
1111		- array('applet', 'codebase'),
1112		- array('area', 'href'),
1113		- array('blockquote', 'cite'),
1114		- array('body', 'background'),
1115		- array('del', 'cite'),
1116		- array('form', 'action'),
1117		- array('frame', 'longdesc'),
1118		- array('frame', 'src'),
1119		- array('iframe', 'longdesc'),
1120		- array('iframe', 'src'),
1121		- array('head', 'profile'),
1122		- array('img', 'longdesc'),
1123		- array('img', 'src'),
1124		- array('img', 'usemap'),
1125		- array('input', 'src'),
1126		- array('input', 'usemap'),
1127		- array('ins', 'cite'),
1128		- array('link', 'href'),
1129		- array('object', 'classid'),
1130		- array('object', 'codebase'),
1131		- array('object', 'data'),
1132		- array('object', 'usemap'),
1133		- array('q', 'cite'),
1134		- array('script', 'src')
1135		- ); /* var SyndicatedPost::$uri_attrs */
1136		-
1137		- var $_base = null;
1138		-
1139		- function resolve_single_relative_uri ($refs) {
1140		- $tag = FeedWordPressHTML::attributeMatch($refs);
1141		- $url = Relative_URI::resolve($tag['value'], $this->_base);
1142		- return $tag['prefix'] . $url . $tag['suffix'];
1143		- } /* function SyndicatedPost::resolve_single_relative_uri() */
1144		-
1145		- function resolve_relative_uris ($content, $obj) {
1146		- # The MagpieRSS upgrade has some `xml:base` support baked in.
1147		- # However, sometimes people do silly things, like putting
1148		- # relative URIs out on a production RSS 2.0 feed or other feeds
1149		- # with no good support for `xml:base`. So we'll do our best to
1150		- # try to catch any remaining relative URIs and resolve them as
1151		- # best we can.
1152		- $obj->_base = $obj->item['link']; // Reset the base for resolving relative URIs
1153		-
1154		- foreach ($obj->uri_attrs as $pair) :
1155		- list($tag, $attr) = $pair;
1156		- $pattern = FeedWordPressHTML::attributeRegex($tag, $attr);
1157		- $content = preg_replace_callback (
1158		- $pattern,
1159		- array(&$obj, 'resolve_single_relative_uri'),
1160		- $content
1161		- );
1162		- endforeach;
1163		-
1164		- return $content;
1165		- } /* function SyndicatedPost::resolve_relative_uris () */
1166		-
1167		- var $strip_attrs = array (
1168		- array('[a-z]+', 'target'),
1169		-// array('[a-z]+', 'style'),
1170		-// array('[a-z]+', 'on[a-z]+'),
1171		- );
1172		-
1173		- function strip_attribute_from_tag ($refs) {
1174		- $tag = FeedWordPressHTML::attributeMatch($refs);
1175		- return $tag['before_attribute'].$tag['after_attribute'];
1176		- }
1177		-
1178		- function sanitize_content ($content, $obj) {
1179		- # This kind of sucks. I intend to replace it with
1180		- # lib_filter sometime soon.
1181		- foreach ($obj->strip_attrs as $pair):
1182		- list($tag,$attr) = $pair;
1183		- $pattern = FeedWordPressHTML::attributeRegex($tag, $attr);
1184		-
1185		- $content = preg_replace_callback (
1186		- $pattern,
1187		- array(&$obj, 'strip_attribute_from_tag'),
1188		- $content
1189		- );
1190		- endforeach;
1191		- return $content;
1192		- }
1193		-} // class SyndicatedPost
	2148	+} /* class SyndicatedPost */
1194	2149