← All changes | syndicatedpost.class.php +1839 -915 2009.1111 → 2013.0504 View file →

		@@ -1,40 +1,115 @@
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 2011.0831
	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;
	33	+ var $_wp_post = null;
13	34
14		- function SyndicatedPost ($item, $link) {
	35	+ /**
	36	+ * SyndicatedPost constructor: Given a feed item and the source from
	37	+ * which it was taken, prepare a post that can be inserted into the
	38	+ * WordPress database on request, or updated in place if it has already
	39	+ * been syndicated.
	40	+ *
	41	+ * @param array $item The item syndicated from the feed.
	42	+ * @param SyndicatedLink $source The feed it was syndicated from.
	43	+ */
	44	+ function SyndicatedPost ($item, &$source) {
15	45	global $wpdb;
16	46
17		- $this->link = $link;
18		- $feedmeta = $link->settings;
19		- $feed = $link->magpie;
	47	+ if (is_array($item)
	48	+ and isset($item['simplepie'])
	49	+ and isset($item['magpie'])) :
	50	+ $this->entry = $item['simplepie'];
	51	+ $this->item = $item['magpie'];
	52	+ $item = $item['magpie'];
	53	+ elseif (is_a($item, 'SimplePie_Item')) :
	54	+ $this->entry = $item;
	55	+
	56	+ // convert to Magpie for compat purposes
	57	+ $mp = new MagpieFromSimplePie($source->simplepie, $this->entry);
	58	+ $this->item = $mp->get_item();
	59	+
	60	+ // done with conversion object
	61	+ $mp = NULL; unset($mp);
	62	+ else :
	63	+ $this->item = $item;
	64	+ endif;
20	65
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.
	66	+ $this->link =& $source;
	67	+ $this->feed = $source->magpie;
	68	+ $this->feedmeta = $source->settings;
	69	+
	70	+ FeedWordPress::diagnostic('feed_items', 'Considering item ['.$this->guid().'] "'.$this->entry->get_title().'"');
	71	+
	72	+ # Dealing with namespaces can get so fucking fucked.
	73	+ $this->xmlns['forward'] = $source->magpie->_XMLNS_FAMILIAR;
	74	+ $this->xmlns['reverse'] = array();
	75	+ foreach ($this->xmlns['forward'] as $url => $ns) :
	76	+ if (!isset($this->xmlns['reverse'][$ns])) :
	77	+ $this->xmlns['reverse'][$ns] = array();
	78	+ endif;
	79	+ $this->xmlns['reverse'][$ns][] = $url;
	80	+ endforeach;
	81	+
	82	+ // Fucking SimplePie.
	83	+ $this->xmlns['reverse']['rss'][] = '';
	84	+
	85	+ # These globals were originally an ugly kludge around a bug in
	86	+ # apply_filters from WordPress 1.5. The bug was fixed in 1.5.1,
	87	+ # and I sure hope at this point that nobody writing filters for
	88	+ # FeedWordPress is still relying on them.
26	89	#
27		- # Cf.: <http://mosquito.wordpress.org/view.php?id=901>
	90	+ # Anyway, I hereby declare them DEPRECATED as of 8 February
	91	+ # 2010. I'll probably remove the globals within 1-2 releases in
	92	+ # the interests of code hygiene and memory usage. If you
	93	+ # currently use them in your filters, I advise you switch off to
	94	+ # accessing the public members SyndicatedPost::feed and
	95	+ # SyndicatedPost::feedmeta.
	96	+
28	97	global $fwp_channel, $fwp_feedmeta;
29		- $fwp_channel = $feed; $fwp_feedmeta = $feedmeta;
	98	+ $fwp_channel = $this->feed; $fwp_feedmeta = $this->feedmeta;
30	99
31		- $this->feed = $feed;
32		- $this->feedmeta = $feedmeta;
33		-
34		- $this->item = $item;
35		- $this->item = apply_filters('syndicated_item', $this->item, $this);
36		-
	100	+ // Trigger global syndicated_item filter.
	101	+ $changed = apply_filters('syndicated_item', $this->item, $this);
	102	+ $this->item = $changed;
	103	+
	104	+ // Allow for feed-specific syndicated_item filters.
	105	+ $changed = apply_filters(
	106	+ "syndicated_item_".$source->uri(),
	107	+ $this->item,
	108	+ $this
	109	+ );
	110	+ $this->item = $changed;
	111	+
37	112	# Filters can halt further processing by returning NULL
38	113	if (is_null($this->item)) :
39	114	$this->post = NULL;
40	115	else :
		@@ -42,62 +117,41 @@
42	117	# That's deliberate. The escaping is done at the point
43	118	# of insertion, not here, to avoid double-escaping and
44	119	# to avoid screwing with syndicated_post filters
45	120
46		- $this->post['post_title'] = apply_filters('syndicated_item_title', $this->item['title'], $this);
	121	+ $this->post['post_title'] = apply_filters(
	122	+ 'syndicated_item_title',
	123	+ $this->entry->get_title(), $this
	124	+ );
47	125
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);
	126	+ $this->named['author'] = apply_filters(
	127	+ 'syndicated_item_author',
	128	+ $this->author(), $this
	129	+ );
	130	+ // This just gives us an alphanumeric name for the author.
	131	+ // We look up (or create) the numeric ID for the author
	132	+ // in SyndicatedPost::add().
52	133
53		- # Identify content and sanitize it.
54		- # ---------------------------------
55		- if (isset($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)):
	134	+ $this->post['post_content'] = apply_filters(
	135	+ 'syndicated_item_content',
	136	+ $this->content(), $this
	137	+ );
	138	+
	139	+ $excerpt = apply_filters('syndicated_item_excerpt', $this->excerpt(), $this);
	140	+ if (!empty($excerpt)):
81	141	$this->post['post_excerpt'] = $excerpt;
82	142	endif;
83		-
84		- // This is unnecessary if we use wp_insert_post
85		- if (!$this->use_api('wp_insert_post')) :
86		- $this->post['post_name'] = sanitize_title($this->post['post_title']);
87		- endif;
88	143
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	144	// Dealing with timestamps in WordPress is so fucking fucked.
94	145	$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());
	146	+ $post_date_gmt = $this->published(array('default' => -1));
	147	+ $post_modified_gmt = $this->updated(array('default' => -1));
99	148
	149	+ $this->post['post_date_gmt'] = gmdate('Y-m-d H:i:s', $post_date_gmt);
	150	+ $this->post['post_date'] = gmdate('Y-m-d H:i:s', $post_date_gmt + $offset);
	151	+ $this->post['post_modified_gmt'] = gmdate('Y-m-d H:i:s', $post_modified_gmt);
	152	+ $this->post['post_modified'] = gmdate('Y-m-d H:i:s', $post_modified_gmt + $offset);
	153	+
100	154	// Use feed-level preferences or the global default.
101	155	$this->post['post_status'] = $this->link->syndicated_status('post', 'publish');
102	156	$this->post['comment_status'] = $this->link->syndicated_status('comment', 'closed');
103	157	$this->post['ping_status'] = $this->link->syndicated_status('ping', 'closed');
		@@ -104,199 +158,1205 @@
104	158
105	159	// Unique ID (hopefully a unique tag: URI); failing that, the permalink
106	160	$this->post['guid'] = apply_filters('syndicated_item_guid', $this->guid(), $this);
107	161
108		- // 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		- $default_custom_settings = get_option('feedwordpress_custom_settings');
110		- if ($default_custom_settings and !is_array($default_custom_settings)) :
111		- $default_custom_settings = unserialize($default_custom_settings);
	162	+ // User-supplied custom settings to apply to each post.
	163	+ // Do first so that FWP-generated custom settings will
	164	+ // overwrite if necessary; thus preventing any munging.
	165	+ $postMetaIn = $this->link->postmeta(array("parsed" => true));
	166	+ $postMetaOut = array();
	167	+
	168	+ foreach ($postMetaIn as $key => $meta) :
	169	+ $postMetaOut[$key] = $meta->do_substitutions($this);
	170	+ endforeach;
	171	+
	172	+ foreach ($postMetaOut as $key => $values) :
	173	+ $this->post['meta'][$key] = array();
	174	+ foreach ($values as $value) :
	175	+ $this->post['meta'][$key][] = apply_filters("syndicated_post_meta_{$key}", $value, $this);
	176	+ endforeach;
	177	+ endforeach;
	178	+
	179	+ // RSS 2.0 / Atom 1.0 enclosure support
	180	+ $enclosures = $this->entry->get_enclosures();
	181	+ if (is_array($enclosures)) : foreach ($enclosures as $enclosure) :
	182	+ $this->post['meta']['enclosure'][] =
	183	+ apply_filters('syndicated_item_enclosure_url', $enclosure->get_link(), $this)."\n".
	184	+ apply_filters('syndicated_item_enclosure_length', $enclosure->get_length(), $this)."\n".
	185	+ apply_filters('syndicated_item_enclosure_type', $enclosure->get_type(), $this);
	186	+ endforeach; endif;
	187	+
	188	+ // In case you want to point back to the blog this was
	189	+ // syndicated from.
	190	+
	191	+ $sourcemeta['syndication_source'] = apply_filters(
	192	+ 'syndicated_item_source_title',
	193	+ $this->link->name(),
	194	+ $this
	195	+ );
	196	+ $sourcemeta['syndication_source_uri'] = apply_filters(
	197	+ 'syndicated_item_source_link',
	198	+ $this->link->homepage(),
	199	+ $this
	200	+ );
	201	+ $sourcemeta['syndication_source_id'] = apply_filters(
	202	+ 'syndicated_item_source_id',
	203	+ $this->link->guid(),
	204	+ $this
	205	+ );
	206	+
	207	+ // Make use of atom:source data, if present in an aggregated feed
	208	+ $entry_source = $this->source();
	209	+ if (!is_null($entry_source)) :
	210	+ foreach ($entry_source as $what => $value) :
	211	+ if (!is_null($value)) :
	212	+ if ($what=='title') : $key = 'syndication_source';
	213	+ elseif ($what=='feed') : $key = 'syndication_feed';
	214	+ else : $key = "syndication_source_${what}";
	215	+ endif;
	216	+
	217	+ $sourcemeta["${key}_original"] = apply_filters(
	218	+ 'syndicated_item_original_source_'.$what,
	219	+ $value,
	220	+ $this
	221	+ );
	222	+ endif;
	223	+ endforeach;
112	224	endif;
113		- if (!is_array($default_custom_settings)) :
114		- $default_custom_settings = array();
115		- endif;
116	225
117		- $custom_settings = (isset($this->link->settings['postmeta']) ? $this->link->settings['postmeta'] : null);
118		- if ($custom_settings and !is_array($custom_settings)) :
119		- $custom_settings = unserialize($custom_settings);
	226	+ foreach ($sourcemeta as $meta_key => $value) :
	227	+ if (!is_null($value)) :
	228	+ $this->post['meta'][$meta_key] = $value;
	229	+ endif;
	230	+ endforeach;
	231	+
	232	+ // Store information on human-readable and machine-readable comment URIs
	233	+
	234	+ // Human-readable comment URI
	235	+ $commentLink = apply_filters('syndicated_item_comments', $this->comment_link(), $this);
	236	+ if (!is_null($commentLink)) : $this->post['meta']['rss:comments'] = $commentLink; endif;
	237	+
	238	+ // Machine-readable content feed URI
	239	+ $commentFeed = apply_filters('syndicated_item_commentrss', $this->comment_feed(), $this);
	240	+ if (!is_null($commentFeed)) : $this->post['meta']['wfw:commentRSS'] = $commentFeed; endif;
	241	+ // Yeah, yeah, now I know that it's supposed to be
	242	+ // wfw:commentRss. Oh well. Path dependence, sucka.
	243	+
	244	+ // Store information to identify the feed that this came from
	245	+ if (isset($this->feedmeta['link/uri'])) :
	246	+ $this->post['meta']['syndication_feed'] = $this->feedmeta['link/uri'];
120	247	endif;
121		- if (!is_array($custom_settings)) :
122		- $custom_settings = array();
	248	+ if (isset($this->feedmeta['link/id'])) :
	249	+ $this->post['meta']['syndication_feed_id'] = $this->feedmeta['link/id'];
123	250	endif;
124		- $this->post['meta'] = array_merge($default_custom_settings, $custom_settings);
125	251
126		- // RSS 2.0 / Atom 1.0 enclosure support
127		- if ( isset($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"], $this);
134		- endfor;
	252	+ if (isset($this->item['source_link_self'])) :
	253	+ $this->post['meta']['syndication_feed_original'] = $this->item['source_link_self'];
135	254	endif;
136	255
137		- // In case you want to point back to the blog this was syndicated from
138		- if (isset($this->feed->channel['title'])) :
139		- $this->post['meta']['syndication_source'] = apply_filters('syndicated_item_source_title', $this->feed->channel['title'], $this);
	256	+ // In case you want to know the external permalink...
	257	+ $this->post['meta']['syndication_permalink'] = apply_filters('syndicated_item_link', $this->permalink());
	258	+
	259	+ // Store a hash of the post content for checking whether something needs to be updated
	260	+ $this->post['meta']['syndication_item_hash'] = $this->update_hash();
	261	+
	262	+ // Categories: start with default categories, if any.
	263	+ $cats = array();
	264	+ if ('no' != $this->link->setting('add/category', NULL, 'yes')) :
	265	+ $fc = get_option("feedwordpress_syndication_cats");
	266	+ if ($fc) :
	267	+ $cats = array_merge($cats, explode("\n", $fc));
	268	+ endif;
140	269	endif;
141	270
142		- if (isset($this->feed->channel['link'])) :
143		- $this->post['meta']['syndication_source_uri'] = apply_filters('syndicated_item_source_link', $this->feed->channel['link'], $this);
	271	+ $fc = $this->link->setting('cats', NULL, array());
	272	+ if (is_array($fc)) :
	273	+ $cats = array_merge($cats, $fc);
144	274	endif;
	275	+ $this->preset_terms['category'] = $cats;
145	276
146		- // Make use of atom:source data, if present in an aggregated feed
147		- if (isset($this->item['source_title'])) :
148		- $this->post['meta']['syndication_source_original'] = $this->item['source_title'];
	277	+ // Now add categories from the post, if we have 'em
	278	+ $cats = array();
	279	+ $post_cats = $this->entry->get_categories();
	280	+ if (is_array($post_cats)) : foreach ($post_cats as $cat) :
	281	+ $cat_name = $cat->get_term();
	282	+ if (!$cat_name) : $cat_name = $cat->get_label(); endif;
	283	+
	284	+ if ($this->link->setting('cat_split', NULL, NULL)) :
	285	+ $pcre = "\007".$this->feedmeta['cat_split']."\007";
	286	+ $cats = array_merge(
	287	+ $cats,
	288	+ preg_split(
	289	+ $pcre,
	290	+ $cat_name,
	291	+ -1 /*=no limit*/,
	292	+ PREG_SPLIT_NO_EMPTY
	293	+ )
	294	+ );
	295	+ else :
	296	+ $cats[] = $cat_name;
	297	+ endif;
	298	+ endforeach; endif;
	299	+
	300	+ $this->feed_terms['category'] = apply_filters('syndicated_item_categories', $cats, $this);
	301	+
	302	+ // Tags: start with default tags, if any
	303	+ $tags = array();
	304	+ if ('no' != $this->link->setting('add/post_tag', NULL, 'yes')) :
	305	+ $ft = get_option("feedwordpress_syndication_tags", NULL);
	306	+ $tags = (is_null($ft) ? array() : explode(FEEDWORDPRESS_CAT_SEPARATOR, $ft));
149	307	endif;
150		-
151		- if (isset($this->item['source_link'])) :
152		- $this->post['meta']['syndication_source_uri_original'] = $this->item['source_link'];
	308	+
	309	+ $ft = $this->link->setting('tags', NULL, array());
	310	+ if (is_array($ft)) :
	311	+ $tags = array_merge($tags, $ft);
153	312	endif;
	313	+ $this->preset_terms['post_tag'] = $tags;
154	314
155		- if (isset($this->item['source_id'])) :
156		- $this->post['meta']['syndication_source_id_original'] = $this->item['source_id'];
	315	+ // Scan post for /a[@rel='tag'] and use as tags if present
	316	+ $tags = $this->inline_tags();
	317	+ $this->feed_terms['post_tag'] = apply_filters('syndicated_item_tags', $tags, $this);
	318	+
	319	+ $taxonomies = $this->link->taxonomies();
	320	+ $feedTerms = $this->link->setting('terms', NULL, array());
	321	+ $globalTerms = get_option('feedwordpress_syndication_terms', array());
	322	+
	323	+ $specials = array('category' => 'cats', 'post_tag' => 'tags');
	324	+ foreach ($taxonomies as $tax) :
	325	+ if (!isset($specials[$tax])) :
	326	+ $terms = array();
	327	+
	328	+ // See if we should get the globals
	329	+ if ('no' != $this->link->setting("add/$tax", NULL, 'yes')) :
	330	+ if (isset($globalTerms[$tax])) :
	331	+ $terms = $globalTerms[$tax];
	332	+ endif;
	333	+ endif;
	334	+
	335	+ // Now merge in the locals
	336	+ if (isset($feedTerms[$tax])) :
	337	+ $terms = array_merge($terms, $feedTerms[$tax]);
	338	+ endif;
	339	+
	340	+ // That's all, folks.
	341	+ $this->preset_terms[$tax] = $terms;
	342	+ endif;
	343	+ endforeach;
	344	+
	345	+ $this->post['post_type'] = apply_filters('syndicated_post_type', $this->link->setting('syndicated post type', 'syndicated_post_type', 'post'), $this);
	346	+ endif;
	347	+ } /* SyndicatedPost::SyndicatedPost() */
	348	+
	349	+ #####################################
	350	+ #### EXTRACT DATA FROM FEED ITEM ####
	351	+ #####################################
	352	+
	353	+ function substitution_function ($name) {
	354	+ $ret = NULL;
	355	+
	356	+ switch ($name) :
	357	+ // Allowed PHP string functions
	358	+ case 'trim':
	359	+ case 'ltrim':
	360	+ case 'rtrim':
	361	+ case 'strtoupper':
	362	+ case 'strtolower':
	363	+ case 'urlencode':
	364	+ case 'urldecode':
	365	+ $ret = $name;
	366	+ endswitch;
	367	+ return $ret;
	368	+ }
	369	+
	370	+ /**
	371	+ * SyndicatedPost::query uses an XPath-like syntax to query arbitrary
	372	+ * elements within the syndicated item.
	373	+ *
	374	+ * @param string $path
	375	+ * @returns array of string values representing contents of matching
	376	+ * elements or attributes
	377	+ */
	378	+ function query ($path) {
	379	+ $urlHash = array();
	380	+
	381	+ // Allow {url} notation for namespaces. URLs will contain : and /, so...
	382	+ preg_match_all('/{([^}]+)}/', $path, $match, PREG_SET_ORDER);
	383	+ foreach ($match as $ref) :
	384	+ $urlHash[md5($ref[1])] = $ref[1];
	385	+ endforeach;
	386	+
	387	+ foreach ($urlHash as $hash => $url) :
	388	+ $path = str_replace('{'.$url.'}', '{#'.$hash.'}', $path);
	389	+ endforeach;
	390	+
	391	+ $path = explode('/', $path);
	392	+ foreach ($path as $index => $node) :
	393	+ if (preg_match('/{#([^}]+)}/', $node, $ref)) :
	394	+ if (isset($urlHash[$ref[1]])) :
	395	+ $path[$index] = str_replace(
	396	+ '{#'.$ref[1].'}',
	397	+ '{'.$urlHash[$ref[1]].'}',
	398	+ $node
	399	+ );
	400	+ endif;
157	401	endif;
	402	+ endforeach;
158	403
159		- // Store information on human-readable and machine-readable comment URIs
160		- if (isset($this->item['comments'])) :
161		- $this->post['meta']['rss:comments'] = apply_filters('syndicated_item_comments', $this->item['comments']);
	404	+ // Start out with a get_item_tags query.
	405	+ $node = '';
	406	+ while (strlen($node)==0 and !is_null($node)) :
	407	+ $node = array_shift($path);
	408	+ endwhile;
	409	+
	410	+ switch ($node) :
	411	+ case 'feed' :
	412	+ case 'channel' :
	413	+ $node = array_shift($path);
	414	+ $data = $this->get_feed_root_element();
	415	+ $data = array_merge($data, $this->get_feed_channel_elements());
	416	+ break;
	417	+ case 'item' :
	418	+ $node = array_shift($path);
	419	+ default :
	420	+ $data = array($this->entry->data);
	421	+ $method = NULL;
	422	+ endswitch;
	423	+
	424	+ while (!is_null($node)) :
	425	+ if (strlen($node) > 0) :
	426	+ $matches = array();
	427	+
	428	+ list($axis, $element) = $this->xpath_name_and_axis($node);
	429	+
	430	+ foreach ($data as $datum) :
	431	+ if (!is_string($datum) and isset($datum[$axis])) :
	432	+ foreach ($datum[$axis] as $ns => $elements) :
	433	+ if (isset($elements[$element])) :
	434	+ // Potential match.
	435	+ // Check namespace.
	436	+ if (is_string($elements[$element])) : // Attribute
	437	+ $addenda = array($elements[$element]);
	438	+ $contexts = array($datum);
	439	+ else : // Element
	440	+ $addenda = $elements[$element];
	441	+ $contexts = $elements[$element];
	442	+ endif;
	443	+
	444	+ foreach ($addenda as $index => $addendum) :
	445	+ $context = $contexts[$index];
	446	+
	447	+ $namespaces = $this->xpath_possible_namespaces($node, $context);
	448	+ if (in_array($ns, $namespaces)) :
	449	+ $matches[] = $addendum;
	450	+ endif;
	451	+ endforeach;
	452	+ endif;
	453	+ endforeach;
	454	+ endif;
	455	+ endforeach;
	456	+
	457	+ $data = $matches;
162	458	endif;
163		-
164		- // RSS 2.0 comment feeds extension
165		- if (isset($this->item['wfw']['commentrss'])) :
166		- $this->post['meta']['wfw:commentRSS'] = apply_filters('syndicated_item_commentrss', $this->item['wfw']['commentrss']);
	459	+ $node = array_shift($path);
	460	+ endwhile;
	461	+
	462	+ $matches = array();
	463	+ foreach ($data as $datum) :
	464	+ if (is_string($datum)) :
	465	+ $matches[] = $datum;
	466	+ elseif (isset($datum['data'])) :
	467	+ $matches[] = $datum['data'];
167	468	endif;
	469	+ endforeach;
	470	+ return $matches;
	471	+ } /* SyndicatedPost::query() */
168	472
169		- // Atom 1.0 comment feeds link-rel
170		- if (isset($this->item['link_replies'])) :
171		- // There may be multiple <link rel="replies"> elements; feeds have a feed MIME type
172		- $N = isset($this->item['link_replies#']) ? $this->item['link_replies#'] : 1;
173		- for ($i = 1; $i <= $N; $i++) :
174		- $currentElement = 'link_replies'.(($i > 1) ? '#'.$i : '');
175		- if (isset($this->item[$currentElement.'@type'])
176		- and preg_match("\007application/(atom|rss|rdf)\+xml\007i", $this->item[$currentElement.'@type'])) :
177		- $this->post['meta']['wfw:commentRSS'] = apply_filters('syndicated_item_commentrss', $this->item[$currentElement]);
178		- endif;
179		- endfor;
	473	+ function get_feed_root_element () {
	474	+ $matches = array();
	475	+ foreach ($this->link->simplepie->data['child'] as $ns => $root) :
	476	+ foreach ($root as $element => $data) :
	477	+ $matches = array_merge($matches, $data);
	478	+ endforeach;
	479	+ endforeach;
	480	+ return $matches;
	481	+ } /* SyndicatedPost::get_feed_root_element() */
	482	+
	483	+ function get_feed_channel_elements () {
	484	+ $rss = array(
	485	+ SIMPLEPIE_NAMESPACE_RSS_090,
	486	+ SIMPLEPIE_NAMESPACE_RSS_10,
	487	+ 'http://backend.userland.com/RSS2',
	488	+ SIMPLEPIE_NAMESPACE_RSS_20,
	489	+ );
	490	+
	491	+ $matches = array();
	492	+ foreach ($rss as $ns) :
	493	+ $data = $this->link->simplepie->get_feed_tags($ns, 'channel');
	494	+ if (!is_null($data)) :
	495	+ $matches = array_merge($matches, $data);
180	496	endif;
	497	+ endforeach;
	498	+ return $matches;
	499	+ } /* SyndicatedPost::get_feed_channel_elements() */
181	500
182		- // Store information to identify the feed that this came from
183		- $this->post['meta']['syndication_feed'] = $this->feedmeta['link/uri'];
184		- $this->post['meta']['syndication_feed_id'] = $this->feedmeta['link/id'];
	501	+ function xpath_default_namespace () {
	502	+ // Get the default namespace.
	503	+ $type = $this->link->simplepie->get_type();
	504	+ if ($type & SIMPLEPIE_TYPE_ATOM_10) :
	505	+ $defaultNS = SIMPLEPIE_NAMESPACE_ATOM_10;
	506	+ elseif ($type & SIMPLEPIE_TYPE_ATOM_03) :
	507	+ $defaultNS = SIMPLEPIE_NAMESPACE_ATOM_03;
	508	+ elseif ($type & SIMPLEPIE_TYPE_RSS_090) :
	509	+ $defaultNS = SIMPLEPIE_NAMESPACE_RSS_090;
	510	+ elseif ($type & SIMPLEPIE_TYPE_RSS_10) :
	511	+ $defaultNS = SIMPLEPIE_NAMESPACE_RSS_10;
	512	+ elseif ($type & SIMPLEPIE_TYPE_RSS_20) :
	513	+ $defaultNS = SIMPLEPIE_NAMESPACE_RSS_20;
	514	+ else :
	515	+ $defaultNS = SIMPLEPIE_NAMESPACE_RSS_20;
	516	+ endif;
	517	+ return $defaultNS;
	518	+ } /* SyndicatedPost::xpath_default_namespace() */
185	519
186		- if (isset($this->item['source_link_self'])) :
187		- $this->post['meta']['syndication_feed_original'] = $this->item['source_link_self'];
	520	+ function xpath_name_and_axis ($node) {
	521	+ $ns = NULL; $element = NULL;
	522	+
	523	+ if (substr($node, 0, 1)=='@') :
	524	+ $axis = 'attribs'; $node = substr($node, 1);
	525	+ else :
	526	+ $axis = 'child';
	527	+ endif;
	528	+
	529	+ if (preg_match('/^{([^}]*)}(.*)$/', $node, $ref)) :
	530	+ $element = $ref[2];
	531	+ elseif (strpos($node, ':') !== FALSE) :
	532	+ list($xmlns, $element) = explode(':', $node, 2);
	533	+ else :
	534	+ $element = $node;
	535	+ endif;
	536	+ return array($axis, $element);
	537	+ } /* SyndicatedPost::xpath_local_name () */
	538	+
	539	+ function xpath_possible_namespaces ($node, $datum = array()) {
	540	+ $ns = NULL; $element = NULL;
	541	+
	542	+ if (substr($node, 0, 1)=='@') :
	543	+ $attr = '@'; $node = substr($node, 1);
	544	+ else :
	545	+ $attr = '';
	546	+ endif;
	547	+
	548	+ if (preg_match('/^{([^}]*)}(.*)$/', $node, $ref)) :
	549	+ $ns = array($ref[1]);
	550	+ elseif (strpos($node, ':') !== FALSE) :
	551	+ list($xmlns, $element) = explode(':', $node, 2);
	552	+
	553	+ if (isset($this->xmlns['reverse'][$xmlns])) :
	554	+ $ns = $this->xmlns['reverse'][$xmlns];
	555	+ else :
	556	+ $ns = array($xmlns);
188	557	endif;
	558	+
	559	+ // Fucking SimplePie. For attributes in default xmlns.
	560	+ $defaultNS = $this->xpath_default_namespace();
	561	+ if (isset($this->xmlns['forward'][$defaultNS])
	562	+ and ($xmlns==$this->xmlns['forward'][$defaultNS])) :
	563	+ $ns[] = '';
	564	+ endif;
	565	+
	566	+ if (isset($datum['xmlns'])) :
	567	+ if (isset($datum['xmlns'][$xmlns])) :
	568	+ $ns[] = $datum['xmlns'][$xmlns];
	569	+ endif;
	570	+ endif;
	571	+ else :
	572	+ // Often in SimplePie, the default namespace gets stored
	573	+ // as an empty string rather than a URL.
	574	+ $ns = array($this->xpath_default_namespace(), '');
	575	+ endif;
	576	+ return array_unique($ns);
	577	+ } /* SyndicatedPost::xpath_possible_namespaces() */
189	578
190		- // In case you want to know the external permalink...
191		- if (isset($this->item['link'])) :
192		- $permalink = $this->item['link'];
	579	+ function content () {
	580	+ $content = NULL;
	581	+ if (isset($this->item['atom_content'])) :
	582	+ $content = $this->item['atom_content'];
	583	+ elseif (isset($this->item['xhtml']['body'])) :
	584	+ $content = $this->item['xhtml']['body'];
	585	+ elseif (isset($this->item['xhtml']['div'])) :
	586	+ $content = $this->item['xhtml']['div'];
	587	+ elseif (isset($this->item['content']['encoded']) and $this->item['content']['encoded']):
	588	+ $content = $this->item['content']['encoded'];
	589	+ elseif (isset($this->item['description'])) :
	590	+ $content = $this->item['description'];
	591	+ endif;
	592	+ return $content;
	593	+ } /* SyndicatedPost::content() */
193	594
194		- // No <link> element. See if this feed has <guid isPermalink="true"> ....
195		- elseif (isset($this->item['guid'])) :
196		- if (isset($this->item['guid@ispermalink']) and strtolower(trim($this->item['guid@ispermalink'])) != 'false') :
197		- $permalink = $this->item['guid'];
	595	+ function excerpt () {
	596	+ # Identify and sanitize excerpt: atom:summary, or rss:description
	597	+ $excerpt = $this->entry->get_description();
	598	+
	599	+ # Many RSS feeds use rss:description, inadvisably, to
	600	+ # carry the entire post (typically with escaped HTML).
	601	+ # If that's what happened, we don't want the full
	602	+ # content for the excerpt.
	603	+ $content = $this->content();
	604	+
	605	+ // Ignore whitespace, case, and tag cruft.
	606	+ $theExcerpt = preg_replace('/\s+/', '', strtolower(strip_tags($excerpt)));
	607	+ $theContent = preg_replace('/\s+/', '', strtolower(strip_tags($content)));
	608	+
	609	+ if ( empty($excerpt) or $theExcerpt == $theContent ) :
	610	+ # If content is available, generate an excerpt.
	611	+ if ( strlen(trim($content)) > 0 ) :
	612	+ $excerpt = strip_tags($content);
	613	+ if (strlen($excerpt) > 255) :
	614	+ $excerpt = substr($excerpt,0,252).'...';
198	615	endif;
199	616	endif;
	617	+ endif;
	618	+ return $excerpt;
	619	+ } /* SyndicatedPost::excerpt() */
200	620
201		- $this->post['meta']['syndication_permalink'] = apply_filters('syndicated_item_link', $permalink);
	621	+ function permalink () {
	622	+ // Handles explicit <link> elements and also RSS 2.0 cases with
	623	+ // <guid isPermaLink="true">, etc. Hooray!
	624	+ $permalink = $this->entry->get_link();
	625	+ return $permalink;
	626	+ }
202	627
203		- // Store a hash of the post content for checking whether something needs to be updated
204		- $this->post['meta']['syndication_item_hash'] = $this->update_hash();
	628	+ function created ($params = array()) {
	629	+ $unfiltered = false; $default = NULL;
	630	+ extract($params);
	631	+
	632	+ $date = '';
	633	+ if (isset($this->item['dc']['created'])) :
	634	+ $date = $this->item['dc']['created'];
	635	+ elseif (isset($this->item['dcterms']['created'])) :
	636	+ $date = $this->item['dcterms']['created'];
	637	+ elseif (isset($this->item['created'])): // Atom 0.3
	638	+ $date = $this->item['created'];
	639	+ endif;
205	640
206		- // Feed-by-feed options for author and category creation
207		- $this->post['named']['unfamiliar']['author'] = (isset($this->feedmeta['unfamiliar author']) ? $this->feedmeta['unfamiliar author'] : null);
208		- $this->post['named']['unfamiliar']['category'] = (isset($this->feedmeta['unfamiliar category']) ? $this->feedmeta['unfamiliar category'] : null);
	641	+ $time = new FeedTime($date);
	642	+ $ts = $time->timestamp();
	643	+ if (!$unfiltered) :
	644	+ apply_filters('syndicated_item_created', $ts, $this);
	645	+ endif;
	646	+ return $ts;
	647	+ } /* SyndicatedPost::created() */
209	648
210		- // Categories: start with default categories, if any
211		- $fc = get_option("feedwordpress_syndication_cats");
212		- if ($fc) :
213		- $this->post['named']['preset/category'] = explode("\n", $fc);
	649	+ function published ($params = array(), $default = NULL) {
	650	+ $fallback = true; $unfiltered = false;
	651	+ if (!is_array($params)) : // Old style
	652	+ $fallback = $params;
	653	+ else : // New style
	654	+ extract($params);
	655	+ endif;
	656	+
	657	+ $date = '';
	658	+ $ts = null;
	659	+
	660	+ # RSS is a fucking mess. Figure out whether we have a date in
	661	+ # <dc:date>, <issued>, <pubDate>, etc., and get it into Unix
	662	+ # epoch format for reformatting. If we can't find anything,
	663	+ # we'll use the last-updated time.
	664	+ if (isset($this->item['dc']['date'])): // Dublin Core
	665	+ $date = $this->item['dc']['date'];
	666	+ elseif (isset($this->item['dcterms']['issued'])) : // Dublin Core extensions
	667	+ $date = $this->item['dcterms']['issued'];
	668	+ elseif (isset($this->item['published'])) : // Atom 1.0
	669	+ $date = $this->item['published'];
	670	+ elseif (isset($this->item['issued'])): // Atom 0.3
	671	+ $date = $this->item['issued'];
	672	+ elseif (isset($this->item['pubdate'])): // RSS 2.0
	673	+ $date = $this->item['pubdate'];
	674	+ endif;
	675	+
	676	+ if (strlen($date) > 0) :
	677	+ $time = new FeedTime($date);
	678	+ $ts = $time->timestamp();
	679	+ elseif ($fallback) : // Fall back to <updated> / <modified> if present
	680	+ $ts = $this->updated(/*fallback=*/ false, /*default=*/ $default);
	681	+ endif;
	682	+
	683	+ # If everything failed, then default to the current time.
	684	+ if (is_null($ts)) :
	685	+ if (-1 == $default) :
	686	+ $ts = time();
214	687	else :
215		- $this->post['named']['preset/category'] = array();
	688	+ $ts = $default;
216	689	endif;
	690	+ endif;
	691	+
	692	+ if (!$unfiltered) :
	693	+ $ts = apply_filters('syndicated_item_published', $ts, $this);
	694	+ endif;
	695	+ return $ts;
	696	+ } /* SyndicatedPost::published() */
217	697
218		- if (isset($this->feedmeta['cats']) and is_array($this->feedmeta['cats'])) :
219		- $this->post['named']['preset/category'] = array_merge($this->post['named']['preset/category'], $this->feedmeta['cats']);
	698	+ function updated ($params = array(), $default = -1) {
	699	+ $fallback = true; $unfiltered = false;
	700	+ if (!is_array($params)) : // Old style
	701	+ $fallback = $params;
	702	+ else : // New style
	703	+ extract($params);
	704	+ endif;
	705	+
	706	+ $date = '';
	707	+ $ts = null;
	708	+
	709	+ # As far as I know, only dcterms and Atom have reliable ways to
	710	+ # specify when something was *modified* last. If neither is
	711	+ # available, then we'll try to get the time of publication.
	712	+ if (isset($this->item['dc']['modified'])) : // Not really correct
	713	+ $date = $this->item['dc']['modified'];
	714	+ elseif (isset($this->item['dcterms']['modified'])) : // Dublin Core extensions
	715	+ $date = $this->item['dcterms']['modified'];
	716	+ elseif (isset($this->item['modified'])): // Atom 0.3
	717	+ $date = $this->item['modified'];
	718	+ elseif (isset($this->item['updated'])): // Atom 1.0
	719	+ $date = $this->item['updated'];
	720	+ endif;
	721	+
	722	+ if (strlen($date) > 0) :
	723	+ $time = new FeedTime($date);
	724	+ $ts = $time->timestamp();
	725	+ elseif ($fallback) : // Fall back to issued / dc:date
	726	+ $ts = $this->published(/*fallback=*/ false, /*default=*/ $default);
	727	+ endif;
	728	+
	729	+ # If everything failed, then default to the current time.
	730	+ if (is_null($ts)) :
	731	+ if (-1 == $default) :
	732	+ $ts = time();
	733	+ else :
	734	+ $ts = $default;
220	735	endif;
	736	+ endif;
221	737
222		- // Now add categories from the post, if we have 'em
223		- $this->post['named']['category'] = array();
224		- if ( isset($this->item['category#']) ) :
225		- for ($i = 1; $i <= $this->item['category#']; $i++) :
226		- $cat_idx = (($i > 1) ? "#{$i}" : "");
227		- $cat = $this->item["category{$cat_idx}"];
	738	+ if (!$unfiltered) :
	739	+ apply_filters('syndicated_item_updated', $ts, $this);
	740	+ endif;
	741	+ return $ts;
	742	+ } /* SyndicatedPost::updated() */
228	743
229		- if ( isset($this->feedmeta['cat_split']) and strlen($this->feedmeta['cat_split']) > 0) :
230		- $pcre = "\007".$this->feedmeta['cat_split']."\007";
231		- $this->post['named']['category'] = array_merge($this->post['named']['category'], preg_split($pcre, $cat, -1 /*=no limit*/, PREG_SPLIT_NO_EMPTY));
	744	+ var $_hashes = array();
	745	+ function stored_hashes ($id = NULL) {
	746	+ if (is_null($id)) :
	747	+ $id = $this->wp_id();
	748	+ endif;
	749	+
	750	+ if (!isset($this->_hashes[$id])) :
	751	+ $this->_hashes[$id] = get_post_custom_values(
	752	+ 'syndication_item_hash', $id
	753	+ );
	754	+ if (is_null($this->_hashes[$id])) :
	755	+ $this->_hashes[$id] = array();
	756	+ endif;
	757	+ endif;
	758	+ return $this->_hashes[$id];
	759	+ }
	760	+
	761	+ function update_hash ($hashed = true) {
	762	+ // Basis for tracking possible changes to item.
	763	+ $hash = array(
	764	+ "title" => $this->entry->get_title(),
	765	+ "link" => $this->permalink(),
	766	+ "content" => $this->content(),
	767	+ "excerpt" => $this->excerpt(),
	768	+ );
	769	+
	770	+ if ($hashed) :
	771	+ $hash = md5(serialize($hash));
	772	+ endif;
	773	+
	774	+ return $hash;
	775	+ } /* SyndicatedPost::update_hash() */
	776	+
	777	+ /*static*/ function normalize_guid_prefix () {
	778	+ return trailingslashit(get_bloginfo('url')).'?guid=';
	779	+ }
	780	+
	781	+ /*static*/ function normalize_guid ($guid) {
	782	+ $guid = trim($guid);
	783	+ if (preg_match('/^[0-9a-z]{32}$/i', $guid)) : // MD5
	784	+ $guid = SyndicatedPost::normalize_guid_prefix().strtolower($guid);
	785	+ elseif ((strlen(esc_url($guid)) == 0) or (esc_url($guid) != $guid)) :
	786	+ $guid = SyndicatedPost::normalize_guid_prefix().md5($guid);
	787	+ endif;
	788	+ $guid = trim($guid);
	789	+ return $guid;
	790	+ } /* SyndicatedPost::normalize_guid() */
	791	+
	792	+ function guid () {
	793	+ $guid = null;
	794	+ if (isset($this->item['id'])): // Atom 0.3 / 1.0
	795	+ $guid = $this->item['id'];
	796	+ elseif (isset($this->item['atom']['id'])) : // Namespaced Atom
	797	+ $guid = $this->item['atom']['id'];
	798	+ elseif (isset($this->item['guid'])) : // RSS 2.0
	799	+ $guid = $this->item['guid'];
	800	+ elseif (isset($this->item['dc']['identifier'])) : // yeah, right
	801	+ $guid = $this->item['dc']['identifier'];
	802	+ endif;
	803	+
	804	+ // Un-set or too long to use as-is. Generate a tag: URI.
	805	+ if (is_null($guid) or strlen($guid) > 250) :
	806	+ // In case we need to check this again
	807	+ $original_guid = $guid;
	808	+
	809	+ // The feed does not seem to have provided us with a
	810	+ // usable unique identifier, so we'll have to cobble
	811	+ // together a tag: URI that might work for us. The base
	812	+ // of the URI will be the host name of the feed source ...
	813	+ $bits = parse_url($this->link->uri());
	814	+ $guid = 'tag:'.$bits['host'];
	815	+
	816	+ // Some ill-mannered feeds (for example, certain feeds
	817	+ // coming from Google Calendar) have extraordinarily long
	818	+ // guids -- so long that they exceed the 255 character
	819	+ // width of the WordPress guid field. But if the string
	820	+ // gets clipped by MySQL, uniqueness tests will fail
	821	+ // forever after and the post will be endlessly
	822	+ // reduplicated. So, instead, Guids Of A Certain Length
	823	+ // are hashed down into a nice, manageable tag: URI.
	824	+ if (!is_null($original_guid)) :
	825	+ $guid .= ',2010-12-03:id.'.md5($original_guid);
	826	+
	827	+ // If we have a date of creation, then we can use that
	828	+ // to uniquely identify the item. (On the other hand, if
	829	+ // the feed producer was consicentious enough to
	830	+ // generate dates of creation, she probably also was
	831	+ // conscientious enough to generate unique identifiers.)
	832	+ elseif (!is_null($this->created())) :
	833	+ $guid .= '://post.'.date('YmdHis', $this->created());
	834	+
	835	+ // Otherwise, use both the URI of the item, *and* the
	836	+ // item's title. We have to use both because titles are
	837	+ // often not unique, and sometimes links aren't unique
	838	+ // either (e.g. Bitch (S)HITLIST, Mozilla Dot Org news,
	839	+ // some podcasts). But it's rare to have *both* the same
	840	+ // title *and* the same link for two different items. So
	841	+ // this is about the best we can do.
	842	+ else :
	843	+ $link = $this->permalink();
	844	+ if (is_null($link)) : $link = $this->link->uri(); endif;
	845	+ $guid .= '://'.md5($link.'/'.$this->item['title']);
	846	+ endif;
	847	+ endif;
	848	+ return $guid;
	849	+ } /* SyndicatedPost::guid() */
	850	+
	851	+ function author () {
	852	+ $author = array ();
	853	+
	854	+ $aa = $this->entry->get_authors();
	855	+ if (count($aa) > 0) :
	856	+ $a = reset($aa);
	857	+
	858	+ $author = array(
	859	+ 'name' => $a->get_name(),
	860	+ 'email' => $a->get_email(),
	861	+ 'uri' => $a->get_link(),
	862	+ );
	863	+ endif;
	864	+
	865	+ if (FEEDWORDPRESS_COMPATIBILITY) :
	866	+ // Search through the MagpieRSS elements: Atom, Dublin Core, RSS
	867	+ if (isset($this->item['author_name'])):
	868	+ $author['name'] = $this->item['author_name'];
	869	+ elseif (isset($this->item['dc']['creator'])):
	870	+ $author['name'] = $this->item['dc']['creator'];
	871	+ elseif (isset($this->item['dc']['contributor'])):
	872	+ $author['name'] = $this->item['dc']['contributor'];
	873	+ elseif (isset($this->feed->channel['dc']['creator'])) :
	874	+ $author['name'] = $this->feed->channel['dc']['creator'];
	875	+ elseif (isset($this->feed->channel['dc']['contributor'])) :
	876	+ $author['name'] = $this->feed->channel['dc']['contributor'];
	877	+ elseif (isset($this->feed->channel['author_name'])) :
	878	+ $author['name'] = $this->feed->channel['author_name'];
	879	+ elseif ($this->feed->is_rss() and isset($this->item['author'])) :
	880	+ // The author element in RSS is allegedly an
	881	+ // e-mail address, but lots of people don't use
	882	+ // it that way. So let's make of it what we can.
	883	+ $author = parse_email_with_realname($this->item['author']);
	884	+
	885	+ if (!isset($author['name'])) :
	886	+ if (isset($author['email'])) :
	887	+ $author['name'] = $author['email'];
232	888	else :
233		- $this->post['named']['category'][] = $cat;
	889	+ $author['name'] = $this->feed->channel['title'];
234	890	endif;
235		- endfor;
	891	+ endif;
236	892	endif;
237		- $this->post['named']['category'] = apply_filters('syndicated_item_categories', $this->post['named']['category'], $this);
	893	+ endif;
	894	+
	895	+ if (!isset($author['name']) or is_null($author['name'])) :
	896	+ // Nothing found. Try some crappy defaults.
	897	+ if ($this->link->name()) :
	898	+ $author['name'] = $this->link->name();
	899	+ else :
	900	+ $url = parse_url($this->link->uri());
	901	+ $author['name'] = $url['host'];
	902	+ endif;
	903	+ endif;
	904	+
	905	+ if (FEEDWORDPRESS_COMPATIBILITY) :
	906	+ if (isset($this->item['author_email'])):
	907	+ $author['email'] = $this->item['author_email'];
	908	+ elseif (isset($this->feed->channel['author_email'])) :
	909	+ $author['email'] = $this->feed->channel['author_email'];
	910	+ endif;
238	911
239		- // Tags: start with default tags, if any
240		- $ft = get_option("feedwordpress_syndication_tags");
241		- if ($ft) :
242		- $this->post['tags_input'] = explode(FEEDWORDPRESS_CAT_SEPARATOR, $ft);
243		- else :
244		- $this->post['tags_input'] = array();
	912	+ if (isset($this->item['author_uri'])):
	913	+ $author['uri'] = $this->item['author_uri'];
	914	+ elseif (isset($this->item['author_url'])):
	915	+ $author['uri'] = $this->item['author_url'];
	916	+ elseif (isset($this->feed->channel['author_uri'])) :
	917	+ $author['uri'] = $this->item['author_uri'];
	918	+ elseif (isset($this->feed->channel['author_url'])) :
	919	+ $author['uri'] = $this->item['author_url'];
	920	+ elseif (isset($this->feed->channel['link'])) :
	921	+ $author['uri'] = $this->feed->channel['link'];
245	922	endif;
	923	+ endif;
	924	+
	925	+ return $author;
	926	+ } /* SyndicatedPost::author() */
	927	+
	928	+ /**
	929	+ * SyndicatedPost::inline_tags: Return a list of all the tags embedded
	930	+ * in post content using the a[@rel="tag"] microformat.
	931	+ *
	932	+ * @since 2010.0630
	933	+ * @return array of string values containing the name of each tag
	934	+ */
	935	+ function inline_tags () {
	936	+ $tags = array();
	937	+ $content = $this->content();
	938	+ $pattern = FeedWordPressHTML::tagWithAttributeRegex('a', 'rel', 'tag');
	939	+ preg_match_all($pattern, $content, $refs, PREG_SET_ORDER);
	940	+ if (count($refs) > 0) :
	941	+ foreach ($refs as $ref) :
	942	+ $tag = FeedWordPressHTML::tagWithAttributeMatch($ref);
	943	+ $tags[] = $tag['content'];
	944	+ endforeach;
	945	+ endif;
	946	+ return $tags;
	947	+ }
	948	+
	949	+ /**
	950	+ * SyndicatedPost::isTaggedAs: Test whether a feed item is
	951	+ * tagged / categorized with a given string. Case and leading and
	952	+ * trailing whitespace are ignored.
	953	+ *
	954	+ * @param string $tag Tag to check for
	955	+ *
	956	+ * @return bool Whether or not at least one of the categories / tags on
	957	+ * $this->item is set to $tag (modulo case and leading and trailing
	958	+ * whitespace)
	959	+ */
	960	+ function isTaggedAs ($tag) {
	961	+ $desiredTag = strtolower(trim($tag)); // Normalize case and whitespace
	962	+
	963	+ // Check to see if this is tagged with $tag
	964	+ $currentCategory = 'category';
	965	+ $currentCategoryNumber = 1;
	966	+
	967	+ // If we have the new MagpieRSS, the number of category elements
	968	+ // on this item is stored under index "category#".
	969	+ if (isset($this->item['category#'])) :
	970	+ $numberOfCategories = (int) $this->item['category#'];
	971	+
	972	+ // We REALLY shouldn't have the old and busted MagpieRSS, but in
	973	+ // case we do, it doesn't support multiple categories, but there
	974	+ // might still be a single value under the "category" index.
	975	+ elseif (isset($this->item['category'])) :
	976	+ $numberOfCategories = 1;
	977	+
	978	+ // No standard category or tag elements on this feed item.
	979	+ else :
	980	+ $numberOfCategories = 0;
	981	+
	982	+ endif;
	983	+
	984	+ $isSoTagged = false; // Innocent until proven guilty
	985	+
	986	+ // Loop through category elements; if there are multiple
	987	+ // elements, they are indexed as category, category#2,
	988	+ // category#3, ... category#N
	989	+ while ($currentCategoryNumber <= $numberOfCategories) :
	990	+ if ($desiredTag == strtolower(trim($this->item[$currentCategory]))) :
	991	+ $isSoTagged = true; // Got it!
	992	+ break;
	993	+ endif;
	994	+
	995	+ $currentCategoryNumber += 1;
	996	+ $currentCategory = 'category#'.$currentCategoryNumber;
	997	+ endwhile;
	998	+
	999	+ return $isSoTagged;
	1000	+ } /* SyndicatedPost::isTaggedAs() */
	1001	+
	1002	+ /**
	1003	+ * SyndicatedPost::enclosures: returns an array with any enclosures
	1004	+ * that may be attached to this syndicated item.
	1005	+ *
	1006	+ * @param string $type If you only want enclosures that match a certain
	1007	+ * MIME type or group of MIME types, you can limit the enclosures
	1008	+ * that will be returned to only those with a MIME type which
	1009	+ * matches this regular expression.
	1010	+ * @return array
	1011	+ */
	1012	+ function enclosures ($type = '/.*/') {
	1013	+ $enclosures = array();
	1014	+
	1015	+ if (isset($this->item['enclosure#'])) :
	1016	+ // Loop through enclosure, enclosure#2, enclosure#3, ....
	1017	+ for ($i = 1; $i <= $this->item['enclosure#']; $i++) :
	1018	+ $eid = (($i > 1) ? "#{$id}" : "");
	1019	+
	1020	+ // Does it match the type we want?
	1021	+ if (preg_match($type, $this->item["enclosure{$eid}@type"])) :
	1022	+ $enclosures[] = array(
	1023	+ "url" => $this->item["enclosure{$eid}@url"],
	1024	+ "type" => $this->item["enclosure{$eid}@type"],
	1025	+ "length" => $this->item["enclosure{$eid}@length"],
	1026	+ );
	1027	+ endif;
	1028	+ endfor;
	1029	+ endif;
	1030	+ return $enclosures;
	1031	+ } /* SyndicatedPost::enclosures() */
	1032	+
	1033	+ function source ($what = NULL) {
	1034	+ $ret = NULL;
	1035	+ $source = $this->entry->get_source();
	1036	+ if ($source) :
	1037	+ $ret = array();
	1038	+ $ret['title'] = $source->get_title();
	1039	+ $ret['uri'] = $source->get_link();
	1040	+ $ret['feed'] = $source->get_link(0, 'self');
246	1041
247		- if (isset($this->feedmeta['tags']) and is_array($this->feedmeta['tags'])) :
248		- $this->post['tags_input'] = array_merge($this->post['tags_input'], $this->feedmeta['tags']);
	1042	+ if ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'id')) :
	1043	+ $ret['id'] = $id_tags[0]['data'];
	1044	+ elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_ATOM_03, 'id')) :
	1045	+ $ret['id'] = $id_tags[0]['data'];
	1046	+ elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_RSS_20, 'guid')) :
	1047	+ $ret['id'] = $id_tags[0]['data'];
	1048	+ elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_RSS_10, 'guid')) :
	1049	+ $ret['id'] = $id_tags[0]['data'];
	1050	+ elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_RSS_090, 'guid')) :
	1051	+ $ret['id'] = $id_tags[0]['data'];
249	1052	endif;
250		- $this->post['tags_input'] = apply_filters('syndicated_item_tags', $this->post['tags_input'], $this);
251	1053	endif;
252		- } // SyndicatedPost::SyndicatedPost()
	1054	+
	1055	+ if (!is_null($what) and is_scalar($what)) :
	1056	+ $ret = $ret[$what];
	1057	+ endif;
	1058	+ return $ret;
	1059	+ }
	1060	+
	1061	+ function comment_link () {
	1062	+ $url = null;
	1063	+
	1064	+ // RSS 2.0 has a standard <comments> element:
	1065	+ // "<comments> is an optional sub-element of <item>. If present,
	1066	+ // it is the url of the comments page for the item."
	1067	+ // <http://cyber.law.harvard.edu/rss/rss.html#ltcommentsgtSubelementOfLtitemgt>
	1068	+ if (isset($this->item['comments'])) :
	1069	+ $url = $this->item['comments'];
	1070	+ endif;
253	1071
	1072	+ // The convention in Atom feeds is to use a standard <link>
	1073	+ // element with @rel="replies" and @type="text/html".
	1074	+ // Unfortunately, SimplePie_Item::get_links() allows us to filter
	1075	+ // by the value of @rel, but not by the value of @type. *sigh*
	1076	+
	1077	+ // Try Atom 1.0 first
	1078	+ $linkElements = $this->entry->get_item_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'link');
	1079	+
	1080	+ // Fall back and try Atom 0.3
	1081	+ if (is_null($linkElements)) : $linkElements = $this->entry->get_item_tags(SIMPLEPIE_NAMESPACE_ATOM_03, 'link'); endif;
	1082	+
	1083	+ // Now loop through the elements, screening by @rel and @type
	1084	+ if (is_array($linkElements)) : foreach ($linkElements as $link) :
	1085	+ $rel = (isset($link['attribs']['']['rel']) ? $link['attribs']['']['rel'] : 'alternate');
	1086	+ $type = (isset($link['attribs']['']['type']) ? $link['attribs']['']['type'] : NULL);
	1087	+ $href = (isset($link['attribs']['']['href']) ? $link['attribs']['']['href'] : NULL);
	1088	+
	1089	+ if (strtolower($rel)=='replies' and $type=='text/html' and !is_null($href)) :
	1090	+ $url = $href;
	1091	+ endif;
	1092	+ endforeach; endif;
	1093	+
	1094	+ return $url;
	1095	+ }
	1096	+
	1097	+ function comment_feed () {
	1098	+ $feed = null;
	1099	+
	1100	+ // Well Formed Web comment feeds extension for RSS 2.0
	1101	+ // <http://www.sellsbrothers.com/spout/default.aspx?content=archive.htm#exposingRssComments>
	1102	+ //
	1103	+ // N.B.: Correct capitalization is wfw:commentRss, but
	1104	+ // wfw:commentRSS is common in the wild (partly due to a typo in
	1105	+ // the original spec). In any case, our item array is normalized
	1106	+ // to all lowercase anyways.
	1107	+ if (isset($this->item['wfw']['commentrss'])) :
	1108	+ $feed = $this->item['wfw']['commentrss'];
	1109	+ endif;
	1110	+
	1111	+ // In Atom 1.0, the convention is to use a standard link element
	1112	+ // with @rel="replies". Sometimes this is also used to pass a
	1113	+ // link to the human-readable comments page, so we also need to
	1114	+ // check link/@type for a feed MIME type.
	1115	+ //
	1116	+ // Which is why I'm not using the SimplePie_Item::get_links()
	1117	+ // method here, incidentally: it doesn't allow you to filter by
	1118	+ // @type. *sigh*
	1119	+ if (isset($this->item['link_replies'])) :
	1120	+ // There may be multiple <link rel="replies"> elements; feeds have a feed MIME type
	1121	+ $N = isset($this->item['link_replies#']) ? $this->item['link_replies#'] : 1;
	1122	+ for ($i = 1; $i <= $N; $i++) :
	1123	+ $currentElement = 'link_replies'.(($i > 1) ? '#'.$i : '');
	1124	+ if (isset($this->item[$currentElement.'@type'])
	1125	+ and preg_match("\007application/(atom|rss|rdf)\+xml\007i", $this->item[$currentElement.'@type'])) :
	1126	+ $feed = $this->item[$currentElement];
	1127	+ endif;
	1128	+ endfor;
	1129	+ endif;
	1130	+ return $feed;
	1131	+ } /* SyndicatedPost::comment_feed() */
	1132	+
	1133	+ ##################################
	1134	+ #### BUILT-IN CONTENT FILTERS ####
	1135	+ ##################################
	1136	+
	1137	+ var $uri_attrs = array (
	1138	+ array('a', 'href'),
	1139	+ array('applet', 'codebase'),
	1140	+ array('area', 'href'),
	1141	+ array('blockquote', 'cite'),
	1142	+ array('body', 'background'),
	1143	+ array('del', 'cite'),
	1144	+ array('form', 'action'),
	1145	+ array('frame', 'longdesc'),
	1146	+ array('frame', 'src'),
	1147	+ array('iframe', 'longdesc'),
	1148	+ array('iframe', 'src'),
	1149	+ array('head', 'profile'),
	1150	+ array('img', 'longdesc'),
	1151	+ array('img', 'src'),
	1152	+ array('img', 'usemap'),
	1153	+ array('input', 'src'),
	1154	+ array('input', 'usemap'),
	1155	+ array('ins', 'cite'),
	1156	+ array('link', 'href'),
	1157	+ array('object', 'classid'),
	1158	+ array('object', 'codebase'),
	1159	+ array('object', 'data'),
	1160	+ array('object', 'usemap'),
	1161	+ array('q', 'cite'),
	1162	+ array('script', 'src')
	1163	+ ); /* var SyndicatedPost::$uri_attrs */
	1164	+
	1165	+ var $_base = null;
	1166	+
	1167	+ function resolve_single_relative_uri ($refs) {
	1168	+ $tag = FeedWordPressHTML::attributeMatch($refs);
	1169	+ $url = SimplePie_Misc::absolutize_url($tag['value'], $this->_base);
	1170	+ return $tag['prefix'] . $url . $tag['suffix'];
	1171	+ } /* function SyndicatedPost::resolve_single_relative_uri() */
	1172	+
	1173	+ function resolve_relative_uris ($content, $obj) {
	1174	+ $set = $obj->link->setting('resolve relative', 'resolve_relative', 'yes');
	1175	+ if ($set and $set != 'no') :
	1176	+ // Fallback: if we don't have anything better, use the
	1177	+ // item link from the feed
	1178	+ $obj->_base = $obj->permalink(); // Reset the base for resolving relative URIs
	1179	+
	1180	+ // What we should do here, properly, is to use
	1181	+ // SimplePie_Item::get_base() -- but that method is
	1182	+ // currently broken. Or getting down and dirty in the
	1183	+ // SimplePie representation of the content tags and
	1184	+ // grabbing the xml_base member for the content element.
	1185	+ // Maybe someday...
	1186	+
	1187	+ foreach ($obj->uri_attrs as $pair) :
	1188	+ list($tag, $attr) = $pair;
	1189	+ $pattern = FeedWordPressHTML::attributeRegex($tag, $attr);
	1190	+ $content = preg_replace_callback (
	1191	+ $pattern,
	1192	+ array($obj, 'resolve_single_relative_uri'),
	1193	+ $content
	1194	+ );
	1195	+ endforeach;
	1196	+ endif;
	1197	+
	1198	+ return $content;
	1199	+ } /* function SyndicatedPost::resolve_relative_uris () */
	1200	+
	1201	+ var $strip_attrs = array (
	1202	+ array('[a-z]+', 'target'),
	1203	+// array('[a-z]+', 'style'),
	1204	+// array('[a-z]+', 'on[a-z]+'),
	1205	+ );
	1206	+
	1207	+ function strip_attribute_from_tag ($refs) {
	1208	+ $tag = FeedWordPressHTML::attributeMatch($refs);
	1209	+ return $tag['before_attribute'].$tag['after_attribute'];
	1210	+ }
	1211	+
	1212	+ function sanitize_content ($content, $obj) {
	1213	+ # This kind of sucks. I intend to replace it with
	1214	+ # lib_filter sometime soon.
	1215	+ foreach ($obj->strip_attrs as $pair):
	1216	+ list($tag,$attr) = $pair;
	1217	+ $pattern = FeedWordPressHTML::attributeRegex($tag, $attr);
	1218	+
	1219	+ $content = preg_replace_callback (
	1220	+ $pattern,
	1221	+ array($obj, 'strip_attribute_from_tag'),
	1222	+ $content
	1223	+ );
	1224	+ endforeach;
	1225	+ return $content;
	1226	+ } /* SyndicatedPost::sanitize() */
	1227	+
	1228	+ #####################
	1229	+ #### POST STATUS ####
	1230	+ #####################
	1231	+
	1232	+ /**
	1233	+ * SyndicatedPost::filtered: check whether or not this post has been
	1234	+ * screened out by a registered filter.
	1235	+ *
	1236	+ * @return bool TRUE iff post has been filtered out by a previous filter
	1237	+ */
254	1238	function filtered () {
255	1239	return is_null($this->post);
256		- }
	1240	+ } /* SyndicatedPost::filtered() */
257	1241
	1242	+ /**
	1243	+ * SyndicatedPost::freshness: check whether post is a new post to be
	1244	+ * inserted, a previously syndicated post that needs to be updated to
	1245	+ * match the latest revision, or a previously syndicated post that is
	1246	+ * still up-to-date.
	1247	+ *
	1248	+ * @return int A status code representing the freshness of the post
	1249	+ * 0 = post already syndicated; no update needed
	1250	+ * 1 = post already syndicated, but needs to be updated to latest
	1251	+ * 2 = post has not yet been syndicated; needs to be created
	1252	+ */
258	1253	function freshness () {
259	1254	global $wpdb;
260	1255
261	1256	if ($this->filtered()) : // This should never happen.
262		- FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__);
	1257	+ FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__, __FILE__);
263	1258	endif;
264	1259
265		- if (is_null($this->_freshness)) :
266		- $guid = $wpdb->escape($this->guid());
	1260	+ if (is_null($this->_freshness)) : // Not yet checked and cached.
	1261	+ $guid = $this->post['guid'];
	1262	+ $eguid = $wpdb->escape($this->post['guid']);
267	1263
268		- $result = $wpdb->get_row("
269		- SELECT id, guid, post_modified_gmt
270		- FROM $wpdb->posts WHERE guid='$guid'
271		- ");
272		-
273		- if (!$result) :
	1264	+ $q = new WP_Query(array(
	1265	+ 'fields' => '_synfresh', // id, guid, post_modified_gmt
	1266	+ 'ignore_sticky_posts' => true,
	1267	+ 'guid' => $guid,
	1268	+ ));
	1269	+
	1270	+ $old_post = NULL;
	1271	+ if ($q->have_posts()) :
	1272	+ while ($q->have_posts()) : $q->the_post();
	1273	+ $old_post = $q->post;
	1274	+ endwhile;
	1275	+ endif;
	1276	+
	1277	+ if (is_null($old_post)) : // No post with this guid
	1278	+ FeedWordPress::diagnostic('feed_items:freshness', 'Item ['.$guid.'] "'.$this->entry->get_title().'" is a NEW POST.');
	1279	+ $this->_wp_id = NULL;
274	1280	$this->_freshness = 2; // New content
275		- else:
276		- $stored_update_hashes = get_post_custom_values('syndication_item_hash', $result->id);
277		- if (count($stored_update_hashes) > 0) :
278		- $stored_update_hash = $stored_update_hashes[0];
279		- $update_hash_changed = ($stored_update_hash != $this->update_hash());
280		- else :
281		- $update_hash_changed = false;
282		- endif;
	1281	+ else :
	1282	+ preg_match('/([0-9]+)-([0-9]+)-([0-9]+) ([0-9]+):([0-9]+):([0-9]+)/', $old_post->post_modified_gmt, $backref);
283	1283
284		- preg_match('/([0-9]+)-([0-9]+)-([0-9]+) ([0-9]+):([0-9]+):([0-9]+)/', $result->post_modified_gmt, $backref);
285		-
286	1284	$last_rev_ts = gmmktime($backref[4], $backref[5], $backref[6], $backref[2], $backref[3], $backref[1]);
287	1285	$updated_ts = $this->updated(/*fallback=*/ true, /*default=*/ NULL);
288		- $updated = ((
	1286	+
	1287	+ // Check timestamps...
	1288	+ $updated = (
289	1289	!is_null($updated_ts)
290	1290	and ($updated_ts > $last_rev_ts)
291		- ) or $update_hash_changed);
	1291	+ );
292	1292
	1293	+ $updatedReason = NULL;
293	1294	if ($updated) :
	1295	+ $updatedReason = preg_replace(
	1296	+ "/\s+/", " ",
	1297	+ 'has been marked with a new timestamp ('
	1298	+ .date('Y-m-d H:i:s', $updated_ts)
	1299	+ ." > "
	1300	+ .date('Y-m-d H:i:s', $last_rev_ts)
	1301	+ .')'
	1302	+ );
	1303	+ else :
	1304	+ // Or the hash...
	1305	+ $hash = $this->update_hash();
	1306	+ $seen = $this->stored_hashes($old_post->ID);
	1307	+ if (count($seen) > 0) :
	1308	+ $updated = !in_array($hash, $seen); // Not seen yet?
	1309	+ else :
	1310	+ $updated = true; // Can't find syndication meta-data
	1311	+ endif;
	1312	+
	1313	+ if ($updated and FeedWordPress::diagnostic_on('feed_items:freshness:reasons')) :
	1314	+ $updatedReason = ' has a not-yet-seen update hash: '
	1315	+ .FeedWordPress::val($hash)
	1316	+ .' not in {'
	1317	+ .implode(", ", array_map(array('FeedWordPress', 'val'), $seen))
	1318	+ .'}. Basis: '
	1319	+ .FeedWordPress::val(array_keys($this->update_hash(false)));
	1320	+ endif;
	1321	+ endif;
	1322	+
	1323	+ $frozen = false;
	1324	+ if ($updated) : // Ignore if the post is frozen
	1325	+ $frozen = ('yes' == $this->link->setting('freeze updates', 'freeze_updates', NULL));
	1326	+ if (!$frozen) :
	1327	+ $frozen_values = get_post_custom_values('_syndication_freeze_updates', $old_post->ID);
	1328	+ $frozen = (count($frozen_values) > 0 and 'yes' == $frozen_values[0]);
	1329	+
	1330	+ if ($frozen) :
	1331	+ $updatedReason = ' IS BLOCKED FROM BEING UPDATED BY A UPDATE LOCK ON THIS POST, EVEN THOUGH IT '.$updatedReason;
	1332	+ endif;
	1333	+ else :
	1334	+ $updatedReason = ' IS BLOCKED FROM BEING UPDATED BY A FEEDWORDPRESS UPDATE LOCK, EVEN THOUGH IT '.$updatedReason;
	1335	+ endif;
	1336	+ endif;
	1337	+ $updated = ($updated and !$frozen);
	1338	+
	1339	+ if ($updated) :
	1340	+ FeedWordPress::diagnostic('feed_items:freshness', 'Item ['.$guid.'] "'.$this->entry->get_title().'" is an update of an existing post.');
	1341	+ if (!is_null($updatedReason)) :
	1342	+ $updatedReason = preg_replace('/\s+/', ' ', $updatedReason);
	1343	+ FeedWordPress::diagnostic('feed_items:freshness:reasons', 'Item ['.$guid.'] "'.$this->entry->get_title().'" '.$updatedReason);
	1344	+ endif;
294	1345	$this->_freshness = 1; // Updated content
295		- $this->_wp_id = $result->id;
	1346	+ $this->_wp_id = $old_post->ID;
	1347	+ $this->_wp_post = $old_post;
	1348	+
	1349	+ // We want this to keep a running list of all the
	1350	+ // processed update hashes.
	1351	+ $this->post['meta']['syndication_item_hash'] = array_merge(
	1352	+ $this->stored_hashes(),
	1353	+ array($this->update_hash())
	1354	+ );
296	1355	else :
	1356	+ FeedWordPress::diagnostic('feed_items:freshness', 'Item ['.$guid.'] "'.$this->entry->get_title().'" is a duplicate of an existing post.');
297	1357	$this->_freshness = 0; // Same old, same old
298		- $this->_wp_id = $result->id;
	1358	+ $this->_wp_id = $old_post->ID;
299	1359	endif;
300	1360	endif;
301	1361	endif;
302	1362	return $this->_freshness;
		@@ -301,11 +1361,15 @@
301	1361	endif;
302	1362	return $this->_freshness;
303	1363	}
304	1364
	1365	+ #################################################
	1366	+ #### INTERNAL STORAGE AND MANAGEMENT METHODS ####
	1367	+ #################################################
	1368	+
305	1369	function wp_id () {
306	1370	if ($this->filtered()) : // This should never happen.
307		- FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__);
	1371	+ FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__, __FILE__);
308	1372	endif;
309	1373
310	1374	if (is_null($this->_wp_id) and is_null($this->_freshness)) :
311	1375	$fresh = $this->freshness(); // sets WP DB id in the process
		@@ -316,9 +1380,9 @@
316	1380	function store () {
317	1381	global $wpdb;
318	1382
319	1383	if ($this->filtered()) : // This should never happen.
320		- FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__);
	1384	+ FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__, __FILE__);
321	1385	endif;
322	1386
323	1387	$freshness = $this->freshness();
324	1388	if ($freshness > 0) :
		@@ -323,229 +1387,254 @@
323	1387	$freshness = $this->freshness();
324	1388	if ($freshness > 0) :
325	1389	# -- Look up, or create, numeric ID for author
326	1390	$this->post['post_author'] = $this->author_id (
327		- FeedWordPress::on_unfamiliar('author', $this->post['named']['unfamiliar']['author'])
	1391	+ $this->link->setting('unfamiliar author', 'unfamiliar_author', 'create')
328	1392	);
329	1393
330	1394	if (is_null($this->post['post_author'])) :
	1395	+ FeedWordPress::diagnostic('feed_items:rejected', 'Filtered out item ['.$this->guid().'] without syndication: no author available');
331	1396	$this->post = NULL;
332	1397	endif;
333	1398	endif;
334	1399
335	1400	if (!$this->filtered() and $freshness > 0) :
336		- # -- Look up, or create, numeric ID for categories
337		- list($pcats, $ptags) = $this->category_ids (
338		- $this->post['named']['category'],
339		- FeedWordPress::on_unfamiliar('category', $this->post['named']['unfamiliar']['category']),
340		- /*tags_too=*/ true
	1401	+ $consider = array(
	1402	+ 'category' => array('abbr' => 'cats', 'domain' => array('category', 'post_tag')),
	1403	+ 'post_tag' => array('abbr' => 'tags', 'domain' => array('post_tag')),
341	1404	);
342	1405
343		- $this->post['post_category'] = $pcats;
344		- $this->post['tags_input'] = array_merge($this->post['tags_input'], $ptags);
	1406	+ $termSet = array(); $valid = null;
	1407	+ foreach ($consider as $what => $taxes) :
	1408	+ if (!is_null($this->post)) : // Not filtered out yet
	1409	+ # -- Look up, or create, numeric ID for categories
	1410	+ $taxonomies = $this->link->setting("match/".$taxes['abbr'], 'match_'.$taxes['abbr'], $taxes['domain']);
	1411	+
	1412	+ // Eliminate dummy variables
	1413	+ $taxonomies = array_filter($taxonomies, 'remove_dummy_zero');
	1414	+
	1415	+ $terms = $this->category_ids (
	1416	+ $this->feed_terms[$what],
	1417	+ $this->link->setting("unfamiliar {$what}", "unfamiliar_{$what}", 'create:'.$what),
	1418	+ /*taxonomies=*/ $taxonomies,
	1419	+ array(
	1420	+ 'singleton' => false, // I don't like surprises
	1421	+ 'filters' => true,
	1422	+ )
	1423	+ );
	1424	+
	1425	+ if (is_null($terms) or is_null($termSet)) :
	1426	+ // filtered out -- no matches
	1427	+ else :
	1428	+ $valid = true;
345	1429
346		- if (is_null($this->post['post_category'])) :
347		- // filter mode on, no matching categories; drop the post
	1430	+ // filter mode off, or at least one match
	1431	+ foreach ($terms as $tax => $term_ids) :
	1432	+ if (!isset($termSet[$tax])) :
	1433	+ $termSet[$tax] = array();
	1434	+ endif;
	1435	+ $termSet[$tax] = array_merge($termSet[$tax], $term_ids);
	1436	+ endforeach;
	1437	+ endif;
	1438	+ endif;
	1439	+ endforeach;
	1440	+
	1441	+ if (is_null($valid)) : // Plonked
348	1442	$this->post = NULL;
349		- else :
350		- // filter mode off or at least one match; now add on the feed and global presets
351		- $this->post['post_category'] = array_merge (
352		- $this->post['post_category'],
353		- $this->category_ids (
354		- $this->post['named']['preset/category'],
355		- 'default'
356		- )
357		- );
	1443	+ else : // We can proceed
	1444	+ $this->post['tax_input'] = array();
	1445	+ foreach ($termSet as $tax => $term_ids) :
	1446	+ if (!isset($this->post['tax_input'][$tax])) :
	1447	+ $this->post['tax_input'][$tax] = array();
	1448	+ endif;
	1449	+ $this->post['tax_input'][$tax] = array_merge(
	1450	+ $this->post['tax_input'][$tax],
	1451	+ $term_ids
	1452	+ );
	1453	+ endforeach;
358	1454
359		- if (count($this->post['post_category']) < 1) :
360		- $this->post['post_category'][] = 1; // Default to category 1 ("Uncategorized" / "General") if nothing else
361		- endif;
	1455	+ // Now let's add on the feed and global presets
	1456	+ foreach ($this->preset_terms as $tax => $term_ids) :
	1457	+ if (!isset($this->post['tax_input'][$tax])) :
	1458	+ $this->post['tax_input'][$tax] = array();
	1459	+ endif;
	1460	+
	1461	+ $this->post['tax_input'][$tax] = array_merge (
	1462	+ $this->post['tax_input'][$tax],
	1463	+ $this->category_ids (
	1464	+ /*terms=*/ $term_ids,
	1465	+ /*unfamiliar=*/ 'create:'.$tax, // These are presets; for those added in a tagbox editor, the tag may not yet exist
	1466	+ /*taxonomies=*/ array($tax),
	1467	+ array(
	1468	+ 'singleton' => true,
	1469	+ ))
	1470	+ );
	1471	+ endforeach;
362	1472	endif;
363	1473	endif;
364	1474
365	1475	if (!$this->filtered() and $freshness > 0) :
366		- unset($this->post['named']);
	1476	+ // Filter some individual fields
	1477	+
	1478	+ // If there already is a post slug (from syndication or by manual
	1479	+ // editing) don't cause WP to overwrite it by sending in a NULL
	1480	+ // post_name. Props Chris Fritz 2012-11-28.
	1481	+ $post_name = (is_null($this->_wp_post) ? NULL : $this->_wp_post->post_name);
	1482	+
	1483	+ // Allow filters to set post slug. Props niska.
	1484	+ $post_name = apply_filters('syndicated_post_slug', $post_name, $this);
	1485	+ if (!empty($post_name)) :
	1486	+ $this->post['post_name'] = $post_name;
	1487	+ endif;
	1488	+
367	1489	$this->post = apply_filters('syndicated_post', $this->post, $this);
	1490	+
	1491	+ // Allow for feed-specific syndicated_post filters.
	1492	+ $this->post = apply_filters(
	1493	+ "syndicated_post_".$this->link->uri(),
	1494	+ $this->post,
	1495	+ $this
	1496	+ );
368	1497	endif;
369	1498
370		- if (!$this->filtered() and $freshness == 2) :
371		- // The item has not yet been added. So let's add it.
372		- $this->insert_new();
373		- $this->add_rss_meta();
374		- do_action('post_syndicated_item', $this->wp_id());
	1499	+ // Hook in early to make sure these get inserted if at all possible
	1500	+ add_action(
	1501	+ /*hook=*/ 'transition_post_status',
	1502	+ /*callback=*/ array($this, 'add_rss_meta'),
	1503	+ /*priority=*/ -10000, /* very early */
	1504	+ /*arguments=*/ 3
	1505	+ );
375	1506
376		- $ret = 'new';
377		- elseif (!$this->filtered() and $freshness == 1) :
378		- $this->post['ID'] = $this->wp_id();
379		- $this->update_existing();
380		- $this->add_rss_meta();
381		- do_action('update_syndicated_item', $this->wp_id());
	1507	+ $retval = array(1 => 'updated', 2 => 'new');
	1508	+
	1509	+ $ret = false;
	1510	+ if (!$this->filtered() and isset($retval[$freshness])) :
	1511	+ $diag = array(
	1512	+ 1 => 'Updating existing post # '.$this->wp_id().', "'.$this->post['post_title'].'"',
	1513	+ 2 => 'Inserting new post "'.$this->post['post_title'].'"',
	1514	+ );
	1515	+ FeedWordPress::diagnostic('syndicated_posts', $diag[$freshness]);
382	1516
383		- $ret = 'updated';
384		- else :
385		- $ret = false;
	1517	+ $this->insert_post(/*update=*/ ($freshness == 1));
	1518	+
	1519	+ $hook = array( 1 => 'update_syndicated_item', 2 => 'post_syndicated_item' );
	1520	+ do_action($hook[$freshness], $this->wp_id(), $this);
	1521	+
	1522	+ $ret = $retval[$freshness];
386	1523	endif;
	1524	+
	1525	+ // If this is a legit, non-filtered post, tag it as found on the feed
	1526	+ // regardless of fresh or stale status
	1527	+ if (!$this->filtered()) :
	1528	+ $key = '_feedwordpress_retire_me_' . $this->link->id;
	1529	+ delete_post_meta($this->wp_id(), $key);
	1530	+
	1531	+ $status = get_post_field('post_status', $this->wp_id());
	1532	+ if ('fwpretired'==$status and $this->link->is_incremental()) :
	1533	+ FeedWordPress::diagnostic('syndicated_posts', "Un-retiring previously retired post # ".$this->wp_id()." due to re-appearance on non-incremental feed.");
	1534	+ set_post_field('post_status', $this->post['post_status'], $this->wp_id());
	1535	+ wp_transition_post_status($this->post['post_status'], $status, $old_status, $this->post);
	1536	+ endif;
	1537	+ endif;
387	1538
	1539	+ // Remove add_rss_meta hook
	1540	+ remove_action(
	1541	+ /*hook=*/ 'transition_post_status',
	1542	+ /*callback=*/ array($this, 'add_rss_meta'),
	1543	+ /*priority=*/ -10000, /* very early */
	1544	+ /*arguments=*/ 3
	1545	+ );
	1546	+
388	1547	return $ret;
389		- } // function SyndicatedPost::store ()
	1548	+ } /* function SyndicatedPost::store () */
390	1549
391		- function insert_new () {
392		- global $wpdb, $wp_db_version;
	1550	+ function insert_post ($update = false) {
	1551	+ global $wpdb;
393	1552
394	1553	$dbpost = $this->normalize_post(/*new=*/ true);
395	1554	if (!is_null($dbpost)) :
396		- if ($this->use_api('wp_insert_post')) :
397		- $dbpost['post_pingback'] = false; // Tell WP 2.1 and 2.2 not to process for pingbacks
	1555	+ $dbpost['post_pingback'] = false; // Tell WP 2.1 and 2.2 not to process for pingbacks
398	1556
399		- // This is a ridiculous fucking kludge necessitated by WordPress 2.6 munging authorship meta-data
400		- add_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
	1557	+ // This is a ridiculous fucking kludge necessitated by WordPress 2.6 munging authorship meta-data
	1558	+ add_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
401	1559
402		- // Kludge to prevent kses filters from stripping the
403		- // content of posts when updating without a logged in
404		- // user who has `unfiltered_html` capability.
405		- add_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11);
	1560	+ // Kludge to prevent kses filters from stripping the
	1561	+ // content of posts when updating without a logged in
	1562	+ // user who has `unfiltered_html` capability.
	1563	+ $mungers = array('wp_filter_kses', 'wp_filter_post_kses');
	1564	+ $removed = array();
	1565	+ foreach ($mungers as $munger) :
	1566	+ if (has_filter('content_save_pre', $munger)) :
	1567	+ remove_filter('content_save_pre', $munger);
	1568	+ $removed[] = $munger;
	1569	+ endif;
	1570	+ endforeach;
	1571	+
	1572	+ if ($update and function_exists('get_post_field')) :
	1573	+ // Don't munge status fields that the user may
	1574	+ // have reset manually
	1575	+ $doNotMunge = array('post_status', 'comment_status', 'ping_status');
406	1576
407		- $this->_wp_id = wp_insert_post($dbpost);
408		-
409		- // Turn off ridiculous fucking kludges #1 and #2
410		- remove_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
411		- remove_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11);
412		-
413		- $this->validate_post_id($dbpost, array(__CLASS__, __FUNCTION__));
414		-
415		- // Unfortunately, as of WordPress 2.3, wp_insert_post()
416		- // *still* offers no way to use a guid of your choice,
417		- // and munges your post modified timestamp, too.
418		- $result = $wpdb->query("
419		- UPDATE $wpdb->posts
420		- SET
421		- guid='{$dbpost['guid']}',
422		- post_modified='{$dbpost['post_modified']}',
423		- post_modified_gmt='{$dbpost['post_modified_gmt']}'
424		- WHERE ID='{$this->_wp_id}'
425		- ");
426		- else :
427		- # The right way to do this is the above. But, alas,
428		- # in earlier versions of WordPress, wp_insert_post has
429		- # too much behavior (mainly related to pings) that can't
430		- # be overridden. In WordPress 1.5, it's enough of a
431		- # resource hog to make PHP segfault after inserting
432		- # 50-100 posts. This can get pretty annoying, especially
433		- # if you are trying to update your feeds for the first
434		- # time.
435		-
436		- $result = $wpdb->query("
437		- INSERT INTO $wpdb->posts
438		- SET
439		- guid = '{$dbpost['guid']}',
440		- post_author = '{$dbpost['post_author']}',
441		- post_date = '{$dbpost['post_date']}',
442		- post_date_gmt = '{$dbpost['post_date_gmt']}',
443		- post_content = '{$dbpost['post_content']}',"
444		- .(isset($dbpost['post_excerpt']) ? "post_excerpt = '{$dbpost['post_excerpt']}'," : "")."
445		- post_title = '{$dbpost['post_title']}',
446		- post_name = '{$dbpost['post_name']}',
447		- post_modified = '{$dbpost['post_modified']}',
448		- post_modified_gmt = '{$dbpost['post_modified_gmt']}',
449		- comment_status = '{$dbpost['comment_status']}',
450		- ping_status = '{$dbpost['ping_status']}',
451		- post_status = '{$dbpost['post_status']}'
452		- ");
453		- $this->_wp_id = $wpdb->insert_id;
454		-
455		- $this->validate_post_id($dbpost, array(__CLASS__, __FUNCTION__));
456		-
457		- // WordPress 1.5.x - 2.0.x
458		- wp_set_post_cats('1', $this->wp_id(), $this->post['post_category']);
459		-
460		- // Since we are not going through official channels, we need to
461		- // manually tell WordPress that we've published a new post.
462		- // We need to make sure to do this in order for FeedWordPress
463		- // to play well with the staticize-reloaded plugin (something
464		- // that a large aggregator website is going to *want* to be
465		- // able to use).
466		- do_action('publish_post', $this->_wp_id);
	1577	+ foreach ($doNotMunge as $field) :
	1578	+ $dbpost[$field] = get_post_field($field, $this->wp_id());
	1579	+ endforeach;
467	1580	endif;
468		- endif;
469		- } /* SyndicatedPost::insert_new() */
	1581	+
	1582	+ // WP3's wp_insert_post scans current_user_can() for the
	1583	+ // tax_input, with no apparent way to override. Ugh.
	1584	+ add_action(
	1585	+ /*hook=*/ 'transition_post_status',
	1586	+ /*callback=*/ array($this, 'add_terms'),
	1587	+ /*priority=*/ -10001, /* very early */
	1588	+ /*arguments=*/ 3
	1589	+ );
	1590	+
	1591	+ // WP3 appears to override whatever you give it for
	1592	+ // post_modified. Ugh.
	1593	+ add_action(
	1594	+ /*hook=*/ 'transition_post_status',
	1595	+ /*callback=*/ array($this, 'fix_post_modified_ts'),
	1596	+ /*priority=*/ -10000, /* very early */
	1597	+ /*arguments=*/ 3
	1598	+ );
470	1599
471		- function update_existing () {
472		- global $wpdb;
	1600	+ if ($update) :
	1601	+ $this->post['ID'] = $this->wp_id();
	1602	+ $dbpost['ID'] = $this->post['ID'];
	1603	+ endif;
	1604	+
	1605	+ $this->_wp_id = wp_insert_post($dbpost, /*return wp_error=*/ true);
473	1606
474		- // Why the fuck doesn't wp_insert_post already do this?
475		- $dbpost = $this->normalize_post(/*new=*/ false);
476		- if (!is_null($dbpost)) :
477		- if ($this->use_api('wp_insert_post')) :
478		- $dbpost['post_pingback'] = false; // Tell WP 2.1 and 2.2 not to process for pingbacks
479		-
480		- // This is a ridiculous fucking kludge necessitated by WordPress 2.6 munging authorship meta-data
481		- add_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
482		-
483		- // Kludge to prevent kses filters from stripping the
484		- // content of posts when updating without a logged in
485		- // user who has `unfiltered_html` capability.
486		- add_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11);
	1607	+ remove_action(
	1608	+ /*hook=*/ 'transition_post_status',
	1609	+ /*callback=*/ array($this, 'add_terms'),
	1610	+ /*priority=*/ -10001, /* very early */
	1611	+ /*arguments=*/ 3
	1612	+ );
487	1613
488		- // Don't munge status fields that the user may have reset manually
489		- if (function_exists('get_post_field')) :
490		- $doNotMunge = array('post_status', 'comment_status', 'ping_status');
491		- foreach ($doNotMunge as $field) :
492		- $dbpost[$field] = get_post_field($field, $this->wp_id());
493		- endforeach;
494		- endif;
	1614	+ remove_action(
	1615	+ /*hook=*/ 'transition_post_status',
	1616	+ /*callback=*/ array($this, 'fix_post_modified_ts'),
	1617	+ /*priority=*/ -10000, /* very early */
	1618	+ /*arguments=*/ 3
	1619	+ );
495	1620
496		- $this->_wp_id = wp_insert_post($dbpost);
	1621	+ // Turn off ridiculous fucking kludges #1 and #2
	1622	+ remove_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
	1623	+ foreach ($removed as $filter) :
	1624	+ add_filter('content_save_pre', $filter);
	1625	+ endforeach;
	1626	+
	1627	+ $this->validate_post_id($dbpost, $update, array(__CLASS__, __FUNCTION__));
	1628	+ endif;
	1629	+ } /* function SyndicatedPost::insert_post () */
497	1630
498		- // Turn off ridiculous fucking kludges #1 and #2
499		- remove_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
500		- remove_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11);
501		-
502		- $this->validate_post_id($dbpost, array(__CLASS__, __FUNCTION__));
503		-
504		- // Unfortunately, as of WordPress 2.3, wp_insert_post()
505		- // munges your post modified timestamp.
506		- $result = $wpdb->query("
507		- UPDATE $wpdb->posts
508		- SET
509		- post_modified='{$dbpost['post_modified']}',
510		- post_modified_gmt='{$dbpost['post_modified_gmt']}'
511		- WHERE ID='{$this->_wp_id}'
512		- ");
513		- else :
514		-
515		- $result = $wpdb->query("
516		- UPDATE $wpdb->posts
517		- SET
518		- post_author = '{$dbpost['post_author']}',
519		- post_content = '{$dbpost['post_content']}',"
520		- .(isset($dbpost['post_excerpt']) ? "post_excerpt = '{$dbpost['post_excerpt']}'," : "")."
521		- post_title = '{$dbpost['post_title']}',
522		- post_name = '{$dbpost['post_name']}',
523		- post_modified = '{$dbpost['post_modified']}',
524		- post_modified_gmt = '{$dbpost['post_modified_gmt']}'
525		- WHERE guid='{$dbpost['guid']}'
526		- ");
527		-
528		- // WordPress 2.1.x and up
529		- if (function_exists('wp_set_post_categories')) :
530		- wp_set_post_categories($this->wp_id(), $this->post['post_category']);
531		- // WordPress 1.5.x - 2.0.x
532		- elseif (function_exists('wp_set_post_cats')) :
533		- wp_set_post_cats('1', $this->wp_id(), $this->post['post_category']);
534		- // This should never happen.
535		- else :
536		- FeedWordPress::critical_bug(__CLASS__.'::'.__FUNCTION.'(): no post categorizing function', array("dbpost" => $dbpost, "this" => $this), __LINE__);
537		- endif;
538		-
539		- // Since we are not going through official channels, we need to
540		- // manually tell WordPress that we've published a new post.
541		- // We need to make sure to do this in order for FeedWordPress
542		- // to play well with the staticize-reloaded plugin (something
543		- // that a large aggregator website is going to *want* to be
544		- // able to use).
545		- do_action('edit_post', $this->post['ID']);
546		- endif;
547		- endif;
	1631	+ function insert_new () {
	1632	+ $this->insert_post(/*update=*/ false);
	1633	+ } /* SyndicatedPost::insert_new() */
	1634	+
	1635	+ function update_existing () {
	1636	+ $this->insert_post(/*update=*/ true);
548	1637	} /* SyndicatedPost::update_existing() */
549	1638
550	1639	/**
551	1640	* SyndicatedPost::normalize_post()
		@@ -555,12 +1644,34 @@
555	1644	*/
556	1645	function normalize_post ($new = true) {
557	1646	global $wpdb;
558	1647
559		- $out = array();
	1648	+ $out = $this->post;
560	1649
	1650	+ $fullPost = $out['post_title'].$out['post_content'];
	1651	+ $fullPost .= (isset($out['post_excerpt']) ? $out['post_excerpt'] : '');
	1652	+ if (strlen($fullPost) < 1) :
	1653	+ // FIXME: Option for filtering out empty posts
	1654	+ endif;
	1655	+ if (strlen($out['post_title'])==0) :
	1656	+ $offset = (int) get_option('gmt_offset') * 60 * 60;
	1657	+ if (isset($this->post['meta']['syndication_source'])) :
	1658	+ $source_title = $this->post['meta']['syndication_source'];
	1659	+ else :
	1660	+ $feed_url = parse_url($this->post['meta']['syndication_feed']);
	1661	+ $source_title = $feed_url['host'];
	1662	+ endif;
	1663	+
	1664	+ $out['post_title'] = $source_title
	1665	+ .' '.gmdate('Y-m-d H:i:s', $this->published() + $offset);
	1666	+ // FIXME: Option for what to fill a blank title with...
	1667	+ endif;
	1668	+
	1669	+ // Normalize the guid if necessary.
	1670	+ $out['guid'] = SyndicatedPost::normalize_guid($out['guid']);
	1671	+
561	1672	// Why the fuck doesn't wp_insert_post already do this?
562		- foreach ($this->post as $key => $value) :
	1673	+ foreach ($out as $key => $value) :
563	1674	if (is_string($value)) :
564	1675	$out[$key] = $wpdb->escape($value);
565	1676	else :
566	1677	$out[$key] = $value;
		@@ -566,19 +1677,8 @@
566	1677	$out[$key] = $value;
567	1678	endif;
568	1679	endforeach;
569	1680
570		- if (strlen($out['post_title'].$out['post_content'].$out['post_excerpt']) == 0) :
571		- // FIXME: Option for filtering out empty posts
572		- endif;
573		- if (strlen($out['post_title'])==0) :
574		- $offset = (int) get_option('gmt_offset') * 60 * 60;
575		- $out['post_title'] =
576		- $this->post['meta']['syndication_source']
577		- .' '.gmdate('Y-m-d H:i:s', $this->published() + $offset);
578		- // FIXME: Option for what to fill a blank title with...
579		- endif;
580		-
581	1681	return $out;
582	1682	}
583	1683
584	1684	/**
		@@ -586,22 +1686,43 @@
586	1686	*
587	1687	* @param array $dbpost An array representing the post we attempted to insert or update
588	1688	* @param mixed $ns A string or array representing the namespace (class, method) whence this method was called.
589	1689	*/
590		- function validate_post_id ($dbpost, $ns) {
	1690	+ function validate_post_id ($dbpost, $is_update, $ns) {
591	1691	if (is_array($ns)) : $ns = implode('::', $ns);
592	1692	else : $ns = (string) $ns; endif;
593		-
	1693	+
594	1694	// This should never happen.
595	1695	if (!is_numeric($this->_wp_id) or ($this->_wp_id == 0)) :
596		- FeedWordPress::critical_bug(
597		- /*name=*/ $ns.'::_wp_id',
	1696	+ $verb = ($is_update ? 'update existing' : 'insert new');
	1697	+ $guid = $this->guid();
	1698	+ $url = $this->permalink();
	1699	+ $feed = $this->link->uri(array('add_params' => true));
	1700	+
	1701	+ // wp_insert_post failed. Diagnostics, or barf up a critical bug
	1702	+ // notice if we are in debug mode.
	1703	+ $mesg = "Failed to $verb item [$guid]. WordPress API returned no valid post ID.\n"
	1704	+ ."\t\tID = ".serialize($this->_wp_id)."\n"
	1705	+ ."\t\tURL = ".FeedWordPress::val($url)
	1706	+ ."\t\tFeed = ".FeedWordPress::val($feed);
	1707	+ FeedWordPress::diagnostic('updated_feeds:errors', "WordPress API error: $mesg");
	1708	+ FeedWordPress::diagnostic('feed_items:rejected', $mesg);
	1709	+
	1710	+ $mesg = <<<EOM
	1711	+The WordPress API returned an invalid post ID
	1712	+ when FeedWordPress tried to $verb item $guid
	1713	+ [URL: $url]
	1714	+ from the feed at $feed
	1715	+
	1716	+$ns::_wp_id
	1717	+EOM;
	1718	+ FeedWordPress::noncritical_bug(
	1719	+ /*message=*/ $mesg,
598	1720	/*var =*/ array(
599	1721	"\$this->_wp_id" => $this->_wp_id,
600	1722	"\$dbpost" => $dbpost,
601		- "\$this" => $this
602	1723	),
603		- /*line # =*/ __LINE__
	1724	+ /*line # =*/ __LINE__, /*filename=*/ __FILE__
604	1725	);
605	1726	endif;
606	1727	} /* SyndicatedPost::validate_post_id() */
607	1728
		@@ -633,100 +1754,171 @@
633	1754	SET post_author={$this->post['post_author']}
634	1755	WHERE post_type = 'revision' AND ID='$revision_id'
635	1756	");
636	1757	} /* SyndicatedPost::fix_revision_meta () */
637		-
	1758	+
638	1759	/**
639		- * SyndicatedPost::avoid_kses_munge() -- If FeedWordPress is processing
640		- * an automatic update, that generally means that wp_insert_post() is
641		- * being called under the user credentials of whoever is viewing the
642		- * blog at the time -- usually meaning no user at all. But if WordPress
643		- * gets a wp_insert_post() when current_user_can('unfiltered_html') is
644		- * false, it will run the content of the post through a kses function
645		- * that strips out lots of HTML tags -- notably <object> and some others.
646		- * This causes problems for syndicating (for example) feeds that contain
647		- * YouTube videos. It also produces an unexpected asymmetry between
648		- * automatically-initiated updates and updates initiated manually from
649		- * the WordPress Dashboard (which are usually initiated under the
650		- * credentials of a logged-in admin, and so don't get run through the
651		- * kses function). So, to avoid the whole mess, what we do here is
652		- * just forcibly disable the kses munging for a single syndicated post,
653		- * by restoring the contents of the `post_content` field.
	1760	+ * SyndicatedPost::add_terms() -- if FeedWordPress is processing an
	1761	+ * automatic update, that generally means that wp_insert_post() is being
	1762	+ * called under the user credentials of whoever is viewing the blog at
	1763	+ * the time -- which usually means no user at all. But wp_insert_post()
	1764	+ * checks current_user_can() before assigning any of the terms in a
	1765	+ * post's tax_input structure -- which is unfortunate, since
	1766	+ * current_user_can() always returns FALSE when there is no current user
	1767	+ * logged in. Meaning that automatic updates get no terms assigned.
654	1768	*
655		- * @param string $content The content of the post, after other filters have gotten to it
656		- * @return string The original content of the post, before other filters had a chance to munge it.
	1769	+ * So, wp_insert_post() is not going to do the term assignments for us.
	1770	+ * If you want something done right....
	1771	+ *
	1772	+ * @param string $new_status Unused action parameter.
	1773	+ * @param string $old_status Unused action parameter.
	1774	+ * @param object $post The database record for the post just inserted.
657	1775	*/
658		- function avoid_kses_munge ($content) {
659		- global $wpdb;
660		- return $wpdb->escape($this->post['post_content']);
661		- }
662		-
663		- // SyndicatedPost::add_rss_meta: adds interesting meta-data to each entry
664		- // using the space for custom keys. The set of keys and values to add is
665		- // specified by the keys and values of $post['meta']. This is used to
666		- // store anything that the WordPress user might want to access from a
667		- // template concerning the post's original source that isn't provided
668		- // for by standard WP meta-data (i.e., any interesting data about the
669		- // syndicated post other than author, title, timestamp, categories, and
670		- // guid). It's also used to hook into WordPress's support for
671		- // enclosures.
672		- function add_rss_meta () {
	1776	+ function add_terms ($new_status, $old_status, $post) {
	1777	+ if ($new_status!='inherit') : // Bail if we are creating a revision.
	1778	+ if ( is_array($this->post) and isset($this->post['tax_input']) and is_array($this->post['tax_input']) ) :
	1779	+ foreach ($this->post['tax_input'] as $taxonomy => $terms) :
	1780	+ if (is_array($terms)) :
	1781	+ $terms = array_filter($terms); // strip out empties
	1782	+ endif;
	1783	+ wp_set_post_terms($post->ID, $terms, $taxonomy);
	1784	+ endforeach;
	1785	+ endif;
	1786	+ endif;
	1787	+ } /* SyndicatedPost::add_terms () */
	1788	+
	1789	+ /**
	1790	+ * SyndicatedPost::fix_post_modified_ts() -- We would like to set
	1791	+ * post_modified and post_modified_gmt to reflect the value of
	1792	+ * <atom:updated> or equivalent elements on the feed. Unfortunately,
	1793	+ * wp_insert_post() refuses to acknowledge explicitly-set post_modified
	1794	+ * fields and overwrites them, either with the post_date (if new) or the
	1795	+ * current timestamp (if updated).
	1796	+ *
	1797	+ * So, wp_insert_post() is not going to do the last-modified assignments
	1798	+ * for us. If you want something done right....
	1799	+ *
	1800	+ * @param string $new_status Unused action parameter.
	1801	+ * @param string $old_status Unused action parameter.
	1802	+ * @param object $post The database record for the post just inserted.
	1803	+ */
	1804	+ function fix_post_modified_ts ($new_status, $old_status, $post) {
673	1805	global $wpdb;
674		- if ( is_array($this->post) and isset($this->post['meta']) and is_array($this->post['meta']) ) :
675		- $postId = $this->wp_id();
676		-
677		- // Aggregated posts should NOT send out pingbacks.
678		- // WordPress 2.1-2.2 claim you can tell them not to
679		- // using $post_pingback, but they don't listen, so we
680		- // make sure here.
681		- $result = $wpdb->query("
682		- DELETE FROM $wpdb->postmeta
683		- WHERE post_id='$postId' AND meta_key='_pingme'
684		- ");
	1806	+ if ($new_status!='inherit') : // Bail if we are creating a revision.
	1807	+ $wpdb->update( $wpdb->posts, /*data=*/ array(
	1808	+ 'post_modified' => $this->post['post_modified'],
	1809	+ 'post_modified_gmt' => $this->post['post_modified_gmt'],
	1810	+ ), /*where=*/ array('ID' => $post->ID) );
	1811	+ endif;
	1812	+ } /* SyndicatedPost::fix_post_modified_ts () */
	1813	+
	1814	+ /**
	1815	+ * SyndicatedPost::add_rss_meta: adds interesting meta-data to each entry
	1816	+ * using the space for custom keys. The set of keys and values to add is
	1817	+ * specified by the keys and values of $post['meta']. This is used to
	1818	+ * store anything that the WordPress user might want to access from a
	1819	+ * template concerning the post's original source that isn't provided
	1820	+ * for by standard WP meta-data (i.e., any interesting data about the
	1821	+ * syndicated post other than author, title, timestamp, categories, and
	1822	+ * guid). It's also used to hook into WordPress's support for
	1823	+ * enclosures.
	1824	+ *
	1825	+ * @param string $new_status Unused action parameter.
	1826	+ * @param string $old_status Unused action parameter.
	1827	+ * @param object $post The database record for the post just inserted.
	1828	+ */
	1829	+ function add_rss_meta ($new_status, $old_status, $post) {
	1830	+ global $wpdb;
	1831	+ if ($new_status!='inherit') : // Bail if we are creating a revision.
	1832	+ FeedWordPress::diagnostic('syndicated_posts:meta_data', 'Adding post meta-data: {'.implode(", ", array_keys($this->post['meta'])).'}');
685	1833
686		- foreach ( $this->post['meta'] as $key => $values ) :
687		-
688		- $key = $wpdb->escape($key);
689		-
690		- // If this is an update, clear out the old
691		- // values to avoid duplication.
	1834	+ if ( is_array($this->post) and isset($this->post['meta']) and is_array($this->post['meta']) ) :
	1835	+ $postId = $post->ID;
	1836	+
	1837	+ // Aggregated posts should NOT send out pingbacks.
	1838	+ // WordPress 2.1-2.2 claim you can tell them not to
	1839	+ // using $post_pingback, but they don't listen, so we
	1840	+ // make sure here.
692	1841	$result = $wpdb->query("
693	1842	DELETE FROM $wpdb->postmeta
694		- WHERE post_id='$postId' AND meta_key='$key'
	1843	+ WHERE post_id='$postId' AND meta_key='_pingme'
695	1844	");
696		-
697		- // Allow for either a single value or an array
698		- if (!is_array($values)) $values = array($values);
699		- foreach ( $values as $value ) :
700		- $value = $wpdb->escape($value);
	1845	+
	1846	+ foreach ( $this->post['meta'] as $key => $values ) :
	1847	+ $eKey = $wpdb->escape($key);
	1848	+
	1849	+ // If this is an update, clear out the old
	1850	+ // values to avoid duplication.
701	1851	$result = $wpdb->query("
702		- INSERT INTO $wpdb->postmeta
703		- SET
704		- post_id='$postId',
705		- meta_key='$key',
706		- meta_value='$value'
	1852	+ DELETE FROM $wpdb->postmeta
	1853	+ WHERE post_id='$postId' AND meta_key='$eKey'
707	1854	");
708		- if (!$result) :
709		- $err = mysql_error();
710		- if (FEEDWORDPRESS_DEBUG) :
711		- echo "[DEBUG:".date('Y-m-d H:i:S')."][feedwordpress]: post metadata insertion FAILED for field '$key' := '$value': [$err]";
712		- endif;
713		- endif;
	1855	+
	1856	+ // Allow for either a single value or an array
	1857	+ if (!is_array($values)) $values = array($values);
	1858	+ foreach ( $values as $value ) :
	1859	+ FeedWordPress::diagnostic('syndicated_posts:meta_data', "Adding post meta-datum to post [$postId]: [$key] = ".FeedWordPress::val($value, /*no newlines=*/ true));
	1860	+ add_post_meta($postId, $key, $value, /*unique=*/ false);
	1861	+ endforeach;
714	1862	endforeach;
715		- endforeach;
	1863	+ endif;
716	1864	endif;
717	1865	} /* SyndicatedPost::add_rss_meta () */
718	1866
719		- // SyndicatedPost::author_id (): get the ID for an author name from
720		- // the feed. Create the author if necessary.
	1867	+ /**
	1868	+ * SyndicatedPost::author_id (): get the ID for an author name from
	1869	+ * the feed. Create the author if necessary.
	1870	+ *
	1871	+ * @param string $unfamiliar_author
	1872	+ *
	1873	+ * @return NULL|int The numeric ID of the author to attribute the post to
	1874	+ * NULL if the post should be filtered out.
	1875	+ */
721	1876	function author_id ($unfamiliar_author = 'create') {
722	1877	global $wpdb;
723	1878
724		- $a = $this->author();
725		- $author = $a['name'];
726		- $email = $a['email'];
727		- $url = $a['uri'];
	1879	+ $a = $this->named['author'];
728	1880
	1881	+ $source = $this->source();
	1882	+ $forbidden = apply_filters('feedwordpress_forbidden_author_names',
	1883	+ array('admin', 'administrator', 'www', 'root'));
	1884	+
	1885	+ // Prepare the list of candidates to try for author name: name from
	1886	+ // feed, original source title (if any), immediate source title live
	1887	+ // from feed, subscription title, prettied version of feed homepage URL,
	1888	+ // prettied version of feed URL, or, failing all, use "unknown author"
	1889	+ // as last resort
	1890	+
	1891	+ $candidates = array();
	1892	+ $candidates[] = $a['name'];
	1893	+ if (!is_null($source)) : $candidates[] = $source['title']; endif;
	1894	+ $candidates[] = $this->link->name(/*fromFeed=*/ true);
	1895	+ $candidates[] = $this->link->name(/*fromFeed=*/ false);
	1896	+ if (strlen($this->link->homepage()) > 0) : $candidates[] = feedwordpress_display_url($this->link->homepage()); endif;
	1897	+ $candidates[] = feedwordpress_display_url($this->link->uri());
	1898	+ $candidates[] = 'unknown author';
	1899	+
	1900	+ // Pick the first one that works from the list, screening against empty
	1901	+ // or forbidden names.
	1902	+
	1903	+ $author = NULL;
	1904	+ while (is_null($author) and ($candidate = each($candidates))) :
	1905	+ if (!is_null($candidate['value'])
	1906	+ and (strlen(trim($candidate['value'])) > 0)
	1907	+ and !in_array(strtolower(trim($candidate['value'])), $forbidden)) :
	1908	+ $author = $candidate['value'];
	1909	+ endif;
	1910	+ endwhile;
	1911	+
	1912	+ $email = (isset($a['email']) ? $a['email'] : NULL);
	1913	+ $authorUrl = (isset($a['uri']) ? $a['uri'] : NULL);
	1914	+
	1915	+
	1916	+ $hostUrl = $this->link->homepage();
	1917	+ if (is_null($hostUrl) or (strlen($hostUrl) < 0)) :
	1918	+ $hostUrl = $this->link->uri();
	1919	+ endif;
	1920	+
729	1921	$match_author_by_email = !('yes' == get_option("feedwordpress_do_not_match_author_by_email"));
730	1922	if ($match_author_by_email and !FeedWordPress::is_null_email($email)) :
731	1923	$test_email = $email;
732	1924	else :
		@@ -734,8 +1926,28 @@
734	1926	endif;
735	1927
736	1928	// Never can be too careful...
737	1929	$login = sanitize_user($author, /*strict=*/ true);
	1930	+
	1931	+ // Possible for, e.g., foreign script author names
	1932	+ if (strlen($login) < 1) :
	1933	+ // No usable characters in author name for a login.
	1934	+ // (Sometimes results from, e.g., foreign scripts.)
	1935	+ //
	1936	+ // We just need *something* in Western alphanumerics,
	1937	+ // so let's try the domain name.
	1938	+ //
	1939	+ // Uniqueness will be guaranteed below if necessary.
	1940	+
	1941	+ $url = parse_url($hostUrl);
	1942	+
	1943	+ $login = sanitize_user($url['host'], /*strict=*/ true);
	1944	+ if (strlen($login) < 1) :
	1945	+ // This isn't working. Frak it.
	1946	+ $login = 'syndicated';
	1947	+ endif;
	1948	+ endif;
	1949	+
738	1950	$login = apply_filters('pre_user_login', $login);
739	1951
740	1952	$nice_author = sanitize_title($author);
741	1953	$nice_author = apply_filters('pre_user_nicename', $nice_author);
		@@ -743,12 +1955,14 @@
743	1955	$reg_author = $wpdb->escape(preg_quote($author));
744	1956	$author = $wpdb->escape($author);
745	1957	$email = $wpdb->escape($email);
746	1958	$test_email = $wpdb->escape($test_email);
747		- $url = $wpdb->escape($url);
	1959	+ $authorUrl = $wpdb->escape($authorUrl);
748	1960
749	1961	// Check for an existing author rule....
750		- if (isset($this->link->settings['map authors']['name'][strtolower(trim($author))])) :
	1962	+ if (isset($this->link->settings['map authors']['name']['*'])) :
	1963	+ $author_rule = $this->link->settings['map authors']['name']['*'];
	1964	+ elseif (isset($this->link->settings['map authors']['name'][strtolower(trim($author))])) :
751	1965	$author_rule = $this->link->settings['map authors']['name'][strtolower(trim($author))];
752	1966	else :
753	1967	$author_rule = NULL;
754	1968	endif;
		@@ -763,65 +1977,34 @@
763	1977
764	1978	else :
765	1979	// Check the database for an existing author record that might fit
766	1980
767		- #-- WordPress 2.0+
768		- if (fwp_test_wp_version(FWP_SCHEMA_HAS_USERMETA)) :
769		-
770		- // First try the user core data table.
	1981	+ // First try the user core data table.
	1982	+ $id = $wpdb->get_var(
	1983	+ "SELECT ID FROM $wpdb->users
	1984	+ WHERE TRIM(LCASE(display_name)) = TRIM(LCASE('$author'))
	1985	+ OR TRIM(LCASE(user_login)) = TRIM(LCASE('$author'))
	1986	+ OR (
	1987	+ LENGTH(TRIM(LCASE(user_email))) > 0
	1988	+ AND TRIM(LCASE(user_email)) = TRIM(LCASE('$test_email'))
	1989	+ )");
	1990	+
	1991	+ // If that fails, look for aliases in the user meta data table
	1992	+ if (is_null($id)) :
771	1993	$id = $wpdb->get_var(
772		- "SELECT ID FROM $wpdb->users
	1994	+ "SELECT user_id FROM $wpdb->usermeta
773	1995	WHERE
774		- TRIM(LCASE(user_login)) = TRIM(LCASE('$login'))
	1996	+ (meta_key = 'description' AND TRIM(LCASE(meta_value)) = TRIM(LCASE('$author')))
775	1997	OR (
776		- LENGTH(TRIM(LCASE(user_email))) > 0
777		- AND TRIM(LCASE(user_email)) = TRIM(LCASE('$test_email'))
778		- )
779		- OR TRIM(LCASE(user_nicename)) = TRIM(LCASE('$nice_author'))
780		- ");
781		-
782		- // If that fails, look for aliases in the user meta data table
783		- if (is_null($id)) :
784		- $id = $wpdb->get_var(
785		- "SELECT user_id FROM $wpdb->usermeta
786		- WHERE
787		- (meta_key = 'description' AND TRIM(LCASE(meta_value)) = TRIM(LCASE('$author')))
788		- OR (
789		- meta_key = 'description'
790		- AND TRIM(LCASE(meta_value))
791		- RLIKE CONCAT(
792		- '(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*',
793		- TRIM(LCASE('$reg_author')),
794		- '( |\\t|\\r)*(\\n|\$)'
795		- )
796		- )
797		- ");
798		- endif;
799		-
800		- #-- WordPress 1.5.x
801		- else :
802		- $id = $wpdb->get_var(
803		- "SELECT ID from $wpdb->users
804		- WHERE
805		- TRIM(LCASE(user_login)) = TRIM(LCASE('$login')) OR
806		- (
807		- LENGTH(TRIM(LCASE(user_email))) > 0
808		- AND TRIM(LCASE(user_email)) = TRIM(LCASE('$test_email'))
809		- ) OR
810		- TRIM(LCASE(user_firstname)) = TRIM(LCASE('$author')) OR
811		- TRIM(LCASE(user_nickname)) = TRIM(LCASE('$author')) OR
812		- TRIM(LCASE(user_nicename)) = TRIM(LCASE('$nice_author')) OR
813		- TRIM(LCASE(user_description)) = TRIM(LCASE('$author')) OR
814		- (
815		- LOWER(user_description)
	1998	+ meta_key = 'description'
	1999	+ AND TRIM(LCASE(meta_value))
816	2000	RLIKE CONCAT(
817	2001	'(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*',
818		- LCASE('$reg_author'),
	2002	+ TRIM(LCASE('$reg_author')),
819	2003	'( |\\t|\\r)*(\\n|\$)'
820	2004	)
821	2005	)
822	2006	");
823		-
824	2007	endif;
825	2008
826	2009	// ... if you don't find one, then do what you need to do
827	2010	if (is_null($id)) :
		@@ -827,8 +2010,16 @@
827	2010	if (is_null($id)) :
828	2011	if ($unfamiliar_author === 'create') :
829	2012	$userdata = array();
830	2013
	2014	+ // WordPress 3 is going to pitch a fit if we attempt to register
	2015	+ // more than one user account with an empty e-mail address, so we
	2016	+ // need *something* here. Ugh.
	2017	+ if (strlen($email) == 0 or FeedWordPress::is_null_email($email)) :
	2018	+ $url = parse_url($hostUrl);
	2019	+ $email = $nice_author.'@'.$url['host'];
	2020	+ endif;
	2021	+
831	2022	#-- user table data
832	2023	$userdata['ID'] = NULL; // new user
833	2024	$userdata['user_login'] = $login;
834	2025	$userdata['user_nicename'] = $nice_author;
		@@ -833,12 +2024,39 @@
833	2024	$userdata['user_login'] = $login;
834	2025	$userdata['user_nicename'] = $nice_author;
835	2026	$userdata['user_pass'] = substr(md5(uniqid(microtime())), 0, 6); // just something random to lock it up
836	2027	$userdata['user_email'] = $email;
837		- $userdata['user_url'] = $url;
	2028	+ $userdata['user_url'] = $authorUrl;
	2029	+ $userdata['nickname'] = $author;
	2030	+ $parts = preg_split('/\s+/', trim($author), 2);
	2031	+ if (isset($parts[0])) : $userdata['first_name'] = $parts[0]; endif;
	2032	+ if (isset($parts[1])) : $userdata['last_name'] = $parts[1]; endif;
838	2033	$userdata['display_name'] = $author;
839		-
840		- $id = wp_insert_user($userdata);
	2034	+ $userdata['role'] = 'contributor';
	2035	+
	2036	+ do { // Keep trying until you get it right. Or until PHP crashes, I guess.
	2037	+ $id = wp_insert_user($userdata);
	2038	+ if (is_wp_error($id)) :
	2039	+ $codes = $id->get_error_code();
	2040	+ switch ($codes) :
	2041	+ case 'empty_user_login' :
	2042	+ case 'existing_user_login' :
	2043	+ // Add a random disambiguator
	2044	+ $userdata['user_login'] .= substr(md5(uniqid(microtime())), 0, 6);
	2045	+ break;
	2046	+ case 'existing_user_email' :
	2047	+ // No disassemble!
	2048	+ $parts = explode('@', $userdata['user_email'], 2);
	2049	+
	2050	+ // Add a random disambiguator as a gmail-style username extension
	2051	+ $parts[0] .= '+'.substr(md5(uniqid(microtime())), 0, 6);
	2052	+
	2053	+ // Reassemble
	2054	+ $userdata['user_email'] = $parts[0].'@'.$parts[1];
	2055	+ break;
	2056	+ endswitch;
	2057	+ endif;
	2058	+ } while (is_wp_error($id));
841	2059	elseif (is_numeric($unfamiliar_author) and get_userdata((int) $unfamiliar_author)) :
842	2060	$id = (int) $unfamiliar_author;
843	2061	elseif ($unfamiliar_author === 'default') :
844	2062	$id = 1;
		@@ -847,16 +2065,38 @@
847	2065	endif;
848	2066
849	2067	if ($id) :
850	2068	$this->link->settings['map authors']['name'][strtolower(trim($author))] = $id;
	2069	+
	2070	+ // Multisite: Check whether the author has been recorded
	2071	+ // on *this* blog before. If not, put her down as a
	2072	+ // Contributor for *this* blog.
	2073	+ $user = new WP_User((int) $id);
	2074	+ if (empty($user->roles)) :
	2075	+ $user->add_role('contributor');
	2076	+ endif;
851	2077	endif;
852	2078	return $id;
853		- } // function SyndicatedPost::author_id ()
	2079	+ } /* function SyndicatedPost::author_id () */
854	2080
855		- // look up (and create) category ids from a list of categories
856		- function category_ids ($cats, $unfamiliar_category = 'create', $tags_too = false) {
857		- global $wpdb;
	2081	+ /**
	2082	+ * category_ids: look up (and create) category ids from a list of categories
	2083	+ *
	2084	+ * @param array $cats
	2085	+ * @param string $unfamiliar_category
	2086	+ * @param array|null $taxonomies
	2087	+ * @return array
	2088	+ */
	2089	+ function category_ids ($cats, $unfamiliar_category = 'create', $taxonomies = NULL, $params = array()) {
	2090	+ $singleton = (isset($params['singleton']) ? $params['singleton'] : true);
	2091	+ $allowFilters = (isset($params['filters']) ? $params['filters'] : false);
	2092	+
	2093	+ $catTax = 'category';
858	2094
	2095	+ if (is_null($taxonomies)) :
	2096	+ $taxonomies = array('category');
	2097	+ endif;
	2098	+
859	2099	// We need to normalize whitespace because (1) trailing
860	2100	// whitespace can cause PHP and MySQL not to see eye to eye on
861	2101	// VARCHAR comparisons for some versions of MySQL (cf.
862	2102	// <http://dev.mysql.com/doc/mysql/en/char.html>), and (2)
		@@ -863,416 +2103,100 @@
863	2103	// because I doubt most people want to make a semantic
864	2104	// distinction between 'Computers' and 'Computers '
865	2105	$cats = array_map('trim', $cats);
866	2106
867		- $tags = array();
	2107	+ $terms = array();
	2108	+ foreach ($taxonomies as $tax) :
	2109	+ $terms[$tax] = array();
	2110	+ endforeach;
	2111	+
	2112	+ foreach ($cats as $cat_name) :
	2113	+ if (preg_match('/^{([^#}]*)#([0-9]+)}$/', $cat_name, $backref)) :
	2114	+ $cat_id = (int) $backref[2];
	2115	+ $tax = $backref[1];
	2116	+ if (strlen($tax) < 1) :
	2117	+ $tax = $catTax;
	2118	+ endif;
868	2119
869		- $cat_ids = array ();
870		- foreach ($cats as $cat_name) :
871		- if (preg_match('/^{#([0-9]+)}$/', $cat_name, $backref)) :
872		- $cat_id = (int) $backref[1];
873		- if (function_exists('is_term') and is_term($cat_id, 'category')) :
874		- $cat_ids[] = $cat_id;
875		- elseif (get_category($cat_id)) :
876		- $cat_ids[] = $cat_id;
	2120	+ $term = term_exists($cat_id, $tax);
	2121	+ if (!is_wp_error($term) and !!$term) :
	2122	+ if (!isset($terms[$tax])) :
	2123	+ $terms[$tax] = array();
	2124	+ endif;
	2125	+ $terms[$tax][] = $cat_id;
877	2126	endif;
878	2127	elseif (strlen($cat_name) > 0) :
879		- $esc = $wpdb->escape($cat_name);
880		- $resc = $wpdb->escape(preg_quote($cat_name));
881		-
882		- // WordPress 2.3+
883		- if (function_exists('is_term')) :
884		- $cat_id = is_term($cat_name, 'category');
885		- if ($cat_id) :
886		- $cat_ids[] = $cat_id['term_id'];
887		- // There must be a better way to do this...
888		- elseif ($results = $wpdb->get_results(
889		- "SELECT term_id
890		- FROM $wpdb->term_taxonomy
891		- WHERE
892		- LOWER(description) RLIKE
893		- CONCAT('(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*', LOWER('{$resc}'), '( |\\t|\\r)*(\\n|\$)')"
894		- )) :
895		- foreach ($results AS $term) :
896		- $cat_ids[] = (int) $term->term_id;
897		- endforeach;
898		- elseif ('tag'==$unfamiliar_category) :
899		- $tags[] = $cat_name;
900		- elseif ('create'===$unfamiliar_category) :
901		- $term = wp_insert_term($cat_name, 'category');
902		- if (is_wp_error($term)) :
903		- FeedWordPress::noncritical_bug('term insertion problem', array('cat_name' => $cat_name, 'term' => $term, 'this' => $this), __LINE__);
904		- else :
905		- $cat_ids[] = $term['term_id'];
	2128	+ $familiar = false;
	2129	+ foreach ($taxonomies as $tax) :
	2130	+ if ($tax!='category' or strtolower($cat_name)!='uncategorized') :
	2131	+ $term = term_exists($cat_name, $tax);
	2132	+ if (!is_wp_error($term) and !!$term) :
	2133	+ $familiar = true;
	2134	+
	2135	+ if (is_array($term)) :
	2136	+ $term_id = (int) $term['term_id'];
	2137	+ else :
	2138	+ $term_id = (int) $term;
	2139	+ endif;
	2140	+
	2141	+ if (!isset($terms[$tax])) :
	2142	+ $terms[$tax] = array();
	2143	+ endif;
	2144	+ $terms[$tax][] = $term_id;
	2145	+ break; // We're done here.
906	2146	endif;
907	2147	endif;
	2148	+ endforeach;
908	2149
909		- // WordPress 1.5.x - 2.2.x
910		- else :
911		- $results = $wpdb->get_results(
912		- "SELECT cat_ID
913		- FROM $wpdb->categories
914		- WHERE
915		- (LOWER(cat_name) = LOWER('$esc'))
916		- OR (LOWER(category_description)
917		- RLIKE CONCAT('(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*', LOWER('{$resc}'), '( |\\t|\\r)*(\\n|\$)'))
918		- ");
919		- if ($results) :
920		- foreach ($results as $term) :
921		- $cat_ids[] = (int) $term->cat_ID;
922		- endforeach;
923		- elseif ('create'===$unfamiliar_category) :
924		- if (function_exists('wp_insert_category')) :
925		- $cat_id = wp_insert_category(array('cat_name' => $esc));
926		- // And into the database we go.
	2150	+ if (!$familiar) :
	2151	+ if ('tag'==$unfamiliar_category) :
	2152	+ $unfamiliar_category = 'create:post_tag';
	2153	+ endif;
	2154	+
	2155	+ if (preg_match('/^create(:(.*))?$/i', $unfamiliar_category, $ref)) :
	2156	+ $tax = $catTax; // Default
	2157	+ if (isset($ref[2]) and strlen($ref[2]) > 2) :
	2158	+ $tax = $ref[2];
	2159	+ endif;
	2160	+ $term = wp_insert_term($cat_name, $tax);
	2161	+ if (is_wp_error($term)) :
	2162	+ FeedWordPress::noncritical_bug('term insertion problem', array('cat_name' => $cat_name, 'term' => $term, 'this' => $this), __LINE__, __FILE__);
927	2163	else :
928		- $nice_kitty = sanitize_title($cat_name);
929		- $wpdb->query(sprintf("
930		- INSERT INTO $wpdb->categories
931		- SET
932		- cat_name='%s',
933		- category_nicename='%s'
934		- ", $esc, $nice_kitty
935		- ));
936		- $cat_id = $wpdb->insert_id;
	2164	+ if (!isset($terms[$tax])) :
	2165	+ $terms[$tax] = array();
	2166	+ endif;
	2167	+ $terms[$tax][] = (int) $term['term_id'];
937	2168	endif;
938		- $cat_ids[] = $cat_id;
939	2169	endif;
940	2170	endif;
941	2171	endif;
942	2172	endforeach;
943	2173
944		- if ((count($cat_ids) == 0) and ($unfamiliar_category === 'filter')) :
945		- $cat_ids = NULL; // Drop the post
946		- else :
947		- $cat_ids = array_unique($cat_ids);
	2174	+ $filtersOn = $allowFilters;
	2175	+ if ($allowFilters) :
	2176	+ $filters = array_filter(
	2177	+ $this->link->setting('match/filter', 'match_filter', array()),
	2178	+ 'remove_dummy_zero'
	2179	+ );
	2180	+ $filtersOn = ($filtersOn and is_array($filters) and (count($filters) > 0));
948	2181	endif;
949	2182
950		- if ($tags_too) : $ret = array($cat_ids, $tags);
951		- else : $ret = $cat_ids;
952		- endif;
953		-
954		- return $ret;
955		- } // function SyndicatedPost::category_ids ()
956		-
957		- function use_api ($tag) {
958		- global $wp_db_version;
959		- switch ($tag) :
960		- case 'wp_insert_post':
961		- // Before 2.2, wp_insert_post does too much of the wrong stuff to use it
962		- // In 1.5 it was such a resource hog it would make PHP segfault on big updates
963		- $ret = (isset($wp_db_version) and $wp_db_version > FWP_SCHEMA_21);
964		- break;
965		- case 'post_status_pending':
966		- $ret = (isset($wp_db_version) and $wp_db_version > FWP_SCHEMA_23);
967		- break;
968		- endswitch;
969		- return $ret;
970		- } // function SyndicatedPost::use_api ()
971		-
972		- #### EXTRACT DATA FROM FEED ITEM ####
973		-
974		- function created () {
975		- $epoch = null;
976		- if (isset($this->item['dc']['created'])) :
977		- $epoch = @parse_w3cdtf($this->item['dc']['created']);
978		- elseif (isset($this->item['dcterms']['created'])) :
979		- $epoch = @parse_w3cdtf($this->item['dcterms']['created']);
980		- elseif (isset($this->item['created'])): // Atom 0.3
981		- $epoch = @parse_w3cdtf($this->item['created']);
982		- endif;
983		- return $epoch;
984		- }
985		- function published ($fallback = true) {
986		- $epoch = null;
987		-
988		- # RSS is a fucking mess. Figure out whether we have a date in
989		- # <dc:date>, <issued>, <pubDate>, etc., and get it into Unix
990		- # epoch format for reformatting. If we can't find anything,
991		- # we'll use the last-updated time.
992		- if (isset($this->item['dc']['date'])): // Dublin Core
993		- $epoch = @parse_w3cdtf($this->item['dc']['date']);
994		- elseif (isset($this->item['dcterms']['issued'])) : // Dublin Core extensions
995		- $epoch = @parse_w3cdtf($this->item['dcterms']['issued']);
996		- elseif (isset($this->item['published'])) : // Atom 1.0
997		- $epoch = @parse_w3cdtf($this->item['published']);
998		- elseif (isset($this->item['issued'])): // Atom 0.3
999		- $epoch = @parse_w3cdtf($this->item['issued']);
1000		- elseif (isset($this->item['pubdate'])): // RSS 2.0
1001		- $epoch = strtotime($this->item['pubdate']);
1002		- elseif ($fallback) : // Fall back to <updated> / <modified> if present
1003		- $epoch = $this->updated(/*fallback=*/ false);
1004		- endif;
1005		-
1006		- # If everything failed, then default to the current time.
1007		- if (is_null($epoch)) :
1008		- if (-1 == $default) :
1009		- $epoch = time();
	2183	+ // Check for filter conditions
	2184	+ foreach ($terms as $tax => $term_ids) :
	2185	+ if ($filtersOn
	2186	+ and (count($term_ids)==0)
	2187	+ and in_array($tax, $filters)) :
	2188	+ $terms = NULL; // Drop the post
	2189	+ break;
1010	2190	else :
1011		- $epoch = $default;
	2191	+ $terms[$tax] = array_unique($term_ids);
1012	2192	endif;
1013		- endif;
1014		-
1015		- return $epoch;
1016		- }
1017		- function updated ($fallback = true, $default = -1) {
1018		- $epoch = null;
	2193	+ endforeach;
1019	2194
1020		- # As far as I know, only dcterms and Atom have reliable ways to
1021		- # specify when something was *modified* last. If neither is
1022		- # available, then we'll try to get the time of publication.
1023		- if (isset($this->item['dc']['modified'])) : // Not really correct
1024		- $epoch = @parse_w3cdtf($this->item['dc']['modified']);
1025		- elseif (isset($this->item['dcterms']['modified'])) : // Dublin Core extensions
1026		- $epoch = @parse_w3cdtf($this->item['dcterms']['modified']);
1027		- elseif (isset($this->item['modified'])): // Atom 0.3
1028		- $epoch = @parse_w3cdtf($this->item['modified']);
1029		- elseif (isset($this->item['updated'])): // Atom 1.0
1030		- $epoch = @parse_w3cdtf($this->item['updated']);
1031		- elseif ($fallback) : // Fall back to issued / dc:date
1032		- $epoch = $this->published(/*fallback=*/ false, /*default=*/ $default);
	2195	+ if ($singleton and count($terms)==1) : // If we only searched one, just return the term IDs
	2196	+ $terms = end($terms);
1033	2197	endif;
1034		-
1035		- # If everything failed, then default to the current time.
1036		- if (is_null($epoch)) :
1037		- if (-1 == $default) :
1038		- $epoch = time();
1039		- else :
1040		- $epoch = $default;
1041		- endif;
1042		- endif;
	2198	+ return $terms;
	2199	+ } // function SyndicatedPost::category_ids ()
1043	2200
1044		- return $epoch;
1045		- }
1046		-
1047		- function update_hash () {
1048		- return md5(serialize($this->item));
1049		- }
1050		-
1051		- function guid () {
1052		- $guid = null;
1053		- if (isset($this->item['id'])): // Atom 0.3 / 1.0
1054		- $guid = $this->item['id'];
1055		- elseif (isset($this->item['atom']['id'])) : // Namespaced Atom
1056		- $guid = $this->item['atom']['id'];
1057		- elseif (isset($this->item['guid'])) : // RSS 2.0
1058		- $guid = $this->item['guid'];
1059		- elseif (isset($this->item['dc']['identifier'])) :// yeah, right
1060		- $guid = $this->item['dc']['identifier'];
1061		- else :
1062		- // The feed does not seem to have provided us with a
1063		- // unique identifier, so we'll have to cobble together
1064		- // a tag: URI that might work for us. The base of the
1065		- // URI will be the host name of the feed source ...
1066		- $bits = parse_url($this->feedmeta['link/uri']);
1067		- $guid = 'tag:'.$bits['host'];
1068		-
1069		- // If we have a date of creation, then we can use that
1070		- // to uniquely identify the item. (On the other hand, if
1071		- // the feed producer was consicentious enough to
1072		- // generate dates of creation, she probably also was
1073		- // conscientious enough to generate unique identifiers.)
1074		- if (!is_null($this->created())) :
1075		- $guid .= '://post.'.date('YmdHis', $this->created());
1076		-
1077		- // Otherwise, use both the URI of the item, *and* the
1078		- // item's title. We have to use both because titles are
1079		- // often not unique, and sometimes links aren't unique
1080		- // either (e.g. Bitch (S)HITLIST, Mozilla Dot Org news,
1081		- // some podcasts). But it's rare to have *both* the same
1082		- // title *and* the same link for two different items. So
1083		- // this is about the best we can do.
1084		- else :
1085		- $guid .= '://'.md5($this->item['link'].'/'.$this->item['title']);
1086		- endif;
1087		- endif;
1088		- return $guid;
1089		- }
1090		-
1091		- function author () {
1092		- $author = array ();
1093		-
1094		- if (isset($this->item['author_name'])):
1095		- $author['name'] = $this->item['author_name'];
1096		- elseif (isset($this->item['dc']['creator'])):
1097		- $author['name'] = $this->item['dc']['creator'];
1098		- elseif (isset($this->item['dc']['contributor'])):
1099		- $author['name'] = $this->item['dc']['contributor'];
1100		- elseif (isset($this->feed->channel['dc']['creator'])) :
1101		- $author['name'] = $this->feed->channel['dc']['creator'];
1102		- elseif (isset($this->feed->channel['dc']['contributor'])) :
1103		- $author['name'] = $this->feed->channel['dc']['contributor'];
1104		- elseif (isset($this->feed->channel['author_name'])) :
1105		- $author['name'] = $this->feed->channel['author_name'];
1106		- elseif ($this->feed->is_rss() and isset($this->item['author'])) :
1107		- // The author element in RSS is allegedly an
1108		- // e-mail address, but lots of people don't use
1109		- // it that way. So let's make of it what we can.
1110		- $author = parse_email_with_realname($this->item['author']);
1111		-
1112		- if (!isset($author['name'])) :
1113		- if (isset($author['email'])) :
1114		- $author['name'] = $author['email'];
1115		- else :
1116		- $author['name'] = $this->feed->channel['title'];
1117		- endif;
1118		- endif;
1119		- else :
1120		- $author['name'] = $this->feed->channel['title'];
1121		- endif;
1122		-
1123		- if (isset($this->item['author_email'])):
1124		- $author['email'] = $this->item['author_email'];
1125		- elseif (isset($this->feed->channel['author_email'])) :
1126		- $author['email'] = $this->feed->channel['author_email'];
1127		- endif;
1128		-
1129		- if (isset($this->item['author_url'])):
1130		- $author['uri'] = $this->item['author_url'];
1131		- elseif (isset($this->feed->channel['author_url'])) :
1132		- $author['uri'] = $this->item['author_url'];
1133		- else:
1134		- $author['uri'] = $this->feed->channel['link'];
1135		- endif;
1136		-
1137		- return $author;
1138		- } // SyndicatedPost::author()
1139		-
1140		- /**
1141		- * SyndicatedPost::isTaggedAs: Test whether a feed item is
1142		- * tagged / categorized with a given string. Case and leading and
1143		- * trailing whitespace are ignored.
1144		- *
1145		- * @param string $tag Tag to check for
1146		- *
1147		- * @return bool Whether or not at least one of the categories / tags on
1148		- * $this->item is set to $tag (modulo case and leading and trailing
1149		- * whitespace)
1150		- */
1151		- function isTaggedAs ($tag) {
1152		- $desiredTag = strtolower(trim($tag)); // Normalize case and whitespace
1153		-
1154		- // Check to see if this is tagged with $tag
1155		- $currentCategory = 'category';
1156		- $currentCategoryNumber = 1;
1157		-
1158		- // If we have the new MagpieRSS, the number of category elements
1159		- // on this item is stored under index "category#".
1160		- if (isset($this->item['category#'])) :
1161		- $numberOfCategories = (int) $this->item['category#'];
1162		-
1163		- // We REALLY shouldn't have the old and busted MagpieRSS, but in
1164		- // case we do, it doesn't support multiple categories, but there
1165		- // might still be a single value under the "category" index.
1166		- elseif (isset($this->item['category'])) :
1167		- $numberOfCategories = 1;
1168		-
1169		- // No standard category or tag elements on this feed item.
1170		- else :
1171		- $numberOfCategories = 0;
1172		-
1173		- endif;
1174		-
1175		- $isSoTagged = false; // Innocent until proven guilty
1176		-
1177		- // Loop through category elements; if there are multiple
1178		- // elements, they are indexed as category, category#2,
1179		- // category#3, ... category#N
1180		- while ($currentCategoryNumber <= $numberOfCategories) :
1181		- if ($desiredTag == strtolower(trim($this->item[$currentCategory]))) :
1182		- $isSoTagged = true; // Got it!
1183		- break;
1184		- endif;
1185		-
1186		- $currentCategoryNumber += 1;
1187		- $currentCategory = 'category#'.$currentCategoryNumber;
1188		- endwhile;
1189		-
1190		- return $isSoTagged;
1191		- } /* SyndicatedPost::isTaggedAs() */
1192		-
1193		- var $uri_attrs = array (
1194		- array('a', 'href'),
1195		- array('applet', 'codebase'),
1196		- array('area', 'href'),
1197		- array('blockquote', 'cite'),
1198		- array('body', 'background'),
1199		- array('del', 'cite'),
1200		- array('form', 'action'),
1201		- array('frame', 'longdesc'),
1202		- array('frame', 'src'),
1203		- array('iframe', 'longdesc'),
1204		- array('iframe', 'src'),
1205		- array('head', 'profile'),
1206		- array('img', 'longdesc'),
1207		- array('img', 'src'),
1208		- array('img', 'usemap'),
1209		- array('input', 'src'),
1210		- array('input', 'usemap'),
1211		- array('ins', 'cite'),
1212		- array('link', 'href'),
1213		- array('object', 'classid'),
1214		- array('object', 'codebase'),
1215		- array('object', 'data'),
1216		- array('object', 'usemap'),
1217		- array('q', 'cite'),
1218		- array('script', 'src')
1219		- ); /* var SyndicatedPost::$uri_attrs */
1220		-
1221		- var $_base = null;
1222		-
1223		- function resolve_single_relative_uri ($refs) {
1224		- $tag = FeedWordPressHTML::attributeMatch($refs);
1225		- $url = Relative_URI::resolve($tag['value'], $this->_base);
1226		- return $tag['prefix'] . $url . $tag['suffix'];
1227		- } /* function SyndicatedPost::resolve_single_relative_uri() */
1228		-
1229		- function resolve_relative_uris ($content, $obj) {
1230		- # The MagpieRSS upgrade has some `xml:base` support baked in.
1231		- # However, sometimes people do silly things, like putting
1232		- # relative URIs out on a production RSS 2.0 feed or other feeds
1233		- # with no good support for `xml:base`. So we'll do our best to
1234		- # try to catch any remaining relative URIs and resolve them as
1235		- # best we can.
1236		- $obj->_base = $obj->item['link']; // Reset the base for resolving relative URIs
1237		-
1238		- foreach ($obj->uri_attrs as $pair) :
1239		- list($tag, $attr) = $pair;
1240		- $pattern = FeedWordPressHTML::attributeRegex($tag, $attr);
1241		- $content = preg_replace_callback (
1242		- $pattern,
1243		- array(&$obj, 'resolve_single_relative_uri'),
1244		- $content
1245		- );
1246		- endforeach;
1247		-
1248		- return $content;
1249		- } /* function SyndicatedPost::resolve_relative_uris () */
1250		-
1251		- var $strip_attrs = array (
1252		- array('[a-z]+', 'target'),
1253		-// array('[a-z]+', 'style'),
1254		-// array('[a-z]+', 'on[a-z]+'),
1255		- );
1256		-
1257		- function strip_attribute_from_tag ($refs) {
1258		- $tag = FeedWordPressHTML::attributeMatch($refs);
1259		- return $tag['before_attribute'].$tag['after_attribute'];
1260		- }
1261		-
1262		- function sanitize_content ($content, $obj) {
1263		- # This kind of sucks. I intend to replace it with
1264		- # lib_filter sometime soon.
1265		- foreach ($obj->strip_attrs as $pair):
1266		- list($tag,$attr) = $pair;
1267		- $pattern = FeedWordPressHTML::attributeRegex($tag, $attr);
1268		-
1269		- $content = preg_replace_callback (
1270		- $pattern,
1271		- array(&$obj, 'strip_attribute_from_tag'),
1272		- $content
1273		- );
1274		- endforeach;
1275		- return $content;
1276		- }
1277		-} // class SyndicatedPost
	2201	+} /* class SyndicatedPost */
1278	2202