← All changes | syndicatedpost.class.php +2029 -974 2009.1111 → 2017.1004 View file →

		@@ -1,103 +1,150 @@
1	1	<?php
	2	+require_once(dirname(__FILE__).'/feedtime.class.php');
	3	+require_once(dirname(__FILE__).'/syndicatedpostterm.class.php');
	4	+require_once(dirname(__FILE__).'/syndicatedpostxpathquery.class.php');
	5	+
	6	+/**
	7	+ * class SyndicatedPost: FeedWordPress uses to manage the conversion of
	8	+ * incoming items from the feed parser into posts for the WordPress
	9	+ * database. It contains several internal management methods primarily
	10	+ * of interest to someone working on the FeedWordPress source, as well
	11	+ * as some utility methods for extracting useful data from many
	12	+ * different feed formats, which may be useful to FeedWordPress users
	13	+ * who make use of feed data in PHP add-ons and filters.
	14	+ *
	15	+ * @version 2017.1004
	16	+ */
2	17	class SyndicatedPost {
3		- var $item = null;
4		-
	18	+ var $item = null; // MagpieRSS representation
	19	+ var $entry = null; // SimplePie_Item representation
	20	+
5	21	var $link = null;
6	22	var $feed = null;
7	23	var $feedmeta = null;
8		-
	24	+
	25	+ var $xmlns = array ();
	26	+
9	27	var $post = array ();
10	28
	29	+ var $named = array ();
	30	+ var $preset_terms = array ();
	31	+ var $feed_terms = array ();
	32	+
11	33	var $_freshness = null;
12	34	var $_wp_id = null;
	35	+ var $_wp_post = null;
13	36
14		- function SyndicatedPost ($item, $link) {
	37	+ /**
	38	+ * SyndicatedPost constructor: Given a feed item and the source from
	39	+ * which it was taken, prepare a post that can be inserted into the
	40	+ * WordPress database on request, or updated in place if it has already
	41	+ * been syndicated.
	42	+ *
	43	+ * @param array $item The item syndicated from the feed.
	44	+ * @param SyndicatedLink $source The feed it was syndicated from.
	45	+ */
	46	+ function __construct ($item, &$source) {
15	47	global $wpdb;
16	48
17		- $this->link = $link;
18		- $feedmeta = $link->settings;
19		- $feed = $link->magpie;
	49	+ if ( empty($item) && empty($source) )
	50	+ return;
20	51
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.
26		- #
27		- # Cf.: <http://mosquito.wordpress.org/view.php?id=901>
28		- global $fwp_channel, $fwp_feedmeta;
29		- $fwp_channel = $feed; $fwp_feedmeta = $feedmeta;
	52	+ if (is_array($item)
	53	+ and isset($item['simplepie'])
	54	+ and isset($item['magpie'])) :
	55	+ $this->entry = $item['simplepie'];
	56	+ $this->item = $item['magpie'];
	57	+ $item = $item['magpie'];
	58	+ elseif (is_a($item, 'SimplePie_Item')) :
	59	+ $this->entry = $item;
30	60
31		- $this->feed = $feed;
32		- $this->feedmeta = $feedmeta;
	61	+ // convert to Magpie for compat purposes
	62	+ $mp = new MagpieFromSimplePie($source->simplepie, $this->entry);
	63	+ $this->item = $mp->get_item();
33	64
34		- $this->item = $item;
35		- $this->item = apply_filters('syndicated_item', $this->item, $this);
	65	+ // done with conversion object
	66	+ $mp = NULL; unset($mp);
	67	+ else :
	68	+ $this->item = $item;
	69	+ endif;
36	70
	71	+ $this->link =& $source;
	72	+ $this->feed = $source->magpie;
	73	+ $this->feedmeta = $source->settings;
	74	+
	75	+ FeedWordPress::diagnostic('feed_items', 'Considering item ['.$this->guid().'] "'.$this->entry->get_title().'"');
	76	+
	77	+ # Dealing with namespaces can get so fucking fucked.
	78	+ $this->xmlns['forward'] = $source->magpie->_XMLNS_FAMILIAR;
	79	+ $this->xmlns['reverse'] = array();
	80	+ foreach ($this->xmlns['forward'] as $url => $ns) :
	81	+ if (!isset($this->xmlns['reverse'][$ns])) :
	82	+ $this->xmlns['reverse'][$ns] = array();
	83	+ endif;
	84	+ $this->xmlns['reverse'][$ns][] = $url;
	85	+ endforeach;
	86	+
	87	+ // Fucking SimplePie.
	88	+ $this->xmlns['reverse']['rss'][] = '';
	89	+
	90	+ // Trigger global syndicated_item filter.
	91	+ $changed = apply_filters('syndicated_item', $this->item, $this);
	92	+ $this->item = $changed;
	93	+
	94	+ // Allow for feed-specific syndicated_item filters.
	95	+ $changed = apply_filters(
	96	+ "syndicated_item_".$source->uri(),
	97	+ $this->item,
	98	+ $this
	99	+ );
	100	+ $this->item = $changed;
	101	+
37	102	# Filters can halt further processing by returning NULL
38	103	if (is_null($this->item)) :
39	104	$this->post = NULL;
40	105	else :
41		- # Note that nothing is run through $wpdb->escape() here.
	106	+
	107	+ # Note that nothing is run through esc_sql() here.
42	108	# That's deliberate. The escaping is done at the point
43	109	# of insertion, not here, to avoid double-escaping and
44	110	# to avoid screwing with syndicated_post filters
45	111
46		- $this->post['post_title'] = apply_filters('syndicated_item_title', $this->item['title'], $this);
	112	+ $this->post['post_title'] = apply_filters(
	113	+ 'syndicated_item_title',
	114	+ $this->entry->get_title(), $this
	115	+ );
47	116
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(), $this);
52	117
53		- # Identify content and sanitize it.
54		- # ---------------------------------
55		- if (isset($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);
	118	+ $this->named['author'] = apply_filters(
	119	+ 'syndicated_item_author',
	120	+ $this->author(), $this
	121	+ );
	122	+ // This just gives us an alphanumeric name for the author.
	123	+ // We look up (or create) the numeric ID for the author
	124	+ // in SyndicatedPost::add().
67	125
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);
	126	+ $this->post['post_content'] = apply_filters(
	127	+ 'syndicated_item_content',
	128	+ $this->content(), $this
	129	+ );
	130	+
	131	+ $excerpt = apply_filters('syndicated_item_excerpt', $this->excerpt(), $this);
79	132
80		- if (!is_null($excerpt)):
	133	+ if (!empty($excerpt)):
81	134	$this->post['post_excerpt'] = $excerpt;
82	135	endif;
83		-
84		- // This is unnecessary if we use wp_insert_post
85		- if (!$this->use_api('wp_insert_post')) :
86		- $this->post['post_name'] = sanitize_title($this->post['post_title']);
87		- endif;
88	136
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	137	// Dealing with timestamps in WordPress is so fucking fucked.
94	138	$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());
	139	+ $post_date_gmt = $this->published(array('default' => -1));
	140	+ $post_modified_gmt = $this->updated(array('default' => -1));
99	141
	142	+ $this->post['post_date_gmt'] = gmdate('Y-m-d H:i:s', $post_date_gmt);
	143	+ $this->post['post_date'] = gmdate('Y-m-d H:i:s', $post_date_gmt + $offset);
	144	+ $this->post['post_modified_gmt'] = gmdate('Y-m-d H:i:s', $post_modified_gmt);
	145	+ $this->post['post_modified'] = gmdate('Y-m-d H:i:s', $post_modified_gmt + $offset);
	146	+
100	147	// Use feed-level preferences or the global default.
101	148	$this->post['post_status'] = $this->link->syndicated_status('post', 'publish');
102	149	$this->post['comment_status'] = $this->link->syndicated_status('comment', 'closed');
103	150	$this->post['ping_status'] = $this->link->syndicated_status('ping', 'closed');
		@@ -104,210 +151,1304 @@
104	151
105	152	// Unique ID (hopefully a unique tag: URI); failing that, the permalink
106	153	$this->post['guid'] = apply_filters('syndicated_item_guid', $this->guid(), $this);
107	154
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);
	155	+ // User-supplied custom settings to apply to each post.
	156	+ // Do first so that FWP-generated custom settings will
	157	+ // overwrite if necessary; thus preventing any munging.
	158	+ $postMetaIn = $this->link->postmeta(array("parsed" => true));
	159	+ $postMetaOut = array();
	160	+
	161	+ foreach ($postMetaIn as $key => $meta) :
	162	+ $postMetaOut[$key] = $meta->do_substitutions($this);
	163	+ endforeach;
	164	+
	165	+ foreach ($postMetaOut as $key => $values) :
	166	+ $this->post['meta'][$key] = array();
	167	+ foreach ($values as $value) :
	168	+ $this->post['meta'][$key][] = apply_filters("syndicated_post_meta_{$key}", $value, $this);
	169	+ endforeach;
	170	+ endforeach;
	171	+
	172	+ // RSS 2.0 / Atom 1.0 enclosure support
	173	+ $enclosures = $this->entry->get_enclosures();
	174	+ if (is_array($enclosures)) : foreach ($enclosures as $enclosure) :
	175	+ $this->post['meta']['enclosure'][] =
	176	+ apply_filters('syndicated_item_enclosure_url', $enclosure->get_link(), $this)."\n".
	177	+ apply_filters('syndicated_item_enclosure_length', $enclosure->get_length(), $this)."\n".
	178	+ apply_filters('syndicated_item_enclosure_type', $enclosure->get_type(), $this);
	179	+ endforeach; endif;
	180	+
	181	+ // In case you want to point back to the blog this was
	182	+ // syndicated from.
	183	+
	184	+ $sourcemeta['syndication_source'] = apply_filters(
	185	+ 'syndicated_item_source_title',
	186	+ $this->link->name(),
	187	+ $this
	188	+ );
	189	+ $sourcemeta['syndication_source_uri'] = apply_filters(
	190	+ 'syndicated_item_source_link',
	191	+ $this->link->homepage(),
	192	+ $this
	193	+ );
	194	+ $sourcemeta['syndication_source_id'] = apply_filters(
	195	+ 'syndicated_item_source_id',
	196	+ $this->link->guid(),
	197	+ $this
	198	+ );
	199	+
	200	+ // Make use of atom:source data, if present in an aggregated feed
	201	+ $entry_source = $this->source();
	202	+ if (!is_null($entry_source)) :
	203	+ foreach ($entry_source as $what => $value) :
	204	+ if (!is_null($value)) :
	205	+ if ($what=='title') : $key = 'syndication_source';
	206	+ elseif ($what=='feed') : $key = 'syndication_feed';
	207	+ else : $key = "syndication_source_${what}";
	208	+ endif;
	209	+
	210	+ $sourcemeta["${key}_original"] = apply_filters(
	211	+ 'syndicated_item_original_source_'.$what,
	212	+ $value,
	213	+ $this
	214	+ );
	215	+ endif;
	216	+ endforeach;
112	217	endif;
113		- if (!is_array($default_custom_settings)) :
114		- $default_custom_settings = array();
	218	+
	219	+ foreach ($sourcemeta as $meta_key => $value) :
	220	+ if (!is_null($value)) :
	221	+ $this->post['meta'][$meta_key] = $value;
	222	+ endif;
	223	+ endforeach;
	224	+
	225	+ // Store information on human-readable and machine-readable comment URIs
	226	+
	227	+ // Human-readable comment URI
	228	+ $commentLink = apply_filters('syndicated_item_comments', $this->comment_link(), $this);
	229	+ if (!is_null($commentLink)) : $this->post['meta']['rss:comments'] = $commentLink; endif;
	230	+
	231	+ // Machine-readable content feed URI
	232	+ $commentFeed = apply_filters('syndicated_item_commentrss', $this->comment_feed(), $this);
	233	+ if (!is_null($commentFeed)) : $this->post['meta']['wfw:commentRSS'] = $commentFeed; endif;
	234	+ // Yeah, yeah, now I know that it's supposed to be
	235	+ // wfw:commentRss. Oh well. Path dependence, sucka.
	236	+
	237	+ // Store information to identify the feed that this came from
	238	+ if (isset($this->feedmeta['link/uri'])) :
	239	+ $this->post['meta']['syndication_feed'] = $this->feedmeta['link/uri'];
115	240	endif;
116		-
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);
	241	+ if (isset($this->feedmeta['link/id'])) :
	242	+ $this->post['meta']['syndication_feed_id'] = $this->feedmeta['link/id'];
120	243	endif;
121		- if (!is_array($custom_settings)) :
122		- $custom_settings = array();
	244	+
	245	+ if (isset($this->item['source_link_self'])) :
	246	+ $this->post['meta']['syndication_feed_original'] = $this->item['source_link_self'];
123	247	endif;
124		- $this->post['meta'] = array_merge($default_custom_settings, $custom_settings);
125	248
126		- // RSS 2.0 / Atom 1.0 enclosure support
127		- if ( isset($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;
	249	+ // In case you want to know the external permalink...
	250	+ $this->post['meta']['syndication_permalink'] = apply_filters('syndicated_item_link', $this->permalink());
	251	+
	252	+ // Store a hash of the post content for checking whether something needs to be updated
	253	+ $this->post['meta']['syndication_item_hash'] = $this->update_hash();
	254	+
	255	+ // Categories, Tags, and other Terms: from settings assignments (global settings, subscription settings),
	256	+ // and from feed assignments (item metadata, post content)
	257	+ $this->preset_terms = apply_filters('syndicated_item_preset_terms', $this->get_terms_from_settings(), $this);
	258	+ $this->feed_terms = apply_filters('syndicated_item_feed_terms', $this->get_terms_from_feeds(), $this);
	259	+
	260	+ $this->post['post_type'] = apply_filters('syndicated_post_type', $this->link->setting('syndicated post type', 'syndicated_post_type', 'post'), $this);
	261	+ endif;
	262	+ } /* SyndicatedPost::__construct() */
	263	+
	264	+ #####################################
	265	+ #### EXTRACT DATA FROM FEED ITEM ####
	266	+ #####################################
	267	+
	268	+ function substitution_function ($name) {
	269	+ $ret = NULL;
	270	+
	271	+ switch ($name) :
	272	+ // Allowed PHP string functions
	273	+ case 'trim':
	274	+ case 'ltrim':
	275	+ case 'rtrim':
	276	+ case 'strtoupper':
	277	+ case 'strtolower':
	278	+ case 'urlencode':
	279	+ case 'urldecode':
	280	+ $ret = $name;
	281	+ endswitch;
	282	+ return $ret;
	283	+ }
	284	+
	285	+ /**
	286	+ * SyndicatedPost::query uses an XPath-like syntax to query arbitrary
	287	+ * elements within the syndicated item.
	288	+ *
	289	+ * @param string $path
	290	+ * @returns array of string values representing contents of matching
	291	+ * elements or attributes
	292	+ */
	293	+ function query ($path) {
	294	+ $xq = new SyndicatedPostXPathQuery(array("path" => $path));
	295	+
	296	+ $feedChannel = array_merge(
	297	+ $this->get_feed_root_element(),
	298	+ $this->get_feed_channel_elements()
	299	+ );
	300	+
	301	+ $matches = $xq->match(array(
	302	+ "type" => $this->link->simplepie->get_type(),
	303	+ "xmlns" => $this->xmlns,
	304	+ "map" => array(
	305	+ "/" => array($this->entry->data),
	306	+ "item" => array($this->entry->data),
	307	+ "feed" => $feedChannel,
	308	+ "channel" => $feedChannel
	309	+ ),
	310	+ "context" => $this->entry->data,
	311	+ "parent" => $feedChannel,
	312	+ ));
	313	+
	314	+ return $matches;
	315	+ } /* SyndicatedPost::query() */
	316	+
	317	+ function get_feed_root_element () {
	318	+ $matches = array();
	319	+ foreach ($this->link->simplepie->data['child'] as $ns => $root) :
	320	+ foreach ($root as $element => $data) :
	321	+ $matches = array_merge($matches, $data);
	322	+ endforeach;
	323	+ endforeach;
	324	+ return $matches;
	325	+ } /* SyndicatedPost::get_feed_root_element() */
	326	+
	327	+ function get_feed_channel_elements () {
	328	+ $rss = array(
	329	+ SIMPLEPIE_NAMESPACE_RSS_090,
	330	+ SIMPLEPIE_NAMESPACE_RSS_10,
	331	+ 'http://backend.userland.com/RSS2',
	332	+ SIMPLEPIE_NAMESPACE_RSS_20,
	333	+ );
	334	+
	335	+ $matches = array();
	336	+ foreach ($rss as $ns) :
	337	+ $data = $this->link->simplepie->get_feed_tags($ns, 'channel');
	338	+ if (!is_null($data)) :
	339	+ $matches = array_merge($matches, $data);
135	340	endif;
	341	+ endforeach;
	342	+ return $matches;
	343	+ } /* SyndicatedPost::get_feed_channel_elements() */
136	344
137		- // In case you want to point back to the blog this was syndicated from
138		- if (isset($this->feed->channel['title'])) :
139		- $this->post['meta']['syndication_source'] = apply_filters('syndicated_item_source_title', $this->feed->channel['title'], $this);
140		- endif;
	345	+ function get_categories ($params = array()) {
	346	+ return $this->entry->get_categories();
	347	+ }
	348	+
	349	+ function title ($params = array()) {
	350	+ return $this->entry->get_title();
	351	+ } /* SyndicatedPost::title () */
	352	+
	353	+ function content ($params = array()) {
141	354
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);
	355	+ $params = wp_parse_args($params, array(
	356	+ "full only" => false,
	357	+ ));
	358	+
	359	+ $content = NULL;
	360	+
	361	+ // FIXME: This is one of the main places in the code still using
	362	+ // the outmoded SimplePie - to - Magpie construction. We could
	363	+ // replace using SimplePie_Item::get_tags() here. (Or if really
	364	+ // ambitious we could attempt to just use
	365	+ // SimplePie_Item::get_content() with content-only set to TRUE
	366	+ // and some sanitization in effect. -CJ 1jul14
	367	+
	368	+ // atom:content, standard method of providing full HTML content
	369	+ // in Atom feeds.
	370	+ if (isset($this->item['atom_content'])) :
	371	+ $content = $this->item['atom_content'];
	372	+ elseif (isset($this->item['atom']['atom_content'])) :
	373	+ $content = $this->item['atom']['atom_content'];
	374	+
	375	+ // Some exotics: back in the day, before widespread convergence
	376	+ // on content:encoding, some RSS feeds took advantage of XML
	377	+ // namespaces to use an inline xhtml:body or xhtml:div element
	378	+ // for full-content HTML. (E.g. Sam Ruby's feed, IIRC.)
	379	+ elseif (isset($this->item['xhtml']['body'])) :
	380	+ $content = $this->item['xhtml']['body'];
	381	+ elseif (isset($this->item['xhtml']['div'])) :
	382	+ $content = $this->item['xhtml']['div'];
	383	+
	384	+ // content:encoded, most common method of providing full HTML in
	385	+ // RSS 2.0 feeds.
	386	+ elseif (isset($this->item['content']['encoded']) and $this->item['content']['encoded']):
	387	+ $content = $this->item['content']['encoded'];
	388	+
	389	+ // Fall back on elements that sometimes may contain full HTML
	390	+ // but sometimes not.
	391	+ elseif (!$params['full only']) :
	392	+
	393	+ // description element is sometimes used for full HTML
	394	+ // sometimes for summary text in RSS. (By the letter of
	395	+ // the standard, it should just be for summary text.)
	396	+ if (isset($this->item['description'])) :
	397	+ $content = $this->item['description'];
144	398	endif;
145	399
146		- // Make use of atom:source data, if present in an aggregated feed
147		- if (isset($this->item['source_title'])) :
148		- $this->post['meta']['syndication_source_original'] = $this->item['source_title'];
	400	+ endif;
	401	+
	402	+ return $content;
	403	+ } /* SyndicatedPost::content() */
	404	+
	405	+ function excerpt () {
	406	+ # Identify and sanitize excerpt: atom:summary, or rss:description
	407	+ $excerpt = $this->entry->get_description();
	408	+
	409	+ # Many RSS feeds use rss:description, inadvisably, to
	410	+ # carry the entire post (typically with escaped HTML).
	411	+ # If that's what happened, we don't want the full
	412	+ # content for the excerpt.
	413	+ $content = $this->content();
	414	+
	415	+ // Ignore whitespace, case, and tag cruft.
	416	+ $theExcerpt = preg_replace('/\s+/', '', strtolower(strip_tags(html_entity_decode($excerpt))));
	417	+ $theContent = preg_replace('/\s+/', '', strtolower(strip_tags(html_entity_decode($content))));
	418	+ if ( empty($excerpt) or $theExcerpt == $theContent ) :
	419	+ # If content is available, generate an excerpt.
	420	+ if ( strlen(trim($content)) > 0 ) :
	421	+ $excerpt = strip_tags($content);
	422	+ if (strlen($excerpt) > 255) :
	423	+ if (is_object($this->link) and is_object($this->link->simplepie)) :
	424	+ $encoding = $this->link->simplepie->get_encoding();
	425	+ else :
	426	+ $encoding = get_option('blog_charset', 'utf8');
	427	+ endif;
	428	+ $excerpt = mb_substr($excerpt,0,252,$encoding).'...';
	429	+ endif;
149	430	endif;
	431	+ endif;
150	432
151		- if (isset($this->item['source_link'])) :
152		- $this->post['meta']['syndication_source_uri_original'] = $this->item['source_link'];
	433	+ return $excerpt;
	434	+ } /* SyndicatedPost::excerpt() */
	435	+
	436	+ function permalink () {
	437	+ // Handles explicit <link> elements and also RSS 2.0 cases with
	438	+ // <guid isPermaLink="true">, etc. Hooray!
	439	+ $permalink = $this->entry->get_link();
	440	+ return $permalink;
	441	+ }
	442	+
	443	+ function created ($params = array()) {
	444	+ $unfiltered = false; $default = NULL;
	445	+ extract($params);
	446	+
	447	+ $date = '';
	448	+ if (isset($this->item['dc']['created'])) :
	449	+ $date = $this->item['dc']['created'];
	450	+ elseif (isset($this->item['dcterms']['created'])) :
	451	+ $date = $this->item['dcterms']['created'];
	452	+ elseif (isset($this->item['created'])): // Atom 0.3
	453	+ $date = $this->item['created'];
	454	+ endif;
	455	+
	456	+ $time = new FeedTime($date);
	457	+ $ts = $time->timestamp();
	458	+ if (!$unfiltered) :
	459	+ apply_filters('syndicated_item_created', $ts, $this);
	460	+ endif;
	461	+ return $ts;
	462	+ } /* SyndicatedPost::created() */
	463	+
	464	+ function published ($params = array(), $default = NULL) {
	465	+ $fallback = true; $unfiltered = false;
	466	+ if (!is_array($params)) : // Old style
	467	+ $fallback = $params;
	468	+ else : // New style
	469	+ extract($params);
	470	+ endif;
	471	+
	472	+ $date = '';
	473	+ $ts = null;
	474	+
	475	+ # RSS is a fucking mess. Figure out whether we have a date in
	476	+ # <dc:date>, <issued>, <pubDate>, etc., and get it into Unix
	477	+ # epoch format for reformatting. If we can't find anything,
	478	+ # we'll use the last-updated time.
	479	+ if (isset($this->item['dc']['date'])): // Dublin Core
	480	+ $date = $this->item['dc']['date'];
	481	+ elseif (isset($this->item['dcterms']['issued'])) : // Dublin Core extensions
	482	+ $date = $this->item['dcterms']['issued'];
	483	+ elseif (isset($this->item['published'])) : // Atom 1.0
	484	+ $date = $this->item['published'];
	485	+ elseif (isset($this->item['issued'])): // Atom 0.3
	486	+ $date = $this->item['issued'];
	487	+ elseif (isset($this->item['pubdate'])): // RSS 2.0
	488	+ $date = $this->item['pubdate'];
	489	+ endif;
	490	+
	491	+ if (strlen($date) > 0) :
	492	+ $time = new FeedTime($date);
	493	+ $ts = $time->timestamp();
	494	+ elseif ($fallback) : // Fall back to <updated> / <modified> if present
	495	+ $ts = $this->updated(/*fallback=*/ false, /*default=*/ $default);
	496	+ endif;
	497	+
	498	+ # If everything failed, then default to the current time.
	499	+ if (is_null($ts)) :
	500	+ if (-1 == $default) :
	501	+ $ts = time();
	502	+ else :
	503	+ $ts = $default;
153	504	endif;
154		-
155		- if (isset($this->item['source_id'])) :
156		- $this->post['meta']['syndication_source_id_original'] = $this->item['source_id'];
	505	+ endif;
	506	+
	507	+ if (!$unfiltered) :
	508	+ $ts = apply_filters('syndicated_item_published', $ts, $this);
	509	+ endif;
	510	+ return $ts;
	511	+ } /* SyndicatedPost::published() */
	512	+
	513	+ function updated ($params = array(), $default = -1) {
	514	+ $fallback = true; $unfiltered = false;
	515	+ if (!is_array($params)) : // Old style
	516	+ $fallback = $params;
	517	+ else : // New style
	518	+ extract($params);
	519	+ endif;
	520	+
	521	+ $date = '';
	522	+ $ts = null;
	523	+
	524	+ # As far as I know, only dcterms and Atom have reliable ways to
	525	+ # specify when something was *modified* last. If neither is
	526	+ # available, then we'll try to get the time of publication.
	527	+ if (isset($this->item['dc']['modified'])) : // Not really correct
	528	+ $date = $this->item['dc']['modified'];
	529	+ elseif (isset($this->item['dcterms']['modified'])) : // Dublin Core extensions
	530	+ $date = $this->item['dcterms']['modified'];
	531	+ elseif (isset($this->item['modified'])): // Atom 0.3
	532	+ $date = $this->item['modified'];
	533	+ elseif (isset($this->item['updated'])): // Atom 1.0
	534	+ $date = $this->item['updated'];
	535	+ endif;
	536	+
	537	+ if (strlen($date) > 0) :
	538	+ $time = new FeedTime($date);
	539	+ $ts = $time->timestamp();
	540	+ elseif ($fallback) : // Fall back to issued / dc:date
	541	+ $ts = $this->published(/*fallback=*/ false, /*default=*/ $default);
	542	+ endif;
	543	+
	544	+ # If everything failed, then default to the current time.
	545	+ if (is_null($ts)) :
	546	+ if (-1 == $default) :
	547	+ $ts = time();
	548	+ else :
	549	+ $ts = $default;
157	550	endif;
	551	+ endif;
158	552
159		- // Store information on human-readable and machine-readable comment URIs
160		- if (isset($this->item['comments'])) :
161		- $this->post['meta']['rss:comments'] = apply_filters('syndicated_item_comments', $this->item['comments']);
	553	+ if (!$unfiltered) :
	554	+ apply_filters('syndicated_item_updated', $ts, $this);
	555	+ endif;
	556	+ return $ts;
	557	+ } /* SyndicatedPost::updated() */
	558	+
	559	+ var $_hashes = array();
	560	+ function stored_hashes ($id = NULL) {
	561	+ if (is_null($id)) :
	562	+ $id = $this->wp_id();
	563	+ endif;
	564	+
	565	+ if (!isset($this->_hashes[$id])) :
	566	+ $this->_hashes[$id] = get_post_custom_values(
	567	+ 'syndication_item_hash', $id
	568	+ );
	569	+ if (is_null($this->_hashes[$id])) :
	570	+ $this->_hashes[$id] = array();
162	571	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']);
	572	+ endif;
	573	+ return $this->_hashes[$id];
	574	+ }
	575	+
	576	+ function update_hash ($hashed = true) {
	577	+ // Basis for tracking possible changes to item.
	578	+ $hash = array(
	579	+ "title" => $this->entry->get_title(),
	580	+ "link" => $this->permalink(),
	581	+ "content" => $this->content(),
	582	+ "excerpt" => $this->excerpt(),
	583	+ );
	584	+
	585	+ if ($hashed) :
	586	+ $hash = md5(serialize($hash));
	587	+ endif;
	588	+
	589	+ return $hash;
	590	+ } /* SyndicatedPost::update_hash() */
	591	+
	592	+ /**
	593	+ * SyndicatedPost::normalize_guid_prefix(): generates a normalized URL
	594	+ * prefix (including scheme, authority, full path, and the beginning of
	595	+ * a query string) for creating guids that conform to WordPress's
	596	+ * internal constraints on the URL space for valid guids. To create a
	597	+ * normalized guid, just concatenate a valid URL query parameter value
	598	+ * to the returned URL.
	599	+ *
	600	+ * @return string The URL prefix generated.
	601	+ *
	602	+ * @uses trailingslashit()
	603	+ * @uses home_url()
	604	+ * @uses apply_filters()
	605	+ */
	606	+ static function normalize_guid_prefix () {
	607	+ $url = trailingslashit(home_url(/*path=*/ '', /*scheme=*/ 'http'));
	608	+ return apply_filters('syndicated_item_guid_normalized_prefix', $url . '?guid=');
	609	+ } /* SyndicatedPost::normalize_guid_prefix() */
	610	+
	611	+ static function normalize_guid ($guid) {
	612	+ $guid = trim($guid);
	613	+ if (preg_match('/^[0-9a-z]{32}$/i', $guid)) : // MD5
	614	+ $guid = SyndicatedPost::normalize_guid_prefix().strtolower($guid);
	615	+ elseif ((strlen(esc_url($guid)) == 0) or (esc_url($guid) != $guid)) :
	616	+ $guid = SyndicatedPost::normalize_guid_prefix().md5($guid);
	617	+ endif;
	618	+ $guid = trim($guid);
	619	+
	620	+ return $guid;
	621	+ } /* SyndicatedPost::normalize_guid() */
	622	+
	623	+ static function alternative_guid_prefix () {
	624	+ $url = trailingslashit(home_url(/*path=*/ '', /*scheme=*/ 'https'));
	625	+ return apply_filters('syndicated_item_guid_normalized_prefix', $url . '?guid=');
	626	+ }
	627	+ static function alternative_guid ($guid) {
	628	+ $guid = trim($guid);
	629	+ if (preg_match('/^[0-9a-z]{32}$/i', $guid)) : // MD5
	630	+ $guid = SyndicatedPost::alternative_guid_prefix().strtolower($guid);
	631	+ elseif ((strlen(esc_url($guid)) == 0) or (esc_url($guid) != $guid)) :
	632	+ $guid = SyndicatedPost::alternative_guid_prefix().md5($guid);
	633	+ endif;
	634	+ $guid = trim($guid);
	635	+
	636	+ return $guid;
	637	+ } /* SyndicatedPost::normalize_guid() */
	638	+
	639	+ function guid () {
	640	+ $guid = null;
	641	+ if (isset($this->item['id'])): // Atom 0.3 / 1.0
	642	+ $guid = $this->item['id'];
	643	+ elseif (isset($this->item['atom']['id'])) : // Namespaced Atom
	644	+ $guid = $this->item['atom']['id'];
	645	+ elseif (isset($this->item['guid'])) : // RSS 2.0
	646	+ $guid = $this->item['guid'];
	647	+ elseif (isset($this->item['dc']['identifier'])) : // yeah, right
	648	+ $guid = $this->item['dc']['identifier'];
	649	+ endif;
	650	+
	651	+ // Un-set or too long to use as-is. Generate a tag: URI.
	652	+ if (is_null($guid) or strlen($guid) > 250) :
	653	+ // In case we need to check this again
	654	+ $original_guid = $guid;
	655	+
	656	+ // The feed does not seem to have provided us with a
	657	+ // usable unique identifier, so we'll have to cobble
	658	+ // together a tag: URI that might work for us. The base
	659	+ // of the URI will be the host name of the feed source ...
	660	+ $bits = parse_url($this->link->uri());
	661	+ $guid = 'tag:'.$bits['host'];
	662	+
	663	+ // Some ill-mannered feeds (for example, certain feeds
	664	+ // coming from Google Calendar) have extraordinarily long
	665	+ // guids -- so long that they exceed the 255 character
	666	+ // width of the WordPress guid field. But if the string
	667	+ // gets clipped by MySQL, uniqueness tests will fail
	668	+ // forever after and the post will be endlessly
	669	+ // reduplicated. So, instead, Guids Of A Certain Length
	670	+ // are hashed down into a nice, manageable tag: URI.
	671	+ if (!is_null($original_guid)) :
	672	+ $guid .= ',2010-12-03:id.'.md5($original_guid);
	673	+
	674	+ // If we have a date of creation, then we can use that
	675	+ // to uniquely identify the item. (On the other hand, if
	676	+ // the feed producer was consicentious enough to
	677	+ // generate dates of creation, she probably also was
	678	+ // conscientious enough to generate unique identifiers.)
	679	+ elseif (!is_null($this->created())) :
	680	+ $guid .= '://post.'.date('YmdHis', $this->created());
	681	+
	682	+ // Otherwise, use both the URI of the item, *and* the
	683	+ // item's title. We have to use both because titles are
	684	+ // often not unique, and sometimes links aren't unique
	685	+ // either (e.g. Bitch (S)HITLIST, Mozilla Dot Org news,
	686	+ // some podcasts). But it's rare to have *both* the same
	687	+ // title *and* the same link for two different items. So
	688	+ // this is about the best we can do.
	689	+ else :
	690	+ $link = $this->permalink();
	691	+ if (is_null($link)) : $link = $this->link->uri(); endif;
	692	+ $guid .= '://'.md5($link.'/'.$this->item['title']);
167	693	endif;
	694	+ endif;
	695	+ return $guid;
	696	+ } /* SyndicatedPost::guid() */
168	697
169		- // Atom 1.0 comment feeds link-rel
170		- if (isset($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;
180		- endif;
	698	+ function author () {
	699	+ $author = array ();
181	700
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'];
	701	+ $aa = $this->entry->get_authors();
	702	+ if (count($aa) > 0) :
	703	+ $a = reset($aa);
185	704
186		- if (isset($this->item['source_link_self'])) :
187		- $this->post['meta']['syndication_feed_original'] = $this->item['source_link_self'];
188		- endif;
	705	+ $author = array(
	706	+ 'name' => $a->get_name(),
	707	+ 'email' => $a->get_email(),
	708	+ 'uri' => $a->get_link(),
	709	+ );
	710	+ endif;
189	711
190		- // In case you want to know the external permalink...
191		- if (isset($this->item['link'])) :
192		- $permalink = $this->item['link'];
	712	+ if (FEEDWORDPRESS_COMPATIBILITY) :
	713	+ // Search through the MagpieRSS elements: Atom, Dublin Core, RSS
	714	+ if (isset($this->item['author_name'])):
	715	+ $author['name'] = $this->item['author_name'];
	716	+ elseif (isset($this->item['dc']['creator'])):
	717	+ $author['name'] = $this->item['dc']['creator'];
	718	+ elseif (isset($this->item['dc']['contributor'])):
	719	+ $author['name'] = $this->item['dc']['contributor'];
	720	+ elseif (isset($this->feed->channel['dc']['creator'])) :
	721	+ $author['name'] = $this->feed->channel['dc']['creator'];
	722	+ elseif (isset($this->feed->channel['dc']['contributor'])) :
	723	+ $author['name'] = $this->feed->channel['dc']['contributor'];
	724	+ elseif (isset($this->feed->channel['author_name'])) :
	725	+ $author['name'] = $this->feed->channel['author_name'];
	726	+ elseif ($this->feed->is_rss() and isset($this->item['author'])) :
	727	+ // The author element in RSS is allegedly an
	728	+ // e-mail address, but lots of people don't use
	729	+ // it that way. So let's make of it what we can.
	730	+ $author = parse_email_with_realname($this->item['author']);
193	731
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 = $this->item['guid'];
	732	+ if (!isset($author['name'])) :
	733	+ if (isset($author['email'])) :
	734	+ $author['name'] = $author['email'];
	735	+ else :
	736	+ $author['name'] = $this->feed->channel['title'];
	737	+ endif;
198	738	endif;
199	739	endif;
	740	+ endif;
200	741
201		- $this->post['meta']['syndication_permalink'] = apply_filters('syndicated_item_link', $permalink);
	742	+ if (!isset($author['name']) or is_null($author['name'])) :
	743	+ // Nothing found. Try some crappy defaults.
	744	+ if ($this->link->name()) :
	745	+ $author['name'] = $this->link->name();
	746	+ else :
	747	+ $url = parse_url($this->link->uri());
	748	+ $author['name'] = $url['host'];
	749	+ endif;
	750	+ endif;
202	751
203		- // Store a hash of the post content for checking whether something needs to be updated
204		- $this->post['meta']['syndication_item_hash'] = $this->update_hash();
	752	+ if (FEEDWORDPRESS_COMPATIBILITY) :
	753	+ if (isset($this->item['author_email'])):
	754	+ $author['email'] = $this->item['author_email'];
	755	+ elseif (isset($this->feed->channel['author_email'])) :
	756	+ $author['email'] = $this->feed->channel['author_email'];
	757	+ endif;
205	758
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']) ? $this->feedmeta['unfamiliar category'] : null);
	759	+ if (isset($this->item['author_uri'])):
	760	+ $author['uri'] = $this->item['author_uri'];
	761	+ elseif (isset($this->item['author_url'])):
	762	+ $author['uri'] = $this->item['author_url'];
	763	+ elseif (isset($this->feed->channel['author_uri'])) :
	764	+ $author['uri'] = $this->item['author_uri'];
	765	+ elseif (isset($this->feed->channel['author_url'])) :
	766	+ $author['uri'] = $this->item['author_url'];
	767	+ elseif (isset($this->feed->channel['link'])) :
	768	+ $author['uri'] = $this->feed->channel['link'];
	769	+ endif;
	770	+ endif;
209	771
210		- // Categories: start with default categories, if any
	772	+ return $author;
	773	+ } /* SyndicatedPost::author() */
	774	+
	775	+ /**
	776	+ * SyndicatedPost::get_terms_from_settings(): Return an array of terms to associate with the incoming
	777	+ * post based on the Categories, Tags, and other terms associated with each new post by the user's
	778	+ * settings (global and feed-specific).
	779	+ *
	780	+ * @since 2016.0331
	781	+ * @return array of lists, each element has the taxonomy for a key ('category', 'post_tag', etc.),
	782	+ * and a list of term codes (either alphanumeric names, or ID numbers encoded in a format that
	783	+ * SyndicatedLink::category_ids() can understand) within that taxonomy
	784	+ *
	785	+ */
	786	+ public function get_terms_from_settings () {
	787	+ // Categories: start with default categories, if any.
	788	+ $cats = array();
	789	+ if ('no' != $this->link->setting('add/category', NULL, 'yes')) :
211	790	$fc = get_option("feedwordpress_syndication_cats");
212	791	if ($fc) :
213		- $this->post['named']['preset/category'] = explode("\n", $fc);
214		- else :
215		- $this->post['named']['preset/category'] = array();
	792	+ $cats = array_merge($cats, explode("\n", $fc));
216	793	endif;
	794	+ endif;
217	795
218		- if (isset($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']);
220		- endif;
	796	+ $fc = $this->link->setting('cats',NULL, array());
	797	+ if (is_array($fc)) :
	798	+ $cats = array_merge($cats, $fc);
	799	+ endif;
	800	+ $preset_terms['category'] = $cats;
221	801
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}"];
	802	+ // Tags: start with default tags, if any
	803	+ $tags = array();
	804	+ if ('no' != $this->link->setting('add/post_tag', NULL, 'yes')) :
	805	+ $ft = get_option("feedwordpress_syndication_tags", NULL);
	806	+ $tags = (is_null($ft) ? array() : explode(FEEDWORDPRESS_CAT_SEPARATOR, $ft));
	807	+ endif;
228	808
229		- if ( isset($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));
232		- else :
233		- $this->post['named']['category'][] = $cat;
	809	+ $ft = $this->link->setting('tags', NULL, array());
	810	+ if (is_array($ft)) :
	811	+ $tags = array_merge($tags, $ft);
	812	+ endif;
	813	+ $preset_terms['post_tag'] = $tags;
	814	+
	815	+ $taxonomies = $this->link->taxonomies();
	816	+ $feedTerms = $this->link->setting('terms', NULL, array());
	817	+ $globalTerms = get_option('feedwordpress_syndication_terms', array());
	818	+ $specials = array('category' => 'cats', 'post_tag' => 'tags');
	819	+
	820	+ foreach ($taxonomies as $tax) :
	821	+ // category and tag settings have already previously been handled
	822	+ // but if this is from another taxonomy, then...
	823	+ if (!isset($specials[$tax])) :
	824	+ $terms = array();
	825	+
	826	+ // See if we should get the globals
	827	+ if ('no' != $this->link->setting("add/$tax", NULL, 'yes')) :
	828	+ if (isset($globalTerms[$tax])) :
	829	+ $terms = $globalTerms[$tax];
234	830	endif;
235		- endfor;
	831	+ endif;
	832	+
	833	+ // Now merge in the locals
	834	+ if (isset($feedTerms[$tax])) :
	835	+ $terms = array_merge($terms, $feedTerms[$tax]);
	836	+ endif;
	837	+
	838	+ // That's all, folks.
	839	+ $preset_terms[$tax] = $terms;
236	840	endif;
237		- $this->post['named']['category'] = apply_filters('syndicated_item_categories', $this->post['named']['category'], $this);
238		-
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);
	841	+ endforeach;
	842	+
	843	+ return $preset_terms;
	844	+ } /* SyndicatedPost::get_terms_from_settings () */
	845	+
	846	+ /**
	847	+ * SyndicatedPost::get_terms_from_feeds(): Return an array of terms to associate with the incoming
	848	+ * post based on the contents of the subscribed feed (atom:category and rss:category elements, dc:subject
	849	+ * elements, tags embedded using microformats in the post content, etc.)
	850	+ *
	851	+ * @since 2016.0331
	852	+ * @return array of lists, each element has the taxonomy for a key ('category', 'post_tag', etc.),
	853	+ * and a list of alphanumeric term names
	854	+ */
	855	+ public function get_terms_from_feeds () {
	856	+ // Now add categories from the post, if we have 'em
	857	+ $cats = array();
	858	+ $post_cats = $this->entry->get_categories();
	859	+ if (is_array($post_cats)) : foreach ($post_cats as $cat) :
	860	+ $cat_name = $cat->get_term();
	861	+ if (!$cat_name) : $cat_name = $cat->get_label(); endif;
	862	+
	863	+ if ($this->link->setting('cat_split', NULL, NULL)) :
	864	+ $pcre = "\007".$this->feedmeta['cat_split']."\007";
	865	+ $cats = array_merge(
	866	+ $cats,
	867	+ preg_split(
	868	+ $pcre,
	869	+ $cat_name,
	870	+ -1 /*=no limit*/,
	871	+ PREG_SPLIT_NO_EMPTY
	872	+ )
	873	+ );
243	874	else :
244		- $this->post['tags_input'] = array();
	875	+ $cats[] = $cat_name;
245	876	endif;
246		-
247		- if (isset($this->feedmeta['tags']) and is_array($this->feedmeta['tags'])) :
248		- $this->post['tags_input'] = array_merge($this->post['tags_input'], $this->feedmeta['tags']);
	877	+ endforeach; endif;
	878	+
	879	+ $feed_terms['category'] = apply_filters('syndicated_item_categories', $cats, $this);
	880	+
	881	+ // Scan post for /a[@rel='tag'] and use as tags if present
	882	+ $tags = $this->inline_tags();
	883	+ $feed_terms['post_tag'] = apply_filters('syndicated_item_tags', $tags, $this);
	884	+
	885	+ return $feed_terms;
	886	+ } /* SyndicatedPost::get_terms_from_feeds () */
	887	+
	888	+ /**
	889	+ * SyndicatedPost::inline_tags: Return a list of all the tags embedded
	890	+ * in post content using the a[@rel="tag"] microformat.
	891	+ *
	892	+ * @since 2010.0630
	893	+ * @return array of string values containing the name of each tag
	894	+ */
	895	+ function inline_tags () {
	896	+ $tags = array();
	897	+ $content = $this->content();
	898	+ $pattern = FeedWordPressHTML::tagWithAttributeRegex('a', 'rel', 'tag');
	899	+ preg_match_all($pattern, $content, $refs, PREG_SET_ORDER);
	900	+ if (count($refs) > 0) :
	901	+ foreach ($refs as $ref) :
	902	+ $tag = FeedWordPressHTML::tagWithAttributeMatch($ref);
	903	+ $tags[] = $tag['content'];
	904	+ endforeach;
	905	+ endif;
	906	+ return $tags;
	907	+ }
	908	+
	909	+ /**
	910	+ * SyndicatedPost::isTaggedAs: Test whether a feed item is
	911	+ * tagged / categorized with a given string. Case and leading and
	912	+ * trailing whitespace are ignored.
	913	+ *
	914	+ * @param string $tag Tag to check for
	915	+ *
	916	+ * @return bool Whether or not at least one of the categories / tags on
	917	+ * $this->item is set to $tag (modulo case and leading and trailing
	918	+ * whitespace)
	919	+ */
	920	+ function isTaggedAs ($tag) {
	921	+ $desiredTag = strtolower(trim($tag)); // Normalize case and whitespace
	922	+
	923	+ // Check to see if this is tagged with $tag
	924	+ $currentCategory = 'category';
	925	+ $currentCategoryNumber = 1;
	926	+
	927	+ // If we have the new MagpieRSS, the number of category elements
	928	+ // on this item is stored under index "category#".
	929	+ if (isset($this->item['category#'])) :
	930	+ $numberOfCategories = (int) $this->item['category#'];
	931	+
	932	+ // We REALLY shouldn't have the old and busted MagpieRSS, but in
	933	+ // case we do, it doesn't support multiple categories, but there
	934	+ // might still be a single value under the "category" index.
	935	+ elseif (isset($this->item['category'])) :
	936	+ $numberOfCategories = 1;
	937	+
	938	+ // No standard category or tag elements on this feed item.
	939	+ else :
	940	+ $numberOfCategories = 0;
	941	+
	942	+ endif;
	943	+
	944	+ $isSoTagged = false; // Innocent until proven guilty
	945	+
	946	+ // Loop through category elements; if there are multiple
	947	+ // elements, they are indexed as category, category#2,
	948	+ // category#3, ... category#N
	949	+ while ($currentCategoryNumber <= $numberOfCategories) :
	950	+ if ($desiredTag == strtolower(trim($this->item[$currentCategory]))) :
	951	+ $isSoTagged = true; // Got it!
	952	+ break;
249	953	endif;
250		- $this->post['tags_input'] = apply_filters('syndicated_item_tags', $this->post['tags_input'], $this);
	954	+
	955	+ $currentCategoryNumber += 1;
	956	+ $currentCategory = 'category#'.$currentCategoryNumber;
	957	+ endwhile;
	958	+
	959	+ return $isSoTagged;
	960	+ } /* SyndicatedPost::isTaggedAs() */
	961	+
	962	+ /**
	963	+ * SyndicatedPost::enclosures: returns an array with any enclosures
	964	+ * that may be attached to this syndicated item.
	965	+ *
	966	+ * @param string $type If you only want enclosures that match a certain
	967	+ * MIME type or group of MIME types, you can limit the enclosures
	968	+ * that will be returned to only those with a MIME type which
	969	+ * matches this regular expression.
	970	+ * @return array
	971	+ */
	972	+ function enclosures ($type = '/.*/') {
	973	+ $enclosures = array();
	974	+
	975	+ if (isset($this->item['enclosure#'])) :
	976	+ // Loop through enclosure, enclosure#2, enclosure#3, ....
	977	+ for ($i = 1; $i <= $this->item['enclosure#']; $i++) :
	978	+ $eid = (($i > 1) ? "#{$id}" : "");
	979	+
	980	+ // Does it match the type we want?
	981	+ if (preg_match($type, $this->item["enclosure{$eid}@type"])) :
	982	+ $enclosures[] = array(
	983	+ "url" => $this->item["enclosure{$eid}@url"],
	984	+ "type" => $this->item["enclosure{$eid}@type"],
	985	+ "length" => $this->item["enclosure{$eid}@length"],
	986	+ );
	987	+ endif;
	988	+ endfor;
251	989	endif;
252		- } // SyndicatedPost::SyndicatedPost()
	990	+ return $enclosures;
	991	+ } /* SyndicatedPost::enclosures() */
253	992
	993	+ function source ($what = NULL) {
	994	+ $ret = NULL;
	995	+ $source = $this->entry->get_source();
	996	+ if ($source) :
	997	+ $ret = array();
	998	+ $ret['title'] = $source->get_title();
	999	+ $ret['uri'] = $source->get_link();
	1000	+ $ret['feed'] = $source->get_link(0, 'self');
	1001	+
	1002	+ if ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'id')) :
	1003	+ $ret['id'] = $id_tags[0]['data'];
	1004	+ elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_ATOM_03, 'id')) :
	1005	+ $ret['id'] = $id_tags[0]['data'];
	1006	+ elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_RSS_20, 'guid')) :
	1007	+ $ret['id'] = $id_tags[0]['data'];
	1008	+ elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_RSS_10, 'guid')) :
	1009	+ $ret['id'] = $id_tags[0]['data'];
	1010	+ elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_RSS_090, 'guid')) :
	1011	+ $ret['id'] = $id_tags[0]['data'];
	1012	+ endif;
	1013	+ endif;
	1014	+
	1015	+ if (!is_null($what) and is_scalar($what)) :
	1016	+ $ret = $ret[$what];
	1017	+ endif;
	1018	+ return $ret;
	1019	+ }
	1020	+
	1021	+ function comment_link () {
	1022	+ $url = null;
	1023	+
	1024	+ // RSS 2.0 has a standard <comments> element:
	1025	+ // "<comments> is an optional sub-element of <item>. If present,
	1026	+ // it is the url of the comments page for the item."
	1027	+ // <http://cyber.law.harvard.edu/rss/rss.html#ltcommentsgtSubelementOfLtitemgt>
	1028	+ if (isset($this->item['comments'])) :
	1029	+ $url = $this->item['comments'];
	1030	+ endif;
	1031	+
	1032	+ // The convention in Atom feeds is to use a standard <link>
	1033	+ // element with @rel="replies" and @type="text/html".
	1034	+ // Unfortunately, SimplePie_Item::get_links() allows us to filter
	1035	+ // by the value of @rel, but not by the value of @type. *sigh*
	1036	+
	1037	+ // Try Atom 1.0 first
	1038	+ $linkElements = $this->entry->get_item_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'link');
	1039	+
	1040	+ // Fall back and try Atom 0.3
	1041	+ if (is_null($linkElements)) : $linkElements = $this->entry->get_item_tags(SIMPLEPIE_NAMESPACE_ATOM_03, 'link'); endif;
	1042	+
	1043	+ // Now loop through the elements, screening by @rel and @type
	1044	+ if (is_array($linkElements)) : foreach ($linkElements as $link) :
	1045	+ $rel = (isset($link['attribs']['']['rel']) ? $link['attribs']['']['rel'] : 'alternate');
	1046	+ $type = (isset($link['attribs']['']['type']) ? $link['attribs']['']['type'] : NULL);
	1047	+ $href = (isset($link['attribs']['']['href']) ? $link['attribs']['']['href'] : NULL);
	1048	+
	1049	+ if (strtolower($rel)=='replies' and $type=='text/html' and !is_null($href)) :
	1050	+ $url = $href;
	1051	+ endif;
	1052	+ endforeach; endif;
	1053	+
	1054	+ return $url;
	1055	+ }
	1056	+
	1057	+ function comment_feed () {
	1058	+ $feed = null;
	1059	+
	1060	+ // Well Formed Web comment feeds extension for RSS 2.0
	1061	+ // <http://www.sellsbrothers.com/spout/default.aspx?content=archive.htm#exposingRssComments>
	1062	+ //
	1063	+ // N.B.: Correct capitalization is wfw:commentRss, but
	1064	+ // wfw:commentRSS is common in the wild (partly due to a typo in
	1065	+ // the original spec). In any case, our item array is normalized
	1066	+ // to all lowercase anyways.
	1067	+ if (isset($this->item['wfw']['commentrss'])) :
	1068	+ $feed = $this->item['wfw']['commentrss'];
	1069	+ endif;
	1070	+
	1071	+ // In Atom 1.0, the convention is to use a standard link element
	1072	+ // with @rel="replies". Sometimes this is also used to pass a
	1073	+ // link to the human-readable comments page, so we also need to
	1074	+ // check link/@type for a feed MIME type.
	1075	+ //
	1076	+ // Which is why I'm not using the SimplePie_Item::get_links()
	1077	+ // method here, incidentally: it doesn't allow you to filter by
	1078	+ // @type. *sigh*
	1079	+ if (isset($this->item['link_replies'])) :
	1080	+ // There may be multiple <link rel="replies"> elements; feeds have a feed MIME type
	1081	+ $N = isset($this->item['link_replies#']) ? $this->item['link_replies#'] : 1;
	1082	+ for ($i = 1; $i <= $N; $i++) :
	1083	+ $currentElement = 'link_replies'.(($i > 1) ? '#'.$i : '');
	1084	+ if (isset($this->item[$currentElement.'@type'])
	1085	+ and preg_match("\007application/(atom|rss|rdf)\+xml\007i", $this->item[$currentElement.'@type'])) :
	1086	+ $feed = $this->item[$currentElement];
	1087	+ endif;
	1088	+ endfor;
	1089	+ endif;
	1090	+ return $feed;
	1091	+ } /* SyndicatedPost::comment_feed() */
	1092	+
	1093	+ ##################################
	1094	+ #### BUILT-IN CONTENT FILTERS ####
	1095	+ ##################################
	1096	+
	1097	+ var $uri_attrs = array (
	1098	+ array('a', 'href'),
	1099	+ array('applet', 'codebase'),
	1100	+ array('area', 'href'),
	1101	+ array('blockquote', 'cite'),
	1102	+ array('body', 'background'),
	1103	+ array('del', 'cite'),
	1104	+ array('form', 'action'),
	1105	+ array('frame', 'longdesc'),
	1106	+ array('frame', 'src'),
	1107	+ array('iframe', 'longdesc'),
	1108	+ array('iframe', 'src'),
	1109	+ array('head', 'profile'),
	1110	+ array('img', 'longdesc'),
	1111	+ array('img', 'src'),
	1112	+ array('img', 'usemap'),
	1113	+ array('input', 'src'),
	1114	+ array('input', 'usemap'),
	1115	+ array('ins', 'cite'),
	1116	+ array('link', 'href'),
	1117	+ array('object', 'classid'),
	1118	+ array('object', 'codebase'),
	1119	+ array('object', 'data'),
	1120	+ array('object', 'usemap'),
	1121	+ array('q', 'cite'),
	1122	+ array('script', 'src')
	1123	+ ); /* var SyndicatedPost::$uri_attrs */
	1124	+
	1125	+ var $_base = null;
	1126	+
	1127	+ function resolve_single_relative_uri ($refs) {
	1128	+ $tag = FeedWordPressHTML::attributeMatch($refs);
	1129	+ $url = SimplePie_Misc::absolutize_url($tag['value'], $this->_base);
	1130	+
	1131	+ return $tag['prefix'] . $url . $tag['suffix'];
	1132	+ } /* function SyndicatedPost::resolve_single_relative_uri() */
	1133	+
	1134	+ static function resolve_relative_uris ($content, $obj) {
	1135	+ $set = $obj->link->setting('resolve relative', 'resolve_relative', 'yes');
	1136	+ if ($set and $set != 'no') :
	1137	+ // Fallback: if we don't have anything better, use the
	1138	+ // item link from the feed
	1139	+ $obj->_base = $obj->permalink(); // Reset the base for resolving relative URIs
	1140	+
	1141	+ // What we should do here, properly, is to use
	1142	+ // SimplePie_Item::get_base() -- but that method is
	1143	+ // currently broken. Or getting down and dirty in the
	1144	+ // SimplePie representation of the content tags and
	1145	+ // grabbing the xml_base member for the content element.
	1146	+ // Maybe someday...
	1147	+
	1148	+ foreach ($obj->uri_attrs as $pair) :
	1149	+ list($tag, $attr) = $pair;
	1150	+ $pattern = FeedWordPressHTML::attributeRegex($tag, $attr);
	1151	+
	1152	+ // FIXME: Encountered issue while testing an extremely long (= 88827 characters) item
	1153	+ // Relying on preg_replace_callback() here can cause a PHP seg fault on my development
	1154	+ // server. preg_match_all() causes a similar problem. Apparently this is a PCRE issue
	1155	+ // Cf. discussion of similar issue <https://bugs.php.net/bug.php?id=65009>
	1156	+ $content = preg_replace_callback (
	1157	+ $pattern,
	1158	+ array($obj, 'resolve_single_relative_uri'),
	1159	+ $content
	1160	+ );
	1161	+
	1162	+ endforeach;
	1163	+ endif;
	1164	+
	1165	+ return $content;
	1166	+ } /* function SyndicatedPost::resolve_relative_uris () */
	1167	+
	1168	+ var $strip_attrs = array (
	1169	+ array('[a-z]+', 'target'),
	1170	+// array('[a-z]+', 'style'),
	1171	+// array('[a-z]+', 'on[a-z]+'),
	1172	+ );
	1173	+
	1174	+ function strip_attribute_from_tag ($refs) {
	1175	+ $tag = FeedWordPressHTML::attributeMatch($refs);
	1176	+ return $tag['before_attribute'].$tag['after_attribute'];
	1177	+ }
	1178	+
	1179	+ static function sanitize_content ($content, $obj) {
	1180	+ # This kind of sucks. I intend to replace it with
	1181	+ # lib_filter sometime soon.
	1182	+ foreach ($obj->strip_attrs as $pair):
	1183	+ list($tag,$attr) = $pair;
	1184	+ $pattern = FeedWordPressHTML::attributeRegex($tag, $attr);
	1185	+
	1186	+ $content = preg_replace_callback (
	1187	+ $pattern,
	1188	+ array($obj, 'strip_attribute_from_tag'),
	1189	+ $content
	1190	+ );
	1191	+ endforeach;
	1192	+ return $content;
	1193	+ } /* SyndicatedPost::sanitize() */
	1194	+
	1195	+ #####################
	1196	+ #### POST STATUS ####
	1197	+ #####################
	1198	+
	1199	+ /**
	1200	+ * SyndicatedPost::filtered: check whether or not this post has been
	1201	+ * screened out by a registered filter.
	1202	+ *
	1203	+ * @return bool TRUE iff post has been filtered out by a previous filter
	1204	+ */
254	1205	function filtered () {
255	1206	return is_null($this->post);
256		- }
	1207	+ } /* SyndicatedPost::filtered() */
257	1208
258		- function freshness () {
	1209	+ /**
	1210	+ * SyndicatedPost::freshness: check whether post is a new post to be
	1211	+ * inserted, a previously syndicated post that needs to be updated to
	1212	+ * match the latest revision, or a previously syndicated post that is
	1213	+ * still up-to-date.
	1214	+ *
	1215	+ * @return int A status code representing the freshness of the post
	1216	+ * -1 = post already syndicated; has a revision that needs to be stored, but not updated to
	1217	+ * 0 = post already syndicated; no update needed
	1218	+ * 1 = post already syndicated, but needs to be updated to latest
	1219	+ * 2 = post has not yet been syndicated; needs to be created
	1220	+ */
	1221	+ function freshness ($format = 'number') {
259	1222	global $wpdb;
260	1223
261	1224	if ($this->filtered()) : // This should never happen.
262		- FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__);
	1225	+ FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__, __FILE__);
263	1226	endif;
264		-
265		- if (is_null($this->_freshness)) :
266		- $guid = $wpdb->escape($this->guid());
267	1227
268		- $result = $wpdb->get_row("
269		- SELECT id, guid, post_modified_gmt
270		- FROM $wpdb->posts WHERE guid='$guid'
271		- ");
	1228	+ if (is_null($this->_freshness)) : // Not yet checked and cached.
	1229	+ $guid = $this->post['guid'];
	1230	+ $eguid = esc_sql($this->post['guid']);
272	1231
273		- if (!$result) :
	1232	+ $q = new WP_Query(array(
	1233	+ 'fields' => '_synfresh', // id, guid, post_modified_gmt
	1234	+ 'ignore_sticky_posts' => true,
	1235	+ 'guid' => $guid,
	1236	+ ));
	1237	+
	1238	+ $old_post = NULL;
	1239	+ if ($q->have_posts()) :
	1240	+ while ($q->have_posts()) : $q->the_post();
	1241	+ $old_post = $q->post;
	1242	+ endwhile;
	1243	+ endif;
	1244	+
	1245	+ if (is_null($old_post)) : // No post with this guid
	1246	+ FeedWordPress::diagnostic('feed_items:freshness', 'Item ['.$guid.'] "'.$this->entry->get_title().'" is a NEW POST.');
	1247	+ $this->_wp_id = NULL;
274	1248	$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());
	1249	+ else :
	1250	+ // Presume there is nothing new until we find
	1251	+ // something new.
	1252	+ $updated = false;
	1253	+ $live = false;
	1254	+
	1255	+ // Pull the list of existing revisions to get
	1256	+ // timestamps.
	1257	+ $revisions = wp_get_post_revisions($old_post->ID);
	1258	+ foreach ($revisions as $rev) :
	1259	+ $revisions_ts[] = mysql2date('G', $rev->post_modified_gmt);
	1260	+ endforeach;
	1261	+
	1262	+ $revisions_ts[] = mysql2date('G', $old_post->post_modified_gmt);
	1263	+ $last_rev_ts = end($revisions_ts);
	1264	+ $updated_ts = $this->updated(/*fallback=*/ true, /*default=*/ NULL);
	1265	+
	1266	+ // If we have an explicit updated timestamp,
	1267	+ // check that against existing stamps.
	1268	+ if (!is_null($updated_ts)) :
	1269	+ $updated = !in_array($updated_ts, $revisions_ts);
	1270	+
	1271	+ // If this a newer revision, make it go
	1272	+ // live. If an older one, just record
	1273	+ // the contents.
	1274	+ $live = ($updated and ($updated_ts > $last_rev_ts));
	1275	+ endif;
	1276	+
	1277	+ // This is a revision we haven't seen before, judging by the date.
	1278	+
	1279	+ $updatedReason = NULL;
	1280	+ if ($updated) :
	1281	+ $updatedReason = preg_replace(
	1282	+ "/\s+/", " ",
	1283	+ 'has been marked with a new timestamp ('
	1284	+ .date('Y-m-d H:i:s', $updated_ts)
	1285	+ ." > "
	1286	+ .date('Y-m-d H:i:s', $last_rev_ts)
	1287	+ .')'
	1288	+ );
	1289	+
	1290	+ // The date does not indicate a new revision, so
	1291	+ // let's check the hash.
280	1292	else :
281		- $update_hash_changed = false;
	1293	+ // Or the hash...
	1294	+ $hash = $this->update_hash();
	1295	+ $seen = $this->stored_hashes($old_post->ID);
	1296	+ if (count($seen) > 0) :
	1297	+ $updated = !in_array($hash, $seen); // Not seen yet?
	1298	+ else :
	1299	+ $updated = true; // Can't find syndication meta-data
	1300	+ endif;
	1301	+
	1302	+ if ($updated and FeedWordPress::diagnostic_on('feed_items:freshness:reasons')) :
	1303	+ // In the absence of definitive
	1304	+ // timestamp information, we
	1305	+ // just have to assume that a
	1306	+ // hash we haven't seen before
	1307	+ // is a newer version.
	1308	+ $live = true;
	1309	+
	1310	+ $updatedReason = ' has a not-yet-seen update hash: '
	1311	+ .MyPHP::val($hash)
	1312	+ .' not in {'
	1313	+ .implode(", ", array_map(array('FeedWordPress', 'val'), $seen))
	1314	+ .'}. Basis: '
	1315	+ .MyPHP::val(array_keys($this->update_hash(false)));
	1316	+ endif;
282	1317	endif;
283	1318
284		- preg_match('/([0-9]+)-([0-9]+)-([0-9]+) ([0-9]+):([0-9]+):([0-9]+)/', $result->post_modified_gmt, $backref);
	1319	+ $frozen = false;
	1320	+ if ($updated) : // Ignore if the post is frozen
	1321	+ $frozen = ('yes' == $this->link->setting('freeze updates', 'freeze_updates', NULL));
	1322	+ if (!$frozen) :
	1323	+ $frozen_values = get_post_custom_values('_syndication_freeze_updates', $old_post->ID);
	1324	+ $frozen = (count($frozen_values) > 0 and 'yes' == $frozen_values[0]);
285	1325
286		- $last_rev_ts = gmmktime($backref[4], $backref[5], $backref[6], $backref[2], $backref[3], $backref[1]);
287		- $updated_ts = $this->updated(/*fallback=*/ true, /*default=*/ NULL);
288		- $updated = ((
289		- !is_null($updated_ts)
290		- and ($updated_ts > $last_rev_ts)
291		- ) or $update_hash_changed);
	1326	+ if ($frozen) :
	1327	+ $updatedReason = ' IS BLOCKED FROM BEING UPDATED BY A UPDATE LOCK ON THIS POST, EVEN THOUGH IT '.$updatedReason;
	1328	+ endif;
	1329	+ else :
	1330	+ $updatedReason = ' IS BLOCKED FROM BEING UPDATED BY A FEEDWORDPRESS UPDATE LOCK, EVEN THOUGH IT '.$updatedReason;
	1331	+ endif;
	1332	+ endif;
	1333	+ $live = ($live and !$frozen);
292	1334
293	1335	if ($updated) :
294		- $this->_freshness = 1; // Updated content
295		- $this->_wp_id = $result->id;
	1336	+ FeedWordPress::diagnostic('feed_items:freshness', 'Item ['.$guid.'] "'.$this->entry->get_title().'" is an update of an existing post.');
	1337	+ if (!is_null($updatedReason)) :
	1338	+ $updatedReason = preg_replace('/\s+/', ' ', $updatedReason);
	1339	+ FeedWordPress::diagnostic('feed_items:freshness:reasons', 'Item ['.$guid.'] "'.$this->entry->get_title().'" '.$updatedReason);
	1340	+ endif;
	1341	+
	1342	+ $this->_freshness = apply_filters('syndicated_item_freshness', ($live ? 1 : -1), $updated, $frozen, $updated_ts, $last_rev_ts, $this);
	1343	+
	1344	+ $this->_wp_id = $old_post->ID;
	1345	+ $this->_wp_post = $old_post;
	1346	+
	1347	+ // We want this to keep a running list of all the
	1348	+ // processed update hashes.
	1349	+ $this->post['meta']['syndication_item_hash'] = array_merge(
	1350	+ $this->stored_hashes(),
	1351	+ array($this->update_hash())
	1352	+ );
296	1353	else :
	1354	+ FeedWordPress::diagnostic('feed_items:freshness', 'Item ['.$guid.'] "'.$this->entry->get_title().'" is a duplicate of an existing post.');
297	1355	$this->_freshness = 0; // Same old, same old
298		- $this->_wp_id = $result->id;
	1356	+ $this->_wp_id = $old_post->ID;
299	1357	endif;
300	1358	endif;
301	1359	endif;
302		- return $this->_freshness;
	1360	+
	1361	+ switch ($format) :
	1362	+ case 'status' :
	1363	+ switch ($this->_freshness) :
	1364	+ case -1:
	1365	+ $ret = 'stored';
	1366	+ break;
	1367	+ case 0:
	1368	+ $ret = NULL;
	1369	+ break;
	1370	+ case 1:
	1371	+ $ret = 'updated';
	1372	+ break;
	1373	+ case 2:
	1374	+ default:
	1375	+ $ret = 'new';
	1376	+ break;
	1377	+ endswitch;
	1378	+ break;
	1379	+ case 'number' :
	1380	+ default :
	1381	+ $ret = $this->_freshness;
	1382	+ endswitch;
	1383	+
	1384	+
	1385	+ return $ret;
	1386	+ } /* SyndicatedPost::freshness () */
	1387	+
	1388	+ function has_fresh_content () {
	1389	+ return ( ! $this->filtered() and $this->freshness() != 0 );
	1390	+ } /* SyndicatedPost::has_fresh_content () */
	1391	+
	1392	+ function this_revision_needs_original_post ($freshness = NULL) {
	1393	+ if (is_null($freshness)) :
	1394	+ $freshness = $this->freshness();
	1395	+ endif;
	1396	+ return ( $freshness >= 2 );
303	1397	}
	1398	+
	1399	+ function this_revision_is_current ($freshness = NULL) {
	1400	+ if (is_null($freshness)) :
	1401	+ $freshness = $this->freshness();
	1402	+ endif;
	1403	+ return ( $freshness >= 1 );
	1404	+ } /* SyndicatedPost::this_revision_is_current () */
	1405	+
	1406	+ function fresh_content_is_update () {
	1407	+ return ($this->freshness() < 2);
	1408	+ } /* SyndicatedPost::fresh_content_is_update () */
	1409	+
	1410	+ function fresh_storage_diagnostic () {
	1411	+ $ret = NULL;
	1412	+ switch ($this->freshness()) :
	1413	+ case -1 :
	1414	+ $ret = 'Storing alternate revision of existing post # '.$this->wp_id().', "'.$this->post['post_title'].'"';
	1415	+ break;
	1416	+ case 1 :
	1417	+ $ret = 'Updating existing post # '.$this->wp_id().', "'.$this->post['post_title'].'"';
	1418	+ break;
	1419	+ case 2 :
	1420	+ default :
	1421	+ $ret = 'Inserting new post "'.$this->post['post_title'].'"';
	1422	+ break;
	1423	+ endswitch;
	1424	+ return $ret;
	1425	+ } /* SyndicatedPost::fresh_storage_diagnostic() */
	1426	+
	1427	+ function fresh_storage_hook () {
	1428	+ $ret = NULL;
	1429	+ switch ($this->freshness()) :
	1430	+ case -1 :
	1431	+ case 1 :
	1432	+ $ret = 'update_syndicated_item';
	1433	+ break;
	1434	+ case 2 :
	1435	+ default :
	1436	+ $ret = 'post_syndicated_item';
	1437	+ break;
	1438	+ endswitch;
	1439	+ return $ret;
	1440	+ } /* SyndicatedPost::fresh_storage_hook () */
	1441	+
	1442	+ #################################################
	1443	+ #### INTERNAL STORAGE AND MANAGEMENT METHODS ####
	1444	+ #################################################
304	1445
305	1446	function wp_id () {
306	1447	if ($this->filtered()) : // This should never happen.
307		- FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__);
	1448	+ FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__, __FILE__);
308	1449	endif;
309		-
	1450	+
310	1451	if (is_null($this->_wp_id) and is_null($this->_freshness)) :
311	1452	$fresh = $this->freshness(); // sets WP DB id in the process
312	1453	endif;
313	1454	return $this->_wp_id;
		@@ -312,240 +1453,340 @@
312	1453	endif;
313	1454	return $this->_wp_id;
314	1455	}
315	1456
316		- function store () {
	1457	+ public function store () {
317	1458	global $wpdb;
318	1459
319	1460	if ($this->filtered()) : // This should never happen.
320		- FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__);
	1461	+ FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__, __FILE__);
321	1462	endif;
322		-
	1463	+
323	1464	$freshness = $this->freshness();
324		- if ($freshness > 0) :
	1465	+ if ($this->has_fresh_content()) :
325	1466	# -- Look up, or create, numeric ID for author
326	1467	$this->post['post_author'] = $this->author_id (
327		- FeedWordPress::on_unfamiliar('author', $this->post['named']['unfamiliar']['author'])
	1468	+ $this->link->setting('unfamiliar author', 'unfamiliar_author', 'create')
328	1469	);
329	1470
330	1471	if (is_null($this->post['post_author'])) :
	1472	+ FeedWordPress::diagnostic('feed_items:rejected', 'Filtered out item ['.$this->guid().'] without syndication: no author available');
331	1473	$this->post = NULL;
332	1474	endif;
333	1475	endif;
334		-
335		- if (!$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
341		- );
342	1476
343		- $this->post['post_category'] = $pcats;
344		- $this->post['tags_input'] = array_merge($this->post['tags_input'], $ptags);
	1477	+ // We have to check again in case post has been filtered during
	1478	+ // the author_id lookup
	1479	+ if ($this->has_fresh_content()) :
	1480	+ $mapping = apply_filters('syndicated_post_terms_mapping', array(
	1481	+ 'category' => array('abbr' => 'cats', 'unfamiliar' => 'category', 'domain' => array('category', 'post_tag')),
	1482	+ 'post_tag' => array('abbr' => 'tags', 'unfamiliar' => 'post_tag', 'domain' => array('post_tag')),
	1483	+ ), $this);
345	1484
346		- if (is_null($this->post['post_category'])) :
347		- // filter mode on, no matching categories; drop the post
	1485	+ $termSet = array(); $valid = null;
	1486	+ foreach ($this->feed_terms as $what => $anTerms) :
	1487	+ // Default to using the inclusive procedures (for cats) rather than exclusive (for inline tags)
	1488	+ $taxes = (isset($mapping[$what]) ? $mapping[$what] : $mapping['category']);
	1489	+ $unfamiliar = $taxes['unfamiliar'];
	1490	+
	1491	+ if (!is_null($this->post)) : // Not filtered out yet
	1492	+ # -- Look up, or create, numeric ID for categories
	1493	+ $taxonomies = $this->link->setting("match/".$taxes['abbr'], 'match_'.$taxes['abbr'], $taxes['domain']);
	1494	+
	1495	+ // Eliminate dummy variables
	1496	+ $taxonomies = array_filter($taxonomies, 'remove_dummy_zero');
	1497	+
	1498	+ // Allow FWP add-on filters to control the taxonomies we use to search for a term
	1499	+ $taxonomies = apply_filters("syndicated_post_terms_match", $taxonomies, $what, $this);
	1500	+ $taxonomies = apply_filters("syndicated_post_terms_match_${what}", $taxonomies, $this);
	1501	+
	1502	+ // Allow FWP add-on filters to control with greater precision what happens on unmatched
	1503	+ $unmatched = apply_filters("syndicated_post_terms_unfamiliar",
	1504	+ $this->link->setting(
	1505	+ "unfamiliar {$unfamiliar}",
	1506	+ "unfamiliar_{$unfamiliar}",
	1507	+ 'create:'.$unfamiliar
	1508	+ ),
	1509	+ $what,
	1510	+ $this
	1511	+ );
	1512	+
	1513	+ $terms = $this->category_ids (
	1514	+ $anTerms,
	1515	+ $unmatched,
	1516	+ /*taxonomies=*/ $taxonomies,
	1517	+ array(
	1518	+ 'singleton' => false, // I don't like surprises
	1519	+ 'filters' => true,
	1520	+ )
	1521	+ );
	1522	+
	1523	+ if (is_null($terms) or is_null($termSet)) :
	1524	+ // filtered out -- no matches
	1525	+ else :
	1526	+ $valid = true;
	1527	+
	1528	+ // filter mode off, or at least one match
	1529	+ foreach ($terms as $tax => $term_ids) :
	1530	+ if (!isset($termSet[$tax])) :
	1531	+ $termSet[$tax] = array();
	1532	+ endif;
	1533	+ $termSet[$tax] = array_merge($termSet[$tax], $term_ids);
	1534	+ endforeach;
	1535	+ endif;
	1536	+ endif;
	1537	+ endforeach;
	1538	+
	1539	+ if (is_null($valid)) : // Plonked
348	1540	$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		- );
	1541	+ else : // We can proceed
	1542	+ $this->post['tax_input'] = array();
	1543	+ foreach ($termSet as $tax => $term_ids) :
	1544	+ if (!isset($this->post['tax_input'][$tax])) :
	1545	+ $this->post['tax_input'][$tax] = array();
	1546	+ endif;
	1547	+ $this->post['tax_input'][$tax] = array_merge(
	1548	+ $this->post['tax_input'][$tax],
	1549	+ $term_ids
	1550	+ );
	1551	+ endforeach;
358	1552
359		- if (count($this->post['post_category']) < 1) :
360		- $this->post['post_category'][] = 1; // Default to category 1 ("Uncategorized" / "General") if nothing else
361		- endif;
	1553	+ // Now let's add on the feed and global presets
	1554	+ foreach ($this->preset_terms as $tax => $term_ids) :
	1555	+ if (!isset($this->post['tax_input'][$tax])) :
	1556	+ $this->post['tax_input'][$tax] = array();
	1557	+ endif;
	1558	+
	1559	+ $this->post['tax_input'][$tax] = array_merge (
	1560	+ $this->post['tax_input'][$tax],
	1561	+ $this->category_ids (
	1562	+ /*terms=*/ $term_ids,
	1563	+ /*unfamiliar=*/ 'create:'.$tax, // These are presets; for those added in a tagbox editor, the tag may not yet exist
	1564	+ /*taxonomies=*/ array($tax),
	1565	+ array(
	1566	+ 'singleton' => true,
	1567	+ ))
	1568	+ );
	1569	+ endforeach;
362	1570	endif;
363	1571	endif;
364		-
365		- if (!$this->filtered() and $freshness > 0) :
366		- unset($this->post['named']);
	1572	+
	1573	+ // We have to check again in case the post has been filtered
	1574	+ // during the category/tags/taxonomy terms lookup
	1575	+ if ($this->has_fresh_content()) :
	1576	+ // Filter some individual fields
	1577	+
	1578	+ // If there already is a post slug (from syndication or by manual
	1579	+ // editing) don't cause WP to overwrite it by sending in a NULL
	1580	+ // post_name. Props Chris Fritz 2012-11-28.
	1581	+ $post_name = (is_null($this->_wp_post) ? NULL : $this->_wp_post->post_name);
	1582	+
	1583	+ // Allow filters to set post slug. Props niska.
	1584	+ $post_name = apply_filters('syndicated_post_slug', $post_name, $this);
	1585	+ if (!empty($post_name)) :
	1586	+ $this->post['post_name'] = $post_name;
	1587	+ endif;
	1588	+
367	1589	$this->post = apply_filters('syndicated_post', $this->post, $this);
	1590	+
	1591	+ // Allow for feed-specific syndicated_post filters.
	1592	+ $this->post = apply_filters(
	1593	+ "syndicated_post_".$this->link->uri(),
	1594	+ $this->post,
	1595	+ $this
	1596	+ );
368	1597	endif;
	1598	+
	1599	+ // Hook in early to make sure these get inserted if at all possible
	1600	+ add_action(
	1601	+ /*hook=*/ 'transition_post_status',
	1602	+ /*callback=*/ array($this, 'add_rss_meta'),
	1603	+ /*priority=*/ -10000, /* very early */
	1604	+ /*arguments=*/ 3
	1605	+ );
369	1606
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());
	1607	+ $ret = false;
	1608	+ if ($this->has_fresh_content()) :
	1609	+ $diag = $this->fresh_storage_diagnostic();
	1610	+ if (!is_null($diag)) :
	1611	+ FeedWordPress::diagnostic('syndicated_posts', $diag);
	1612	+ endif;
	1613	+
	1614	+ $this->insert_post(/*update=*/ $this->fresh_content_is_update(), $this->freshness());
375	1615
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());
	1616	+ $hook = $this->fresh_storage_hook();
	1617	+ if (!is_null($hook)) :
	1618	+ do_action($hook, $this->wp_id(), $this);
	1619	+ endif;
	1620	+
	1621	+ $ret = $this->freshness('status');
	1622	+ endif;
382	1623
383		- $ret = 'updated';
384		- else :
385		- $ret = false;
	1624	+ // If this is a legit, non-filtered post, tag it as found on the
	1625	+ // feed regardless of fresh or stale status
	1626	+ if (!$this->filtered()) :
	1627	+ $key = '_feedwordpress_retire_me_' . $this->link->id;
	1628	+ delete_post_meta($this->wp_id(), $key);
	1629	+
	1630	+ $status = get_post_field('post_status', $this->wp_id());
	1631	+ if ('fwpretired'==$status and $this->link->is_non_incremental()) :
	1632	+ FeedWordPress::diagnostic('syndicated_posts', "Un-retiring previously retired post # ".$this->wp_id()." due to re-appearance on non-incremental feed.");
	1633	+ set_post_field('post_status', $this->post['post_status'], $this->wp_id());
	1634	+ wp_transition_post_status($this->post['post_status'], $status, $this->post);
	1635	+ elseif ('fwpzapped'==$status) :
	1636	+ // Set this new revision up to be
	1637	+ // blanked on the next update.
	1638	+ add_post_meta($this->wp_id(), '_feedwordpress_zapped_blank_me', 2, /*single=*/ true);
	1639	+ endif;
386	1640	endif;
387		-
	1641	+
	1642	+ // Remove add_rss_meta hook
	1643	+ remove_action(
	1644	+ /*hook=*/ 'transition_post_status',
	1645	+ /*callback=*/ array($this, 'add_rss_meta'),
	1646	+ /*priority=*/ -10000, /* very early */
	1647	+ /*arguments=*/ 3
	1648	+ );
	1649	+
388	1650	return $ret;
389		- } // function SyndicatedPost::store ()
390		-
391		- function insert_new () {
392		- global $wpdb, $wp_db_version;
	1651	+ } /* function SyndicatedPost::store () */
393	1652
	1653	+ function insert_post ($update = false, $freshness = 2) {
	1654	+ global $wpdb;
	1655	+
394	1656	$dbpost = $this->normalize_post(/*new=*/ true);
	1657	+
	1658	+ $ret = null;
	1659	+
395	1660	if (!is_null($dbpost)) :
396		- if ($this->use_api('wp_insert_post')) :
397		- $dbpost['post_pingback'] = false; // Tell WP 2.1 and 2.2 not to process for pingbacks
398		-
399		- // This is a ridiculous fucking kludge necessitated by WordPress 2.6 munging authorship meta-data
400		- add_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
	1661	+ $dbpost['post_pingback'] = false; // Tell WP 2.1 and 2.2 not to process for pingbacks
	1662	+
	1663	+ // This is a ridiculous fucking kludge necessitated by WordPress 2.6 munging authorship meta-data
	1664	+ add_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
	1665	+
	1666	+ // Kludge to prevent kses filters from stripping the
	1667	+ // content of posts when updating without a logged in
	1668	+ // user who has `unfiltered_html` capability.
	1669	+ $mungers = array('wp_filter_kses', 'wp_filter_post_kses');
	1670	+ $removed = array();
	1671	+ foreach ($mungers as $munger) :
	1672	+ if (has_filter('content_save_pre', $munger)) :
	1673	+ remove_filter('content_save_pre', $munger);
	1674	+ $removed[] = $munger;
	1675	+ endif;
	1676	+ endforeach;
	1677	+
	1678	+ if ($update and function_exists('get_post_field')) :
	1679	+ // Don't munge status fields that the user may
	1680	+ // have reset manually
	1681	+ $doNotMunge = array('post_status', 'comment_status', 'ping_status');
	1682	+
	1683	+ foreach ($doNotMunge as $field) :
	1684	+ $dbpost[$field] = get_post_field($field, $this->wp_id());
	1685	+ endforeach;
	1686	+ endif;
	1687	+
	1688	+ // WP3's wp_insert_post scans current_user_can() for the
	1689	+ // tax_input, with no apparent way to override. Ugh.
	1690	+ add_action(
	1691	+ /*hook=*/ 'transition_post_status',
	1692	+ /*callback=*/ array($this, 'add_terms'),
	1693	+ /*priority=*/ -10001, /* very early */
	1694	+ /*arguments=*/ 3
	1695	+ );
	1696	+
	1697	+ // WP3 appears to override whatever you give it for
	1698	+ // post_modified. Ugh.
	1699	+ add_action(
	1700	+ /*hook=*/ 'transition_post_status',
	1701	+ /*callback=*/ array($this, 'fix_post_modified_ts'),
	1702	+ /*priority=*/ -10000, /* very early */
	1703	+ /*arguments=*/ 3
	1704	+ );
	1705	+
	1706	+ if ($update) :
	1707	+ $this->post['ID'] = $this->wp_id();
	1708	+ $dbpost['ID'] = $this->post['ID'];
	1709	+ endif;
	1710	+
	1711	+ // O.K., is this a new post? If so, we need to create
	1712	+ // the basic post record before we do anything else.
	1713	+ if ($this->this_revision_needs_original_post()) :
	1714	+ // *sigh*, for handling inconsistent slash expectations < 3.6
	1715	+ $sdbpost = $this->db_sanitize_post($dbpost);
401	1716
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);
	1717	+ // Go ahead and insert the first post record to
	1718	+ // anchor the revision history.
	1719	+
	1720	+ $this->_wp_id = wp_insert_post($sdbpost, /*return wp_error=*/ true);
406	1721
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);
	1722	+ $dbpost['ID'] = $this->_wp_id;
467	1723	endif;
468		- endif;
469		- } /* SyndicatedPost::insert_new() */
	1724	+
	1725	+ // Sanity check: if the attempt to insert post
	1726	+ // returned an error, then feeding that error
	1727	+ // object in to _wp_put_post_revision() would
	1728	+ // cause a fatal error. Better to break out.
	1729	+ if (!is_wp_error($this->_wp_id)) :
	1730	+ // Now that we've made sure the original exists, insert
	1731	+ // this version here as a revision.
	1732	+ $revision_id = _wp_put_post_revision($dbpost, /*autosave=*/ false);
470	1733
471		- function update_existing () {
472		- global $wpdb;
	1734	+ if (!$this->this_revision_needs_original_post()) :
	1735	+
	1736	+ if ($this->this_revision_is_current()) :
473	1737
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($this, 'avoid_kses_munge'), 11);
	1738	+ wp_restore_post_revision($revision_id);
487	1739
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;
	1740	+ else :
495	1741
496		- $this->_wp_id = wp_insert_post($dbpost);
497		-
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__);
	1742	+ // If we do not activate this revision, then the
	1743	+ // add_rss_meta will not be called, which is
	1744	+ // more or less as it should be, but that means
	1745	+ // we have to actively record this revision's
	1746	+ // update hash from here.
	1747	+ $postId = $this->post['ID'];
	1748	+ $key = 'syndication_item_hash';
	1749	+ $hash = $this->update_hash();
	1750	+ FeedWordPress::diagnostic('syndicated_posts:meta_data', "Adding post meta-datum to post [$postId]: [$key] = ".FeedWordPress::val($hash, /*no newlines=*/ true));
	1751	+ add_post_meta( $postId, $key, $hash, /*unique=*/ false );
	1752	+ endif;
537	1753	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', $this->post['ID']);
546	1754	endif;
	1755	+
	1756	+ remove_action(
	1757	+ /*hook=*/ 'transition_post_status',
	1758	+ /*callback=*/ array($this, 'add_terms'),
	1759	+ /*priority=*/ -10001, /* very early */
	1760	+ /*arguments=*/ 3
	1761	+ );
	1762	+
	1763	+ remove_action(
	1764	+ /*hook=*/ 'transition_post_status',
	1765	+ /*callback=*/ array($this, 'fix_post_modified_ts'),
	1766	+ /*priority=*/ -10000, /* very early */
	1767	+ /*arguments=*/ 3
	1768	+ );
	1769	+
	1770	+ // Turn off ridiculous fucking kludges #1 and #2
	1771	+ remove_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
	1772	+ foreach ($removed as $filter) :
	1773	+ add_filter('content_save_pre', $filter);
	1774	+ endforeach;
	1775	+
	1776	+ $this->validate_post_id($dbpost, $update, array(__CLASS__, __FUNCTION__));
	1777	+
	1778	+ $ret = $this->_wp_id;
547	1779	endif;
	1780	+ return $ret;
	1781	+ } /* function SyndicatedPost::insert_post () */
	1782	+
	1783	+ function insert_new () {
	1784	+ $this->insert_post(/*update=*/ false, 1);
	1785	+ } /* SyndicatedPost::insert_new() */
	1786	+
	1787	+ function update_existing () {
	1788	+ $this->insert_post(/*update=*/ true, 2);
548	1789	} /* SyndicatedPost::update_existing() */
549	1790
550	1791	/**
551	1792	* SyndicatedPost::normalize_post()
		@@ -555,33 +1796,104 @@
555	1796	*/
556	1797	function normalize_post ($new = true) {
557	1798	global $wpdb;
558	1799
559		- $out = array();
	1800	+ $out = $this->post;
560	1801
561		- // Why the fuck doesn't wp_insert_post already do this?
562		- foreach ($this->post as $key => $value) :
563		- if (is_string($value)) :
564		- $out[$key] = $wpdb->escape($value);
565		- else :
566		- $out[$key] = $value;
567		- endif;
568		- endforeach;
569		-
570		- if (strlen($out['post_title'].$out['post_content'].$out['post_excerpt']) == 0) :
	1802	+ $fullPost = $out['post_title'].$out['post_content'];
	1803	+ $fullPost .= (isset($out['post_excerpt']) ? $out['post_excerpt'] : '');
	1804	+ if (strlen($fullPost) < 1) :
571	1805	// FIXME: Option for filtering out empty posts
572	1806	endif;
573	1807	if (strlen($out['post_title'])==0) :
574	1808	$offset = (int) get_option('gmt_offset') * 60 * 60;
575		- $out['post_title'] =
576		- $this->post['meta']['syndication_source']
	1809	+ if (isset($this->post['meta']['syndication_source'])) :
	1810	+ $source_title = $this->post['meta']['syndication_source'];
	1811	+ else :
	1812	+ $feed_url = parse_url($this->post['meta']['syndication_feed']);
	1813	+ $source_title = $feed_url['host'];
	1814	+ endif;
	1815	+
	1816	+ $out['post_title'] = $source_title
577	1817	.' '.gmdate('Y-m-d H:i:s', $this->published() + $offset);
578	1818	// FIXME: Option for what to fill a blank title with...
579	1819	endif;
580	1820
	1821	+ // Normalize the guid if necessary.
	1822	+ $out['guid'] = SyndicatedPost::normalize_guid($out['guid']);
	1823	+
581	1824	return $out;
582	1825	}
583	1826
	1827	+ public function db_sanitize_post_check_encoding ($out) {
	1828	+ // Check encoding recursively: every string field needs to be checked
	1829	+ // for character encoding issues. This is a bit problematic because we
	1830	+ // *should* be using DB_CHARSET, but DB_CHARSET sometimes has values
	1831	+ // that work for MySQL but not for PHP mb_check_encoding. So instead
	1832	+ // we must rely on WordPress setting blog_charset and hope that the user
	1833	+ // has got their database encoding set up to roughly match
	1834	+ $charset = get_option('blog_charset', 'utf8');
	1835	+
	1836	+ foreach ($out as $key => $value) :
	1837	+ if (is_string($value)) :
	1838	+
	1839	+ if (!function_exists('mb_check_encoding') or mb_check_encoding($value, $charset)) :
	1840	+ $out[$key] = $value;
	1841	+ else :
	1842	+ $fromCharset = mb_detect_encoding($value, mb_detect_order(), /*strict=*/ true);
	1843	+ $out[$key] = mb_convert_encoding($value, $charset, $fromCharset);
	1844	+ endif;
	1845	+
	1846	+ elseif (is_array($value)) :
	1847	+ $out[$key] = $this->db_sanitize_post_check_encoding($value);
	1848	+
	1849	+ else :
	1850	+ $out[$key] = $value;
	1851	+ endif;
	1852	+
	1853	+ endforeach;
	1854	+
	1855	+ return $out;
	1856	+ } /* SyndicatedPost::db_sanitize_post_check_encoding () */
	1857	+
	1858	+ function db_sanitize_post ($out) {
	1859	+ global $wp_db_version;
	1860	+
	1861	+ $out = $this->db_sanitize_post_check_encoding($out);
	1862	+
	1863	+ // < 3.6. Core API, including `wp_insert_post()`, expects
	1864	+ // properly slashed data. If `wp_slash()` exists, then
	1865	+ // this is after the big change-over in how data slashing
	1866	+ // was handled.
	1867	+ if (!function_exists('wp_slash')) :
	1868	+
	1869	+ foreach ($out as $key => $value) :
	1870	+ if (is_string($value)) :
	1871	+ $out[$key] = esc_sql($value);
	1872	+ else :
	1873	+ $out[$key] = $value;
	1874	+ endif;
	1875	+ endforeach;
	1876	+
	1877	+ // For revisions [@23416,@23554), core API expects
	1878	+ // unslashed data. Cf. <https://core.trac.wordpress.org/browser/trunk/wp-includes/post.php?rev=23416>
	1879	+ // NOOP for those revisions.
	1880	+
	1881	+ // In revisions @23554 to present, `wp_insert_post()`
	1882	+ // expects slashed data once again.
	1883	+ // Cf. <https://core.trac.wordpress.org/changeset/23554/trunk/wp-includes/post.php?contextall=1>
	1884	+ // But at least now we can use the wp_slash API function to do that.
	1885	+ // Hooray.
	1886	+
	1887	+ elseif ($wp_db_version >= 23524) :
	1888	+
	1889	+ $out = wp_slash($out);
	1890	+
	1891	+ endif;
	1892	+
	1893	+ return $out;
	1894	+ }
	1895	+
584	1896	/**
585	1897	* SyndicatedPost::validate_post_id()
586	1898	*
587	1899	* @param array $dbpost An array representing the post we attempted to insert or update
		@@ -586,35 +1898,57 @@
586	1898	*
587	1899	* @param array $dbpost An array representing the post we attempted to insert or update
588	1900	* @param mixed $ns A string or array representing the namespace (class, method) whence this method was called.
589	1901	*/
590		- function validate_post_id ($dbpost, $ns) {
	1902	+ function validate_post_id ($dbpost, $is_update, $ns) {
591	1903	if (is_array($ns)) : $ns = implode('::', $ns);
592	1904	else : $ns = (string) $ns; endif;
593		-
	1905	+
594	1906	// This should never happen.
595	1907	if (!is_numeric($this->_wp_id) or ($this->_wp_id == 0)) :
596		- FeedWordPress::critical_bug(
597		- /*name=*/ $ns.'::_wp_id',
	1908	+ $verb = ($is_update ? 'update existing' : 'insert new');
	1909	+ $guid = $this->guid();
	1910	+ $url = $this->permalink();
	1911	+ $feed = $this->link->uri(array('add_params' => true));
	1912	+
	1913	+ // wp_insert_post failed. Diagnostics, or barf up a critical bug
	1914	+ // notice if we are in debug mode.
	1915	+ $mesg = "Failed to $verb item [$guid]. WordPress API returned no valid post ID.\n"
	1916	+ ."\t\tID = ".serialize($this->_wp_id)."\n"
	1917	+ ."\t\tURL = ".MyPHP::val($url)
	1918	+ ."\t\tFeed = ".MyPHP::val($feed);
	1919	+
	1920	+ FeedWordPress::diagnostic('updated_feeds:errors', "WordPress API error: $mesg");
	1921	+ FeedWordPress::diagnostic('feed_items:rejected', $mesg);
	1922	+
	1923	+ $mesg = <<<EOM
	1924	+The WordPress API returned an invalid post ID
	1925	+ when FeedWordPress tried to $verb item $guid
	1926	+ [URL: $url]
	1927	+ from the feed at $feed
	1928	+
	1929	+$ns::_wp_id
	1930	+EOM;
	1931	+ FeedWordPress::noncritical_bug(
	1932	+ /*message=*/ $mesg,
598	1933	/*var =*/ array(
599	1934	"\$this->_wp_id" => $this->_wp_id,
600	1935	"\$dbpost" => $dbpost,
601		- "\$this" => $this
602	1936	),
603		- /*line # =*/ __LINE__
	1937	+ /*line # =*/ __LINE__, /*filename=*/ __FILE__
604	1938	);
605	1939	endif;
606	1940	} /* SyndicatedPost::validate_post_id() */
607		-
	1941	+
608	1942	/**
609		- * SyndicatedPost::fix_revision_meta() - Fixes the way WP 2.6+ fucks up
610		- * meta-data (authorship, etc.) when storing revisions of an updated
611		- * syndicated post.
	1943	+ * SyndicatedPost::fix_revision_meta() - Ensures that we get the meta
	1944	+ * data (authorship, guid, etc.) that we want when storing revisions of
	1945	+ * a syndicated post.
612	1946	*
613		- * In their infinite wisdom, the WordPress coders have made it completely
614		- * impossible for a plugin that uses wp_insert_post() to set certain
615		- * meta-data (such as the author) when you store an old revision of an
616		- * updated post. Instead, it uses the WordPress defaults (= currently
	1947	+ * In their infinite wisdom, the WordPress coders seem to have made it
	1948	+ * completely impossible for a plugin that uses wp_insert_post() to set
	1949	+ * certain meta-data (such as the author) when you store an old revision
	1950	+ * of an updated post. Instead, it uses the WordPress defaults (= cur.
617	1951	* active user ID if the process is running with a user logged in, or
618	1952	* = #0 if there is no user logged in). This results in bogus authorship
619	1953	* data for revisions that are syndicated from off the feed, unless we
620	1954	* use a ridiculous kludge like this to end-run the munging of meta-data
		@@ -623,110 +1957,203 @@
623	1957	* @param int $revision_id The revision ID to fix up meta-data
624	1958	*/
625	1959	function fix_revision_meta ($revision_id) {
626	1960	global $wpdb;
	1961	+
	1962	+ $post_author = (int) $this->post['post_author'];
	1963	+
	1964	+ $revision_id = (int) $revision_id;
627	1965
628		- $post_author = (int) $this->post['post_author'];
	1966	+ // Let's fix the author.
	1967	+ set_post_field('post_author', $this->post['post_author'], $revision_id);
629	1968
630		- $revision_id = (int) $revision_id;
631		- $wpdb->query("
632		- UPDATE $wpdb->posts
633		- SET post_author={$this->post['post_author']}
634		- WHERE post_type = 'revision' AND ID='$revision_id'
635		- ");
	1969	+ // Let's fix the GUID to a dummy URL with the update hash.
	1970	+ set_post_field('guid', 'http://feedwordpress.radgeek.com/?rev='.$this->update_hash(), $revision_id);
	1971	+
	1972	+ // Let's fire an event for add-ons and filters
	1973	+ do_action('syndicated_post_fix_revision_meta', $revision_id, $this);
	1974	+
636	1975	} /* SyndicatedPost::fix_revision_meta () */
637	1976
638	1977	/**
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.
	1978	+ * SyndicatedPost::add_terms() -- if FeedWordPress is processing an
	1979	+ * automatic update, that generally means that wp_insert_post() is being
	1980	+ * called under the user credentials of whoever is viewing the blog at
	1981	+ * the time -- which usually means no user at all. But wp_insert_post()
	1982	+ * checks current_user_can() before assigning any of the terms in a
	1983	+ * post's tax_input structure -- which is unfortunate, since
	1984	+ * current_user_can() always returns FALSE when there is no current user
	1985	+ * logged in. Meaning that automatic updates get no terms assigned.
654	1986	*
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.
	1987	+ * So, wp_insert_post() is not going to do the term assignments for us.
	1988	+ * If you want something done right....
	1989	+ *
	1990	+ * @param string $new_status Unused action parameter.
	1991	+ * @param string $old_status Unused action parameter.
	1992	+ * @param object $post The database record for the post just inserted.
657	1993	*/
658		- function avoid_kses_munge ($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 $post['meta']. This is used to
666		- // store anything that the WordPress user might want to access from a
667		- // template concerning the post's original source that isn't provided
668		- // for by standard WP meta-data (i.e., any interesting data about the
669		- // syndicated post other than author, title, timestamp, categories, and
670		- // guid). It's also used to hook into WordPress's support for
671		- // enclosures.
672		- function add_rss_meta () {
	1994	+ function add_terms ($new_status, $old_status, $post) {
	1995	+
	1996	+ if ($new_status!='inherit') : // Bail if we are creating a revision.
	1997	+ if ( is_array($this->post) and isset($this->post['tax_input']) and is_array($this->post['tax_input']) ) :
	1998	+ foreach ($this->post['tax_input'] as $taxonomy => $terms) :
	1999	+ if (is_array($terms)) :
	2000	+ $terms = array_filter($terms); // strip out empties
	2001	+ endif;
	2002	+
	2003	+ $res = wp_set_post_terms(
	2004	+ /*post_id=*/ $post->ID,
	2005	+ /*terms=*/ $terms,
	2006	+ /*taxonomy=*/ $taxonomy
	2007	+ );
	2008	+
	2009	+ FeedWordPress::diagnostic(
	2010	+ 'syndicated_posts:categories',
	2011	+ 'Category: post('.json_encode($post->ID).') '.$taxonomy
	2012	+ .' := '
	2013	+ .json_encode($terms)
	2014	+ .' / result: '
	2015	+ .json_encode($res)
	2016	+ );
	2017	+
	2018	+ endforeach;
	2019	+ endif;
	2020	+ endif;
	2021	+
	2022	+ } /* SyndicatedPost::add_terms () */
	2023	+
	2024	+ /**
	2025	+ * SyndicatedPost::fix_post_modified_ts() -- We would like to set
	2026	+ * post_modified and post_modified_gmt to reflect the value of
	2027	+ * <atom:updated> or equivalent elements on the feed. Unfortunately,
	2028	+ * wp_insert_post() refuses to acknowledge explicitly-set post_modified
	2029	+ * fields and overwrites them, either with the post_date (if new) or the
	2030	+ * current timestamp (if updated).
	2031	+ *
	2032	+ * So, wp_insert_post() is not going to do the last-modified assignments
	2033	+ * for us. If you want something done right....
	2034	+ *
	2035	+ * @param string $new_status Unused action parameter.
	2036	+ * @param string $old_status Unused action parameter.
	2037	+ * @param object $post The database record for the post just inserted.
	2038	+ */
	2039	+ function fix_post_modified_ts ($new_status, $old_status, $post) {
673	2040	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		- ");
	2041	+ if ($new_status!='inherit') : // Bail if we are creating a revision.
	2042	+ $wpdb->update( $wpdb->posts, /*data=*/ array(
	2043	+ 'post_modified' => $this->post['post_modified'],
	2044	+ 'post_modified_gmt' => $this->post['post_modified_gmt'],
	2045	+ ), /*where=*/ array('ID' => $post->ID) );
	2046	+ endif;
	2047	+ } /* SyndicatedPost::fix_post_modified_ts () */
685	2048
686		- foreach ( $this->post['meta'] as $key => $values ) :
	2049	+ /**
	2050	+ * SyndicatedPost::add_rss_meta: adds interesting meta-data to each entry
	2051	+ * using the space for custom keys. The set of keys and values to add is
	2052	+ * specified by the keys and values of $post['meta']. This is used to
	2053	+ * store anything that the WordPress user might want to access from a
	2054	+ * template concerning the post's original source that isn't provided
	2055	+ * for by standard WP meta-data (i.e., any interesting data about the
	2056	+ * syndicated post other than author, title, timestamp, categories, and
	2057	+ * guid). It's also used to hook into WordPress's support for
	2058	+ * enclosures.
	2059	+ *
	2060	+ * @param string $new_status Unused action parameter.
	2061	+ * @param string $old_status Unused action parameter.
	2062	+ * @param object $post The database record for the post just inserted.
	2063	+ */
	2064	+ function add_rss_meta ($new_status, $old_status, $post) {
	2065	+ global $wpdb;
	2066	+ if ($new_status!='inherit') : // Bail if we are creating a revision.
	2067	+ FeedWordPress::diagnostic('syndicated_posts:meta_data', 'Adding post meta-data: {'.implode(", ", array_keys($this->post['meta'])).'}');
687	2068
688		- $key = $wpdb->escape($key);
	2069	+ if ( is_array($this->post) and isset($this->post['meta']) and is_array($this->post['meta']) ) :
	2070	+ $postId = $post->ID;
689	2071
690		- // If this is an update, clear out the old
691		- // values to avoid duplication.
	2072	+ // Aggregated posts should NOT send out pingbacks.
	2073	+ // WordPress 2.1-2.2 claim you can tell them not to
	2074	+ // using $post_pingback, but they don't listen, so we
	2075	+ // make sure here.
692	2076	$result = $wpdb->query("
693	2077	DELETE FROM $wpdb->postmeta
694		- WHERE post_id='$postId' AND meta_key='$key'
	2078	+ WHERE post_id='$postId' AND meta_key='_pingme'
695	2079	");
696	2080
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);
	2081	+ foreach ( $this->post['meta'] as $key => $values ) :
	2082	+ $eKey = esc_sql($key);
	2083	+
	2084	+ // If this is an update, clear out the old
	2085	+ // values to avoid duplication.
701	2086	$result = $wpdb->query("
702		- INSERT INTO $wpdb->postmeta
703		- SET
704		- post_id='$postId',
705		- meta_key='$key',
706		- meta_value='$value'
	2087	+ DELETE FROM $wpdb->postmeta
	2088	+ WHERE post_id='$postId' AND meta_key='$eKey'
707	2089	");
708		- if (!$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;
	2090	+
	2091	+ // Allow for either a single value or an array
	2092	+ if (!is_array($values)) $values = array($values);
	2093	+ foreach ( $values as $value ) :
	2094	+ FeedWordPress::diagnostic('syndicated_posts:meta_data', "Adding post meta-datum to post [$postId]: [$key] = ".MyPHP::val($value, /*no newlines=*/ true));
	2095	+ add_post_meta($postId, $key, $value, /*unique=*/ false);
	2096	+ endforeach;
714	2097	endforeach;
715		- endforeach;
	2098	+ endif;
716	2099	endif;
717	2100	} /* SyndicatedPost::add_rss_meta () */
718	2101
719		- // SyndicatedPost::author_id (): get the ID for an author name from
720		- // the feed. Create the author if necessary.
	2102	+ /**
	2103	+ * SyndicatedPost::author_id (): get the ID for an author name from
	2104	+ * the feed. Create the author if necessary.
	2105	+ *
	2106	+ * @param string $unfamiliar_author
	2107	+ *
	2108	+ * @return NULL|int The numeric ID of the author to attribute the post to
	2109	+ * NULL if the post should be filtered out.
	2110	+ */
721	2111	function author_id ($unfamiliar_author = 'create') {
722	2112	global $wpdb;
723	2113
724		- $a = $this->author();
725		- $author = $a['name'];
726		- $email = $a['email'];
727		- $url = $a['uri'];
	2114	+ $a = $this->named['author'];
728	2115
	2116	+ $source = $this->source();
	2117	+ $forbidden = apply_filters('feedwordpress_forbidden_author_names',
	2118	+ array('admin', 'administrator', 'www', 'root'));
	2119	+
	2120	+ // Prepare the list of candidates to try for author name: name from
	2121	+ // feed, original source title (if any), immediate source title live
	2122	+ // from feed, subscription title, prettied version of feed homepage URL,
	2123	+ // prettied version of feed URL, or, failing all, use "unknown author"
	2124	+ // as last resort
	2125	+
	2126	+ $candidates = array();
	2127	+ $candidates[] = $a['name'];
	2128	+ if (!is_null($source)) : $candidates[] = $source['title']; endif;
	2129	+ $candidates[] = $this->link->name(/*fromFeed=*/ true);
	2130	+ $candidates[] = $this->link->name(/*fromFeed=*/ false);
	2131	+ if (strlen($this->link->homepage()) > 0) : $candidates[] = feedwordpress_display_url($this->link->homepage()); endif;
	2132	+ $candidates[] = feedwordpress_display_url($this->link->uri());
	2133	+ $candidates[] = 'unknown author';
	2134	+
	2135	+ // Pick the first one that works from the list, screening against empty
	2136	+ // or forbidden names.
	2137	+
	2138	+ $author = NULL;
	2139	+ while (is_null($author) and ($candidate = each($candidates))) :
	2140	+ if (!is_null($candidate['value'])
	2141	+ and (strlen(trim($candidate['value'])) > 0)
	2142	+ and !in_array(strtolower(trim($candidate['value'])), $forbidden)) :
	2143	+ $author = $candidate['value'];
	2144	+ endif;
	2145	+ endwhile;
	2146	+
	2147	+ $email = (isset($a['email']) ? $a['email'] : NULL);
	2148	+ $authorUrl = (isset($a['uri']) ? $a['uri'] : NULL);
	2149	+
	2150	+
	2151	+ $hostUrl = $this->link->homepage();
	2152	+ if (is_null($hostUrl) or (strlen($hostUrl) < 0)) :
	2153	+ $hostUrl = $this->link->uri();
	2154	+ endif;
	2155	+
729	2156	$match_author_by_email = !('yes' == get_option("feedwordpress_do_not_match_author_by_email"));
730	2157	if ($match_author_by_email and !FeedWordPress::is_null_email($email)) :
731	2158	$test_email = $email;
732	2159	else :
		@@ -734,21 +2161,43 @@
734	2161	endif;
735	2162
736	2163	// Never can be too careful...
737	2164	$login = sanitize_user($author, /*strict=*/ true);
	2165	+
	2166	+ // Possible for, e.g., foreign script author names
	2167	+ if (strlen($login) < 1) :
	2168	+ // No usable characters in author name for a login.
	2169	+ // (Sometimes results from, e.g., foreign scripts.)
	2170	+ //
	2171	+ // We just need *something* in Western alphanumerics,
	2172	+ // so let's try the domain name.
	2173	+ //
	2174	+ // Uniqueness will be guaranteed below if necessary.
	2175	+
	2176	+ $url = parse_url($hostUrl);
	2177	+
	2178	+ $login = sanitize_user($url['host'], /*strict=*/ true);
	2179	+ if (strlen($login) < 1) :
	2180	+ // This isn't working. Frak it.
	2181	+ $login = 'syndicated';
	2182	+ endif;
	2183	+ endif;
	2184	+
738	2185	$login = apply_filters('pre_user_login', $login);
739	2186
740	2187	$nice_author = sanitize_title($author);
741	2188	$nice_author = apply_filters('pre_user_nicename', $nice_author);
742	2189
743		- $reg_author = $wpdb->escape(preg_quote($author));
744		- $author = $wpdb->escape($author);
745		- $email = $wpdb->escape($email);
746		- $test_email = $wpdb->escape($test_email);
747		- $url = $wpdb->escape($url);
	2190	+ $reg_author = esc_sql(preg_quote($author));
	2191	+ $author = esc_sql($author);
	2192	+ $email = esc_sql($email);
	2193	+ $test_email = esc_sql($test_email);
	2194	+ $authorUrl = esc_sql($authorUrl);
748	2195
749	2196	// Check for an existing author rule....
750		- if (isset($this->link->settings['map authors']['name'][strtolower(trim($author))])) :
	2197	+ if (isset($this->link->settings['map authors']['name']['*'])) :
	2198	+ $author_rule = $this->link->settings['map authors']['name']['*'];
	2199	+ elseif (isset($this->link->settings['map authors']['name'][strtolower(trim($author))])) :
751	2200	$author_rule = $this->link->settings['map authors']['name'][strtolower(trim($author))];
752	2201	else :
753	2202	$author_rule = NULL;
754	2203	endif;
		@@ -759,69 +2208,38 @@
759	2208
760	2209	// User name is filtered out
761	2210	elseif ('filter' == $author_rule) :
762	2211	$id = NULL;
763		-
	2212	+
764	2213	else :
765	2214	// Check the database for an existing author record that might fit
766	2215
767		- #-- WordPress 2.0+
768		- if (fwp_test_wp_version(FWP_SCHEMA_HAS_USERMETA)) :
	2216	+ // First try the user core data table.
	2217	+ $id = $wpdb->get_var(
	2218	+ "SELECT ID FROM $wpdb->users
	2219	+ WHERE TRIM(LCASE(display_name)) = TRIM(LCASE('$author'))
	2220	+ OR TRIM(LCASE(user_login)) = TRIM(LCASE('$author'))
	2221	+ OR (
	2222	+ LENGTH(TRIM(LCASE(user_email))) > 0
	2223	+ AND TRIM(LCASE(user_email)) = TRIM(LCASE('$test_email'))
	2224	+ )");
769	2225
770		- // First try the user core data table.
	2226	+ // If that fails, look for aliases in the user meta data table
	2227	+ if (is_null($id)) :
771	2228	$id = $wpdb->get_var(
772		- "SELECT ID FROM $wpdb->users
	2229	+ "SELECT user_id FROM $wpdb->usermeta
773	2230	WHERE
774		- TRIM(LCASE(user_login)) = TRIM(LCASE('$login'))
	2231	+ (meta_key = 'description' AND TRIM(LCASE(meta_value)) = TRIM(LCASE('$author')))
775	2232	OR (
776		- LENGTH(TRIM(LCASE(user_email))) > 0
777		- AND TRIM(LCASE(user_email)) = TRIM(LCASE('$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('$author')) OR
814		- (
815		- LOWER(user_description)
	2233	+ meta_key = 'description'
	2234	+ AND TRIM(LCASE(meta_value))
816	2235	RLIKE CONCAT(
817	2236	'(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*',
818		- LCASE('$reg_author'),
	2237	+ TRIM(LCASE('$reg_author')),
819	2238	'( |\\t|\\r)*(\\n|\$)'
820	2239	)
821	2240	)
822	2241	");
823		-
824	2242	endif;
825	2243
826	2244	// ... if you don't find one, then do what you need to do
827	2245	if (is_null($id)) :
		@@ -827,8 +2245,16 @@
827	2245	if (is_null($id)) :
828	2246	if ($unfamiliar_author === 'create') :
829	2247	$userdata = array();
830	2248
	2249	+ // WordPress 3 is going to pitch a fit if we attempt to register
	2250	+ // more than one user account with an empty e-mail address, so we
	2251	+ // need *something* here. Ugh.
	2252	+ if (strlen($email) == 0 or FeedWordPress::is_null_email($email)) :
	2253	+ $url = parse_url($hostUrl);
	2254	+ $email = $nice_author.'@'.$url['host'];
	2255	+ endif;
	2256	+
831	2257	#-- user table data
832	2258	$userdata['ID'] = NULL; // new user
833	2259	$userdata['user_login'] = $login;
834	2260	$userdata['user_nicename'] = $nice_author;
		@@ -833,12 +2259,43 @@
833	2259	$userdata['user_login'] = $login;
834	2260	$userdata['user_nicename'] = $nice_author;
835	2261	$userdata['user_pass'] = substr(md5(uniqid(microtime())), 0, 6); // just something random to lock it up
836	2262	$userdata['user_email'] = $email;
837		- $userdata['user_url'] = $url;
	2263	+ $userdata['user_url'] = $authorUrl;
	2264	+ $userdata['nickname'] = $author;
	2265	+ $parts = preg_split('/\s+/', trim($author), 2);
	2266	+ if (isset($parts[0])) : $userdata['first_name'] = $parts[0]; endif;
	2267	+ if (isset($parts[1])) : $userdata['last_name'] = $parts[1]; endif;
838	2268	$userdata['display_name'] = $author;
839		-
840		- $id = wp_insert_user($userdata);
	2269	+ $userdata['role'] = 'contributor';
	2270	+
	2271	+ do { // Keep trying until you get it right. Or until PHP crashes, I guess.
	2272	+ $id = wp_insert_user($userdata);
	2273	+ if (is_wp_error($id)) :
	2274	+ $codes = $id->get_error_code();
	2275	+ switch ($codes) :
	2276	+ case 'empty_user_login' :
	2277	+ case 'existing_user_login' :
	2278	+ // Add a random disambiguator
	2279	+ $userdata['user_login'] .= substr(md5(uniqid(microtime())), 0, 6);
	2280	+ break;
	2281	+ case 'user_nicename_too_long' :
	2282	+ // Add a limited 50 caracters user_nicename based on user_login
	2283	+ $userdata['user_nicename'] = mb_substr( $userdata['user_login'], 0, 50 );
	2284	+ break;
	2285	+ case 'existing_user_email' :
	2286	+ // No disassemble!
	2287	+ $parts = explode('@', $userdata['user_email'], 2);
	2288	+
	2289	+ // Add a random disambiguator as a gmail-style username extension
	2290	+ $parts[0] .= '+'.substr(md5(uniqid(microtime())), 0, 6);
	2291	+
	2292	+ // Reassemble
	2293	+ $userdata['user_email'] = $parts[0].'@'.$parts[1];
	2294	+ break;
	2295	+ endswitch;
	2296	+ endif;
	2297	+ } while (is_wp_error($id));
841	2298	elseif (is_numeric($unfamiliar_author) and get_userdata((int) $unfamiliar_author)) :
842	2299	$id = (int) $unfamiliar_author;
843	2300	elseif ($unfamiliar_author === 'default') :
844	2301	$id = 1;
		@@ -847,432 +2304,30 @@
847	2304	endif;
848	2305
849	2306	if ($id) :
850	2307	$this->link->settings['map authors']['name'][strtolower(trim($author))] = $id;
851		- endif;
852		- return $id;
853		- } // function SyndicatedPost::author_id ()
854	2308
855		- // look up (and create) category ids from a list of categories
856		- function category_ids ($cats, $unfamiliar_category = 'create', $tags_too = false) {
857		- global $wpdb;
858		-
859		- // We need to normalize whitespace because (1) trailing
860		- // whitespace can cause PHP and MySQL not to see eye to eye on
861		- // VARCHAR comparisons for some versions of MySQL (cf.
862		- // <http://dev.mysql.com/doc/mysql/en/char.html>), and (2)
863		- // because I doubt most people want to make a semantic
864		- // distinction between 'Computers' and 'Computers '
865		- $cats = array_map('trim', $cats);
866		-
867		- $tags = array();
868		-
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[] = $cat_id;
877		- endif;
878		- elseif (strlen($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'];
906		- endif;
907		- endif;
908		-
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.
927		- 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;
937		- endif;
938		- $cat_ids[] = $cat_id;
939		- endif;
940		- endif;
	2309	+ // Multisite: Check whether the author has been recorded
	2310	+ // on *this* blog before. If not, put her down as a
	2311	+ // Contributor for *this* blog.
	2312	+ $user = new WP_User((int) $id);
	2313	+ if (empty($user->roles)) :
	2314	+ $user->add_role('contributor');
941	2315	endif;
942		- endforeach;
943		-
944		- if ((count($cat_ids) == 0) and ($unfamiliar_category === 'filter')) :
945		- $cat_ids = NULL; // Drop the post
946		- else :
947		- $cat_ids = array_unique($cat_ids);
948	2316	endif;
949		-
950		- if ($tags_too) : $ret = array($cat_ids, $tags);
951		- else : $ret = $cat_ids;
952		- endif;
	2317	+ return $id;
	2318	+ } /* function SyndicatedPost::author_id () */
953	2319
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();
1010		- else :
1011		- $epoch = $default;
1012		- endif;
1013		- endif;
1014		-
1015		- return $epoch;
1016		- }
1017		- function updated ($fallback = true, $default = -1) {
1018		- $epoch = null;
1019		-
1020		- # As far as I know, only dcterms and Atom have reliable ways to
1021		- # specify when something was *modified* last. If neither is
1022		- # available, then we'll try to get the time of publication.
1023		- if (isset($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=*/ $default);
1033		- endif;
1034		-
1035		- # If everything failed, then default to the current time.
1036		- if (is_null($epoch)) :
1037		- if (-1 == $default) :
1038		- $epoch = time();
1039		- else :
1040		- $epoch = $default;
1041		- endif;
1042		- endif;
1043		-
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	2320	/**
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.
	2321	+ * category_ids: look up (and create) category ids from a list of
	2322	+ * categories
1144	2323	*
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)
	2324	+ * @param array $cats
	2325	+ * @param string $unfamiliar_category
	2326	+ * @param array|null $taxonomies
	2327	+ * @return array
1150	2328	*/
1151		- function isTaggedAs ($tag) {
1152		- $desiredTag = strtolower(trim($tag)); // Normalize case and whitespace
	2329	+ function category_ids ($cats, $unfamiliar_category = 'create', $taxonomies = NULL, $params = array()) {
	2330	+ return $this->link->category_ids($this, $cats, $unfamiliar_category, $taxonomies, $params);
	2331	+ } /* SyndicatedPost::category_ids () */
1153	2332
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
1278		-
	2333	+} /* class SyndicatedPost */