← All changes | syndicatedpost.class.php +2171 -925 2009.0707 → 2025.1211 View file →

		@@ -1,288 +1,1470 @@
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 2017.1018
	16	+ */
2	17	class SyndicatedPost {
	18	+ /** @var MagpieRSS|null MagpieRSS representation. */
3	19	var $item = null;
4		-
	20	+ /** @var SimplePie_Item|null SimplePie_Item representation. */
	21	+ var $entry = null;
	22	+
5	23	var $link = null;
6	24	var $feed = null;
7	25	var $feedmeta = null;
8		-
9		- var $post = array ();
10	26
	27	+ var $xmlns = array();
	28	+
	29	+ var $post = array();
	30	+
	31	+ var $named = array();
	32	+ var $preset_terms = array();
	33	+ var $feed_terms = array();
	34	+
11	35	var $_freshness = null;
12	36	var $_wp_id = null;
	37	+ var $_wp_post = null;
13	38
14		- function SyndicatedPost ($item, $link) {
15		- global $wpdb;
	39	+ /**
	40	+ * SyndicatedPost constructor: Given a feed item and the source from
	41	+ * which it was taken, prepare a post that can be inserted into the
	42	+ * WordPress database on request, or updated in place if it has already
	43	+ * been syndicated.
	44	+ *
	45	+ * @param array $item The item syndicated from the feed.
	46	+ * @param SyndicatedLink $source The feed it was syndicated from.
	47	+ */
	48	+ public function __construct( $item, $source ) {
	49	+ if ( empty( $item ) and empty( $source ) )
	50	+ return;
16	51
17		- $this->link = $link;
18		- $feedmeta = $link->settings;
19		- $feed = $link->magpie;
	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;
20	60
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;
	61	+ // convert to Magpie for compat purposes
	62	+ $mpie = new MagpieFromSimplePie( $source->simplepie, $this->entry );
	63	+ $this->item = $mpie->get_item();
30	64
31		- $this->feed = $feed;
32		- $this->feedmeta = $feedmeta;
	65	+ // done with conversion object
	66	+ $mpie = NULL; unset( $mpie );
	67	+ else :
	68	+ $this->item = $item;
	69	+ endif;
33	70
34		- $this->item = $item;
35		- $this->item = apply_filters('syndicated_item', $this->item, $this);
	71	+ $this->link = $source;
	72	+ $this->feed = $source->magpie;
	73	+ $this->feedmeta = $source->settings;
36	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 = $changed;
	101	+
37	102	# Filters can halt further processing by returning NULL
38		- if (is_null($this->item)) :
	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(),
	115	+ $this
	116	+ );
47	117
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);
	118	+ $this->named['author'] = apply_filters(
	119	+ 'syndicated_item_author',
	120	+ $this->author(),
	121	+ $this
	122	+ );
	123	+ // This just gives us an alphanumeric name for the author.
	124	+ // We look up (or create) the numeric ID for the author
	125	+ // in SyndicatedPost::add().
52	126
53		- # Identify content and sanitize it.
54		- # ---------------------------------
55		- if (isset($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);
	127	+ $this->post['post_content'] = apply_filters(
	128	+ 'syndicated_item_content',
	129	+ $this->content(),
	130	+ $this
	131	+ );
67	132
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);
	133	+ $excerpt = apply_filters( 'syndicated_item_excerpt', $this->excerpt(), $this );
79	134
80		- if (!is_null($excerpt)):
	135	+ if ( !empty( $excerpt ) ):
81	136	$this->post['post_excerpt'] = $excerpt;
82	137	endif;
83		-
84		- // This is unnecessary if we use wp_insert_post
85		- if (!$this->use_api('wp_insert_post')) :
86		- $this->post['post_name'] = sanitize_title($this->post['post_title']);
87		- endif;
88	138
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);
	139	+ // Dealing with timestamps in WordPress is so fucking fucked.
	140	+ $offset = (int) get_option( 'gmt_offset' ) * 60 * 60;
	141	+ $post_date_gmt = $this->published( array( 'default' => -1 ) );
	142	+ $post_modified_gmt = $this->updated( array( 'default' => -1 ) );
92	143
93		- // Dealing with timestamps in WordPress is so fucking fucked.
94		- $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());
	144	+ $this->post['post_date_gmt'] = gmdate( 'Y-m-d H:i:s', $post_date_gmt );
	145	+ $this->post['post_date'] = gmdate( 'Y-m-d H:i:s', $post_date_gmt + $offset );
	146	+ $this->post['post_modified_gmt'] = gmdate( 'Y-m-d H:i:s', $post_modified_gmt );
	147	+ $this->post['post_modified'] = gmdate( 'Y-m-d H:i:s', $post_modified_gmt + $offset );
99	148
100	149	// Use feed-level preferences or the global default.
101		- $this->post['post_status'] = $this->link->syndicated_status('post', 'publish');
102		- $this->post['comment_status'] = $this->link->syndicated_status('comment', 'closed');
103		- $this->post['ping_status'] = $this->link->syndicated_status('ping', 'closed');
	150	+ $this->post['post_status'] = $this->link->syndicated_status( 'post', 'publish' );
	151	+ $this->post['comment_status'] = $this->link->syndicated_status( 'comment', 'closed' );
	152	+ $this->post['ping_status'] = $this->link->syndicated_status( 'ping', 'closed' );
104	153
105	154	// Unique ID (hopefully a unique tag: URI); failing that, the permalink
106		- $this->post['guid'] = apply_filters('syndicated_item_guid', $this->guid(), $this);
	155	+ $this->post['guid'] = apply_filters( 'syndicated_item_guid', $this->guid(), $this );
107	156
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) :
111		- $default_custom_settings = unserialize($default_custom_settings);
	157	+ // User-supplied custom settings to apply to each post.
	158	+ // Do first so that FWP-generated custom settings will
	159	+ // overwrite if necessary; thus preventing any munging.
	160	+ $postMetaIn = $this->link->postmeta( array( "parsed" => true ) );
	161	+ $postMetaOut = array();
	162	+
	163	+ foreach ( $postMetaIn as $key => $meta ) :
	164	+ $postMetaOut[ $key ] = $meta->do_substitutions( $this );
	165	+ endforeach;
	166	+
	167	+ foreach ( $postMetaOut as $key => $values ) :
	168	+ if ( is_null( $values ) ) { // have chosen to replace value with empty string
	169	+ $values = [ '' ];
	170	+ }
	171	+ $this->post['meta'][ $key ] = array();
	172	+ foreach ( $values as $value ) :
	173	+ $this->post['meta'][ $key ][] = apply_filters( "syndicated_post_meta_{$key}", $value, $this );
	174	+ endforeach;
	175	+ endforeach;
	176	+
	177	+ // RSS 2.0 / Atom 1.0 enclosure support
	178	+ $enclosures = $this->entry->get_enclosures();
	179	+ if ( is_array( $enclosures ) ) : foreach ( $enclosures as $enclosure ) :
	180	+ $this->post['meta']['enclosure'][] =
	181	+ apply_filters( 'syndicated_item_enclosure_url', $enclosure->get_link(), $this ) . "\n".
	182	+ apply_filters( 'syndicated_item_enclosure_length', $enclosure->get_length(), $this ) . "\n".
	183	+ apply_filters( 'syndicated_item_enclosure_type', $enclosure->get_type(), $this );
	184	+ endforeach; endif;
	185	+
	186	+ // In case you want to point back to the blog this was
	187	+ // syndicated from.
	188	+
	189	+ $sourcemeta['syndication_source'] = apply_filters(
	190	+ 'syndicated_item_source_title',
	191	+ $this->link->name(),
	192	+ $this
	193	+ );
	194	+ $sourcemeta['syndication_source_uri'] = apply_filters(
	195	+ 'syndicated_item_source_link',
	196	+ $this->link->homepage(),
	197	+ $this
	198	+ );
	199	+ $sourcemeta['syndication_source_id'] = apply_filters(
	200	+ 'syndicated_item_source_id',
	201	+ $this->link->guid(),
	202	+ $this
	203	+ );
	204	+
	205	+ // Make use of atom:source data, if present in an aggregated feed
	206	+ $entry_source = $this->source();
	207	+ if ( ! is_null( $entry_source ) ) :
	208	+ foreach ( $entry_source as $what => $value ) :
	209	+ if ( ! is_null( $value ) ) :
	210	+ if ( 'title' == $what ) : $key = 'syndication_source';
	211	+ elseif ( 'feed' == $what ) : $key = 'syndication_feed';
	212	+ else : $key = "syndication_source_{$what}";
	213	+ endif;
	214	+
	215	+ $sourcemeta["{$key}_original"] = apply_filters(
	216	+ 'syndicated_item_original_source_' . $what,
	217	+ $value,
	218	+ $this
	219	+ );
	220	+ endif;
	221	+ endforeach;
112	222	endif;
113		- if (!is_array($default_custom_settings)) :
114		- $default_custom_settings = array();
	223	+
	224	+ foreach ( $sourcemeta as $meta_key => $value ) :
	225	+ if ( !is_null( $value ) ) :
	226	+ $this->post['meta'][ $meta_key ] = $value;
	227	+ endif;
	228	+ endforeach;
	229	+
	230	+ // Store information on human-readable and machine-readable comment URIs
	231	+
	232	+ // Human-readable comment URI
	233	+ $commentLink = apply_filters( 'syndicated_item_comments', $this->comment_link(), $this );
	234	+ if ( ! is_null( $commentLink) ) : $this->post['meta']['rss:comments'] = $commentLink; endif;
	235	+
	236	+ // Machine-readable content feed URI
	237	+ $commentFeed = apply_filters( 'syndicated_item_commentrss', $this->comment_feed(), $this );
	238	+ if ( ! is_null( $commentFeed ) ) : $this->post['meta']['wfw:commentRSS'] = $commentFeed; endif;
	239	+ // Yeah, yeah, now I know that it's supposed to be
	240	+ // wfw:commentRss. Oh well. Path dependence, sucka.
	241	+
	242	+ // Store information to identify the feed that this came from
	243	+ if ( isset( $this->feedmeta['link/uri'] ) ) :
	244	+ $this->post['meta']['syndication_feed'] = $this->feedmeta['link/uri'];
115	245	endif;
116		-
117		- $custom_settings = (isset($this->link->settings['postmeta']) ? $this->link->settings['postmeta'] : null);
118		- if ($custom_settings) :
119		- $custom_settings = unserialize($custom_settings);
	246	+ if ( isset( $this->feedmeta['link/id'] ) ) :
	247	+ $this->post['meta']['syndication_feed_id'] = $this->feedmeta['link/id'];
120	248	endif;
121		- if (!is_array($custom_settings)) :
122		- $custom_settings = array();
	249	+
	250	+ if ( isset( $this->item['source_link_self'] ) ) :
	251	+ $this->post['meta']['syndication_feed_original'] = $this->item['source_link_self'];
123	252	endif;
124		- $this->post['meta'] = array_merge($default_custom_settings, $custom_settings);
125	253
126		- // RSS 2.0 / Atom 1.0 enclosure support
127		- if ( isset($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;
	254	+ // In case you want to know the external permalink...
	255	+ $this->post['meta']['syndication_permalink'] = apply_filters( 'syndicated_item_link', $this->permalink() );
	256	+
	257	+ // Store a hash of the post content for checking whether something needs to be updated
	258	+ $this->post['meta']['syndication_item_hash'] = $this->update_hash();
	259	+
	260	+ // Categories, Tags, and other Terms: from settings assignments (global settings, subscription settings),
	261	+ // and from feed assignments (item metadata, post content)
	262	+ $this->preset_terms = apply_filters( 'syndicated_item_preset_terms', $this->get_terms_from_settings(), $this );
	263	+ $this->feed_terms = apply_filters( 'syndicated_item_feed_terms', $this->get_terms_from_feeds(), $this );
	264	+
	265	+ $this->post['post_type'] = apply_filters(
	266	+ 'syndicated_post_type',
	267	+ $this->link->setting( 'syndicated post type', 'syndicated_post_type', 'post' ),
	268	+ $this
	269	+ );
	270	+ endif;
	271	+ } /* SyndicatedPost::__construct() */
	272	+
	273	+ #####################################
	274	+ #### EXTRACT DATA FROM FEED ITEM ####
	275	+ #####################################
	276	+
	277	+ function substitution_function ($name) {
	278	+ $ret = NULL;
	279	+
	280	+ switch ($name) :
	281	+ // Allowed PHP string functions
	282	+ case 'trim':
	283	+ case 'ltrim':
	284	+ case 'rtrim':
	285	+ case 'strtoupper':
	286	+ case 'strtolower':
	287	+ case 'urlencode':
	288	+ case 'urldecode':
	289	+ $ret = $name;
	290	+ endswitch;
	291	+ return $ret;
	292	+ }
	293	+
	294	+ /**
	295	+ * SyndicatedPost::query uses an XPath-like syntax to query arbitrary
	296	+ * elements within the syndicated item.
	297	+ *
	298	+ * @param string $path
	299	+ * @returns array of string values representing contents of matching
	300	+ * elements or attributes
	301	+ */
	302	+ public function query ($path) {
	303	+ $xq = new SyndicatedPostXPathQuery(array("path" => $path));
	304	+
	305	+ $feedChannel = array_merge(
	306	+ $this->get_feed_root_element(),
	307	+ $this->get_feed_channel_elements()
	308	+ );
	309	+
	310	+ $matches = $xq->match(array(
	311	+ "type" => $this->link->simplepie->get_type(),
	312	+ "xmlns" => $this->xmlns,
	313	+ "map" => array(
	314	+ "/" => array($this->entry->data),
	315	+ "item" => array($this->entry->data),
	316	+ "feed" => $feedChannel,
	317	+ "channel" => $feedChannel
	318	+ ),
	319	+ "context" => $this->entry->data,
	320	+ "parent" => $feedChannel,
	321	+ ));
	322	+
	323	+ return $matches;
	324	+ } /* SyndicatedPost::query() */
	325	+
	326	+ function get_feed_root_element () {
	327	+ $matches = array();
	328	+ foreach ($this->link->simplepie->data['child'] as $ns => $root) :
	329	+ foreach ($root as $element => $data) :
	330	+ $matches = array_merge($matches, $data);
	331	+ endforeach;
	332	+ endforeach;
	333	+ return $matches;
	334	+ } /* SyndicatedPost::get_feed_root_element() */
	335	+
	336	+ function get_feed_channel_elements () {
	337	+ $rss = array(
	338	+ SIMPLEPIE_NAMESPACE_RSS_090,
	339	+ SIMPLEPIE_NAMESPACE_RSS_10,
	340	+ 'http://backend.userland.com/RSS2',
	341	+ SIMPLEPIE_NAMESPACE_RSS_20,
	342	+ );
	343	+
	344	+ $matches = array();
	345	+ foreach ($rss as $ns) :
	346	+ $data = $this->link->simplepie->get_feed_tags($ns, 'channel');
	347	+ if ( !is_null($data)) :
	348	+ $matches = array_merge($matches, $data);
135	349	endif;
	350	+ endforeach;
	351	+ return $matches;
	352	+ } /* SyndicatedPost::get_feed_channel_elements() */
136	353
137		- // In case you want to point back to the blog this was syndicated from
138		- if (isset($this->feed->channel['title'])) :
139		- $this->post['meta']['syndication_source'] = apply_filters('syndicated_item_source_title', $this->feed->channel['title'], $this);
	354	+ public function get_categories ($params = array()) {
	355	+ return $this->entry->get_categories();
	356	+ }
	357	+
	358	+ public function title ($params = array()) {
	359	+ return $this->entry->get_title();
	360	+ } /* SyndicatedPost::title () */
	361	+
	362	+ public function content ($params = array())
	363	+ {
	364	+ $params = wp_parse_args($params, array(
	365	+ "full only" => false,
	366	+ ));
	367	+
	368	+ $content = NULL;
	369	+
	370	+ // FIXME: This is one of the main places in the code still using
	371	+ // the outmoded SimplePie - to - Magpie construction. We could
	372	+ // replace using SimplePie_Item::get_tags() here. (Or if really
	373	+ // ambitious we could attempt to just use
	374	+ // SimplePie_Item::get_content() with content-only set to TRUE
	375	+ // and some sanitization in effect. -CJ 1jul14
	376	+
	377	+ // atom:content, standard method of providing full HTML content
	378	+ // in Atom feeds.
	379	+ if (isset($this->item['atom_content'])) :
	380	+ $content = $this->item['atom_content'];
	381	+ elseif (isset($this->item['atom']['atom_content'])) :
	382	+ $content = $this->item['atom']['atom_content'];
	383	+
	384	+ // Some exotics: back in the day, before widespread convergence
	385	+ // on content:encoding, some RSS feeds took advantage of XML
	386	+ // namespaces to use an inline xhtml:body or xhtml:div element
	387	+ // for full-content HTML. (E.g. Sam Ruby's feed, IIRC.)
	388	+ elseif (isset($this->item['xhtml']['body'])) :
	389	+ $content = $this->item['xhtml']['body'];
	390	+ elseif (isset($this->item['xhtml']['div'])) :
	391	+ $content = $this->item['xhtml']['div'];
	392	+
	393	+ // content:encoded, most common method of providing full HTML in
	394	+ // RSS 2.0 feeds.
	395	+ elseif (isset($this->item['content']['encoded']) and $this->item['content']['encoded']):
	396	+ $content = $this->item['content']['encoded'];
	397	+
	398	+ // Fall back on elements that sometimes may contain full HTML
	399	+ // but sometimes not.
	400	+ elseif ( ! $params['full only']) :
	401	+
	402	+ // description element is sometimes used for full HTML
	403	+ // sometimes for summary text in RSS. (By the letter of
	404	+ // the standard, it should just be for summary text.)
	405	+ if (isset($this->item['description'])) :
	406	+ $content = $this->item['description'];
140	407	endif;
141	408
142		- if (isset($this->feed->channel['link'])) :
143		- $this->post['meta']['syndication_source_uri'] = apply_filters('syndicated_item_source_link', $this->feed->channel['link'], $this);
	409	+ endif;
	410	+
	411	+ return $content;
	412	+ } /* SyndicatedPost::content() */
	413	+
	414	+ public function excerpt () {
	415	+ # Identify and sanitize excerpt: atom:summary, or rss:description
	416	+ $excerpt = $this->entry->get_description();
	417	+
	418	+ # Many RSS feeds use rss:description, inadvisably, to
	419	+ # carry the entire post (typically with escaped HTML).
	420	+ # If that's what happened, we don't want the full
	421	+ # content for the excerpt.
	422	+ $content = $this->content();
	423	+
	424	+ // Ignore whitespace, case, and tag cruft.
	425	+ $theExcerpt = preg_replace('/\s+/', '', strtolower(strip_tags(html_entity_decode($excerpt))));
	426	+ $theContent = preg_replace('/\s+/', '', strtolower(strip_tags(html_entity_decode($content))));
	427	+ if ( empty($excerpt) or $theExcerpt == $theContent ) :
	428	+ # If content is available, generate an excerpt.
	429	+ if ( strlen(trim($content)) > 0 ) :
	430	+ $excerpt = strip_tags($content);
	431	+ if (strlen($excerpt) > 255) :
	432	+ if (is_object($this->link) and is_object($this->link->simplepie)) :
	433	+ $encoding = $this->link->simplepie->get_encoding();
	434	+ else :
	435	+ $encoding = get_option('blog_charset', 'utf8');
	436	+ endif;
	437	+ $excerpt = mb_substr($excerpt,0,252,$encoding).'...';
	438	+ endif;
144	439	endif;
145		-
146		- // Make use of atom:source data, if present in an aggregated feed
147		- if (isset($this->item['source_title'])) :
148		- $this->post['meta']['syndication_source_original'] = $this->item['source_title'];
	440	+ endif;
	441	+
	442	+ return $excerpt;
	443	+ } /* SyndicatedPost::excerpt() */
	444	+
	445	+ /**
	446	+ * SyndicatedPost::permalink: returns the permalink for the post, as provided by the
	447	+ * source feed.
	448	+ *
	449	+ * @return string The URL of the original permalink for this syndicated post
	450	+ */
	451	+ public function permalink () {
	452	+ // Handles explicit <link> elements and also RSS 2.0 cases with
	453	+ // <guid isPermaLink="true">, etc. Hooray!
	454	+ $permalink = $this->entry->get_link();
	455	+ return $permalink;
	456	+ } /* SyndicatedPost::permalink () */
	457	+
	458	+ public function created ($params = array()) {
	459	+ $unfiltered = false; // $default = NULL; // seems to be unused on this function (gwyneth 20230916)
	460	+ extract($params);
	461	+
	462	+ $date = '';
	463	+ if (isset($this->item['dc']['created'])) :
	464	+ $date = $this->item['dc']['created'];
	465	+ elseif (isset($this->item['dcterms']['created'])) :
	466	+ $date = $this->item['dcterms']['created'];
	467	+ elseif (isset($this->item['created'])): // Atom 0.3
	468	+ $date = $this->item['created'];
	469	+ endif;
	470	+
	471	+ $time = new FeedTime($date);
	472	+ $tstamp = $time->timestamp();
	473	+ if ( ! $unfiltered) :
	474	+ apply_filters('syndicated_item_created', $tstamp, $this);
	475	+ endif;
	476	+ return $tstamp;
	477	+ } /* SyndicatedPost::created() */
	478	+
	479	+ public function published ($params = array(), $default = NULL) {
	480	+ $fallback = true; $unfiltered = false;
	481	+ if ( !is_array($params)) : // Old style
	482	+ $fallback = $params;
	483	+ else : // New style
	484	+ extract($params);
	485	+ endif;
	486	+
	487	+ $date = '';
	488	+ $tstamp = null;
	489	+
	490	+ # RSS is a fucking mess. Figure out whether we have a date in
	491	+ # <dc:date>, <issued>, <pubDate>, etc., and get it into Unix
	492	+ # epoch format for reformatting. If we can't find anything,
	493	+ # we'll use the last-updated time.
	494	+ if (isset($this->item['dc']['date'])): // Dublin Core
	495	+ $date = $this->item['dc']['date'];
	496	+ elseif (isset($this->item['dcterms']['issued'])) : // Dublin Core extensions
	497	+ $date = $this->item['dcterms']['issued'];
	498	+ elseif (isset($this->item['published'])) : // Atom 1.0
	499	+ $date = $this->item['published'];
	500	+ elseif (isset($this->item['issued'])): // Atom 0.3
	501	+ $date = $this->item['issued'];
	502	+ elseif (isset($this->item['pubdate'])): // RSS 2.0
	503	+ $date = $this->item['pubdate'];
	504	+ endif;
	505	+
	506	+ if (strlen($date) > 0) :
	507	+ $time = new FeedTime($date);
	508	+ $tstamp = $time->timestamp();
	509	+ elseif ($fallback) : // Fall back to <updated> / <modified> if present
	510	+ $tstamp = $this->updated(/*fallback=*/ false, /*default=*/ $default);
	511	+ endif;
	512	+
	513	+ # If everything failed, then default to the current time.
	514	+ if (is_null($tstamp)) :
	515	+ if (-1 == $default) :
	516	+ $tstamp = time();
	517	+ else :
	518	+ $tstamp = $default;
149	519	endif;
	520	+ endif;
150	521
151		- if (isset($this->item['source_link'])) :
152		- $this->post['meta']['syndication_source_uri_original'] = $this->item['source_link'];
	522	+ if ( ! $unfiltered) :
	523	+ $tstamp = apply_filters('syndicated_item_published', $tstamp, $this);
	524	+ endif;
	525	+ return $tstamp;
	526	+ } /* SyndicatedPost::published() */
	527	+
	528	+ public function updated ($params = array(), $default = -1) {
	529	+ $fallback = true; $unfiltered = false;
	530	+ if ( !is_array($params)) : // Old style
	531	+ $fallback = $params;
	532	+ else : // New style
	533	+ extract($params);
	534	+ endif;
	535	+
	536	+ $date = '';
	537	+ $tstamp = null;
	538	+
	539	+ # As far as I know, only dcterms and Atom have reliable ways to
	540	+ # specify when something was *modified* last. If neither is
	541	+ # available, then we'll try to get the time of publication.
	542	+ if (isset($this->item['dc']['modified'])) : // Not really correct
	543	+ $date = $this->item['dc']['modified'];
	544	+ elseif (isset($this->item['dcterms']['modified'])) : // Dublin Core extensions
	545	+ $date = $this->item['dcterms']['modified'];
	546	+ elseif (isset($this->item['modified'])): // Atom 0.3
	547	+ $date = $this->item['modified'];
	548	+ elseif (isset($this->item['updated'])): // Atom 1.0
	549	+ $date = $this->item['updated'];
	550	+ endif;
	551	+
	552	+ if (strlen($date) > 0) :
	553	+ $time = new FeedTime($date);
	554	+ $tstamp = $time->timestamp();
	555	+ elseif ($fallback) : // Fall back to issued / dc:date
	556	+ $tstamp = $this->published(/*fallback=*/ false, /*default=*/ $default);
	557	+ endif;
	558	+
	559	+ # If everything failed, then default to the current time.
	560	+ if (is_null($tstamp)) :
	561	+ if (-1 == $default) :
	562	+ $tstamp = time();
	563	+ else :
	564	+ $tstamp = $default;
153	565	endif;
154		-
155		- if (isset($this->item['source_id'])) :
156		- $this->post['meta']['syndication_source_id_original'] = $this->item['source_id'];
	566	+ endif;
	567	+
	568	+ if ( ! $unfiltered) :
	569	+ $tstamp = apply_filters('syndicated_item_updated', $tstamp, $this);
	570	+ endif;
	571	+ return $tstamp;
	572	+ } /* SyndicatedPost::updated() */
	573	+
	574	+ var $_hashes = array();
	575	+ function stored_hashes ($id = NULL) {
	576	+ if (is_null($id)) :
	577	+ $id = $this->wp_id();
	578	+ endif;
	579	+
	580	+ if ( !isset($this->_hashes[$id])) :
	581	+ $this->_hashes[$id] = get_post_custom_values(
	582	+ 'syndication_item_hash', $id
	583	+ );
	584	+ if (is_null($this->_hashes[$id])) :
	585	+ $this->_hashes[$id] = array();
157	586	endif;
	587	+ endif;
	588	+ return $this->_hashes[$id];
	589	+ }
158	590
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', $this->item['comments']);
	591	+ function update_hash ($hashed = true) {
	592	+ // Basis for tracking possible changes to item.
	593	+ $hash = array(
	594	+ "title" => $this->entry->get_title(),
	595	+ "link" => $this->permalink(),
	596	+ "content" => $this->content(),
	597	+ "excerpt" => $this->excerpt(),
	598	+ );
	599	+
	600	+ if ($hashed) :
	601	+ $hash = md5(serialize($hash));
	602	+ endif;
	603	+
	604	+ return $hash;
	605	+ } /* SyndicatedPost::update_hash() */
	606	+
	607	+ /**
	608	+ * SyndicatedPost::normalize_guid_prefix(): generates a normalized URL
	609	+ * prefix (including scheme, authority, full path, and the beginning of
	610	+ * a query string) for creating guids that conform to WordPress's
	611	+ * internal constraints on the URL space for valid guids. To create a
	612	+ * normalized guid, just concatenate a valid URL query parameter value
	613	+ * to the returned URL.
	614	+ *
	615	+ * @return string The URL prefix generated.
	616	+ *
	617	+ * @uses trailingslashit()
	618	+ * @uses home_url()
	619	+ * @uses apply_filters()
	620	+ */
	621	+ static function normalize_guid_prefix () {
	622	+ $url = trailingslashit(home_url(/*path=*/ '', /*scheme=*/ 'http'));
	623	+ return apply_filters('syndicated_item_guid_normalized_prefix', $url . '?guid=');
	624	+ } /* SyndicatedPost::normalize_guid_prefix() */
	625	+
	626	+ static function normalize_guid ($guid) {
	627	+ $guid = trim($guid);
	628	+ if (preg_match('/^[0-9a-z]{32}$/i', $guid)) : // MD5
	629	+ $guid = SyndicatedPost::normalize_guid_prefix().strtolower($guid);
	630	+ elseif ((strlen(esc_url($guid)) == 0) or (esc_url($guid) != $guid)) :
	631	+ $guid = SyndicatedPost::normalize_guid_prefix().md5($guid);
	632	+ endif;
	633	+ $guid = trim($guid);
	634	+
	635	+ return $guid;
	636	+ } /* SyndicatedPost::normalize_guid() */
	637	+
	638	+ static function alternative_guid_prefix () {
	639	+ $url = trailingslashit(home_url(/*path=*/ '', /*scheme=*/ 'https'));
	640	+ return apply_filters('syndicated_item_guid_normalized_prefix', $url . '?guid=');
	641	+ }
	642	+ static function alternative_guid ($guid) {
	643	+ $guid = trim($guid);
	644	+ if (preg_match('/^[0-9a-z]{32}$/i', $guid)) : // MD5
	645	+ $guid = SyndicatedPost::alternative_guid_prefix().strtolower($guid);
	646	+ elseif ((strlen(esc_url($guid)) == 0) or (esc_url($guid) != $guid)) :
	647	+ $guid = SyndicatedPost::alternative_guid_prefix().md5($guid);
	648	+ endif;
	649	+ $guid = trim($guid);
	650	+
	651	+ return $guid;
	652	+ } /* SyndicatedPost::normalize_guid() */
	653	+
	654	+ public function guid () {
	655	+ $guid = null;
	656	+ if (isset($this->item['id'])): // Atom 0.3 / 1.0
	657	+ $guid = $this->item['id'];
	658	+ elseif (isset($this->item['atom']['id'])) : // Namespaced Atom
	659	+ $guid = $this->item['atom']['id'];
	660	+ elseif (isset($this->item['guid'])) : // RSS 2.0
	661	+ $guid = $this->item['guid'];
	662	+ elseif (isset($this->item['dc']['identifier'])) : // yeah, right
	663	+ $guid = $this->item['dc']['identifier'];
	664	+ endif;
	665	+
	666	+ // Un-set or too long to use as-is. Generate a tag: URI.
	667	+ if (is_null($guid) or strlen($guid) > 250) :
	668	+ // In case we need to check this again
	669	+ $original_guid = $guid;
	670	+
	671	+ // The feed does not seem to have provided us with a
	672	+ // usable unique identifier, so we'll have to cobble
	673	+ // together a tag: URI that might work for us. The base
	674	+ // of the URI will be the host name of the feed source ...
	675	+ $bits = parse_url($this->link->uri());
	676	+ $guid = 'tag:'.$bits['host'];
	677	+
	678	+ // Some ill-mannered feeds (for example, certain feeds
	679	+ // coming from Google Calendar) have extraordinarily long
	680	+ // guids -- so long that they exceed the 255 character
	681	+ // width of the WordPress guid field. But if the string
	682	+ // gets clipped by MySQL, uniqueness tests will fail
	683	+ // forever after and the post will be endlessly
	684	+ // reduplicated. So, instead, Guids Of A Certain Length
	685	+ // are hashed down into a nice, manageable tag: URI.
	686	+ if ( !is_null($original_guid)) :
	687	+ $guid .= ',2010-12-03:id.'.md5($original_guid);
	688	+
	689	+ // If we have a date of creation, then we can use that
	690	+ // to uniquely identify the item. (On the other hand, if
	691	+ // the feed producer was consicentious enough to
	692	+ // generate dates of creation, she probably also was
	693	+ // conscientious enough to generate unique identifiers.)
	694	+ elseif ( !is_null($this->created())) :
	695	+ $guid .= '://post.'.date('YmdHis', $this->created());
	696	+
	697	+ // Otherwise, use both the URI of the item, *and* the
	698	+ // item's title. We have to use both because titles are
	699	+ // often not unique, and sometimes links aren't unique
	700	+ // either (e.g. Bitch (S)HITLIST, Mozilla Dot Org news,
	701	+ // some podcasts). But it's rare to have *both* the same
	702	+ // title *and* the same link for two different items. So
	703	+ // this is about the best we can do.
	704	+ else :
	705	+ $link = $this->permalink();
	706	+ if (is_null($link)) : $link = $this->link->uri(); endif;
	707	+ $guid .= '://'.md5($link.'/'.$this->title());
162	708	endif;
163		- if (isset($this->item['wfw']['commentrss'])) :
164		- $this->post['meta']['wfw:commentRSS'] = apply_filters('syndicated_item_commentrss', $this->item['wfw']['commentrss']);
	709	+ endif;
	710	+ return $guid;
	711	+ } /* SyndicatedPost::guid() */
	712	+
	713	+ public function author () {
	714	+ $author = array ();
	715	+
	716	+ $aa = $this->entry->get_authors();
	717	+ if (is_countable($aa) and count($aa) > 0) :
	718	+ $a = reset($aa);
	719	+
	720	+ $author = array(
	721	+ 'name' => $a->get_name(),
	722	+ 'email' => $a->get_email(),
	723	+ 'uri' => $a->get_link(),
	724	+ );
	725	+ endif;
	726	+
	727	+ if (FEEDWORDPRESS_COMPATIBILITY) :
	728	+ // Search through the MagpieRSS elements: Atom, Dublin Core, RSS
	729	+ if (isset($this->item['author_name'])):
	730	+ $author['name'] = $this->item['author_name'];
	731	+ elseif (isset($this->item['dc']['creator'])):
	732	+ $author['name'] = $this->item['dc']['creator'];
	733	+ elseif (isset($this->item['dc']['contributor'])):
	734	+ $author['name'] = $this->item['dc']['contributor'];
	735	+ elseif (isset($this->feed->channel['dc']['creator'])) :
	736	+ $author['name'] = $this->feed->channel['dc']['creator'];
	737	+ elseif (isset($this->feed->channel['dc']['contributor'])) :
	738	+ $author['name'] = $this->feed->channel['dc']['contributor'];
	739	+ elseif (isset($this->feed->channel['author_name'])) :
	740	+ $author['name'] = $this->feed->channel['author_name'];
	741	+ elseif ($this->feed->is_rss() and isset($this->item['author'])) :
	742	+ // The author element in RSS is allegedly an
	743	+ // e-mail address, but lots of people don't use
	744	+ // it that way. So let's make of it what we can.
	745	+ $author = parse_email_with_realname($this->item['author']);
	746	+
	747	+ if ( !isset($author['name'])) :
	748	+ if (isset($author['email'])) :
	749	+ $author['name'] = $author['email'];
	750	+ else :
	751	+ $author['name'] = $this->feed->channel['title'];
	752	+ endif;
	753	+ endif;
165	754	endif;
	755	+ endif;
166	756
167		- // Store information to identify the feed that this came from
168		- $this->post['meta']['syndication_feed'] = $this->feedmeta['link/uri'];
169		- $this->post['meta']['syndication_feed_id'] = $this->feedmeta['link/id'];
	757	+ if ( !isset($author['name']) or is_null($author['name'])) :
	758	+ // Nothing found. Try some crappy defaults.
	759	+ if ($this->link->name()) :
	760	+ $author['name'] = $this->link->name();
	761	+ else :
	762	+ $url = parse_url($this->link->uri());
	763	+ $author['name'] = $url['host'];
	764	+ endif;
	765	+ endif;
170	766
171		- if (isset($this->item['source_link_self'])) :
172		- $this->post['meta']['syndication_feed_original'] = $this->item['source_link_self'];
	767	+ if (FEEDWORDPRESS_COMPATIBILITY) :
	768	+ if (isset($this->item['author_email'])):
	769	+ $author['email'] = $this->item['author_email'];
	770	+ elseif (isset($this->feed->channel['author_email'])) :
	771	+ $author['email'] = $this->feed->channel['author_email'];
173	772	endif;
174	773
175		- // In case you want to know the external permalink...
176		- $this->post['meta']['syndication_permalink'] = apply_filters('syndicated_item_link', $this->item['link']);
	774	+ if (isset($this->item['author_uri'])):
	775	+ $author['uri'] = $this->item['author_uri'];
	776	+ elseif (isset($this->item['author_url'])):
	777	+ $author['uri'] = $this->item['author_url'];
	778	+ elseif (isset($this->feed->channel['author_uri'])) :
	779	+ $author['uri'] = $this->item['author_uri'];
	780	+ elseif (isset($this->feed->channel['author_url'])) :
	781	+ $author['uri'] = $this->item['author_url'];
	782	+ elseif (isset($this->feed->channel['link'])) :
	783	+ $author['uri'] = $this->feed->channel['link'];
	784	+ endif;
	785	+ endif;
177	786
178		- // Store a hash of the post content for checking whether something needs to be updated
179		- $this->post['meta']['syndication_item_hash'] = $this->update_hash();
	787	+ return $author;
	788	+ } /* SyndicatedPost::author() */
180	789
181		- // Feed-by-feed options for author and category creation
182		- $this->post['named']['unfamiliar']['author'] = (isset($this->feedmeta['unfamiliar author']) ? $this->feedmeta['unfamiliar author'] : null);
183		- $this->post['named']['unfamiliar']['category'] = (isset($this->feedmeta['unfamiliar category']) ? $this->feedmeta['unfamiliar category'] : null);
184		-
185		- // Categories: start with default categories, if any
	790	+ /**
	791	+ * SyndicatedPost::get_terms_from_settings(): Return an array of terms to associate with the incoming
	792	+ * post based on the Categories, Tags, and other terms associated with each new post by the user's
	793	+ * settings (global and feed-specific).
	794	+ *
	795	+ * @since 2016.0331
	796	+ * @return array of lists, each element has the taxonomy for a key ('category', 'post_tag', etc.),
	797	+ * and a list of term codes (either alphanumeric names, or ID numbers encoded in a format that
	798	+ * SyndicatedLink::category_ids() can understand) within that taxonomy
	799	+ *
	800	+ */
	801	+ public function get_terms_from_settings () {
	802	+ // Categories: start with default categories, if any.
	803	+ $cats = array();
	804	+ if ('no' != $this->link->setting('add/category', NULL, 'yes')) :
186	805	$fc = get_option("feedwordpress_syndication_cats");
187	806	if ($fc) :
188		- $this->post['named']['preset/category'] = explode("\n", $fc);
189		- else :
190		- $this->post['named']['preset/category'] = array();
	807	+ $cats = array_merge($cats, explode("\n", $fc));
191	808	endif;
	809	+ endif;
192	810
193		- if (isset($this->feedmeta['cats']) and is_array($this->feedmeta['cats'])) :
194		- $this->post['named']['preset/category'] = array_merge($this->post['named']['preset/category'], $this->feedmeta['cats']);
195		- endif;
	811	+ $fc = $this->link->setting('cats',NULL, array());
	812	+ if (is_array($fc)) :
	813	+ $cats = array_merge($cats, $fc);
	814	+ endif;
	815	+ $preset_terms['category'] = $cats;
196	816
197		- // Now add categories from the post, if we have 'em
198		- $this->post['named']['category'] = array();
199		- if ( isset($this->item['category#']) ) :
200		- for ($i = 1; $i <= $this->item['category#']; $i++) :
201		- $cat_idx = (($i > 1) ? "#{$i}" : "");
202		- $cat = $this->item["category{$cat_idx}"];
	817	+ // Tags: start with default tags, if any
	818	+ $tags = array();
	819	+ if ('no' != $this->link->setting('add/post_tag', NULL, 'yes')) :
	820	+ $ft = get_option("feedwordpress_syndication_tags", NULL);
	821	+ $tags = (is_null($ft) ? array() : explode(FEEDWORDPRESS_CAT_SEPARATOR, $ft));
	822	+ endif;
203	823
204		- if ( isset($this->feedmeta['cat_split']) and strlen($this->feedmeta['cat_split']) > 0) :
205		- $pcre = "\007".$this->feedmeta['cat_split']."\007";
206		- $this->post['named']['category'] = array_merge($this->post['named']['category'], preg_split($pcre, $cat, -1 /*=no limit*/, PREG_SPLIT_NO_EMPTY));
207		- else :
208		- $this->post['named']['category'][] = $cat;
	824	+ $ft = $this->link->setting('tags', NULL, array());
	825	+ if (is_array($ft)) :
	826	+ $tags = array_merge($tags, $ft);
	827	+ endif;
	828	+ $preset_terms['post_tag'] = $tags;
	829	+
	830	+ $taxonomies = $this->link->taxonomies();
	831	+ $feedTerms = $this->link->setting('terms', NULL, array());
	832	+ $globalTerms = get_option('feedwordpress_syndication_terms', array());
	833	+ $specials = array('category' => 'cats', 'post_tag' => 'tags');
	834	+
	835	+ foreach ($taxonomies as $tax) :
	836	+ // category and tag settings have already previously been handled
	837	+ // but if this is from another taxonomy, then...
	838	+ if ( !isset($specials[$tax])) :
	839	+ $terms = array();
	840	+
	841	+ // See if we should get the globals
	842	+ if ('no' != $this->link->setting("add/$tax", NULL, 'yes')) :
	843	+ if (isset($globalTerms[$tax])) :
	844	+ $terms = $globalTerms[$tax];
209	845	endif;
210		- endfor;
	846	+ endif;
	847	+
	848	+ // Now merge in the locals
	849	+ if (isset($feedTerms[$tax])) :
	850	+ $terms = array_merge($terms, $feedTerms[$tax]);
	851	+ endif;
	852	+
	853	+ // That's all, folks.
	854	+ $preset_terms[$tax] = $terms;
211	855	endif;
212		- $this->post['named']['category'] = apply_filters('syndicated_item_categories', $this->post['named']['category'], $this);
213		-
214		- // Tags: start with default tags, if any
215		- $ft = get_option("feedwordpress_syndication_tags");
216		- if ($ft) :
217		- $this->post['tags_input'] = explode(FEEDWORDPRESS_CAT_SEPARATOR, $ft);
	856	+ endforeach;
	857	+
	858	+ return $preset_terms;
	859	+ } /* SyndicatedPost::get_terms_from_settings () */
	860	+
	861	+ /**
	862	+ * SyndicatedPost::get_terms_from_feeds(): Return an array of terms to associate with the incoming
	863	+ * post based on the contents of the subscribed feed (atom:category and rss:category elements, dc:subject
	864	+ * elements, tags embedded using microformats in the post content, etc.)
	865	+ *
	866	+ * @since 2016.0331
	867	+ * @return array of lists, each element has the taxonomy for a key ('category', 'post_tag', etc.),
	868	+ * and a list of alphanumeric term names
	869	+ */
	870	+ public function get_terms_from_feeds () {
	871	+ // Now add categories from the post, if we have 'em
	872	+ $cats = array();
	873	+ $post_cats = $this->entry->get_categories();
	874	+ if (is_array($post_cats)) : foreach ($post_cats as $cat) :
	875	+ $cat_name = $cat->get_term();
	876	+ if ( ! $cat_name) : $cat_name = $cat->get_label(); endif;
	877	+
	878	+ if ($this->link->setting('cat_split', NULL, NULL)) :
	879	+ $pcre = "\007".$this->feedmeta['cat_split']."\007";
	880	+ $cats = array_merge(
	881	+ $cats,
	882	+ preg_split(
	883	+ $pcre,
	884	+ $cat_name,
	885	+ -1 /*=no limit*/,
	886	+ PREG_SPLIT_NO_EMPTY
	887	+ )
	888	+ );
218	889	else :
219		- $this->post['tags_input'] = array();
	890	+ $cats[] = $cat_name;
220	891	endif;
221		-
222		- if (isset($this->feedmeta['tags']) and is_array($this->feedmeta['tags'])) :
223		- $this->post['tags_input'] = array_merge($this->post['tags_input'], $this->feedmeta['tags']);
	892	+ endforeach; endif;
	893	+
	894	+ $feed_terms['category'] = apply_filters('syndicated_item_categories', $cats, $this);
	895	+
	896	+ // Scan post for /a[@rel='tag'] and use as tags if present
	897	+ $tags = $this->inline_tags();
	898	+ $feed_terms['post_tag'] = apply_filters('syndicated_item_tags', $tags, $this);
	899	+
	900	+ return $feed_terms;
	901	+ } /* SyndicatedPost::get_terms_from_feeds () */
	902	+
	903	+ /**
	904	+ * SyndicatedPost::inline_tags: Return a list of all the tags embedded
	905	+ * in post content using the a[@rel="tag"] microformat.
	906	+ *
	907	+ * @since 2010.0630
	908	+ * @return array of string values containing the name of each tag
	909	+ */
	910	+ function inline_tags () {
	911	+ $tags = array();
	912	+ $content = $this->content();
	913	+ $pattern = FeedWordPressHTML::tagWithAttributeRegex('a', 'rel', 'tag');
	914	+ preg_match_all($pattern, $content, $refs, PREG_SET_ORDER);
	915	+ if (is_countable($refs) and count($refs) > 0) :
	916	+ foreach ($refs as $ref) :
	917	+ $tag = FeedWordPressHTML::tagWithAttributeMatch($ref);
	918	+ $tags[] = $tag['content'];
	919	+ endforeach;
	920	+ endif;
	921	+ return $tags;
	922	+ }
	923	+
	924	+ /**
	925	+ * SyndicatedPost::isTaggedAs: Test whether a feed item is
	926	+ * tagged / categorized with a given string. Case and leading and
	927	+ * trailing whitespace are ignored.
	928	+ *
	929	+ * @param string $tag Tag to check for
	930	+ *
	931	+ * @return bool Whether or not at least one of the categories / tags on
	932	+ * $this->item is set to $tag (modulo case and leading and trailing
	933	+ * whitespace)
	934	+ */
	935	+ function isTaggedAs ($tag) {
	936	+ $desiredTag = strtolower(trim($tag)); // Normalize case and whitespace
	937	+
	938	+ // Check to see if this is tagged with $tag
	939	+ $currentCategory = 'category';
	940	+ $currentCategoryNumber = 1;
	941	+
	942	+ // If we have the new MagpieRSS, the number of category elements
	943	+ // on this item is stored under index "category#".
	944	+ if (isset($this->item['category#'])) :
	945	+ $numberOfCategories = (int) $this->item['category#'];
	946	+
	947	+ // We REALLY shouldn't have the old and busted MagpieRSS, but in
	948	+ // case we do, it doesn't support multiple categories, but there
	949	+ // might still be a single value under the "category" index.
	950	+ elseif (isset($this->item['category'])) :
	951	+ $numberOfCategories = 1;
	952	+
	953	+ // No standard category or tag elements on this feed item.
	954	+ else :
	955	+ $numberOfCategories = 0;
	956	+
	957	+ endif;
	958	+
	959	+ $isSoTagged = false; // Innocent until proven guilty
	960	+
	961	+ // Loop through category elements; if there are multiple
	962	+ // elements, they are indexed as category, category#2,
	963	+ // category#3, ... category#N
	964	+ while ($currentCategoryNumber <= $numberOfCategories) :
	965	+ if ($desiredTag == strtolower(trim($this->item[$currentCategory]))) :
	966	+ $isSoTagged = true; // Got it!
	967	+ break;
224	968	endif;
225	969
	970	+ $currentCategoryNumber += 1;
	971	+ $currentCategory = 'category#'.$currentCategoryNumber;
	972	+ endwhile;
	973	+
	974	+ return $isSoTagged;
	975	+ } /* SyndicatedPost::isTaggedAs() */
	976	+
	977	+ /**
	978	+ * SyndicatedPost::enclosures: returns an array with any enclosures
	979	+ * that may be attached to this syndicated item.
	980	+ *
	981	+ * @param string $type If you only want enclosures that match a certain
	982	+ * MIME type or group of MIME types, you can limit the enclosures
	983	+ * that will be returned to only those with a MIME type which
	984	+ * matches this regular expression.
	985	+ * @return array
	986	+ */
	987	+ function enclosures( $type = '/.*/' ) {
	988	+ $enclosures = array();
	989	+
	990	+ if ( isset( $this->item['enclosure#'] ) ) :
	991	+ // Loop through enclosure, enclosure#2, enclosure#3, ....
	992	+ for ( $i = 1; $i <= $this->item['enclosure#']; $i++ ) :
	993	+ $eid = ( ( $i > 1 ) ? "#{$i}" : "" ); // this was set as #{$id} — was that a typo? (gwyneth 20230919)
	994	+
	995	+ // Does it match the type we want?
	996	+ if ( preg_match($type, $this->item["enclosure{$eid}@type"] ) ) :
	997	+ $enclosures[] = array(
	998	+ "url" => $this->item["enclosure{$eid}@url"],
	999	+ "type" => $this->item["enclosure{$eid}@type"],
	1000	+ "length" => $this->item["enclosure{$eid}@length"],
	1001	+ );
	1002	+ endif;
	1003	+ endfor;
226	1004	endif;
227		- } // SyndicatedPost::SyndicatedPost()
	1005	+ return $enclosures;
	1006	+ } /* SyndicatedPost::enclosures() */
228	1007
	1008	+ function source ($what = NULL) {
	1009	+ $ret = NULL;
	1010	+ $source = $this->entry->get_source();
	1011	+ if ($source) :
	1012	+ $ret = array();
	1013	+ $ret['title'] = $source->get_title();
	1014	+ $ret['uri'] = $source->get_link();
	1015	+ $ret['feed'] = $source->get_link(0, 'self');
	1016	+
	1017	+ if ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'id')) :
	1018	+ $ret['id'] = $id_tags[0]['data'];
	1019	+ elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_ATOM_03, 'id')) :
	1020	+ $ret['id'] = $id_tags[0]['data'];
	1021	+ elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_RSS_20, 'guid')) :
	1022	+ $ret['id'] = $id_tags[0]['data'];
	1023	+ elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_RSS_10, 'guid')) :
	1024	+ $ret['id'] = $id_tags[0]['data'];
	1025	+ elseif ($id_tags = $source->get_source_tags(SIMPLEPIE_NAMESPACE_RSS_090, 'guid')) :
	1026	+ $ret['id'] = $id_tags[0]['data'];
	1027	+ endif;
	1028	+ endif;
	1029	+
	1030	+ if ( !is_null($what) and is_scalar($what)) :
	1031	+ $ret = $ret[$what];
	1032	+ endif;
	1033	+ return $ret;
	1034	+ }
	1035	+
	1036	+ function comment_link () {
	1037	+ $url = null;
	1038	+
	1039	+ // RSS 2.0 has a standard <comments> element:
	1040	+ // "<comments> is an optional sub-element of <item>. If present,
	1041	+ // it is the url of the comments page for the item."
	1042	+ // <http://cyber.law.harvard.edu/rss/rss.html#ltcommentsgtSubelementOfLtitemgt>
	1043	+ if (isset($this->item['comments'])) :
	1044	+ $url = $this->item['comments'];
	1045	+ endif;
	1046	+
	1047	+ // The convention in Atom feeds is to use a standard <link>
	1048	+ // element with @rel="replies" and @type="text/html".
	1049	+ // Unfortunately, SimplePie_Item::get_links() allows us to filter
	1050	+ // by the value of @rel, but not by the value of @type. *sigh*
	1051	+
	1052	+ // Try Atom 1.0 first
	1053	+ $linkElements = $this->entry->get_item_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'link');
	1054	+
	1055	+ // Fall back and try Atom 0.3
	1056	+ if (is_null($linkElements)) : $linkElements = $this->entry->get_item_tags(SIMPLEPIE_NAMESPACE_ATOM_03, 'link'); endif;
	1057	+
	1058	+ // Now loop through the elements, screening by @rel and @type
	1059	+ if (is_array($linkElements)) : foreach ($linkElements as $link) :
	1060	+ $rel = (isset($link['attribs']['']['rel']) ? $link['attribs']['']['rel'] : 'alternate');
	1061	+ $type = (isset($link['attribs']['']['type']) ? $link['attribs']['']['type'] : NULL);
	1062	+ $href = (isset($link['attribs']['']['href']) ? $link['attribs']['']['href'] : NULL);
	1063	+
	1064	+ if (strtolower($rel)=='replies' and $type=='text/html' and !is_null($href)) :
	1065	+ $url = $href;
	1066	+ endif;
	1067	+ endforeach; endif;
	1068	+
	1069	+ return $url;
	1070	+ }
	1071	+
	1072	+ function comment_feed () {
	1073	+ $feed = null;
	1074	+
	1075	+ // Well Formed Web comment feeds extension for RSS 2.0
	1076	+ // <http://www.sellsbrothers.com/spout/default.aspx?content=archive.htm#exposingRssComments>
	1077	+ //
	1078	+ // N.B.: Correct capitalization is wfw:commentRss, but
	1079	+ // wfw:commentRSS is common in the wild (partly due to a typo in
	1080	+ // the original spec). In any case, our item array is normalized
	1081	+ // to all lowercase anyways.
	1082	+ if (isset($this->item['wfw']['commentrss'])) :
	1083	+ $feed = $this->item['wfw']['commentrss'];
	1084	+ endif;
	1085	+
	1086	+ // In Atom 1.0, the convention is to use a standard link element
	1087	+ // with @rel="replies". Sometimes this is also used to pass a
	1088	+ // link to the human-readable comments page, so we also need to
	1089	+ // check link/@type for a feed MIME type.
	1090	+ //
	1091	+ // Which is why I'm not using the SimplePie_Item::get_links()
	1092	+ // method here, incidentally: it doesn't allow you to filter by
	1093	+ // @type. *sigh*
	1094	+ if (isset($this->item['link_replies'])) :
	1095	+ // There may be multiple <link rel="replies"> elements; feeds have a feed MIME type
	1096	+ $N = isset($this->item['link_replies#']) ? $this->item['link_replies#'] : 1;
	1097	+ for ($i = 1; $i <= $N; $i++) :
	1098	+ $currentElement = 'link_replies'.(($i > 1) ? '#'.$i : '');
	1099	+ if (isset($this->item[$currentElement.'@type'])
	1100	+ and preg_match("\007application/(atom|rss|rdf)\+xml\007i", $this->item[$currentElement.'@type'])) :
	1101	+ $feed = $this->item[$currentElement];
	1102	+ endif;
	1103	+ endfor;
	1104	+ endif;
	1105	+ return $feed;
	1106	+ } /* SyndicatedPost::comment_feed() */
	1107	+
	1108	+ ##################################
	1109	+ #### BUILT-IN CONTENT FILTERS ####
	1110	+ ##################################
	1111	+
	1112	+ var $uri_attrs = array (
	1113	+ array('a', 'href'),
	1114	+ array('applet', 'codebase'),
	1115	+ array('area', 'href'),
	1116	+ array('blockquote', 'cite'),
	1117	+ array('body', 'background'),
	1118	+ array('del', 'cite'),
	1119	+ array('form', 'action'),
	1120	+ array('frame', 'longdesc'),
	1121	+ array('frame', 'src'),
	1122	+ array('iframe', 'longdesc'),
	1123	+ array('iframe', 'src'),
	1124	+ array('head', 'profile'),
	1125	+ array('img', 'longdesc'),
	1126	+ array('img', 'src'),
	1127	+ array('img', 'usemap'),
	1128	+ array('input', 'src'),
	1129	+ array('input', 'usemap'),
	1130	+ array('ins', 'cite'),
	1131	+ array('link', 'href'),
	1132	+ array('object', 'classid'),
	1133	+ array('object', 'codebase'),
	1134	+ array('object', 'data'),
	1135	+ array('object', 'usemap'),
	1136	+ array('q', 'cite'),
	1137	+ array('script', 'src')
	1138	+ ); /* var SyndicatedPost::$uri_attrs */
	1139	+
	1140	+ var $_base = null;
	1141	+
	1142	+ function resolve_single_relative_uri ($refs) {
	1143	+ $tag = FeedWordPressHTML::attributeMatch($refs);
	1144	+ $url = SimplePie_Misc::absolutize_url($tag['value'], $this->_base);
	1145	+
	1146	+ return $tag['prefix'] . $url . $tag['suffix'];
	1147	+ } /* function SyndicatedPost::resolve_single_relative_uri() */
	1148	+
	1149	+ static function resolve_relative_uris ($content, $obj) {
	1150	+ $set = $obj->link->setting('resolve relative', 'resolve_relative', 'yes');
	1151	+ if ($set and $set != 'no') :
	1152	+ // Fallback: if we don't have anything better, use the
	1153	+ // item link from the feed
	1154	+ $obj->_base = $obj->permalink(); // Reset the base for resolving relative URIs
	1155	+
	1156	+ // What we should do here, properly, is to use
	1157	+ // SimplePie_Item::get_base() -- but that method is
	1158	+ // currently broken. Or getting down and dirty in the
	1159	+ // SimplePie representation of the content tags and
	1160	+ // grabbing the xml_base member for the content element.
	1161	+ // Maybe someday...
	1162	+
	1163	+ foreach ($obj->uri_attrs as $pair) :
	1164	+ list($tag, $attr) = $pair;
	1165	+ $pattern = FeedWordPressHTML::attributeRegex($tag, $attr);
	1166	+
	1167	+ // FIXME: Encountered issue while testing an extremely long (= 88827 characters) item
	1168	+ // Relying on preg_replace_callback() here can cause a PHP seg fault on my development
	1169	+ // server. preg_match_all() causes a similar problem. Apparently this is a PCRE issue
	1170	+ // Cf. discussion of similar issue <https://bugs.php.net/bug.php?id=65009>
	1171	+ $content = preg_replace_callback (
	1172	+ $pattern,
	1173	+ array($obj, 'resolve_single_relative_uri'),
	1174	+ $content
	1175	+ );
	1176	+
	1177	+ endforeach;
	1178	+ endif;
	1179	+
	1180	+ return $content;
	1181	+ } /* function SyndicatedPost::resolve_relative_uris () */
	1182	+
	1183	+ var $strip_attrs = array (
	1184	+ array('[a-z]+', 'target'),
	1185	+// array('[a-z]+', 'style'),
	1186	+// array('[a-z]+', 'on[a-z]+'),
	1187	+ );
	1188	+
	1189	+ function strip_attribute_from_tag ($refs) {
	1190	+ $tag = FeedWordPressHTML::attributeMatch($refs);
	1191	+ return $tag['before_attribute'].$tag['after_attribute'];
	1192	+ }
	1193	+
	1194	+ static function sanitize_content ($content, $obj) {
	1195	+ # This kind of sucks. I intend to replace it with
	1196	+ # lib_filter sometime soon.
	1197	+ foreach ($obj->strip_attrs as $pair):
	1198	+ list($tag,$attr) = $pair;
	1199	+ $pattern = FeedWordPressHTML::attributeRegex($tag, $attr);
	1200	+
	1201	+ $content = preg_replace_callback (
	1202	+ $pattern,
	1203	+ array($obj, 'strip_attribute_from_tag'),
	1204	+ $content
	1205	+ );
	1206	+ endforeach;
	1207	+ return $content;
	1208	+ } /* SyndicatedPost::sanitize() */
	1209	+
	1210	+ #####################
	1211	+ #### POST STATUS ####
	1212	+ #####################
	1213	+
	1214	+ /**
	1215	+ * SyndicatedPost::filtered: check whether or not this post has been
	1216	+ * screened out by a registered filter.
	1217	+ *
	1218	+ * @return bool TRUE iff post has been filtered out by a previous filter
	1219	+ */
229	1220	function filtered () {
230	1221	return is_null($this->post);
231		- }
	1222	+ } /* SyndicatedPost::filtered() */
232	1223
233		- function freshness () {
234		- global $wpdb;
	1224	+ /**
	1225	+ * SyndicatedPost::freshness: check whether post is a new post to be
	1226	+ * inserted, a previously syndicated post that needs to be updated to
	1227	+ * match the latest revision, or a previously syndicated post that is
	1228	+ * still up-to-date.
	1229	+ *
	1230	+ * @return int A status code representing the freshness of the post
	1231	+ * -1 = post already syndicated; has a revision that needs to be stored, but not updated to
	1232	+ * 0 = post already syndicated; no update needed
	1233	+ * 1 = post already syndicated, but needs to be updated to latest
	1234	+ * 2 = post has not yet been syndicated; needs to be created
	1235	+ */
	1236	+ function freshness ($format = 'number') {
235	1237
236	1238	if ($this->filtered()) : // This should never happen.
237		- FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__);
	1239	+ FeedWordPressDiagnostic::critical_bug('SyndicatedPost', $this, __LINE__, __FILE__);
238	1240	endif;
239		-
240		- if (is_null($this->_freshness)) :
241		- $guid = $wpdb->escape($this->guid());
242	1241
243		- $result = $wpdb->get_row("
244		- SELECT id, guid, post_modified_gmt
245		- FROM $wpdb->posts WHERE guid='$guid'
246		- ");
	1242	+ if (is_null($this->_freshness)) : // Not yet checked and cached.
	1243	+ $guid = $this->post['guid'];
	1244	+ $eguid = esc_sql($this->post['guid']);
247	1245
248		- if (!$result) :
	1246	+ $q = new WP_Query(array(
	1247	+ 'fields' => '_synfresh', // id, guid, post_modified_gmt
	1248	+ 'ignore_sticky_posts' => true,
	1249	+ 'guid' => $eguid, // it's always better to use the escaped version, right? (gwyneth 20230915)
	1250	+ ));
	1251	+
	1252	+ $old_post = NULL;
	1253	+ if ($q->have_posts()) :
	1254	+ while ($q->have_posts()) : $q->the_post();
	1255	+ if (get_post_type($q->post->ID) == $this->post['post_type']):
	1256	+ $old_post = $q->post;
	1257	+ endif;
	1258	+ endwhile;
	1259	+ endif;
	1260	+
	1261	+ if (is_null($old_post)) : // No post with this guid
	1262	+ FeedWordPress::diagnostic('feed_items:freshness', 'Item ['.$guid.'] "'.$this->entry->get_title().'" is a NEW POST.');
	1263	+ $this->_wp_id = NULL;
249	1264	$this->_freshness = 2; // New content
250		- else:
251		- $stored_update_hashes = get_post_custom_values('syndication_item_hash', $result->id);
252		- if (count($stored_update_hashes) > 0) :
253		- $stored_update_hash = $stored_update_hashes[0];
254		- $update_hash_changed = ($stored_update_hash != $this->update_hash());
	1265	+ else :
	1266	+ // Presume there is nothing new until we find
	1267	+ // something new.
	1268	+ $updated = false;
	1269	+ $live = false;
	1270	+
	1271	+ // Pull the list of existing revisions to get
	1272	+ // timestamps.
	1273	+ $revisions = wp_get_post_revisions($old_post->ID);
	1274	+ foreach ($revisions as $rev) :
	1275	+ $revisions_ts[] = mysql2date('G', $rev->post_modified_gmt);
	1276	+ endforeach;
	1277	+
	1278	+ $revisions_ts[] = mysql2date('G', $old_post->post_modified_gmt);
	1279	+ $last_rev_ts = end($revisions_ts);
	1280	+ $updated_ts = $this->updated(/*fallback=*/ true, /*default=*/ NULL);
	1281	+
	1282	+ // If we have an explicit updated timestamp,
	1283	+ // check that against existing stamps.
	1284	+ if ( !is_null($updated_ts)) :
	1285	+ $updated = !in_array($updated_ts, $revisions_ts);
	1286	+
	1287	+ // If this a newer revision, make it go
	1288	+ // live. If an older one, just record
	1289	+ // the contents.
	1290	+ $live = ($updated and ($updated_ts > $last_rev_ts));
	1291	+ endif;
	1292	+
	1293	+ // This is a revision we haven't seen before, judging by the date.
	1294	+
	1295	+ $updatedReason = NULL;
	1296	+ if ($updated) :
	1297	+ $updatedReason = preg_replace(
	1298	+ "/\s+/", " ",
	1299	+ 'has been marked with a new timestamp ('
	1300	+ .date('Y-m-d H:i:s', $updated_ts)
	1301	+ ." > "
	1302	+ .date('Y-m-d H:i:s', $last_rev_ts)
	1303	+ .')'
	1304	+ );
	1305	+
	1306	+ // The date does not indicate a new revision, so
	1307	+ // let's check the hash.
255	1308	else :
256		- $update_hash_changed = false;
	1309	+ // Or the hash...
	1310	+ $hash = $this->update_hash();
	1311	+ $seen = $this->stored_hashes($old_post->ID);
	1312	+ if (is_countable($seen) and count($seen) > 0) :
	1313	+ $updated = !in_array($hash, $seen); // Not seen yet?
	1314	+ else :
	1315	+ $updated = true; // Can't find syndication meta-data
	1316	+ endif;
	1317	+
	1318	+ if ($updated and FeedWordPressDiagnostic::is_on('feed_items:freshness:reasons')) :
	1319	+ // In the absence of definitive
	1320	+ // timestamp information, we
	1321	+ // just have to assume that a
	1322	+ // hash we haven't seen before
	1323	+ // is a newer version.
	1324	+ $live = true;
	1325	+
	1326	+ $updatedReason = ' has a not-yet-seen update hash: '
	1327	+ .MyPHP::val($hash)
	1328	+ .' not in {'
	1329	+ .implode(", ", array_map(array('FeedWordPress', 'val'), $seen))
	1330	+ .'}. Basis: '
	1331	+ .MyPHP::val(array_keys($this->update_hash(false)));
	1332	+ endif;
257	1333	endif;
258	1334
259		- preg_match('/([0-9]+)-([0-9]+)-([0-9]+) ([0-9]+):([0-9]+):([0-9]+)/', $result->post_modified_gmt, $backref);
	1335	+ $frozen = false;
	1336	+ if ($updated) : // Ignore if the post is frozen
	1337	+ $frozen = ('yes' == $this->link->setting('freeze updates', 'freeze_updates', NULL));
	1338	+ if ( ! $frozen) :
	1339	+ $frozen_value = get_post_meta($old_post->ID, '_syndication_freeze_updates', /*single=*/ true);
	1340	+ $frozen = ( !is_null($frozen_value) and ('yes' == $frozen_value));
260	1341
261		- $last_rev_ts = gmmktime($backref[4], $backref[5], $backref[6], $backref[2], $backref[3], $backref[1]);
262		- $updated_ts = $this->updated(/*fallback=*/ true, /*default=*/ NULL);
263		- $updated = ((
264		- !is_null($updated_ts)
265		- and ($updated_ts > $last_rev_ts)
266		- ) or $update_hash_changed);
	1342	+ if ($frozen) :
	1343	+ $updatedReason = ' IS BLOCKED FROM BEING UPDATED BY A UPDATE LOCK ON THIS POST, EVEN THOUGH IT '.$updatedReason;
	1344	+ endif;
	1345	+ else :
	1346	+ $updatedReason = ' IS BLOCKED FROM BEING UPDATED BY A FEEDWORDPRESS UPDATE LOCK, EVEN THOUGH IT '.$updatedReason;
	1347	+ endif;
	1348	+ endif;
	1349	+ $live = ($live and ! $frozen);
267	1350
268	1351	if ($updated) :
269		- $this->_freshness = 1; // Updated content
270		- $this->_wp_id = $result->id;
	1352	+ FeedWordPress::diagnostic('feed_items:freshness', 'Item ['.$guid.'] "'.$this->entry->get_title().'" is an update of an existing post.');
	1353	+ if ( !is_null($updatedReason)) :
	1354	+ $updatedReason = preg_replace('/\s+/', ' ', $updatedReason);
	1355	+ FeedWordPress::diagnostic('feed_items:freshness:reasons', 'Item ['.$guid.'] "'.$this->entry->get_title().'" '.$updatedReason);
	1356	+ endif;
	1357	+
	1358	+ $this->_freshness = apply_filters('syndicated_item_freshness', ($live ? 1 : -1), $updated, $frozen, $updated_ts, $last_rev_ts, $this);
	1359	+
	1360	+ $this->_wp_id = $old_post->ID;
	1361	+ $this->_wp_post = $old_post;
	1362	+
	1363	+ // We want this to keep a running list of all the
	1364	+ // processed update hashes.
	1365	+ $this->post['meta']['syndication_item_hash'] = array_merge(
	1366	+ $this->stored_hashes(),
	1367	+ array($this->update_hash())
	1368	+ );
271	1369	else :
	1370	+ FeedWordPress::diagnostic('feed_items:freshness', 'Item ['.$guid.'] "'.$this->entry->get_title().'" is a duplicate of an existing post.');
272	1371	$this->_freshness = 0; // Same old, same old
273		- $this->_wp_id = $result->id;
	1372	+ $this->_wp_id = $old_post->ID;
274	1373	endif;
275	1374	endif;
276	1375	endif;
277		- return $this->_freshness;
	1376	+
	1377	+ switch ($format) :
	1378	+ case 'status' :
	1379	+ switch ($this->_freshness) :
	1380	+ case -1:
	1381	+ $ret = 'stored';
	1382	+ break;
	1383	+ case 0:
	1384	+ $ret = NULL;
	1385	+ break;
	1386	+ case 1:
	1387	+ $ret = 'updated';
	1388	+ break;
	1389	+ case 2:
	1390	+ default:
	1391	+ $ret = 'new';
	1392	+ break;
	1393	+ endswitch;
	1394	+ break;
	1395	+ case 'number' :
	1396	+ default :
	1397	+ $ret = $this->_freshness;
	1398	+ endswitch;
	1399	+
	1400	+
	1401	+ return $ret;
	1402	+ } /* SyndicatedPost::freshness () */
	1403	+
	1404	+ function has_fresh_content () {
	1405	+ return ( ! $this->filtered() and $this->freshness() != 0 );
	1406	+ } /* SyndicatedPost::has_fresh_content () */
	1407	+
	1408	+ function this_revision_needs_original_post ($freshness = NULL) {
	1409	+ if (is_null($freshness)) :
	1410	+ $freshness = $this->freshness();
	1411	+ endif;
	1412	+ return ( $freshness >= 2 );
278	1413	}
279	1414
	1415	+ function this_revision_is_current ($freshness = NULL) {
	1416	+ if (is_null($freshness)) :
	1417	+ $freshness = $this->freshness();
	1418	+ endif;
	1419	+ return ( $freshness >= 1 );
	1420	+ } /* SyndicatedPost::this_revision_is_current () */
	1421	+
	1422	+ function fresh_content_is_update () {
	1423	+ return ($this->freshness() < 2);
	1424	+ } /* SyndicatedPost::fresh_content_is_update () */
	1425	+
	1426	+ function fresh_storage_diagnostic () {
	1427	+ $ret = NULL;
	1428	+ switch ($this->freshness()) :
	1429	+ case -1 :
	1430	+ $ret = 'Storing alternate revision of existing post # '.$this->wp_id().', "'.$this->post['post_title'].'"';
	1431	+ break;
	1432	+ case 1 :
	1433	+ $ret = 'Updating existing post # '.$this->wp_id().', "'.$this->post['post_title'].'"';
	1434	+ break;
	1435	+ case 2 :
	1436	+ default :
	1437	+ $ret = 'Inserting new post "'.$this->post['post_title'].'"';
	1438	+ break;
	1439	+ endswitch;
	1440	+ return $ret;
	1441	+ } /* SyndicatedPost::fresh_storage_diagnostic() */
	1442	+
	1443	+ function fresh_storage_hook () {
	1444	+ $ret = NULL;
	1445	+ switch ($this->freshness()) :
	1446	+ case -1 :
	1447	+ case 1 :
	1448	+ $ret = 'update_syndicated_item';
	1449	+ break;
	1450	+ case 2 :
	1451	+ default :
	1452	+ $ret = 'post_syndicated_item';
	1453	+ break;
	1454	+ endswitch;
	1455	+ return $ret;
	1456	+ } /* SyndicatedPost::fresh_storage_hook () */
	1457	+
	1458	+ #################################################
	1459	+ #### INTERNAL STORAGE AND MANAGEMENT METHODS ####
	1460	+ #################################################
	1461	+
280	1462	function wp_id () {
281	1463	if ($this->filtered()) : // This should never happen.
282		- FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__);
	1464	+ FeedWordPressDiagnostic::critical_bug('SyndicatedPost', $this, __LINE__, __FILE__);
283	1465	endif;
284		-
	1466	+
285	1467	if (is_null($this->_wp_id) and is_null($this->_freshness)) :
286	1468	$fresh = $this->freshness(); // sets WP DB id in the process
287	1469	endif;
288	1470	return $this->_wp_id;
		@@ -287,272 +1469,492 @@
287	1469	endif;
288	1470	return $this->_wp_id;
289	1471	}
290	1472
291		- function store () {
292		- global $wpdb;
	1473	+ /**
	1474	+ * SyndicatedPost::secure_author_id(). Look up, or create, a numeric ID
	1475	+ * for the author of the incoming post.
	1476	+ *
	1477	+ * side effect: int|NULL stored in $this->post['post_author']
	1478	+ * side effect: IF no valid author is found, NULL stored in $this->post
	1479	+ * side effect: diagnostic output in case item is rejected with NULL author
	1480	+ *
	1481	+ * @used-by SyndicatedPost::store
	1482	+ *
	1483	+ * @uses SyndicatedPost::post
	1484	+ * @uses SyndicatedPost::author_id
	1485	+ * @uses SyndicatedLink::setting
	1486	+ * @uses FeedWordPress::diagnostic
	1487	+ */
	1488	+ protected function secure_author_id () {
	1489	+ # -- Look up, or create, numeric ID for author
	1490	+ $this->post['post_author'] = $this->author_id (
	1491	+ $this->link->setting('unfamiliar author', 'unfamiliar_author', 'create')
	1492	+ );
293	1493
294		- if ($this->filtered()) : // This should never happen.
295		- FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__);
	1494	+ if (is_null($this->post['post_author'])) :
	1495	+ FeedWordPress::diagnostic('feed_items:rejected', 'Filtered out item ['.$this->guid().'] without syndication: no author available');
	1496	+ $this->post = NULL;
296	1497	endif;
297		-
298		- $freshness = $this->freshness();
299		- if ($freshness > 0) :
300		- # -- Look up, or create, numeric ID for author
301		- $this->post['post_author'] = $this->author_id (
302		- FeedWordPress::on_unfamiliar('author', $this->post['named']['unfamiliar']['author'])
303		- );
	1498	+ } /* SyndicatedPost::secure_author_id() */
304	1499
305		- if (is_null($this->post['post_author'])) :
306		- $this->post = NULL;
307		- endif;
308		- endif;
309		-
310		- if (!$this->filtered() and $freshness > 0) :
311		- # -- Look up, or create, numeric ID for categories
312		- list($pcats, $ptags) = $this->category_ids (
313		- $this->post['named']['category'],
314		- FeedWordPress::on_unfamiliar('category', $this->post['named']['unfamiliar']['category']),
315		- /*tags_too=*/ true
316		- );
	1500	+ /**
	1501	+ * SyndicatedPost::secure_term_ids(). Look up, or create, numeric IDs
	1502	+ * for the terms (categories, tags, etc.) assigned to the incoming post,
	1503	+ * whether by global settings, feed settings, or by the tags on the feed.
	1504	+ *
	1505	+ * side effect: array of term ids stored in $this->post['tax_input']
	1506	+ * side effect: IF settings or filters determine post should be filtered out,
	1507	+ * NULL stored in $this->post
	1508	+ *
	1509	+ * @used-by SyndicatedPost::store
	1510	+ *
	1511	+ * @uses apply_filters
	1512	+ * @uses SyndicatedLink::setting
	1513	+ * @uses SyndicatedPost::category_ids
	1514	+ * @uses SyndicatedPost::preset_terms
	1515	+ * @uses SyndicatedPost::post
	1516	+ */
	1517	+ protected function secure_term_ids () {
	1518	+ $mapping = apply_filters('syndicated_post_terms_mapping', array(
	1519	+ 'category' => array('abbr' => 'cats', 'unfamiliar' => 'category', 'domain' => array('category', 'post_tag')),
	1520	+ 'post_tag' => array('abbr' => 'tags', 'unfamiliar' => 'post_tag', 'domain' => array('post_tag')),
	1521	+ ), $this);
317	1522
318		- $this->post['post_category'] = $pcats;
319		- $this->post['tags_input'] = array_merge($this->post['tags_input'], $ptags);
	1523	+ $termSet = array(); $valid = null;
	1524	+ foreach ($this->feed_terms as $what => $anTerms) :
	1525	+ // Default to using the inclusive procedures (for cats) rather than exclusive (for inline tags)
	1526	+ $taxes = (isset($mapping[$what]) ? $mapping[$what] : $mapping['category']);
	1527	+ $unfamiliar = $taxes['unfamiliar'];
320	1528
321		- if (is_null($this->post['post_category'])) :
322		- // filter mode on, no matching categories; drop the post
323		- $this->post = NULL;
324		- else :
325		- // filter mode off or at least one match; now add on the feed and global presets
326		- $this->post['post_category'] = array_merge (
327		- $this->post['post_category'],
328		- $this->category_ids (
329		- $this->post['named']['preset/category'],
330		- 'default'
	1529	+ if ( !is_null($this->post)) : // Not filtered out yet
	1530	+ # -- Look up, or create, numeric ID for categories
	1531	+ $taxonomies = $this->link->setting("match/".$taxes['abbr'], 'match_'.$taxes['abbr'], $taxes['domain']);
	1532	+
	1533	+ // Eliminate dummy variables
	1534	+ $taxonomies = array_filter( $taxonomies, 'remove_dummy_zero' );
	1535	+
	1536	+ // Allow FWP add-on filters to control the taxonomies we use to search for a term
	1537	+ $taxonomies = apply_filters("syndicated_post_terms_match", $taxonomies, $what, $this);
	1538	+ $taxonomies = apply_filters("syndicated_post_terms_match_{$what}", $taxonomies, $this);
	1539	+
	1540	+ // Allow FWP add-on filters to control with greater precision what happens on unmatched
	1541	+ $unmatched = apply_filters("syndicated_post_terms_unfamiliar",
	1542	+ $this->link->setting(
	1543	+ "unfamiliar {$unfamiliar}",
	1544	+ "unfamiliar_{$unfamiliar}",
	1545	+ 'create:'.$unfamiliar
	1546	+ ),
	1547	+ $what,
	1548	+ $this
	1549	+ );
	1550	+
	1551	+ $terms = $this->category_ids (
	1552	+ $anTerms,
	1553	+ $unmatched,
	1554	+ /*taxonomies=*/ $taxonomies,
	1555	+ array(
	1556	+ 'singleton' => false, // I don't like surprises
	1557	+ 'filters' => true,
331	1558	)
332	1559	);
333	1560
334		- if (count($this->post['post_category']) < 1) :
335		- $this->post['post_category'][] = 1; // Default to category 1 ("Uncategorized" / "General") if nothing else
	1561	+ if (is_null($terms) or is_null($termSet)) :
	1562	+ // filtered out -- no matches
	1563	+ else :
	1564	+ $valid = true;
	1565	+
	1566	+ // filter mode off, or at least one match
	1567	+ foreach ($terms as $tax => $term_ids) :
	1568	+ if ( !isset($termSet[$tax])) :
	1569	+ $termSet[$tax] = array();
	1570	+ endif;
	1571	+ $termSet[$tax] = array_merge($termSet[$tax], $term_ids);
	1572	+ endforeach;
336	1573	endif;
337	1574	endif;
	1575	+ endforeach;
	1576	+
	1577	+ if (is_null($valid)) : // Plonked
	1578	+ $this->post = NULL;
	1579	+ else : // We can proceed
	1580	+ $this->post['tax_input'] = array();
	1581	+ foreach ($termSet as $tax => $term_ids) :
	1582	+ if ( !isset($this->post['tax_input'][$tax])) :
	1583	+ $this->post['tax_input'][$tax] = array();
	1584	+ endif;
	1585	+ $this->post['tax_input'][$tax] = array_merge(
	1586	+ $this->post['tax_input'][$tax],
	1587	+ $term_ids
	1588	+ );
	1589	+ endforeach;
	1590	+
	1591	+ // Now let's add on the feed and global presets
	1592	+ foreach ($this->preset_terms as $tax => $term_ids) :
	1593	+ if ( !isset($this->post['tax_input'][$tax])) :
	1594	+ $this->post['tax_input'][$tax] = array();
	1595	+ endif;
	1596	+
	1597	+ $this->post['tax_input'][$tax] = array_merge (
	1598	+ $this->post['tax_input'][$tax],
	1599	+ $this->category_ids (
	1600	+ /*terms=*/ $term_ids,
	1601	+ /*unfamiliar=*/ 'create:'.$tax, // These are presets; for those added in a tagbox editor, the tag may not yet exist
	1602	+ /*taxonomies=*/ array($tax),
	1603	+ array(
	1604	+ 'singleton' => true,
	1605	+ ))
	1606	+ );
	1607	+ endforeach;
338	1608	endif;
339		-
340		- if (!$this->filtered() and $freshness > 0) :
341		- unset($this->post['named']);
	1609	+ } /* SyndicatedPost::secure_term_ids() */
	1610	+
	1611	+ /**
	1612	+ * SyndicatedPost::store
	1613	+ *
	1614	+ * @uses SyndicatedPost::secure_author_id
	1615	+ */
	1616	+ public function store () {
	1617	+ if ($this->filtered()) : // This should never happen.
	1618	+ FeedWordPressDiagnostic::critical_bug('SyndicatedPost', $this, __LINE__, __FILE__);
	1619	+ endif;
	1620	+
	1621	+ $freshness = $this->freshness(); // see above: this sets WP DB id in the process (gwyneth 20230919)
	1622	+ if ($this->has_fresh_content()) :
	1623	+ $this->secure_author_id();
	1624	+ endif;
	1625	+
	1626	+ if ($this->has_fresh_content()) : // Was this filtered during author_id lookup?
	1627	+ $this->secure_term_ids();
	1628	+ endif;
	1629	+
	1630	+ // We have to check again in case the post has been filtered
	1631	+ // during the category/tags/taxonomy terms lookup
	1632	+ if ($this->has_fresh_content()) :
	1633	+ // Filter some individual fields
	1634	+
	1635	+ // If there already is a post slug (from syndication or by manual
	1636	+ // editing) don't cause WP to overwrite it by sending in a NULL
	1637	+ // post_name. Props Chris Fritz 2012-11-28.
	1638	+ $post_name = (is_null($this->_wp_post) ? NULL : $this->_wp_post->post_name);
	1639	+
	1640	+ // Allow filters to set post slug. Props niska.
	1641	+ $post_name = apply_filters('syndicated_post_slug', $post_name, $this);
	1642	+ if ( !empty($post_name)) :
	1643	+ $this->post['post_name'] = $post_name;
	1644	+ endif;
	1645	+
342	1646	$this->post = apply_filters('syndicated_post', $this->post, $this);
	1647	+
	1648	+ // Allow for feed-specific syndicated_post filters.
	1649	+ $this->post = apply_filters(
	1650	+ "syndicated_post_".$this->link->uri(),
	1651	+ $this->post,
	1652	+ $this
	1653	+ );
343	1654	endif;
344		-
345		- if (!$this->filtered() and $freshness == 2) :
346		- // The item has not yet been added. So let's add it.
347		- $this->insert_new();
348		- $this->add_rss_meta();
349		- do_action('post_syndicated_item', $this->wp_id());
350	1655
351		- $ret = 'new';
352		- elseif (!$this->filtered() and $freshness == 1) :
353		- $this->post['ID'] = $this->wp_id();
354		- $this->update_existing();
355		- $this->add_rss_meta();
356		- do_action('update_syndicated_item', $this->wp_id());
	1656	+ // Hook in early to make sure these get inserted if at all possible
	1657	+ add_action(
	1658	+ /*hook=*/ 'transition_post_status',
	1659	+ /*callback=*/ array($this, 'add_rss_meta'),
	1660	+ /*priority=*/ -10000, /* very early */
	1661	+ /*arguments=*/ 3
	1662	+ );
357	1663
358		- $ret = 'updated';
359		- else :
360		- $ret = false;
	1664	+ $ret = false;
	1665	+ if ($this->has_fresh_content()) :
	1666	+ $diag = $this->fresh_storage_diagnostic();
	1667	+ if ( !is_null($diag)) :
	1668	+ FeedWordPress::diagnostic('syndicated_posts', $diag);
	1669	+ endif;
	1670	+
	1671	+ $this->insert_post(/*update=*/ $this->fresh_content_is_update(), $this->freshness());
	1672	+
	1673	+ $hook = $this->fresh_storage_hook();
	1674	+ if ( !is_null($hook)) :
	1675	+ do_action($hook, $this->wp_id(), $this);
	1676	+ endif;
	1677	+
	1678	+ $ret = $this->freshness('status');
361	1679	endif;
362		-
	1680	+
	1681	+ // If this is a legit, non-filtered post, tag it as found on the
	1682	+ // feed regardless of fresh or stale status
	1683	+ if ( ! $this->filtered()) :
	1684	+ $key = '_feedwordpress_retire_me_' . $this->link->id;
	1685	+ delete_post_meta($this->wp_id(), $key);
	1686	+
	1687	+ $status = get_post_field('post_status', $this->wp_id());
	1688	+ if ('fwpretired'==$status and $this->link->is_non_incremental()) :
	1689	+ FeedWordPress::diagnostic('syndicated_posts', "Un-retiring previously retired post # ".$this->wp_id()." due to re-appearance on non-incremental feed.");
	1690	+ set_post_field('post_status', $this->post['post_status'], $this->wp_id());
	1691	+ wp_transition_post_status($this->post['post_status'], $status, $this->post);
	1692	+ elseif ('fwpzapped'==$status) :
	1693	+ // Set this new revision up to be
	1694	+ // blanked on the next update.
	1695	+ add_post_meta($this->wp_id(), '_feedwordpress_zapped_blank_me', 2, /*single=*/ true);
	1696	+ endif;
	1697	+ endif;
	1698	+
	1699	+ // Remove add_rss_meta hook
	1700	+ remove_action(
	1701	+ /*hook=*/ 'transition_post_status',
	1702	+ /*callback=*/ array($this, 'add_rss_meta'),
	1703	+ /*priority=*/ -10000, /* very early */
	1704	+ /*arguments=*/ 3
	1705	+ );
	1706	+
363	1707	return $ret;
364		- } // function SyndicatedPost::store ()
365		-
366		- function insert_new () {
367		- global $wpdb, $wp_db_version;
	1708	+ } /* function SyndicatedPost::store () */
368	1709
	1710	+ function insert_post ($update = false, $freshness = 2) {
369	1711	$dbpost = $this->normalize_post(/*new=*/ true);
370		- if (!is_null($dbpost)) :
371		- if ($this->use_api('wp_insert_post')) :
372		- $dbpost['post_pingback'] = false; // Tell WP 2.1 and 2.2 not to process for pingbacks
373		-
374		- // This is a ridiculous fucking kludge necessitated by WordPress 2.6 munging authorship meta-data
375		- add_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
376		-
377		- // Kludge to prevent kses filters from stripping the
378		- // content of posts when updating without a logged in
379		- // user who has `unfiltered_html` capability.
380		- add_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11);
381		-
382		- $this->_wp_id = wp_insert_post($dbpost);
383		-
384		- // Turn off ridiculous fucking kludges #1 and #2
385		- remove_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
386		- remove_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11);
387		-
388		- $this->validate_post_id($dbpost, array(__CLASS__, __FUNCTION__));
389		-
390		- // Unfortunately, as of WordPress 2.3, wp_insert_post()
391		- // *still* offers no way to use a guid of your choice,
392		- // and munges your post modified timestamp, too.
393		- $result = $wpdb->query("
394		- UPDATE $wpdb->posts
395		- SET
396		- guid='{$dbpost['guid']}',
397		- post_modified='{$dbpost['post_modified']}',
398		- post_modified_gmt='{$dbpost['post_modified_gmt']}'
399		- WHERE ID='{$this->_wp_id}'
400		- ");
401		- else :
402		- # The right way to do this is the above. But, alas,
403		- # in earlier versions of WordPress, wp_insert_post has
404		- # too much behavior (mainly related to pings) that can't
405		- # be overridden. In WordPress 1.5, it's enough of a
406		- # resource hog to make PHP segfault after inserting
407		- # 50-100 posts. This can get pretty annoying, especially
408		- # if you are trying to update your feeds for the first
409		- # time.
410		-
411		- $result = $wpdb->query("
412		- INSERT INTO $wpdb->posts
413		- SET
414		- guid = '{$dbpost['guid']}',
415		- post_author = '{$dbpost['post_author']}',
416		- post_date = '{$dbpost['post_date']}',
417		- post_date_gmt = '{$dbpost['post_date_gmt']}',
418		- post_content = '{$dbpost['post_content']}',"
419		- .(isset($dbpost['post_excerpt']) ? "post_excerpt = '{$dbpost['post_excerpt']}'," : "")."
420		- post_title = '{$dbpost['post_title']}',
421		- post_name = '{$dbpost['post_name']}',
422		- post_modified = '{$dbpost['post_modified']}',
423		- post_modified_gmt = '{$dbpost['post_modified_gmt']}',
424		- comment_status = '{$dbpost['comment_status']}',
425		- ping_status = '{$dbpost['ping_status']}',
426		- post_status = '{$dbpost['post_status']}'
427		- ");
428		- $this->_wp_id = $wpdb->insert_id;
429		-
430		- $this->validate_post_id($dbpost, array(__CLASS__, __FUNCTION__));
431		-
432		- // WordPress 1.5.x - 2.0.x
433		- wp_set_post_cats('1', $this->wp_id(), $this->post['post_category']);
434		-
435		- // Since we are not going through official channels, we need to
436		- // manually tell WordPress that we've published a new post.
437		- // We need to make sure to do this in order for FeedWordPress
438		- // to play well with the staticize-reloaded plugin (something
439		- // that a large aggregator website is going to *want* to be
440		- // able to use).
441		- do_action('publish_post', $this->_wp_id);
	1712	+
	1713	+ $ret = null;
	1714	+
	1715	+ if ( !is_null($dbpost)) :
	1716	+ $dbpost['post_pingback'] = false; // Tell WP 2.1 and 2.2 not to process for pingbacks
	1717	+
	1718	+ // This is a ridiculous fucking kludge necessitated by WordPress 2.6 munging authorship meta-data
	1719	+ add_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
	1720	+
	1721	+ // Kludge to prevent kses filters from stripping the
	1722	+ // content of posts when updating without a logged in
	1723	+ // user who has `unfiltered_html` capability.
	1724	+ $mungers = array('wp_filter_kses', 'wp_filter_post_kses');
	1725	+ $removed = array();
	1726	+ foreach ($mungers as $munger) :
	1727	+ if (has_filter('content_save_pre', $munger)) :
	1728	+ remove_filter('content_save_pre', $munger);
	1729	+ $removed[] = $munger;
	1730	+ endif;
	1731	+ endforeach;
	1732	+
	1733	+ if ($update and function_exists('get_post_field')) :
	1734	+ // Don't munge status fields that the user may
	1735	+ // have reset manually
	1736	+ $doNotMunge = array('post_status', 'comment_status', 'ping_status');
	1737	+
	1738	+ foreach ($doNotMunge as $field) :
	1739	+ $dbpost[$field] = get_post_field($field, $this->wp_id());
	1740	+ endforeach;
442	1741	endif;
443		- endif;
444		- } /* SyndicatedPost::insert_new() */
445	1742
446		- function update_existing () {
447		- global $wpdb;
	1743	+ // WP3's wp_insert_post scans current_user_can() for the
	1744	+ // tax_input, with no apparent way to override. Ugh.
	1745	+ add_action(
	1746	+ /*hook=*/ 'transition_post_status',
	1747	+ /*callback=*/ array($this, 'add_terms'),
	1748	+ /*priority=*/ -10001, /* very early */
	1749	+ /*arguments=*/ 3
	1750	+ );
448	1751
449		- // Why the fuck doesn't wp_insert_post already do this?
450		- $dbpost = $this->normalize_post(/*new=*/ false);
451		- if (!is_null($dbpost)) :
452		- if ($this->use_api('wp_insert_post')) :
453		- $dbpost['post_pingback'] = false; // Tell WP 2.1 and 2.2 not to process for pingbacks
454		-
455		- // This is a ridiculous fucking kludge necessitated by WordPress 2.6 munging authorship meta-data
456		- add_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
457		-
458		- // Kludge to prevent kses filters from stripping the
459		- // content of posts when updating without a logged in
460		- // user who has `unfiltered_html` capability.
461		- add_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11);
	1752	+ // WP3 appears to override whatever you give it for
	1753	+ // post_modified. Ugh.
	1754	+ add_action(
	1755	+ /*hook=*/ 'transition_post_status',
	1756	+ /*callback=*/ array($this, 'fix_post_modified_ts'),
	1757	+ /*priority=*/ -10000, /* very early */
	1758	+ /*arguments=*/ 3
	1759	+ );
462	1760
463		- // Don't munge status fields that the user may have reset manually
464		- if (function_exists('get_post_field')) :
465		- $doNotMunge = array('post_status', 'comment_status', 'ping_status');
466		- foreach ($doNotMunge as $field) :
467		- $dbpost[$field] = get_post_field($field, $this->wp_id());
468		- endforeach;
469		- endif;
	1761	+ if ($update) :
	1762	+ $this->post['ID'] = $this->wp_id();
	1763	+ $dbpost['ID'] = $this->post['ID'];
	1764	+ endif;
470	1765
471		- $this->_wp_id = wp_insert_post($dbpost);
472		-
473		- // Turn off ridiculous fucking kludges #1 and #2
474		- remove_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
475		- remove_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11);
476		-
477		- $this->validate_post_id($dbpost, array(__CLASS__, __FUNCTION__));
478		-
479		- // Unfortunately, as of WordPress 2.3, wp_insert_post()
480		- // munges your post modified timestamp.
481		- $result = $wpdb->query("
482		- UPDATE $wpdb->posts
483		- SET
484		- post_modified='{$dbpost['post_modified']}',
485		- post_modified_gmt='{$dbpost['post_modified_gmt']}'
486		- WHERE ID='{$this->_wp_id}'
487		- ");
488		- else :
489		-
490		- $result = $wpdb->query("
491		- UPDATE $wpdb->posts
492		- SET
493		- post_author = '{$dbpost['post_author']}',
494		- post_content = '{$dbpost['post_content']}',"
495		- .(isset($dbpost['post_excerpt']) ? "post_excerpt = '{$dbpost['post_excerpt']}'," : "")."
496		- post_title = '{$dbpost['post_title']}',
497		- post_name = '{$dbpost['post_name']}',
498		- post_modified = '{$dbpost['post_modified']}',
499		- post_modified_gmt = '{$dbpost['post_modified_gmt']}'
500		- WHERE guid='{$dbpost['guid']}'
501		- ");
502		-
503		- // WordPress 2.1.x and up
504		- if (function_exists('wp_set_post_categories')) :
505		- wp_set_post_categories($this->wp_id(), $this->post['post_category']);
506		- // WordPress 1.5.x - 2.0.x
507		- elseif (function_exists('wp_set_post_cats')) :
508		- wp_set_post_cats('1', $this->wp_id(), $this->post['post_category']);
509		- // This should never happen.
510		- else :
511		- FeedWordPress::critical_bug(__CLASS__.'::'.__FUNCTION.'(): no post categorizing function', array("dbpost" => $dbpost, "this" => $this), __LINE__);
	1766	+ // O.K., is this a new post? If so, we need to create
	1767	+ // the basic post record before we do anything else.
	1768	+ if ($this->this_revision_needs_original_post()) :
	1769	+ // *sigh*, for handling inconsistent slash expectations < 3.6
	1770	+ $sdbpost = $this->db_sanitize_post($dbpost);
	1771	+
	1772	+ // Go ahead and insert the first post record to
	1773	+ // anchor the revision history.
	1774	+
	1775	+ $this->_wp_id = wp_insert_post($sdbpost, /*return wp_error=*/ true);
	1776	+
	1777	+ $dbpost['ID'] = $this->_wp_id;
	1778	+ endif;
	1779	+
	1780	+ // Sanity check: if the attempt to insert post
	1781	+ // returned an error, then feeding that error
	1782	+ // object in to _wp_put_post_revision() would
	1783	+ // cause a fatal error. Better to break out.
	1784	+ if ( !is_wp_error($this->_wp_id)) :
	1785	+ // Now that we've made sure the original exists, insert
	1786	+ // this version here as a revision.
	1787	+ $revision_id = _wp_put_post_revision($dbpost, /*autosave=*/ false);
	1788	+
	1789	+ if ( ! $this->this_revision_needs_original_post()) :
	1790	+
	1791	+ if ($this->this_revision_is_current()) :
	1792	+
	1793	+ wp_restore_post_revision($revision_id);
	1794	+
	1795	+ else :
	1796	+
	1797	+ // If we do not activate this revision, then the
	1798	+ // add_rss_meta will not be called, which is
	1799	+ // more or less as it should be, but that means
	1800	+ // we have to actively record this revision's
	1801	+ // update hash from here.
	1802	+ $postId = $this->post['ID'];
	1803	+ $key = 'syndication_item_hash';
	1804	+ $hash = $this->update_hash();
	1805	+ FeedWordPress::diagnostic('syndicated_posts:meta_data', "Adding post meta-datum to post [$postId]: [$key] = ".FeedWordPress::val($hash, /*no newlines=*/ true));
	1806	+ add_post_meta( $postId, $key, $hash, /*unique=*/ false );
	1807	+ endif;
512	1808	endif;
513		-
514		- // Since we are not going through official channels, we need to
515		- // manually tell WordPress that we've published a new post.
516		- // We need to make sure to do this in order for FeedWordPress
517		- // to play well with the staticize-reloaded plugin (something
518		- // that a large aggregator website is going to *want* to be
519		- // able to use).
520		- do_action('edit_post', $this->post['ID']);
521	1809	endif;
	1810	+
	1811	+ remove_action(
	1812	+ /*hook=*/ 'transition_post_status',
	1813	+ /*callback=*/ array($this, 'add_terms'),
	1814	+ /*priority=*/ -10001, /* very early */
	1815	+ /*arguments=*/ 3
	1816	+ );
	1817	+
	1818	+ remove_action(
	1819	+ /*hook=*/ 'transition_post_status',
	1820	+ /*callback=*/ array($this, 'fix_post_modified_ts'),
	1821	+ /*priority=*/ -10000, /* very early */
	1822	+ /*arguments=*/ 3
	1823	+ );
	1824	+
	1825	+ // Turn off ridiculous fucking kludges #1 and #2
	1826	+ remove_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
	1827	+ foreach ($removed as $filter) :
	1828	+ add_filter('content_save_pre', $filter);
	1829	+ endforeach;
	1830	+
	1831	+ $this->validate_post_id($dbpost, $update, array(__CLASS__, __FUNCTION__));
	1832	+
	1833	+ $ret = $this->_wp_id;
522	1834	endif;
	1835	+ return $ret;
	1836	+ } /* function SyndicatedPost::insert_post () */
	1837	+
	1838	+ /**
	1839	+ * SyndicatedPost::insert_new(). Uses the data collected in this post object to insert
	1840	+ * a new post into the wp_posts table.
	1841	+ *
	1842	+ * @uses SyndicatedPost::insert_post
	1843	+ */
	1844	+ function insert_new () {
	1845	+ $this->insert_post(/*update=*/ false, 1);
	1846	+ } /* SyndicatedPost::insert_new() */
	1847	+
	1848	+ /**
	1849	+ * SyndicatedPost::insert_new(). Uses the data collected in this post object to update
	1850	+ * an existing post in the wp_posts table.
	1851	+ *
	1852	+ * @uses SyndicatedPost::insert_post
	1853	+ */
	1854	+ function update_existing () {
	1855	+ $this->insert_post(/*update=*/ true, 2);
523	1856	} /* SyndicatedPost::update_existing() */
524	1857
525	1858	/**
526	1859	* SyndicatedPost::normalize_post()
527	1860	*
528		- * @param bool $new If true, this post is to be inserted anew. If false, it is an update of an existing post.
	1861	+ * @param bool $new If true, this post is to be inserted anew. If false, it is an update of an existing post. (Unused)
529	1862	* @return array A normalized representation of the post ready to be inserted into the database or sent to the WordPress API functions
530	1863	*/
531		- function normalize_post ($new = true) {
532		- global $wpdb;
	1864	+ function normalize_post( $new = true ) {
	1865	+ $out = $this->post;
533	1866
534		- $out = array();
	1867	+ $fullPost = $out['post_title'].$out['post_content'];
	1868	+ $fullPost .= (isset($out['post_excerpt']) ? $out['post_excerpt'] : '');
	1869	+ if (strlen($fullPost) < 1) :
	1870	+ // FIXME: Option for filtering out empty posts
	1871	+ endif;
	1872	+ if (strlen($out['post_title'])==0) :
	1873	+ $offset = (int) get_option('gmt_offset') * 60 * 60;
	1874	+ if (isset($this->post['meta']['syndication_source'])) :
	1875	+ $source_title = $this->post['meta']['syndication_source'];
	1876	+ else :
	1877	+ $feed_url = parse_url($this->post['meta']['syndication_feed']);
	1878	+ $source_title = $feed_url['host'];
	1879	+ endif;
535	1880
536		- // Why the fuck doesn't wp_insert_post already do this?
537		- foreach ($this->post as $key => $value) :
	1881	+ $out['post_title'] = $source_title
	1882	+ .' '.gmdate('Y-m-d H:i:s', $this->published() + $offset);
	1883	+ // FIXME: Option for what to fill a blank title with...
	1884	+ endif;
	1885	+
	1886	+ // Normalize the guid if necessary.
	1887	+ $out['guid'] = SyndicatedPost::normalize_guid($out['guid']);
	1888	+
	1889	+ return $out;
	1890	+ }
	1891	+
	1892	+ public function db_sanitize_post_check_encoding ($out) {
	1893	+ // Check encoding recursively: every string field needs to be checked
	1894	+ // for character encoding issues. This is a bit problematic because we
	1895	+ // *should* be using DB_CHARSET, but DB_CHARSET sometimes has values
	1896	+ // that work for MySQL but not for PHP mb_check_encoding. So instead
	1897	+ // we must rely on WordPress setting blog_charset and hope that the user
	1898	+ // has got their database encoding set up to roughly match
	1899	+ $charset = get_option('blog_charset', 'utf8');
	1900	+
	1901	+ foreach ($out as $key => $value) :
538	1902	if (is_string($value)) :
539		- $out[$key] = $wpdb->escape($value);
	1903	+
	1904	+ if ( !function_exists('mb_check_encoding') or mb_check_encoding($value, $charset)) :
	1905	+ $out[$key] = $value;
	1906	+ else :
	1907	+ $fromCharset = mb_detect_encoding($value, mb_detect_order(), /*strict=*/ true);
	1908	+ $out[$key] = mb_convert_encoding($value, $charset, $fromCharset);
	1909	+ endif;
	1910	+
	1911	+ elseif (is_array($value)) :
	1912	+ $out[$key] = $this->db_sanitize_post_check_encoding($value);
	1913	+
540	1914	else :
541	1915	$out[$key] = $value;
542	1916	endif;
	1917	+
543	1918	endforeach;
544	1919
545		- if (strlen($out['post_title'].$out['post_content'].$out['post_excerpt']) == 0) :
546		- // FIXME: Option for filtering out empty posts
	1920	+ return $out;
	1921	+ } /* SyndicatedPost::db_sanitize_post_check_encoding () */
	1922	+
	1923	+ function db_sanitize_post ($out) {
	1924	+ global $wp_db_version;
	1925	+
	1926	+ $out = $this->db_sanitize_post_check_encoding($out);
	1927	+
	1928	+ // < 3.6. Core API, including `wp_insert_post()`, expects
	1929	+ // properly slashed data. If `wp_slash()` exists, then
	1930	+ // this is after the big change-over in how data slashing
	1931	+ // was handled.
	1932	+ if ( !function_exists('wp_slash')) :
	1933	+
	1934	+ foreach ($out as $key => $value) :
	1935	+ if (is_string($value)) :
	1936	+ $out[$key] = esc_sql($value);
	1937	+ else :
	1938	+ $out[$key] = $value;
	1939	+ endif;
	1940	+ endforeach;
	1941	+
	1942	+ // For revisions [@23416,@23554), core API expects
	1943	+ // unslashed data. Cf. <https://core.trac.wordpress.org/browser/trunk/wp-includes/post.php?rev=23416>
	1944	+ // NOOP for those revisions.
	1945	+
	1946	+ // In revisions @23554 to present, `wp_insert_post()`
	1947	+ // expects slashed data once again.
	1948	+ // Cf. <https://core.trac.wordpress.org/changeset/23554/trunk/wp-includes/post.php?contextall=1>
	1949	+ // But at least now we can use the wp_slash API function to do that.
	1950	+ // Hooray.
	1951	+
	1952	+ elseif ($wp_db_version >= 23524) :
	1953	+
	1954	+ $out = wp_slash($out);
	1955	+
547	1956	endif;
548		- if (strlen($out['post_title'])==0) :
549		- $offset = (int) get_option('gmt_offset') * 60 * 60;
550		- $out['post_title'] =
551		- $this->post['meta']['syndication_source']
552		- .' '.gmdate('Y-m-d H:i:s', $this->published() + $offset);
553		- // FIXME: Option for what to fill a blank title with...
554		- endif;
555	1957
556	1958	return $out;
557	1959	}
558	1960
		@@ -561,35 +1963,57 @@
561	1963	*
562	1964	* @param array $dbpost An array representing the post we attempted to insert or update
563	1965	* @param mixed $ns A string or array representing the namespace (class, method) whence this method was called.
564	1966	*/
565		- function validate_post_id ($dbpost, $ns) {
	1967	+ function validate_post_id ($dbpost, $is_update, $ns) {
566	1968	if (is_array($ns)) : $ns = implode('::', $ns);
567	1969	else : $ns = (string) $ns; endif;
568		-
	1970	+
569	1971	// This should never happen.
570		- if (!is_numeric($this->_wp_id) or ($this->_wp_id == 0)) :
571		- FeedWordPress::critical_bug(
572		- /*name=*/ $ns.'::_wp_id',
	1972	+ if ( !is_numeric($this->_wp_id) or ($this->_wp_id == 0)) :
	1973	+ $verb = ($is_update ? 'update existing' : 'insert new');
	1974	+ $guid = $this->guid();
	1975	+ $url = $this->permalink();
	1976	+ $feed = $this->link->uri(array('add_params' => true));
	1977	+
	1978	+ // wp_insert_post failed. Diagnostics, or barf up a critical bug
	1979	+ // notice if we are in debug mode.
	1980	+ $mesg = "Failed to $verb item [$guid]. WordPress API returned no valid post ID.\n"
	1981	+ ."\t\tID = ".serialize($this->_wp_id)."\n"
	1982	+ ."\t\tURL = ".MyPHP::val($url)
	1983	+ ."\t\tFeed = ".MyPHP::val($feed);
	1984	+
	1985	+ FeedWordPress::diagnostic('updated_feeds:errors', "WordPress API error: $mesg");
	1986	+ FeedWordPress::diagnostic('feed_items:rejected', $mesg);
	1987	+
	1988	+ $mesg = <<<EOM
	1989	+The WordPress API returned an invalid post ID
	1990	+ when FeedWordPress tried to $verb item $guid
	1991	+ [URL: $url]
	1992	+ from the feed at $feed
	1993	+
	1994	+$ns::_wp_id
	1995	+EOM;
	1996	+ FeedWordPressDiagnostic::noncritical_bug(
	1997	+ /*message=*/ $mesg,
573	1998	/*var =*/ array(
574	1999	"\$this->_wp_id" => $this->_wp_id,
575	2000	"\$dbpost" => $dbpost,
576		- "\$this" => $this
577	2001	),
578		- /*line # =*/ __LINE__
	2002	+ /*line # =*/ __LINE__, /*filename=*/ __FILE__
579	2003	);
580	2004	endif;
581	2005	} /* SyndicatedPost::validate_post_id() */
582		-
	2006	+
583	2007	/**
584		- * SyndicatedPost::fix_revision_meta() - Fixes the way WP 2.6+ fucks up
585		- * meta-data (authorship, etc.) when storing revisions of an updated
586		- * syndicated post.
	2008	+ * SyndicatedPost::fix_revision_meta() - Ensures that we get the meta
	2009	+ * data (authorship, guid, etc.) that we want when storing revisions of
	2010	+ * a syndicated post.
587	2011	*
588		- * In their infinite wisdom, the WordPress coders have made it completely
589		- * impossible for a plugin that uses wp_insert_post() to set certain
590		- * meta-data (such as the author) when you store an old revision of an
591		- * updated post. Instead, it uses the WordPress defaults (= currently
	2012	+ * In their infinite wisdom, the WordPress coders seem to have made it
	2013	+ * completely impossible for a plugin that uses wp_insert_post() to set
	2014	+ * certain meta-data (such as the author) when you store an old revision
	2015	+ * of an updated post. Instead, it uses the WordPress defaults (= cur.
592	2016	* active user ID if the process is running with a user logged in, or
593	2017	* = #0 if there is no user logged in). This results in bogus authorship
594	2018	* data for revisions that are syndicated from off the feed, unless we
595	2019	* use a ridiculous kludge like this to end-run the munging of meta-data
		@@ -596,106 +2020,209 @@
596	2020	* by _wp_put_post_revision.
597	2021	*
598	2022	* @param int $revision_id The revision ID to fix up meta-data
599	2023	*/
600		- function fix_revision_meta ($revision_id) {
601		- global $wpdb;
602		-
603		- $post_author = (int) $this->post['post_author'];
604		-
	2024	+ function fix_revision_meta( $revision_id ) {
	2025	+ $post_author = (int) $this->post['post_author']; // is this a global? And if it is, where is it defined? (gwyneth 20230919)
	2026	+
605	2027	$revision_id = (int) $revision_id;
606		- $wpdb->query("
607		- UPDATE $wpdb->posts
608		- SET post_author={$this->post['post_author']}
609		- WHERE post_type = 'revision' AND ID='$revision_id'
610		- ");
	2028	+
	2029	+ // Let's fix the author.
	2030	+ set_post_field( 'post_author', $this->post['post_author'], $revision_id );
	2031	+
	2032	+ // Let's fix the GUID to a dummy URL with the update hash.
	2033	+ set_post_field( 'guid', 'http://feedwordpress.radgeek.com/?rev=' . $this->update_hash(), $revision_id );
	2034	+
	2035	+ // Let's fire an event for add-ons and filters
	2036	+ do_action( 'syndicated_post_fix_revision_meta', $revision_id, $this );
	2037	+
611	2038	} /* SyndicatedPost::fix_revision_meta () */
612	2039
613	2040	/**
614		- * SyndicatedPost::avoid_kses_munge() -- If FeedWordPress is processing
615		- * an automatic update, that generally means that wp_insert_post() is
616		- * being called under the user credentials of whoever is viewing the
617		- * blog at the time -- usually meaning no user at all. But if WordPress
618		- * gets a wp_insert_post() when current_user_can('unfiltered_html') is
619		- * false, it will run the content of the post through a kses function
620		- * that strips out lots of HTML tags -- notably <object> and some others.
621		- * This causes problems for syndicating (for example) feeds that contain
622		- * YouTube videos. It also produces an unexpected asymmetry between
623		- * automatically-initiated updates and updates initiated manually from
624		- * the WordPress Dashboard (which are usually initiated under the
625		- * credentials of a logged-in admin, and so don't get run through the
626		- * kses function). So, to avoid the whole mess, what we do here is
627		- * just forcibly disable the kses munging for a single syndicated post,
628		- * by restoring the contents of the `post_content` field.
	2041	+ * SyndicatedPost::add_terms() -- if FeedWordPress is processing an
	2042	+ * automatic update, that generally means that wp_insert_post() is being
	2043	+ * called under the user credentials of whoever is viewing the blog at
	2044	+ * the time -- which usually means no user at all. But wp_insert_post()
	2045	+ * checks current_user_can() before assigning any of the terms in a
	2046	+ * post's tax_input structure -- which is unfortunate, since
	2047	+ * current_user_can() always returns FALSE when there is no current user
	2048	+ * logged in. Meaning that automatic updates get no terms assigned.
629	2049	*
630		- * @param string $content The content of the post, after other filters have gotten to it
631		- * @return string The original content of the post, before other filters had a chance to munge it.
	2050	+ * So, wp_insert_post() is not going to do the term assignments for us.
	2051	+ * If you want something done right....
	2052	+ *
	2053	+ * @param string $new_status Unused action parameter.
	2054	+ * @param string $old_status Unused action parameter.
	2055	+ * @param object $post The database record for the post just inserted.
632	2056	*/
633		- function avoid_kses_munge ($content) {
634		- global $wpdb;
635		- return $wpdb->escape($this->post['post_content']);
636		- }
637		-
638		- // SyndicatedPost::add_rss_meta: adds interesting meta-data to each entry
639		- // using the space for custom keys. The set of keys and values to add is
640		- // specified by the keys and values of $post['meta']. This is used to
641		- // store anything that the WordPress user might want to access from a
642		- // template concerning the post's original source that isn't provided
643		- // for by standard WP meta-data (i.e., any interesting data about the
644		- // syndicated post other than author, title, timestamp, categories, and
645		- // guid). It's also used to hook into WordPress's support for
646		- // enclosures.
647		- function add_rss_meta () {
	2057	+ function add_terms ($new_status, $old_status, $post) {
	2058	+
	2059	+ if ($new_status!='inherit') : // Bail if we are creating a revision.
	2060	+ if ( is_array($this->post) and isset($this->post['tax_input']) and is_array($this->post['tax_input']) ) :
	2061	+ foreach ($this->post['tax_input'] as $taxonomy => $terms) :
	2062	+ if (is_array($terms)) :
	2063	+ $terms = array_filter($terms); // strip out empties
	2064	+ endif;
	2065	+
	2066	+ $res = wp_set_post_terms(
	2067	+ /*post_id=*/ $post->ID,
	2068	+ /*terms=*/ $terms,
	2069	+ /*taxonomy=*/ $taxonomy
	2070	+ );
	2071	+
	2072	+ FeedWordPress::diagnostic(
	2073	+ 'syndicated_posts:categories',
	2074	+ 'Category: post('.json_encode($post->ID).') '.$taxonomy
	2075	+ .' := '
	2076	+ .json_encode($terms)
	2077	+ .' / result: '
	2078	+ .json_encode($res)
	2079	+ );
	2080	+
	2081	+ endforeach;
	2082	+ endif;
	2083	+ endif;
	2084	+
	2085	+ } /* SyndicatedPost::add_terms () */
	2086	+
	2087	+ /**
	2088	+ * SyndicatedPost::fix_post_modified_ts() -- We would like to set
	2089	+ * post_modified and post_modified_gmt to reflect the value of
	2090	+ * <atom:updated> or equivalent elements on the feed. Unfortunately,
	2091	+ * wp_insert_post() refuses to acknowledge explicitly-set post_modified
	2092	+ * fields and overwrites them, either with the post_date (if new) or the
	2093	+ * current timestamp (if updated).
	2094	+ *
	2095	+ * So, wp_insert_post() is not going to do the last-modified assignments
	2096	+ * for us. If you want something done right....
	2097	+ *
	2098	+ * @param string $new_status Unused action parameter.
	2099	+ * @param string $old_status Unused action parameter.
	2100	+ * @param object $post The database record for the post just inserted.
	2101	+ */
	2102	+ function fix_post_modified_ts ($new_status, $old_status, $post) {
648	2103	global $wpdb;
649		- if ( is_array($this->post) and isset($this->post['meta']) and is_array($this->post['meta']) ) :
650		- $postId = $this->wp_id();
651		-
652		- // Aggregated posts should NOT send out pingbacks.
653		- // WordPress 2.1-2.2 claim you can tell them not to
654		- // using $post_pingback, but they don't listen, so we
655		- // make sure here.
656		- $result = $wpdb->query("
657		- DELETE FROM $wpdb->postmeta
658		- WHERE post_id='$postId' AND meta_key='_pingme'
659		- ");
	2104	+ if ($new_status!='inherit') : // Bail if we are creating a revision.
	2105	+ $wpdb->update( $wpdb->posts, /*data=*/ array(
	2106	+ 'post_modified' => $this->post['post_modified'],
	2107	+ 'post_modified_gmt' => $this->post['post_modified_gmt'],
	2108	+ ), /*where=*/ array('ID' => $post->ID) );
	2109	+ endif;
	2110	+ } /* SyndicatedPost::fix_post_modified_ts () */
660	2111
661		- foreach ( $this->post['meta'] as $key => $values ) :
	2112	+ /**
	2113	+ * SyndicatedPost::add_rss_meta: adds interesting meta-data to each entry
	2114	+ * using the space for custom keys. The set of keys and values to add is
	2115	+ * specified by the keys and values of $post['meta']. This is used to
	2116	+ * store anything that the WordPress user might want to access from a
	2117	+ * template concerning the post's original source that isn't provided
	2118	+ * for by standard WP meta-data (i.e., any interesting data about the
	2119	+ * syndicated post other than author, title, timestamp, categories, and
	2120	+ * guid). It's also used to hook into WordPress's support for
	2121	+ * enclosures.
	2122	+ *
	2123	+ * @param string $new_status Unused action parameter.
	2124	+ * @param string $old_status Unused action parameter.
	2125	+ * @param object $post The database record for the post just inserted.
	2126	+ */
	2127	+ function add_rss_meta($new_status, $old_status, $post) {
	2128	+ global $wpdb;
	2129	+ if ($new_status!='inherit') : // Bail if we are creating a revision.
	2130	+ FeedWordPress::diagnostic('syndicated_posts:meta_data', 'Adding post meta-data: {'.implode(", ", array_keys($this->post['meta'])).'}');
662	2131
663		- $key = $wpdb->escape($key);
	2132	+ if ( is_array($this->post) and isset($this->post['meta']) and is_array($this->post['meta']) ) :
	2133	+ $postId = $post->ID;
664	2134
665		- // If this is an update, clear out the old
666		- // values to avoid duplication.
	2135	+ // Aggregated posts should NOT send out pingbacks.
	2136	+ // WordPress 2.1-2.2 claim you can tell them not to
	2137	+ // using $post_pingback, but they don't listen, so we
	2138	+ // make sure here.
667	2139	$result = $wpdb->query("
668	2140	DELETE FROM $wpdb->postmeta
669		- WHERE post_id='$postId' AND meta_key='$key'
	2141	+ WHERE post_id='$postId' AND meta_key='_pingme'
670	2142	");
671	2143
672		- // Allow for either a single value or an array
673		- if (!is_array($values)) $values = array($values);
674		- foreach ( $values as $value ) :
675		- $value = $wpdb->escape($value);
	2144	+ foreach ( $this->post['meta'] as $key => $values ) :
	2145	+ $eKey = esc_sql($key);
	2146	+
	2147	+ // If this is an update, clear out the old
	2148	+ // values to avoid duplication.
676	2149	$result = $wpdb->query("
677		- INSERT INTO $wpdb->postmeta
678		- SET
679		- post_id='$postId',
680		- meta_key='$key',
681		- meta_value='$value'
	2150	+ DELETE FROM $wpdb->postmeta
	2151	+ WHERE post_id='$postId' AND meta_key='$eKey'
682	2152	");
	2153	+
	2154	+ // Allow for either a single value or an array
	2155	+ if ( !is_array($values)) $values = array($values);
	2156	+ foreach ( $values as $value ) :
	2157	+ FeedWordPress::diagnostic('syndicated_posts:meta_data', "Adding post meta-datum to post [$postId]: [$key] = ".MyPHP::val($value, /*no newlines=*/ true));
	2158	+ add_post_meta($postId, $key, $value, /*unique=*/ false);
	2159	+ endforeach;
683	2160	endforeach;
684		- endforeach;
	2161	+
	2162	+ if ( $result === false)
	2163	+ {
	2164	+ error_log( "delete query failed: " . $wpdb->last_error );
	2165	+ }
	2166	+ endif;
685	2167	endif;
686	2168	} /* SyndicatedPost::add_rss_meta () */
687	2169
688		- // SyndicatedPost::author_id (): get the ID for an author name from
689		- // the feed. Create the author if necessary.
	2170	+ /**
	2171	+ * SyndicatedPost::author_id (): get the ID for an author name from
	2172	+ * the feed. Create the author if necessary.
	2173	+ *
	2174	+ * @param string $unfamiliar_author
	2175	+ *
	2176	+ * @return NULL|int The numeric ID of the author to attribute the post to
	2177	+ * NULL if the post should be filtered out.
	2178	+ */
690	2179	function author_id ($unfamiliar_author = 'create') {
691	2180	global $wpdb;
692	2181
693		- $a = $this->author();
694		- $author = $a['name'];
695		- $email = $a['email'];
696		- $url = $a['uri'];
	2182	+ $a = $this->named['author'];
697	2183
	2184	+ $source = $this->source();
	2185	+ $forbidden = apply_filters('feedwordpress_forbidden_author_names',
	2186	+ array('admin', 'administrator', 'www', 'root'));
	2187	+
	2188	+ // Prepare the list of candidates to try for author name: name from
	2189	+ // feed, original source title (if any), immediate source title live
	2190	+ // from feed, subscription title, prettied version of feed homepage URL,
	2191	+ // prettied version of feed URL, or, failing all, use "unknown author"
	2192	+ // as last resort
	2193	+
	2194	+ $candidates = array();
	2195	+ $candidates[] = $a['name'];
	2196	+ if ( !is_null($source)) : $candidates[] = $source['title']; endif;
	2197	+ $candidates[] = $this->link->name(/*fromFeed=*/ true);
	2198	+ $candidates[] = $this->link->name(/*fromFeed=*/ false);
	2199	+ if (strlen($this->link->homepage()) > 0) : $candidates[] = feedwordpress_display_url($this->link->homepage()); endif;
	2200	+ $candidates[] = feedwordpress_display_url($this->link->uri());
	2201	+ $candidates[] = 'unknown author';
	2202	+
	2203	+ // Pick the first one that works from the list, screening against empty
	2204	+ // or forbidden names.
	2205	+
	2206	+ $author = NULL;
	2207	+ foreach ($candidates as $candidate) {
	2208	+ if ( !is_null($candidate)
	2209	+ and (strlen(trim($candidate)) > 0)
	2210	+ and !in_array(strtolower(trim($candidate)), $forbidden)) :
	2211	+ $author = $candidate;
	2212	+ break;
	2213	+ endif;
	2214	+ }
	2215	+
	2216	+ $email = (isset($a['email']) ? $a['email'] : NULL);
	2217	+ $authorUrl = (isset($a['uri']) ? $a['uri'] : NULL);
	2218	+
	2219	+
	2220	+ $hostUrl = $this->link->homepage();
	2221	+ if (is_null($hostUrl) or (strlen($hostUrl) < 0)) :
	2222	+ $hostUrl = $this->link->uri();
	2223	+ endif;
	2224	+
698	2225	$match_author_by_email = !('yes' == get_option("feedwordpress_do_not_match_author_by_email"));
699	2226	if ($match_author_by_email and !FeedWordPress::is_null_email($email)) :
700	2227	$test_email = $email;
701	2228	else :
		@@ -703,21 +2230,43 @@
703	2230	endif;
704	2231
705	2232	// Never can be too careful...
706	2233	$login = sanitize_user($author, /*strict=*/ true);
	2234	+
	2235	+ // Possible for, e.g., foreign script author names
	2236	+ if (strlen($login) < 1) :
	2237	+ // No usable characters in author name for a login.
	2238	+ // (Sometimes results from, e.g., foreign scripts.)
	2239	+ //
	2240	+ // We just need *something* in Western alphanumerics,
	2241	+ // so let's try the domain name.
	2242	+ //
	2243	+ // Uniqueness will be guaranteed below if necessary.
	2244	+
	2245	+ $url = parse_url($hostUrl);
	2246	+
	2247	+ $login = sanitize_user($url['host'], /*strict=*/ true);
	2248	+ if (strlen($login) < 1) :
	2249	+ // This isn't working. Frak it.
	2250	+ $login = 'syndicated';
	2251	+ endif;
	2252	+ endif;
	2253	+
707	2254	$login = apply_filters('pre_user_login', $login);
708	2255
709	2256	$nice_author = sanitize_title($author);
710	2257	$nice_author = apply_filters('pre_user_nicename', $nice_author);
711	2258
712		- $reg_author = $wpdb->escape(preg_quote($author));
713		- $author = $wpdb->escape($author);
714		- $email = $wpdb->escape($email);
715		- $test_email = $wpdb->escape($test_email);
716		- $url = $wpdb->escape($url);
	2259	+ $reg_author = esc_sql(preg_quote($author ?: "(unknown)"));
	2260	+ $author = esc_sql($author);
	2261	+ $email = esc_sql($email);
	2262	+ $test_email = esc_sql($test_email);
	2263	+ $authorUrl = esc_sql($authorUrl);
717	2264
718	2265	// Check for an existing author rule....
719		- if (isset($this->link->settings['map authors']['name'][strtolower(trim($author))])) :
	2266	+ if (isset($this->link->settings['map authors']['name']['*'])) :
	2267	+ $author_rule = $this->link->settings['map authors']['name']['*'];
	2268	+ elseif (isset($this->link->settings['map authors']['name'][strtolower(trim($author))])) :
720	2269	$author_rule = $this->link->settings['map authors']['name'][strtolower(trim($author))];
721	2270	else :
722	2271	$author_rule = NULL;
723	2272	endif;
		@@ -722,75 +2271,46 @@
722	2271	$author_rule = NULL;
723	2272	endif;
724	2273
725	2274	// User name is mapped to a particular author. If that author ID exists, use it.
726		- if (is_numeric($author_rule) and get_userdata((int) $author_rule)) :
	2275	+ // Note: get_userdata() is a pluggable function, so it might not be available yet;
	2276	+ // check if it exists before using it, or WP will throw a fatal error. (gwyneth 20220223)
	2277	+ if (is_numeric($author_rule) and function_exists('get_userdata') and get_userdata((int) $author_rule)) :
727	2278	$id = (int) $author_rule;
728	2279
729	2280	// User name is filtered out
730	2281	elseif ('filter' == $author_rule) :
731	2282	$id = NULL;
732		-
	2283	+
733	2284	else :
734	2285	// Check the database for an existing author record that might fit
735	2286
736		- #-- WordPress 2.0+
737		- if (fwp_test_wp_version(FWP_SCHEMA_HAS_USERMETA)) :
	2287	+ // First try the user core data table.
	2288	+ $id = $wpdb->get_var(
	2289	+ "SELECT ID FROM $wpdb->users
	2290	+ WHERE TRIM(LCASE(display_name)) = TRIM(LCASE('$author'))
	2291	+ OR TRIM(LCASE(user_login)) = TRIM(LCASE('$author'))
	2292	+ OR (
	2293	+ LENGTH(TRIM(LCASE(user_email))) > 0
	2294	+ AND TRIM(LCASE(user_email)) = TRIM(LCASE('$test_email'))
	2295	+ )");
738	2296
739		- // First try the user core data table.
	2297	+ // If that fails, look for aliases in the user meta data table
	2298	+ if (is_null($id)) :
740	2299	$id = $wpdb->get_var(
741		- "SELECT ID FROM $wpdb->users
	2300	+ "SELECT user_id FROM $wpdb->usermeta
742	2301	WHERE
743		- TRIM(LCASE(user_login)) = TRIM(LCASE('$login'))
	2302	+ (meta_key = 'description' AND TRIM(LCASE(meta_value)) = TRIM(LCASE('$author')))
744	2303	OR (
745		- LENGTH(TRIM(LCASE(user_email))) > 0
746		- AND TRIM(LCASE(user_email)) = TRIM(LCASE('$test_email'))
747		- )
748		- OR TRIM(LCASE(user_nicename)) = TRIM(LCASE('$nice_author'))
749		- ");
750		-
751		- // If that fails, look for aliases in the user meta data table
752		- if (is_null($id)) :
753		- $id = $wpdb->get_var(
754		- "SELECT user_id FROM $wpdb->usermeta
755		- WHERE
756		- (meta_key = 'description' AND TRIM(LCASE(meta_value)) = TRIM(LCASE('$author')))
757		- OR (
758		- meta_key = 'description'
759		- AND TRIM(LCASE(meta_value))
760		- RLIKE CONCAT(
761		- '(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*',
762		- TRIM(LCASE('$reg_author')),
763		- '( |\\t|\\r)*(\\n|\$)'
764		- )
765		- )
766		- ");
767		- endif;
768		-
769		- #-- WordPress 1.5.x
770		- else :
771		- $id = $wpdb->get_var(
772		- "SELECT ID from $wpdb->users
773		- WHERE
774		- TRIM(LCASE(user_login)) = TRIM(LCASE('$login')) OR
775		- (
776		- LENGTH(TRIM(LCASE(user_email))) > 0
777		- AND TRIM(LCASE(user_email)) = TRIM(LCASE('$test_email'))
778		- ) OR
779		- TRIM(LCASE(user_firstname)) = TRIM(LCASE('$author')) OR
780		- TRIM(LCASE(user_nickname)) = TRIM(LCASE('$author')) OR
781		- TRIM(LCASE(user_nicename)) = TRIM(LCASE('$nice_author')) OR
782		- TRIM(LCASE(user_description)) = TRIM(LCASE('$author')) OR
783		- (
784		- LOWER(user_description)
	2304	+ meta_key = 'description'
	2305	+ AND TRIM(LCASE(meta_value))
785	2306	RLIKE CONCAT(
786	2307	'(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*',
787		- LCASE('$reg_author'),
	2308	+ TRIM(LCASE('$reg_author')),
788	2309	'( |\\t|\\r)*(\\n|\$)'
789	2310	)
790	2311	)
791	2312	");
792		-
793	2313	endif;
794	2314
795	2315	// ... if you don't find one, then do what you need to do
796	2316	if (is_null($id)) :
		@@ -796,8 +2316,16 @@
796	2316	if (is_null($id)) :
797	2317	if ($unfamiliar_author === 'create') :
798	2318	$userdata = array();
799	2319
	2320	+ #-- we need *something* for the email here or WordPress
	2321	+ #-- is liable to pitch a fit. So, make something up if
	2322	+ #-- necessary. (Ugh.)
	2323	+ if (strlen($email) == 0 or FeedWordPress::is_null_email($email)) :
	2324	+ $url = parse_url($hostUrl);
	2325	+ $email = $nice_author.'@'.$url['host'];
	2326	+ endif;
	2327	+
800	2328	#-- user table data
801	2329	$userdata['ID'] = NULL; // new user
802	2330	$userdata['user_login'] = $login;
803	2331	$userdata['user_nicename'] = $nice_author;
		@@ -802,393 +2330,111 @@
802	2330	$userdata['user_login'] = $login;
803	2331	$userdata['user_nicename'] = $nice_author;
804	2332	$userdata['user_pass'] = substr(md5(uniqid(microtime())), 0, 6); // just something random to lock it up
805	2333	$userdata['user_email'] = $email;
806		- $userdata['user_url'] = $url;
	2334	+ $userdata['user_url'] = $authorUrl;
	2335	+ $userdata['nickname'] = $author;
	2336	+
	2337	+ $parts = preg_split('/\s+/', trim($author), 2);
	2338	+ if (isset($parts[0])) : $userdata['first_name'] = $parts[0]; endif;
	2339	+ if (isset($parts[1])) : $userdata['last_name'] = $parts[1]; endif;
	2340	+
807	2341	$userdata['display_name'] = $author;
808		-
809		- $id = wp_insert_user($userdata);
810		- elseif (is_numeric($unfamiliar_author) and get_userdata((int) $unfamiliar_author)) :
811		- $id = (int) $unfamiliar_author;
812		- elseif ($unfamiliar_author === 'default') :
813		- $id = 1;
814		- endif;
815		- endif;
816		- endif;
	2342	+ $userdata['role'] = 'contributor';
817	2343
818		- if ($id) :
819		- $this->link->settings['map authors']['name'][strtolower(trim($author))] = $id;
820		- endif;
821		- return $id;
822		- } // function SyndicatedPost::author_id ()
	2344	+ #-- loop. Keep trying to add the user until you get it
	2345	+ #-- right. Or until PHP crashes, I guess.
	2346	+ $insanity = 0;
	2347	+ do {
	2348	+ $id = wp_insert_user($userdata);
	2349	+ if (is_wp_error($id)) :
	2350	+ $codes = $id->get_error_code();
	2351	+ switch ($codes) :
	2352	+ case 'empty_user_login' :
	2353	+ case 'existing_user_login' :
	2354	+ case 'invalid_username' :
	2355	+ // Add a random disambiguator
	2356	+ $userdata['user_login'] .= substr(md5(uniqid(microtime())), 0, 6);
	2357	+ break;
	2358	+ case 'user_login_too_long' :
	2359	+ // Limit length to 53 characters; if we end up needing a random disambiguator,
	2360	+ // we should still have space to add it.
	2361	+ $userdata['user_login'] = mb_substr( $userdata['user_login'], 0, 53 );
	2362	+ break;
	2363	+ case 'user_nicename_too_long' :
	2364	+ // Add a limited 50 characters user_nicename based on user_login
	2365	+ $userdata['user_nicename'] = mb_substr( $userdata['user_login'], 0, 50 );
	2366	+ break;
	2367	+ case 'existing_user_email' :
	2368	+ // Disassemble email for username, host
	2369	+ $parts = explode('@', $userdata['user_email'], 2);
823	2370
824		- // look up (and create) category ids from a list of categories
825		- function category_ids ($cats, $unfamiliar_category = 'create', $tags_too = false) {
826		- global $wpdb;
	2371	+ // Add a random disambiguator as a gmail-style username extension
	2372	+ $parts[0] .= '+'.substr(md5(uniqid(microtime())), 0, 6);
827	2373
828		- // We need to normalize whitespace because (1) trailing
829		- // whitespace can cause PHP and MySQL not to see eye to eye on
830		- // VARCHAR comparisons for some versions of MySQL (cf.
831		- // <http://dev.mysql.com/doc/mysql/en/char.html>), and (2)
832		- // because I doubt most people want to make a semantic
833		- // distinction between 'Computers' and 'Computers '
834		- $cats = array_map('trim', $cats);
	2374	+ // Reassemble
	2375	+ $userdata['user_email'] = $parts[0].'@'.$parts[1];
	2376	+ break;
	2377	+ default :
	2378	+ if ( $insanity > 10 ) :
	2379	+ // Try some settings that are unlikely to cause complaint...
	2380	+ $url = parse_url($hostUrl);
835	2381
836		- $tags = array();
	2382	+ $userdata['user_login'] = substr(md5(uniqid(microtime())), 0, 6);
	2383	+ $userdata['user_nicename'] = $userdata['user_login'];
	2384	+ $userdata['user_email'] = 'noreply@' . $url['host'];
	2385	+ elseif ( $insanity > 50 ) :
	2386	+ // Stop doing the same thing and expecting a different result
	2387	+ break;
	2388	+ endif;
	2389	+ endswitch;
	2390	+ endif;
	2391	+ $insanity = $insanity + 1;
	2392	+ } while (is_wp_error($id));
837	2393
838		- $cat_ids = array ();
839		- foreach ($cats as $cat_name) :
840		- if (preg_match('/^{#([0-9]+)}$/', $cat_name, $backref)) :
841		- $cat_id = (int) $backref[1];
842		- if (function_exists('is_term') and is_term($cat_id, 'category')) :
843		- $cat_ids[] = $cat_id;
844		- elseif (get_category($cat_id)) :
845		- $cat_ids[] = $cat_id;
846		- endif;
847		- elseif (strlen($cat_name) > 0) :
848		- $esc = $wpdb->escape($cat_name);
849		- $resc = $wpdb->escape(preg_quote($cat_name));
850		-
851		- // WordPress 2.3+
852		- if (function_exists('is_term')) :
853		- $cat_id = is_term($cat_name, 'category');
854		- if ($cat_id) :
855		- $cat_ids[] = $cat_id['term_id'];
856		- // There must be a better way to do this...
857		- elseif ($results = $wpdb->get_results(
858		- "SELECT term_id
859		- FROM $wpdb->term_taxonomy
860		- WHERE
861		- LOWER(description) RLIKE
862		- CONCAT('(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*', LOWER('{$resc}'), '( |\\t|\\r)*(\\n|\$)')"
863		- )) :
864		- foreach ($results AS $term) :
865		- $cat_ids[] = (int) $term->term_id;
866		- endforeach;
867		- elseif ('tag'==$unfamiliar_category) :
868		- $tags[] = $cat_name;
869		- elseif ('create'===$unfamiliar_category) :
870		- $term = wp_insert_term($cat_name, 'category');
871		- if (is_wp_error($term)) :
872		- FeedWordPress::noncritical_bug('term insertion problem', array('cat_name' => $cat_name, 'term' => $term, 'this' => $this), __LINE__);
873		- else :
874		- $cat_ids[] = $term['term_id'];
875		- endif;
	2394	+ // $id should now contain the numeric ID of a newly minted
	2395	+ // user account. Let's mark them as having been generated
	2396	+ // by FeedWordPress in the usermeta table, as per the
	2397	+ // suggestion of @boonebgorges, in case we need to process,
	2398	+ // winnow, filter, or merge syndicated author accounts, &c.
	2399	+ if ( !is_wp_error($id)) :
	2400	+ add_user_meta($id, 'feedwordpress_generated', 1);
	2401	+ else :
	2402	+ $id = null;
876	2403	endif;
877		-
878		- // WordPress 1.5.x - 2.2.x
879		- else :
880		- $results = $wpdb->get_results(
881		- "SELECT cat_ID
882		- FROM $wpdb->categories
883		- WHERE
884		- (LOWER(cat_name) = LOWER('$esc'))
885		- OR (LOWER(category_description)
886		- RLIKE CONCAT('(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*', LOWER('{$resc}'), '( |\\t|\\r)*(\\n|\$)'))
887		- ");
888		- if ($results) :
889		- foreach ($results as $term) :
890		- $cat_ids[] = (int) $term->cat_ID;
891		- endforeach;
892		- elseif ('create'===$unfamiliar_category) :
893		- if (function_exists('wp_insert_category')) :
894		- $cat_id = wp_insert_category(array('cat_name' => $cat_name));
895		- // And into the database we go.
896		- else :
897		- $nice_kitty = sanitize_title($cat_name);
898		- $wpdb->query(sprintf("
899		- INSERT INTO $wpdb->categories
900		- SET
901		- cat_name='%s',
902		- category_nicename='%s'
903		- ", $wpdb->escape($cat_name), $nice_kitty
904		- ));
905		- $cat_id = $wpdb->insert_id;
906		- endif;
907		- $cat_ids[] = $cat_id;
908		- endif;
	2404	+ // also see comment above regarding get_userdata(). (gwyneth 20220223)
	2405	+ elseif ( is_numeric( $unfamiliar_author ) and function_exists( 'get_userdata ') and get_userdata((int) $unfamiliar_author ) ) :
	2406	+ $id = (int) $unfamiliar_author;
	2407	+ elseif ( $unfamiliar_author === 'default' ) :
	2408	+ $id = 1;
909	2409	endif;
910	2410	endif;
911		- endforeach;
912		-
913		- if ((count($cat_ids) == 0) and ($unfamiliar_category === 'filter')) :
914		- $cat_ids = NULL; // Drop the post
915		- else :
916		- $cat_ids = array_unique($cat_ids);
917	2411	endif;
918		-
919		- if ($tags_too) : $ret = array($cat_ids, $tags);
920		- else : $ret = $cat_ids;
921		- endif;
922	2412
923		- return $ret;
924		- } // function SyndicatedPost::category_ids ()
	2413	+ if ($id) :
	2414	+ $this->link->settings['map authors']['name'][strtolower(trim($author))] = $id;
925	2415
926		- function use_api ($tag) {
927		- global $wp_db_version;
928		- switch ($tag) :
929		- case 'wp_insert_post':
930		- // Before 2.2, wp_insert_post does too much of the wrong stuff to use it
931		- // In 1.5 it was such a resource hog it would make PHP segfault on big updates
932		- $ret = (isset($wp_db_version) and $wp_db_version > FWP_SCHEMA_21);
933		- break;
934		- case 'post_status_pending':
935		- $ret = (isset($wp_db_version) and $wp_db_version > FWP_SCHEMA_23);
936		- break;
937		- endswitch;
938		- return $ret;
939		- } // function SyndicatedPost::use_api ()
940		-
941		- #### EXTRACT DATA FROM FEED ITEM ####
942		-
943		- function created () {
944		- $epoch = null;
945		- if (isset($this->item['dc']['created'])) :
946		- $epoch = @parse_w3cdtf($this->item['dc']['created']);
947		- elseif (isset($this->item['dcterms']['created'])) :
948		- $epoch = @parse_w3cdtf($this->item['dcterms']['created']);
949		- elseif (isset($this->item['created'])): // Atom 0.3
950		- $epoch = @parse_w3cdtf($this->item['created']);
951		- endif;
952		- return $epoch;
953		- }
954		- function published ($fallback = true) {
955		- $epoch = null;
956		-
957		- # RSS is a fucking mess. Figure out whether we have a date in
958		- # <dc:date>, <issued>, <pubDate>, etc., and get it into Unix
959		- # epoch format for reformatting. If we can't find anything,
960		- # we'll use the last-updated time.
961		- if (isset($this->item['dc']['date'])): // Dublin Core
962		- $epoch = @parse_w3cdtf($this->item['dc']['date']);
963		- elseif (isset($this->item['dcterms']['issued'])) : // Dublin Core extensions
964		- $epoch = @parse_w3cdtf($this->item['dcterms']['issued']);
965		- elseif (isset($this->item['published'])) : // Atom 1.0
966		- $epoch = @parse_w3cdtf($this->item['published']);
967		- elseif (isset($this->item['issued'])): // Atom 0.3
968		- $epoch = @parse_w3cdtf($this->item['issued']);
969		- elseif (isset($this->item['pubdate'])): // RSS 2.0
970		- $epoch = strtotime($this->item['pubdate']);
971		- elseif ($fallback) : // Fall back to <updated> / <modified> if present
972		- $epoch = $this->updated(/*fallback=*/ false);
973		- endif;
974		-
975		- # If everything failed, then default to the current time.
976		- if (is_null($epoch)) :
977		- if (-1 == $default) :
978		- $epoch = time();
979		- else :
980		- $epoch = $default;
	2416	+ // Multisite: Check whether the author has been recorded
	2417	+ // on *this* blog before. If not, put her down as a
	2418	+ // Contributor for *this* blog.
	2419	+ $user = new WP_User((int) $id);
	2420	+ if (empty($user->roles)) :
	2421	+ $user->add_role('contributor');
981	2422	endif;
982	2423	endif;
983		-
984		- return $epoch;
985		- }
986		- function updated ($fallback = true, $default = -1) {
987		- $epoch = null;
	2424	+ return $id;
	2425	+ } /* function SyndicatedPost::author_id () */
988	2426
989		- # As far as I know, only dcterms and Atom have reliable ways to
990		- # specify when something was *modified* last. If neither is
991		- # available, then we'll try to get the time of publication.
992		- if (isset($this->item['dc']['modified'])) : // Not really correct
993		- $epoch = @parse_w3cdtf($this->item['dc']['modified']);
994		- elseif (isset($this->item['dcterms']['modified'])) : // Dublin Core extensions
995		- $epoch = @parse_w3cdtf($this->item['dcterms']['modified']);
996		- elseif (isset($this->item['modified'])): // Atom 0.3
997		- $epoch = @parse_w3cdtf($this->item['modified']);
998		- elseif (isset($this->item['updated'])): // Atom 1.0
999		- $epoch = @parse_w3cdtf($this->item['updated']);
1000		- elseif ($fallback) : // Fall back to issued / dc:date
1001		- $epoch = $this->published(/*fallback=*/ false, /*default=*/ $default);
1002		- endif;
1003		-
1004		- # If everything failed, then default to the current time.
1005		- if (is_null($epoch)) :
1006		- if (-1 == $default) :
1007		- $epoch = time();
1008		- else :
1009		- $epoch = $default;
1010		- endif;
1011		- endif;
	2427	+ /**
	2428	+ * category_ids: look up (and create) category ids from a list of
	2429	+ * categories
	2430	+ *
	2431	+ * @param array $cats
	2432	+ * @param string $unfamiliar_category
	2433	+ * @param array|null $taxonomies
	2434	+ * @return array
	2435	+ */
	2436	+ function category_ids ($cats, $unfamiliar_category = 'create', $taxonomies = NULL, $params = array()) {
	2437	+ return $this->link->category_ids($this, $cats, $unfamiliar_category, $taxonomies, $params);
	2438	+ } /* SyndicatedPost::category_ids () */
1012	2439
1013		- return $epoch;
1014		- }
1015		-
1016		- function update_hash () {
1017		- return md5(serialize($this->item));
1018		- }
1019		-
1020		- function guid () {
1021		- $guid = null;
1022		- if (isset($this->item['id'])): // Atom 0.3 / 1.0
1023		- $guid = $this->item['id'];
1024		- elseif (isset($this->item['atom']['id'])) : // Namespaced Atom
1025		- $guid = $this->item['atom']['id'];
1026		- elseif (isset($this->item['guid'])) : // RSS 2.0
1027		- $guid = $this->item['guid'];
1028		- elseif (isset($this->item['dc']['identifier'])) :// yeah, right
1029		- $guid = $this->item['dc']['identifier'];
1030		- else :
1031		- // The feed does not seem to have provided us with a
1032		- // unique identifier, so we'll have to cobble together
1033		- // a tag: URI that might work for us. The base of the
1034		- // URI will be the host name of the feed source ...
1035		- $bits = parse_url($this->feedmeta['link/uri']);
1036		- $guid = 'tag:'.$bits['host'];
1037		-
1038		- // If we have a date of creation, then we can use that
1039		- // to uniquely identify the item. (On the other hand, if
1040		- // the feed producer was consicentious enough to
1041		- // generate dates of creation, she probably also was
1042		- // conscientious enough to generate unique identifiers.)
1043		- if (!is_null($this->created())) :
1044		- $guid .= '://post.'.date('YmdHis', $this->created());
1045		-
1046		- // Otherwise, use both the URI of the item, *and* the
1047		- // item's title. We have to use both because titles are
1048		- // often not unique, and sometimes links aren't unique
1049		- // either (e.g. Bitch (S)HITLIST, Mozilla Dot Org news,
1050		- // some podcasts). But it's rare to have *both* the same
1051		- // title *and* the same link for two different items. So
1052		- // this is about the best we can do.
1053		- else :
1054		- $guid .= '://'.md5($this->item['link'].'/'.$this->item['title']);
1055		- endif;
1056		- endif;
1057		- return $guid;
1058		- }
1059		-
1060		- function author () {
1061		- $author = array ();
1062		-
1063		- if (isset($this->item['author_name'])):
1064		- $author['name'] = $this->item['author_name'];
1065		- elseif (isset($this->item['dc']['creator'])):
1066		- $author['name'] = $this->item['dc']['creator'];
1067		- elseif (isset($this->item['dc']['contributor'])):
1068		- $author['name'] = $this->item['dc']['contributor'];
1069		- elseif (isset($this->feed->channel['dc']['creator'])) :
1070		- $author['name'] = $this->feed->channel['dc']['creator'];
1071		- elseif (isset($this->feed->channel['dc']['contributor'])) :
1072		- $author['name'] = $this->feed->channel['dc']['contributor'];
1073		- elseif (isset($this->feed->channel['author_name'])) :
1074		- $author['name'] = $this->feed->channel['author_name'];
1075		- elseif ($this->feed->is_rss() and isset($this->item['author'])) :
1076		- // The author element in RSS is allegedly an
1077		- // e-mail address, but lots of people don't use
1078		- // it that way. So let's make of it what we can.
1079		- $author = parse_email_with_realname($this->item['author']);
1080		-
1081		- if (!isset($author['name'])) :
1082		- if (isset($author['email'])) :
1083		- $author['name'] = $author['email'];
1084		- else :
1085		- $author['name'] = $this->feed->channel['title'];
1086		- endif;
1087		- endif;
1088		- else :
1089		- $author['name'] = $this->feed->channel['title'];
1090		- endif;
1091		-
1092		- if (isset($this->item['author_email'])):
1093		- $author['email'] = $this->item['author_email'];
1094		- elseif (isset($this->feed->channel['author_email'])) :
1095		- $author['email'] = $this->feed->channel['author_email'];
1096		- endif;
1097		-
1098		- if (isset($this->item['author_url'])):
1099		- $author['uri'] = $this->item['author_url'];
1100		- elseif (isset($this->feed->channel['author_url'])) :
1101		- $author['uri'] = $this->item['author_url'];
1102		- else:
1103		- $author['uri'] = $this->feed->channel['link'];
1104		- endif;
1105		-
1106		- return $author;
1107		- } // SyndicatedPost::author()
1108		-
1109		- var $uri_attrs = array (
1110		- array('a', 'href'),
1111		- array('applet', 'codebase'),
1112		- array('area', 'href'),
1113		- array('blockquote', 'cite'),
1114		- array('body', 'background'),
1115		- array('del', 'cite'),
1116		- array('form', 'action'),
1117		- array('frame', 'longdesc'),
1118		- array('frame', 'src'),
1119		- array('iframe', 'longdesc'),
1120		- array('iframe', 'src'),
1121		- array('head', 'profile'),
1122		- array('img', 'longdesc'),
1123		- array('img', 'src'),
1124		- array('img', 'usemap'),
1125		- array('input', 'src'),
1126		- array('input', 'usemap'),
1127		- array('ins', 'cite'),
1128		- array('link', 'href'),
1129		- array('object', 'classid'),
1130		- array('object', 'codebase'),
1131		- array('object', 'data'),
1132		- array('object', 'usemap'),
1133		- array('q', 'cite'),
1134		- array('script', 'src')
1135		- ); /* var SyndicatedPost::$uri_attrs */
1136		-
1137		- var $_base = null;
1138		-
1139		- function resolve_single_relative_uri ($refs) {
1140		- $tag = FeedWordPressHTML::attributeMatch($refs);
1141		- $url = Relative_URI::resolve($tag['value'], $this->_base);
1142		- return $tag['prefix'] . $url . $tag['suffix'];
1143		- } /* function SyndicatedPost::resolve_single_relative_uri() */
1144		-
1145		- function resolve_relative_uris ($content, $obj) {
1146		- # The MagpieRSS upgrade has some `xml:base` support baked in.
1147		- # However, sometimes people do silly things, like putting
1148		- # relative URIs out on a production RSS 2.0 feed or other feeds
1149		- # with no good support for `xml:base`. So we'll do our best to
1150		- # try to catch any remaining relative URIs and resolve them as
1151		- # best we can.
1152		- $obj->_base = $obj->item['link']; // Reset the base for resolving relative URIs
1153		-
1154		- foreach ($obj->uri_attrs as $pair) :
1155		- list($tag, $attr) = $pair;
1156		- $pattern = FeedWordPressHTML::attributeRegex($tag, $attr);
1157		- $content = preg_replace_callback (
1158		- $pattern,
1159		- array(&$obj, 'resolve_single_relative_uri'),
1160		- $content
1161		- );
1162		- endforeach;
1163		-
1164		- return $content;
1165		- } /* function SyndicatedPost::resolve_relative_uris () */
1166		-
1167		- var $strip_attrs = array (
1168		- array('[a-z]+', 'target'),
1169		-// array('[a-z]+', 'style'),
1170		-// array('[a-z]+', 'on[a-z]+'),
1171		- );
1172		-
1173		- function strip_attribute_from_tag ($refs) {
1174		- $tag = FeedWordPressHTML::attributeMatch($refs);
1175		- return $tag['before_attribute'].$tag['after_attribute'];
1176		- }
1177		-
1178		- function sanitize_content ($content, $obj) {
1179		- # This kind of sucks. I intend to replace it with
1180		- # lib_filter sometime soon.
1181		- foreach ($obj->strip_attrs as $pair):
1182		- list($tag,$attr) = $pair;
1183		- $pattern = FeedWordPressHTML::attributeRegex($tag, $attr);
1184		-
1185		- $content = preg_replace_callback (
1186		- $pattern,
1187		- array(&$obj, 'strip_attribute_from_tag'),
1188		- $content
1189		- );
1190		- endforeach;
1191		- return $content;
1192		- }
1193		-} // class SyndicatedPost
1194		-
	2440	+} /* class SyndicatedPost */