← All changes | includes/content-graph/graph-builder.php +52 -68 0.9.7 → 1.1.10 View file →

		@@ -1,7 +1,7 @@
1	1	<?php
2	2	/**
3		- * Desktop Mode — Content Graph: graph builder.
	3	+ * OpenStation — Content Graph: graph builder.
4	4	*
5	5	* Walks the WordPress post store to produce two arrays for the bundle:
6	6	*
7	7	* - `nodes` — one entry per post matching the requested types.
		@@ -19,10 +19,9 @@
19	19	* also explicitly bust the cache on `save_post` and `deleted_post` so
20	20	* editors don't see stale data even if some other process pre-warms
21	21	* the transient.
22	22	*
23		- * @package WPDesktopMode
24		- * @since 0.8.2
	23	+ * @package OpenStation
25	24	*/
26	25
27	26	defined( 'ABSPATH' ) || exit;
28	27
		@@ -30,17 +29,22 @@
30	29	// transients still alive at upgrade time would otherwise return the
31	30	// previous schema (e.g. missing per-node `contributor_ids`) and
32	31	// surface as runtime errors on the client. Each bump is a one-time
33	32	// cache miss for every site that updates the plugin.
34		-const DESKTOP_MODE_CONTENT_GRAPH_TRANSIENT_PREFIX = 'desktop_mode_cg3_';
35		-const DESKTOP_MODE_CONTENT_GRAPH_TRANSIENT_TTL = 6 * HOUR_IN_SECONDS;
	33	+/**
	34	+ * The VALUE keeps its pre-rebrand spelling on purpose: it is a
	35	+ * persisted or externally-visible identifier, so renaming it would
	36	+ * orphan data already written by live installs (or break a live
	37	+ * URL). The mismatch between this constant's name and its value is
	38	+ * deliberate — it is NOT a half-finished rename.
	39	+ */
	40	+const OPENSTATION_CONTENT_GRAPH_TRANSIENT_PREFIX = 'desktop_mode_cg3_';
	41	+const OPENSTATION_CONTENT_GRAPH_TRANSIENT_TTL = 6 * HOUR_IN_SECONDS;
36	42
37	43	/**
38	44	* Build (or reuse the cached version of) the graph payload for the
39	45	* requested post types.
40	46	*
41		- * @since 0.8.2
42		- *
43	47	* @param string[] $types Post type slugs. Filtered against the public
44	48	* post-type registry, attachments excluded.
45	49	* @return array{
46	50	* nodes: array<int, array{
		@@ -58,10 +62,10 @@
58	62	* },
59	63	* stats: array{ nodes: int, edges: int, generated_at: int }
60	64	* }
61	65	*/
62		-function desktop_mode_content_graph_build( array $types ) {
63		- $types = desktop_mode_content_graph_normalize_types( $types );
	66	+function openstation_content_graph_build( array $types ) {
	67	+ $types = openstation_content_graph_normalize_types( $types );
64	68	if ( empty( $types ) ) {
65	69	return array(
66	70	'nodes' => array(),
67	71	'edges' => array(),
		@@ -77,15 +81,15 @@
77	81	),
78	82	);
79	83	}
80	84
81		- $cache_key = desktop_mode_content_graph_cache_key( $types );
	85	+ $cache_key = openstation_content_graph_cache_key( $types );
82	86	$cached = get_transient( $cache_key );
83	87	if ( is_array( $cached ) && isset( $cached['nodes'], $cached['edges'] ) ) {
84	88	return $cached;
85	89	}
86	90
87		- $rows = desktop_mode_content_graph_fetch_rows( $types );
	91	+ $rows = openstation_content_graph_fetch_rows( $types );
88	92
89	93	$post_ids = array();
90	94	foreach ( $rows as $row ) {
91	95	$post_ids[] = (int) $row->ID;
		@@ -90,14 +94,14 @@
90	94	foreach ( $rows as $row ) {
91	95	$post_ids[] = (int) $row->ID;
92	96	}
93	97
94		- $terms_by_post = desktop_mode_content_graph_collect_post_terms( $post_ids );
	98	+ $terms_by_post = openstation_content_graph_collect_post_terms( $post_ids );
95	99	// Distinct revision authors per post — used to pull collaborator
96	100	// posts toward both the primary author's cluster AND the
97	101	// contributor's clusters when grouping by author. Bulk-queried so
98	102	// we don't N+1 `wp_get_post_revisions` per node.
99		- $contribs_by_post = desktop_mode_content_graph_collect_post_contributors( $post_ids );
	103	+ $contribs_by_post = openstation_content_graph_collect_post_contributors( $post_ids );
100	104
101	105	$default_category = max( 1, (int) get_option( 'default_category', 1 ) );
102	106
103	107	$nodes = array();
		@@ -105,12 +109,12 @@
105	109	$author_ids = array();
106	110	$cat_ids = array();
107	111	$tag_ids = array();
108	112	foreach ( $rows as $row ) {
109		- $id = (int) $row->ID;
110		- $author_id = (int) $row->post_author;
111		- $year = 0;
112		- $year_month = '';
	113	+ $id = (int) $row->ID;
	114	+ $author_id = (int) $row->post_author;
	115	+ $year = 0;
	116	+ $year_month = '';
113	117	if ( ! empty( $row->post_date ) ) {
114	118	// Use post_date (site-local) rather than post_date_gmt for the
115	119	// "year published" / "year-month published" buckets — editors
116	120	// think in their own timezone.
		@@ -132,9 +136,9 @@
132	136	}
133	137	$post_tags = isset( $terms_by_post[ $id ]['post_tag'] )
134	138	? $terms_by_post[ $id ]['post_tag']
135	139	: array();
136		- $contribs = isset( $contribs_by_post[ $id ] )
	140	+ $contribs = isset( $contribs_by_post[ $id ] )
137	141	? $contribs_by_post[ $id ]
138	142	: array();
139	143	// Strip the primary author from the contributor list — the
140	144	// node carries that separately in `author_id`, and surfacing
		@@ -150,9 +154,9 @@
150	154	)
151	155	);
152	156	}
153	157
154		- $node = array(
	158	+ $node = array(
155	159	'id' => $id,
156	160	'type' => (string) $row->post_type,
157	161	'title' => (string) get_the_title( $row ),
158	162	'status' => (string) $row->post_status,
		@@ -183,11 +187,11 @@
183	187	}
184	188	}
185	189
186	190	$groups = array(
187		- 'authors' => desktop_mode_content_graph_format_author_catalog( array_keys( $author_ids ) ),
188		- 'categories' => desktop_mode_content_graph_format_term_catalog( array_keys( $cat_ids ), 'category' ),
189		- 'tags' => desktop_mode_content_graph_format_term_catalog( array_keys( $tag_ids ), 'post_tag' ),
	191	+ 'authors' => openstation_content_graph_format_author_catalog( array_keys( $author_ids ) ),
	192	+ 'categories' => openstation_content_graph_format_term_catalog( array_keys( $cat_ids ), 'category' ),
	193	+ 'tags' => openstation_content_graph_format_term_catalog( array_keys( $tag_ids ), 'post_tag' ),
190	194	);
191	195
192	196	$edges_seen = array();
193	197	$edges = array();
		@@ -192,9 +196,9 @@
192	196	$edges_seen = array();
193	197	$edges = array();
194	198	foreach ( $rows as $row ) {
195	199	$from = (int) $row->ID;
196		- $tos = desktop_mode_content_graph_extract_internal_links( (string) $row->post_content );
	200	+ $tos = openstation_content_graph_extract_internal_links( (string) $row->post_content );
197	201	foreach ( $tos as $to ) {
198	202	if ( $to === $from ) {
199	203	continue;
200	204	}
		@@ -226,9 +230,9 @@
226	230	'generated_at' => time(),
227	231	),
228	232	);
229	233
230		- set_transient( $cache_key, $payload, DESKTOP_MODE_CONTENT_GRAPH_TRANSIENT_TTL );
	234	+ set_transient( $cache_key, $payload, OPENSTATION_CONTENT_GRAPH_TRANSIENT_TTL );
231	235
232	236	return $payload;
233	237	}
234	238
		@@ -235,18 +239,16 @@
235	239	/**
236	240	* Filter, sanitize, and uniquify the requested type slugs against the
237	241	* public post-type registry. Returned slugs are guaranteed to exist
238	242	* AND to be among the slugs declared by
239		- * `desktop_mode_content_graph_post_types()`.
	243	+ * `openstation_content_graph_post_types()`.
240	244	*
241		- * @since 0.8.2
242		- *
243	245	* @param string[] $types
244	246	* @return string[]
245	247	*/
246		-function desktop_mode_content_graph_normalize_types( array $types ) {
	248	+function openstation_content_graph_normalize_types( array $types ) {
247	249	$allowed = array();
248		- foreach ( desktop_mode_content_graph_post_types() as $entry ) {
	250	+ foreach ( openstation_content_graph_post_types() as $entry ) {
249	251	if ( ! empty( $entry['slug'] ) ) {
250	252	$allowed[ (string) $entry['slug'] ] = true;
251	253	}
252	254	}
		@@ -274,14 +276,12 @@
274	276	* The `key` element encodes the resulting privilege tier (and, when
275	277	* the own-author clause is active, the user id) so cached payloads
276	278	* are never served across privilege levels.
277	279	*
278		- * @since 0.9.2
279		- *
280	280	* @param string[] $types Already normalized.
281	281	* @return array{ where: string, values: array, key: string }
282	282	*/
283		-function desktop_mode_content_graph_visibility_sql( array $types ) {
	283	+function openstation_content_graph_visibility_sql( array $types ) {
284	284	$placeholders = implode( ',', array_fill( 0, count( $types ), '%s' ) );
285	285	$values = $types;
286	286
287	287	$priv_types = array();
		@@ -326,16 +326,14 @@
326	326	* relevant edit busts the cache implicitly, plus the viewer's
327	327	* privilege-tier signature so a payload built for a user who can read
328	328	* private posts is never served to one who can't (and vice versa).
329	329	*
330		- * @since 0.8.2
331		- *
332	330	* @param string[] $types Already normalized.
333	331	* @return string
334	332	*/
335		-function desktop_mode_content_graph_cache_key( array $types ) {
	333	+function openstation_content_graph_cache_key( array $types ) {
336	334	global $wpdb;
337		- $visibility = desktop_mode_content_graph_visibility_sql( $types );
	335	+ $visibility = openstation_content_graph_visibility_sql( $types );
338	336	// phpcs:disable WordPress.DB.PreparedSQL.NotPrepared, WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching
339	337	$hash = (string) $wpdb->get_var(
340	338	$wpdb->prepare(
341	339	"SELECT MD5( GROUP_CONCAT( CONCAT( ID, ':', post_modified_gmt ) ORDER BY ID ) )
		@@ -347,9 +345,9 @@
347	345	// phpcs:enable
348	346	if ( '' === $hash || null === $hash ) {
349	347	$hash = 'empty';
350	348	}
351		- return DESKTOP_MODE_CONTENT_GRAPH_TRANSIENT_PREFIX . substr( md5( implode( ',', $types ) . '|' . $visibility['key'] . '|' . $hash ), 0, 24 );
	349	+ return OPENSTATION_CONTENT_GRAPH_TRANSIENT_PREFIX . substr( md5( implode( ',', $types ) . '|' . $visibility['key'] . '|' . $hash ), 0, 24 );
352	350	}
353	351
354	352	/**
355	353	* Fetch the participating posts in a single query. Returns full rows
		@@ -358,18 +356,16 @@
358	356	*
359	357	* Rows are scoped to what the current user can read: published posts,
360	358	* plus private posts only where the user holds the type's
361	359	* `read_private_posts` capability (or authored the post). See
362		- * `desktop_mode_content_graph_visibility_sql()`.
	360	+ * `openstation_content_graph_visibility_sql()`.
363	361	*
364		- * @since 0.8.2
365		- *
366	362	* @param string[] $types Already normalized.
367	363	* @return WP_Post[]
368	364	*/
369		-function desktop_mode_content_graph_fetch_rows( array $types ) {
	365	+function openstation_content_graph_fetch_rows( array $types ) {
370	366	global $wpdb;
371		- $visibility = desktop_mode_content_graph_visibility_sql( $types );
	367	+ $visibility = openstation_content_graph_visibility_sql( $types );
372	368	// phpcs:disable WordPress.DB.PreparedSQL.NotPrepared, WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching
373	369	$rows = $wpdb->get_results(
374	370	$wpdb->prepare(
375	371	"SELECT ID, post_type, post_status, post_title, post_name, post_content, post_author, post_date
		@@ -403,14 +399,12 @@
403	399	* Pull every internal-target post id out of a chunk of post_content.
404	400	* Uses DOMDocument for robustness against malformed HTML, then
405	401	* `url_to_postid()` to resolve each href.
406	402	*
407		- * @since 0.8.2
408		- *
409	403	* @param string $content
410	404	* @return int[] Unique target post ids (order preserved).
411	405	*/
412		-function desktop_mode_content_graph_extract_internal_links( $content ) {
	406	+function openstation_content_graph_extract_internal_links( $content ) {
413	407	if ( '' === trim( (string) $content ) ) {
414	408	return array();
415	409	}
416	410
		@@ -454,15 +448,13 @@
454	448	}
455	449
456	450	/**
457	451	* Cache invalidation. Any post-type change wipes every transient
458		- * carrying the `desktop_mode_cg_` prefix. We don't have a per-type
	452	+ * carrying the `openstation_cg_` prefix. We don't have a per-type
459	453	* index so we wipe globally, the cost is one extra build on next
460	454	* open which dominates the time-savings on subsequent opens.
461		- *
462		- * @since 0.8.2
463	455	*/
464		-function desktop_mode_content_graph_flush_cache() {
	456	+function openstation_content_graph_flush_cache() {
465	457	global $wpdb;
466	458	// Enumerate matching transient option names first, then route each
467	459	// through `delete_transient()` so WP's transient/options cache
468	460	// invalidation runs alongside the wp_options delete. The earlier
		@@ -471,9 +463,9 @@
471	463	// the in-memory options cache and returned the stale payload.
472	464	// This bit the `set_object_terms` invalidation path because that
473	465	// hook doesn't update `post_modified_gmt`, so the cache key stayed
474	466	// identical and `get_transient` (cache hit) returned pre-retag data.
475		- $prefix_like = $wpdb->esc_like( '_transient_' . DESKTOP_MODE_CONTENT_GRAPH_TRANSIENT_PREFIX ) . '%';
	467	+ $prefix_like = $wpdb->esc_like( '_transient_' . OPENSTATION_CONTENT_GRAPH_TRANSIENT_PREFIX ) . '%';
476	468	// phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching
477	469	$option_names = $wpdb->get_col(
478	470	$wpdb->prepare(
479	471	"SELECT option_name FROM {$wpdb->options} WHERE option_name LIKE %s",
		@@ -491,14 +483,14 @@
491	483	delete_transient( $transient );
492	484	}
493	485	}
494	486	}
495		-add_action( 'save_post', 'desktop_mode_content_graph_flush_cache' );
496		-add_action( 'deleted_post', 'desktop_mode_content_graph_flush_cache' );
	487	+add_action( 'save_post', 'openstation_content_graph_flush_cache' );
	488	+add_action( 'deleted_post', 'openstation_content_graph_flush_cache' );
497	489	// Term assignments can change outside the post-edit path (CLI, bulk
498	490	// quick-edit, REST). Without this the per-node `category_ids` / `tag_ids`
499	491	// the group-by UI reads would go stale until the 6h TTL expires.
500		-add_action( 'set_object_terms', 'desktop_mode_content_graph_flush_cache' );
	492	+add_action( 'set_object_terms', 'openstation_content_graph_flush_cache' );
501	493
502	494	/**
503	495	* Bulk-fetch every (post_id → taxonomy → term_ids[]) mapping for the
504	496	* `category` and `post_tag` taxonomies in a single query. Used to
		@@ -504,15 +496,13 @@
504	496	* `category` and `post_tag` taxonomies in a single query. Used to
505	497	* populate the per-node `category_ids` / `tag_ids` arrays without N+1
506	498	* `wp_get_object_terms` calls.
507	499	*
508		- * @since 0.8.6
509		- *
510	500	* @param int[] $post_ids
511	501	* @return array<int, array<string, int[]>> Outer key = post id; inner
512	502	* key = taxonomy slug; value = term ids the post is in.
513	503	*/
514		-function desktop_mode_content_graph_collect_post_terms( array $post_ids ) {
	504	+function openstation_content_graph_collect_post_terms( array $post_ids ) {
515	505	$post_ids = array_values( array_filter( array_map( 'intval', $post_ids ) ) );
516	506	if ( empty( $post_ids ) ) {
517	507	return array();
518	508	}
		@@ -551,9 +541,9 @@
551	541	}
552	542
553	543	/**
554	544	* Bulk-fetch distinct revision authors per post for the requested
555		- * post ids. Used by `desktop_mode_content_graph_build()` to populate
	545	+ * post ids. Used by `openstation_content_graph_build()` to populate
556	546	* each node's `contributor_ids` array so the cluster-attractor force
557	547	* can pull collaborator posts toward both the primary author's
558	548	* cluster AND each contributor's cluster (weighted so the primary
559	549	* still wins).
		@@ -561,14 +551,12 @@
561	551	* Returns the distinct `post_author` values from each in-scope post's
562	552	* revision children. Includes the primary author if they also
563	553	* authored a revision; the caller is expected to filter that out.
564	554	*
565		- * @since 0.8.6
566		- *
567	555	* @param int[] $post_ids
568	556	* @return array<int, int[]> post_id => list of contributor user ids.
569	557	*/
570		-function desktop_mode_content_graph_collect_post_contributors( array $post_ids ) {
	558	+function openstation_content_graph_collect_post_contributors( array $post_ids ) {
571	559	$post_ids = array_values( array_filter( array_map( 'intval', $post_ids ) ) );
572	560	if ( empty( $post_ids ) ) {
573	561	return array();
574	562	}
		@@ -605,14 +593,12 @@
605	593	/**
606	594	* Build a `{ id => { name } }` catalog for the given author ids.
607	595	* Uses one `WP_User_Query` rather than per-id `get_userdata` calls.
608	596	*
609		- * @since 0.8.6
610		- *
611	597	* @param int[] $author_ids
612	598	* @return array<int, array{ name: string }>
613	599	*/
614		-function desktop_mode_content_graph_format_author_catalog( array $author_ids ) {
	600	+function openstation_content_graph_format_author_catalog( array $author_ids ) {
615	601	$author_ids = array_values( array_unique( array_filter( array_map( 'intval', $author_ids ) ) ) );
616	602	if ( empty( $author_ids ) ) {
617	603	return array();
618	604	}
		@@ -622,9 +608,9 @@
622	608	'fields' => array( 'ID', 'display_name' ),
623	609	'number' => count( $author_ids ),
624	610	)
625	611	);
626		- $out = array();
	612	+ $out = array();
627	613	foreach ( (array) $query->get_results() as $user ) {
628	614	$out[ (int) $user->ID ] = array(
629	615	'name' => (string) $user->display_name,
630	616	);
		@@ -636,15 +622,13 @@
636	622	* Build a `{ id => { name } }` catalog for the given term ids in a
637	623	* single taxonomy. Uses `get_terms` with `include` so the names come
638	624	* back in one query.
639	625	*
640		- * @since 0.8.6
641		- *
642	626	* @param int[] $term_ids
643	627	* @param string $taxonomy
644	628	* @return array<int, array{ name: string }>
645	629	*/
646		-function desktop_mode_content_graph_format_term_catalog( array $term_ids, $taxonomy ) {
	630	+function openstation_content_graph_format_term_catalog( array $term_ids, $taxonomy ) {
647	631	$term_ids = array_values( array_unique( array_filter( array_map( 'intval', $term_ids ) ) ) );
648	632	if ( empty( $term_ids ) ) {
649	633	return array();
650	634	}
		@@ -655,9 +639,9 @@
655	639	'hide_empty' => false,
656	640	'number' => count( $term_ids ),
657	641	)
658	642	);
659		- $out = array();
	643	+ $out = array();
660	644	if ( is_wp_error( $terms ) || ! is_array( $terms ) ) {
661	645	return $out;
662	646	}
663	647	foreach ( $terms as $term ) {