← All changes | src/Fonts/UsageTracker.php +213 -51 2.0.2 → 2.0.4 View file →

		@@ -49,9 +49,15 @@
49	49	*/
50	50	public static function route_for( string $url ): string {
51	51	$path = (string) wp_parse_url( $url, PHP_URL_PATH );
52	52
53		- return '' === $path ? '/' : $path;
	53	+ if ( '' === $path ) {
	54	+ return '/';
	55	+ }
	56	+
	57	+ // Guarantee a canonical leading slash — malformed input (e.g. a bare
	58	+ // scheme string) must never mint a route outside the path namespace.
	59	+ return '/' === $path[0] ? $path : '/' . $path;
54	60	}
55	61
56	62	/**
57	63	* Canonical route key for the current request.
		@@ -75,9 +81,9 @@
75	81	if ( empty( $records ) ) {
76	82	return;
77	83	}
78	84
79		- $now = current_time( 'mysql' );
	85	+ $now = gmdate( 'Y-m-d H:i:s' ); // UTC — compared against UTC_TIMESTAMP() in gc().
80	86	$table = $this->usage_table();
81	87
82	88	// Collapse to one row per (page, family, weight, style); merge any
83	89	// duplicates within this batch (strongest rendered/above_fold wins,
		@@ -129,9 +135,13 @@
129	135	$placeholders = array();
130	136	$args = array();
131	137
132	138	foreach ( $by_key as $row ) {
133		- $placeholders[] = '(%s,%s,%d,%s,%s,%s,%s,%d,%d,1,%s)';
	139	+ // A beacon measurement that saw the variant loaded but not rendered
	140	+ // counts one "miss" toward the scope-out corroboration threshold.
	141	+ $miss = ( 'beacon' === $row['origin'] && 0 === $row['rendered'] ) ? 1 : 0;
	142	+
	143	+ $placeholders[] = '(%s,%s,%d,%s,%s,%s,%s,%d,%d,1,%d,%s)';
134	144	array_push(
135	145	$args,
136	146	$row['key'],
137	147	$page_url,
		@@ -141,21 +151,29 @@
141	151	$row['style'],
142	152	$row['origin'],
143	153	$row['rendered'],
144	154	$row['above_fold'],
	155	+ $miss,
145	156	$now
146	157	);
147	158	}
148	159
	160	+ $alias = $this->dup_alias();
	161	+
149	162	$sql = "INSERT INTO {$table}
150		- (usage_key, page_url, page_id, family, weight, style, origin, rendered, above_fold, hits, last_seen)
151		- VALUES " . implode( ',', $placeholders ) . "
	163	+ (usage_key, page_url, page_id, family, weight, style, origin, rendered, above_fold, hits, beacon_misses, last_seen)
	164	+ VALUES " . implode( ',', $placeholders ) . "{$alias}
152	165	ON DUPLICATE KEY UPDATE
153	166	hits = hits + 1,
154		- rendered = GREATEST(rendered, VALUES(rendered)),
155		- above_fold = GREATEST(above_fold, VALUES(above_fold)),
156		- origin = VALUES(origin),
157		- last_seen = VALUES(last_seen)";
	167	+ rendered = GREATEST(rendered, {$this->dup( 'rendered' )}),
	168	+ above_fold = GREATEST(above_fold, {$this->dup( 'above_fold' )}),
	169	+ origin = {$this->dup( 'origin' )},
	170	+ beacon_misses = CASE
	171	+ WHEN {$this->dup( 'origin' )} = 'beacon' AND {$this->dup( 'rendered' )} = 1 THEN 0
	172	+ WHEN {$this->dup( 'origin' )} = 'beacon' AND {$this->dup( 'rendered' )} = 0 THEN beacon_misses + 1
	173	+ ELSE beacon_misses
	174	+ END,
	175	+ last_seen = {$this->dup( 'last_seen' )}";
158	176
159	177	// phpcs:ignore WordPress.DB.DirectDatabaseQuery, WordPress.DB.PreparedSQL.NotPrepared
160	178	$wpdb->query( $wpdb->prepare( $sql, $args ) );
161	179	}
		@@ -198,15 +216,20 @@
198	216	* are family-level and based on the page's authorial font-family, so they're
199	217	* robust to weight/style differences and self-correcting.
200	218	*
201	219	* - `rendered` drives page scoping (which fonts load on this page).
202		- * - `above_fold` drives preload (which of them to fetch early).
	220	+ * - `above_fold` the families above the fold.
	221	+ * - `fold_faces` drives preload: for every family rendered here, the faces
	222	+ * ('weight|style') the beacon saw above the fold — empty when the family
	223	+ * renders only further down. (2.0.4) Preload used to take the whole
	224	+ * family, so a family with many weights filled every slot with weights
	225	+ * the page never shows.
203	226	*
204	227	* Empty `rendered` means the route hasn't been measured yet — callers must
205	228	* neither scope nor preload from no data.
206	229	*
207	230	* @param string $route Canonical route (path).
208		- * @return array{rendered:array<string,bool>,above_fold:array<string,bool>,scope_out:array<string,bool>,measured:bool}
	231	+ * @return array{rendered:array<string,bool>,above_fold:array<string,bool>,fold_faces:array<string,array<string,bool>>,scope_out:array<string,bool>,measured:bool}
209	232	*/
210	233	public function route_render_profile( string $route ): array {
211	234	global $wpdb;
212	235
		@@ -212,8 +235,9 @@
212	235
213	236	$out = array(
214	237	'rendered' => array(),
215	238	'above_fold' => array(),
	239	+ 'fold_faces' => array(),
216	240	'scope_out' => array(),
217	241	'measured' => false,
218	242	);
219	243
		@@ -228,43 +252,72 @@
228	252	// confirmed scope-out ONLY when a beacon has actually measured it and
229	253	// never saw it render. A family that hasn't been beacon-measured yet
230	254	// (e.g. a font just added/replaced in the Customizer) is NEVER scoped
231	255	// out — it loads normally until a beacon has had a chance to judge it.
	256	+ // Grouped per face (still one query); folded per family below.
232	257	// phpcs:ignore WordPress.DB.DirectDatabaseQuery
233	258	$rows = $wpdb->get_results(
234	259	$wpdb->prepare(
235		- "SELECT family,
	260	+ "SELECT family, weight, style,
236	261	MAX(rendered) AS rnd,
237	262	MAX(CASE WHEN rendered = 1 THEN above_fold ELSE 0 END) AS af,
238		- MAX(CASE WHEN origin = 'beacon' THEN 1 ELSE 0 END) AS beaconed
	263	+ MAX(CASE WHEN origin = 'beacon' THEN 1 ELSE 0 END) AS beaconed,
	264	+ MAX(beacon_misses) AS misses
239	265	FROM {$this->usage_table()}
240	266	WHERE page_url = %s
241		- GROUP BY family",
	267	+ GROUP BY family, weight, style",
242	268	$route
243	269	),
244	270	ARRAY_A
245	271	);
246	272
247		- if ( $rows ) {
248		- foreach ( $rows as $r ) {
249		- $fam = strtolower( (string) $r['family'] );
250		- $rendered = ! empty( $r['rnd'] );
251		- $beaconed = ! empty( $r['beaconed'] );
	273	+ $families = array();
252	274
253		- if ( $beaconed ) {
254		- $out['measured'] = true;
255		- }
	275	+ foreach ( (array) $rows as $r ) {
	276	+ $fam = strtolower( (string) $r['family'] );
	277	+ $agg = $families[ $fam ] ?? array( 'rnd' => 0, 'af' => 0, 'beaconed' => 0, 'misses' => 0, 'faces' => array() );
256	278
257		- if ( $rendered ) {
258		- $out['rendered'][ $fam ] = true;
	279	+ $agg['rnd'] = max( $agg['rnd'], (int) $r['rnd'] );
	280	+ $agg['af'] = max( $agg['af'], (int) $r['af'] );
	281	+ $agg['beaconed'] = max( $agg['beaconed'], (int) $r['beaconed'] );
	282	+ $agg['misses'] = max( $agg['misses'], (int) $r['misses'] );
259	283
260		- if ( ! empty( $r['af'] ) ) {
261		- $out['above_fold'][ $fam ] = true;
262		- }
263		- } elseif ( $beaconed ) {
264		- // Beacon measured it and never saw it render → scope out.
265		- $out['scope_out'][ $fam ] = true;
	284	+ if ( ! empty( $r['af'] ) ) {
	285	+ $agg['faces'][ $r['weight'] . '|' . $r['style'] ] = true;
	286	+ }
	287	+
	288	+ $families[ $fam ] = $agg;
	289	+ }
	290	+
	291	+ /**
	292	+ * Consistent beacon measurements required before a never-rendered
	293	+ * family is scoped off a page. Corroboration means one bad or forged
	294	+ * report can't remove a font (beacons are throttled hourly, so this is
	295	+ * also a minimum observation window).
	296	+ *
	297	+ * @param int $min
	298	+ */
	299	+ $min_misses = max( 1, (int) apply_filters( 'easyfonts_scope_out_min_measurements', 2 ) );
	300	+
	301	+ foreach ( $families as $fam => $r ) {
	302	+ $rendered = ! empty( $r['rnd'] );
	303	+ $beaconed = ! empty( $r['beaconed'] );
	304	+
	305	+ if ( $beaconed ) {
	306	+ $out['measured'] = true;
	307	+ }
	308	+
	309	+ if ( $rendered ) {
	310	+ $out['rendered'][ $fam ] = true;
	311	+ $out['fold_faces'][ $fam ] = $r['faces'];
	312	+
	313	+ if ( ! empty( $r['af'] ) ) {
	314	+ $out['above_fold'][ $fam ] = true;
266	315	}
	316	+ } elseif ( $beaconed && (int) $r['misses'] >= $min_misses ) {
	317	+ // Multiple beacons measured it and never saw it render →
	318	+ // confirmed scope-out.
	319	+ $out['scope_out'][ $fam ] = true;
267	320	}
268	321	}
269	322
270	323	return $out;
		@@ -276,14 +329,27 @@
276	329	* wherever they appear without consulting per-route data (and even before a
277	330	* given route has been measured), which is the cheap path: no per-route
278	331	* check needed for the fonts whose answer is the same everywhere.
279	332	*
280		- * @return array<string,bool> Lowercased family => true.
	333	+ * (2.0.4) With the faces ('weight|style') those families render in. A value
	334	+ * of `true` is a family stored by 2.0.3 without its faces — it stands for
	335	+ * the whole family until the next recompute (next beacon, or weekly GC).
	336	+ *
	337	+ * @return array<string,array<string,bool>|true> Lowercased family => faces.
281	338	*/
282		- public function global_preload_families(): array {
	339	+ public function global_preload_faces(): array {
283	340	$list = get_option( 'easyfonts_global_preload', array() );
	341	+ $out = array();
284	342
285		- return is_array( $list ) ? array_fill_keys( array_map( 'strval', $list ), true ) : array();
	343	+ foreach ( is_array( $list ) ? $list : array() as $key => $value ) {
	344	+ if ( is_array( $value ) ) {
	345	+ $out[ (string) $key ] = array_fill_keys( array_map( 'strval', $value ), true );
	346	+ } elseif ( is_int( $key ) ) {
	347	+ $out[ strtolower( (string) $value ) ] = true;
	348	+ }
	349	+ }
	350	+
	351	+ return $out;
286	352	}
287	353
288	354	/**
289	355	* Recompute and cache the uniformly-above-the-fold family set. A family
		@@ -288,10 +354,11 @@
288	354	/**
289	355	* Recompute and cache the uniformly-above-the-fold family set. A family
290	356	* qualifies when every measured row for it (across all routes/variants where
291	357	* it renders) is above the fold — so its preload answer doesn't vary by page
292		- * and we can skip per-route checks for it. Cheap: one grouped scan of the
293		- * (capped) usage table; called when a beacon is accepted and from GC.
	358	+ * and we can skip per-route checks for it. Stored with the faces it renders
	359	+ * in, so preload stays weight-level. Cheap: one grouped scan of the (capped)
	360	+ * usage table; called when a beacon is accepted and from GC.
294	361	*
295	362	* @return void
296	363	*/
297	364	public function recompute_global_preload(): void {
		@@ -297,18 +364,37 @@
297	364	public function recompute_global_preload(): void {
298	365	global $wpdb;
299	366
300	367	// phpcs:ignore WordPress.DB.DirectDatabaseQuery, WordPress.DB.PreparedSQL.NotPrepared
301		- $families = $wpdb->get_col(
302		- "SELECT family FROM {$this->usage_table()}
	368	+ $rows = $wpdb->get_results(
	369	+ "SELECT family, weight, style, MIN(above_fold) AS af FROM {$this->usage_table()}
303	370	WHERE rendered = 1
304		- GROUP BY family
305		- HAVING MIN(above_fold) = 1"
	371	+ GROUP BY family, weight, style",
	372	+ ARRAY_A
306	373	);
307	374
308		- $families = $families ? array_values( array_unique( array_map( 'strtolower', $families ) ) ) : array();
	375	+ $faces = array();
	376	+ $below = array();
309	377
310		- update_option( 'easyfonts_global_preload', $families, true );
	378	+ foreach ( (array) $rows as $r ) {
	379	+ $fam = strtolower( (string) $r['family'] );
	380	+
	381	+ if ( empty( $r['af'] ) ) {
	382	+ $below[ $fam ] = true;
	383	+ }
	384	+
	385	+ $faces[ $fam ][ $r['weight'] . '|' . $r['style'] ] = true;
	386	+ }
	387	+
	388	+ $global = array();
	389	+
	390	+ foreach ( $faces as $fam => $set ) {
	391	+ if ( ! isset( $below[ $fam ] ) ) {
	392	+ $global[ $fam ] = array_keys( $set );
	393	+ }
	394	+ }
	395	+
	396	+ update_option( 'easyfonts_global_preload', $global, true );
311	397	}
312	398
313	399	/**
314	400	* Forget the stored decision for a route (all devices), so the next beacon
		@@ -339,24 +425,26 @@
339	425	global $wpdb;
340	426
341	427	$key = sha1( $route . '|' . $device );
342	428
	429	+ $alias = $this->dup_alias();
	430	+
343	431	// phpcs:ignore WordPress.DB.DirectDatabaseQuery
344	432	$wpdb->query(
345	433	$wpdb->prepare(
346	434	"INSERT INTO {$this->decisions_table()}
347	435	(route_key, route, device, preload, unload, updated_at)
348		- VALUES (%s,%s,%s,%s,%s,%s)
	436	+ VALUES (%s,%s,%s,%s,%s,%s){$alias}
349	437	ON DUPLICATE KEY UPDATE
350		- preload = VALUES(preload),
351		- unload = VALUES(unload),
352		- updated_at = VALUES(updated_at)",
	438	+ preload = {$this->dup( 'preload' )},
	439	+ unload = {$this->dup( 'unload' )},
	440	+ updated_at = {$this->dup( 'updated_at' )}",
353	441	$key,
354	442	$route,
355	443	$device,
356	444	wp_json_encode( array_values( $preload ) ),
357	445	wp_json_encode( array_values( $unload ) ),
358		- current_time( 'mysql' )
	446	+ gmdate( 'Y-m-d H:i:s' ) // UTC — parsed as UTC in has_fresh_decision().
359	447	)
360	448	);
361	449	}
362	450
		@@ -457,19 +545,20 @@
457	545	$days = max( 1, $days );
458	546	$usage = $this->usage_table();
459	547	$dec = $this->decisions_table();
460	548
461		- // 1. Time-based prune (usage + decisions).
	549	+ // 1. Time-based prune (usage + decisions). Timestamps are stored in
	550	+ // UTC, so compare against UTC_TIMESTAMP() (not session-TZ NOW()).
462	551	// phpcs:disable WordPress.DB.DirectDatabaseQuery
463	552	$wpdb->query(
464	553	$wpdb->prepare(
465		- "DELETE FROM {$usage} WHERE last_seen IS NOT NULL AND last_seen < ( NOW() - INTERVAL %d DAY )",
	554	+ "DELETE FROM {$usage} WHERE last_seen IS NOT NULL AND last_seen < ( UTC_TIMESTAMP() - INTERVAL %d DAY )",
466	555	$days
467	556	)
468	557	);
469	558	$wpdb->query(
470	559	$wpdb->prepare(
471		- "DELETE FROM {$dec} WHERE updated_at IS NOT NULL AND updated_at < ( NOW() - INTERVAL %d DAY )",
	560	+ "DELETE FROM {$dec} WHERE updated_at IS NOT NULL AND updated_at < ( UTC_TIMESTAMP() - INTERVAL %d DAY )",
472	561	$days
473	562	)
474	563	);
475	564	// phpcs:enable
		@@ -474,8 +563,10 @@
474	563	);
475	564	// phpcs:enable
476	565
477	566	// 2. Hard cap: if usage still exceeds the cap, drop the oldest surplus.
	567	+ // The id-bounded form is deterministic (replication-safe), unlike a bare
	568	+ // DELETE ... ORDER BY ... LIMIT.
478	569	// phpcs:ignore WordPress.DB.DirectDatabaseQuery, WordPress.DB.PreparedSQL.NotPrepared
479	570	$total = (int) $wpdb->get_var( "SELECT COUNT(*) FROM {$usage}" );
480	571
481	572	if ( $total > $max_rows ) {
		@@ -483,14 +574,40 @@
483	574
484	575	// phpcs:ignore WordPress.DB.DirectDatabaseQuery
485	576	$wpdb->query(
486	577	$wpdb->prepare(
487		- "DELETE FROM {$usage} ORDER BY last_seen ASC LIMIT %d",
	578	+ "DELETE FROM {$usage} WHERE id IN (
	579	+ SELECT id FROM ( SELECT id FROM {$usage} ORDER BY last_seen ASC, id ASC LIMIT %d ) AS ef_tmp
	580	+ )",
488	581	$surplus
489	582	)
490	583	);
491	584	}
492	585
	586	+ // 3. Hard cap on decisions rows too — beacon routes are public input, so
	587	+ // without a cap the table could be grown without bound between prunes.
	588	+ /**
	589	+ * Hard cap on decision rows after time-based pruning.
	590	+ *
	591	+ * @param int $max
	592	+ */
	593	+ $max_dec = max( 100, (int) apply_filters( 'easyfonts_gc_max_decision_rows', 2000 ) );
	594	+
	595	+ // phpcs:ignore WordPress.DB.DirectDatabaseQuery, WordPress.DB.PreparedSQL.NotPrepared
	596	+ $total_dec = (int) $wpdb->get_var( "SELECT COUNT(*) FROM {$dec}" );
	597	+
	598	+ if ( $total_dec > $max_dec ) {
	599	+ // phpcs:ignore WordPress.DB.DirectDatabaseQuery
	600	+ $wpdb->query(
	601	+ $wpdb->prepare(
	602	+ "DELETE FROM {$dec} WHERE id IN (
	603	+ SELECT id FROM ( SELECT id FROM {$dec} ORDER BY updated_at ASC, id ASC LIMIT %d ) AS ef_tmp
	604	+ )",
	605	+ $total_dec - $max_dec
	606	+ )
	607	+ );
	608	+ }
	609	+
493	610	// Pruning may have changed which families are uniformly above the fold.
494	611	$this->recompute_global_preload();
495	612	}
496	613
		@@ -522,9 +639,54 @@
522	639	if ( ! $updated ) {
523	640	return false;
524	641	}
525	642
526		- return ( time() - strtotime( (string) $updated ) ) < ( max( 1, $fresh_hours ) * HOUR_IN_SECONDS );
	643	+ // Stored in UTC — parse it as UTC regardless of PHP's default timezone.
	644	+ return ( time() - strtotime( (string) $updated . ' UTC' ) ) < ( max( 1, $fresh_hours ) * HOUR_IN_SECONDS );
	645	+ }
	646	+
	647	+ /**
	648	+ * ON DUPLICATE KEY UPDATE row reference, portable across servers.
	649	+ *
	650	+ * MySQL 8.0.20 deprecated VALUES() in ON DUPLICATE KEY UPDATE in favour of
	651	+ * a row alias — which MariaDB does not support. Detect the server once and
	652	+ * emit whichever syntax is native, so neither engine logs deprecation
	653	+ * warnings nor breaks when VALUES() is eventually removed.
	654	+ *
	655	+ * @param string $column Column name.
	656	+ * @return string SQL fragment referencing the incoming row's value.
	657	+ */
	658	+ private function dup( string $column ): string {
	659	+ return $this->use_row_alias() ? "ef_new.{$column}" : "VALUES({$column})";
	660	+ }
	661	+
	662	+ /**
	663	+ * Row-alias suffix for the INSERT (empty when using VALUES()).
	664	+ *
	665	+ * @return string
	666	+ */
	667	+ private function dup_alias(): string {
	668	+ return $this->use_row_alias() ? ' AS ef_new' : '';
	669	+ }
	670	+
	671	+ /**
	672	+ * Should upserts use the MySQL 8.0.19+ row-alias syntax?
	673	+ *
	674	+ * @return bool
	675	+ */
	676	+ private function use_row_alias(): bool {
	677	+ static $use = null;
	678	+
	679	+ if ( null === $use ) {
	680	+ global $wpdb;
	681	+
	682	+ $info = method_exists( $wpdb, 'db_server_info' ) ? (string) $wpdb->db_server_info() : '';
	683	+ $is_mariadb = false !== stripos( $info, 'mariadb' );
	684	+
	685	+ $use = ! $is_mariadb && '' !== $info && version_compare( (string) $wpdb->db_version(), '8.0.19', '>=' );
	686	+ }
	687	+
	688	+ return $use;
527	689	}
528	690
529	691	/**
530	692	* Empty both tables.