404-solution / includes / DataAccessTrait_ViewQueriesHitsLifecycle.php

DataAccessTrait_ViewQueriesHitsLifecycle.php in 404 Solution 4.1.19, at includes/DataAccessTrait_ViewQueriesHitsLifecycle.php

344 lines 14.2 KB

Raw Download Zip

1	<?php
2
3	if (!defined('ABSPATH')) {
4	exit;
5	}
6
7	/**
8	* Hits-table lifecycle helpers extracted from
9	* ABJ_404_Solution_DataAccess_ViewQueriesTrait to keep the parent trait under
10	* the ModularityTest CLASS_LIMIT. Owns the *scheduling, locking, and
11	* existence-probe* concerns for the `{wp_abj404_logs_hits}` rollup table,
12	* plus the small log-id watermark helpers (getMaxLogId / getMinLogId /
13	* getStoredMaxLogId) the scheduling path consults. The rebuild path
14	* itself lives in ABJ_404_Solution_DataAccess_LogsHitsRebuildTrait
15	* (DataAccessTrait_LogsHitsRebuild.php).
16	*
17	* Composed into ABJ_404_Solution_DataAccess alongside the other DAO traits.
18	* Depends on host-class state and methods provided by the rest of DAO:
19	* `queryAndGetResults`, `doTableNameReplacements`, `getRuntimeFlag`,
20	* `setRuntimeFlag`, `getLowercasePrefix`, `shouldSkipNonEssentialDbWrites`,
21	* `recordLogsHitsRollupStalenessSignal`, `hitsTableNeedsRebuild`,
22	* `createRedirectsForViewHitsTable`, `getLogsHitsTableStatusRow`, the
23	* `$logger` property, the `$hitsTableRebuildScheduled` static, and the
24	* HITS_TABLE_* constants on the host class.
25	*/
26	trait ABJ_404_Solution_DataAccess_ViewQueriesHitsLifecycleTrait {
27
28	/** @return void */
29	function maybeUpdateRedirectsForViewHitsTable(): void {
30	// Record that we checked during this request (used for admin tooltip UX).
31	$this->setRuntimeFlag(self::HITS_TABLE_LAST_CHECKED_FLAG, time(), 86400);
32
33	// Piggyback on the captured-404s tab render: also schedule a
34	// 15-second logsv2.canonical_url backfill at shutdown if there's
35	// legacy NULL-row backlog. The shutdown handler holds a worker
36	// for the budget but the admin response is already flushed by
37	// fastcgi_finish_request, so the user doesn't perceive the wait.
38	// The function is internally deduped + gated on column existence,
39	// probe results, and the backfill-complete option, so calling it
40	// unconditionally is cheap.
41	if (function_exists('abj_service')) {
42	$upgradesEtc = abj_service('database_upgrades');
43	if (is_object($upgradesEtc) && method_exists($upgradesEtc, 'scheduleLogsv2CanonicalUrlBackfill')) {
44	$upgradesEtc->scheduleLogsv2CanonicalUrlBackfill();
45	}
46	}
47
48	if ($this->shouldSkipNonEssentialDbWrites()) {
49	$this->logger->debugMessage(__FUNCTION__ . " skipped due to temporary DB write cooldown.");
50	$this->setRuntimeFlag(self::HITS_TABLE_LAST_DECISION_FLAG, 'paused', 86400);
51	return;
52	}
53
54	// Check if the table exists
55	if (!$this->logsHitsTableExists()) {
56	// Defer creation to shutdown hook so the admin page loads immediately.
57	// The view query gracefully falls back to null hits columns when the
58	// table doesn't exist (getRedirectsForViewQuery checks logsHitsTableExists).
59	// On sites with large logsv2 tables the INSERT...SELECT that populates
60	// the hits table can take minutes, which exceeds proxy timeouts (e.g.
61	// Cloudflare's 100-second limit → HTTP 524).
62	$this->logger->debugMessage(__FUNCTION__ . " table doesn't exist, deferring creation to shutdown hook.");
63	$this->scheduleHitsTableRebuild();
64	return;
65	}
66
67	// Diagnostic: track the max_log_id age signal so a stalled rollup
68	// surfaces a broken-cron admin notice instead of silently showing
69	// stale hit-count columns. Self-heals when the gap closes.
70	$this->recordLogsHitsRollupStalenessSignal();
71
72	// Check if rebuild is needed (logs have changed since last build)
73	if (!$this->hitsTableNeedsRebuild()) {
74	// No new log entries - skip rebuild to reduce server load
75	$this->setRuntimeFlag(self::HITS_TABLE_LAST_DECISION_FLAG, 'not_needed', 86400);
76	return;
77	}
78
79	// Table exists and logs have changed - defer to shutdown hook
80	$this->scheduleHitsTableRebuild();
81	}
82
83	/**
84	* Schedule the hits table to be rebuilt at shutdown.
85	*
86	* Uses a static flag to ensure the hook is only registered once per request,
87	* even if multiple calls to getRedirectsForView with hits sorting occur.
88	*
89	* The shutdown hook runs after the response is sent, so the admin sees the page
90	* immediately with existing data, and fresh data is available on next load.
91	*/
92	/** @return void */
93	function scheduleHitsTableRebuild(): void {
94	if ($this->shouldSkipNonEssentialDbWrites()) {
95	$this->logger->debugMessage(__FUNCTION__ . " skipped due to temporary DB write cooldown.");
96	$this->setRuntimeFlag(self::HITS_TABLE_LAST_DECISION_FLAG, 'paused', 86400);
97	return;
98	}
99	if (!self::$hitsTableRebuildScheduled) {
100	if ($this->isHitsTableRebuildLocked()) {
101	$this->logger->debugMessage(__FUNCTION__ . " skipping scheduling because another rebuild is already running.");
102	$this->setRuntimeFlag(self::HITS_TABLE_LAST_DECISION_FLAG, 'running', 86400);
103	return;
104	}
105
106	$rawScheduledFlag = $this->getRuntimeFlag(self::HITS_TABLE_LAST_SCHEDULED_FLAG);
107	$lastScheduled = is_scalar($rawScheduledFlag) ? (int)$rawScheduledFlag : 0;
108	if ($lastScheduled > 0 && (time() - $lastScheduled) < self::HITS_TABLE_SCHEDULE_COOLDOWN_SECONDS) {
109	$this->logger->debugMessage(__FUNCTION__ . " skipping scheduling due to cooldown.");
110	$this->setRuntimeFlag(self::HITS_TABLE_LAST_DECISION_FLAG, 'cooldown', 86400);
111	return;
112	}
113
114	self::$hitsTableRebuildScheduled = true;
115	$this->setRuntimeFlag(self::HITS_TABLE_LAST_SCHEDULED_FLAG, time(), 86400);
116	$this->setRuntimeFlag(self::HITS_TABLE_LAST_DECISION_FLAG, 'scheduled', 86400);
117	if ($this->shouldScheduleHitsTableRebuildViaCron()) {
118	$this->logger->debugMessage(__FUNCTION__ . " scheduling hits table rebuild via WP-Cron.");
119	if (function_exists('wp_schedule_single_event')) {
120	wp_schedule_single_event(time() + 5, 'abj404_updateLogsHitsTableAction');
121	}
122	return;
123	}
124
125	$this->logger->debugMessage(__FUNCTION__ . " scheduling hits table rebuild for shutdown hook.");
126	add_action('shutdown', function(): void { $this->createRedirectsForViewHitsTable(); });
127	}
128	}
129
130	/** @return bool */
131	private function shouldScheduleHitsTableRebuildViaCron(): bool {
132	if (function_exists('wp_doing_ajax') && wp_doing_ajax()) {
133	return true;
134	}
135	$scriptName = isset($_SERVER['SCRIPT_NAME']) && is_string($_SERVER['SCRIPT_NAME'])
136	? $_SERVER['SCRIPT_NAME'] : '';
137	if ($scriptName !== '' && basename($scriptName) === 'admin-ajax.php') {
138	return true;
139	}
140	$pagenow = isset($GLOBALS['pagenow']) && is_string($GLOBALS['pagenow'])
141	? $GLOBALS['pagenow'] : '';
142	return $pagenow === 'admin-ajax.php';
143	}
144
145	private function getHitsTableRebuildLockOptionName(): string {
146	return $this->getLowercasePrefix() . 'abj404_logs_hits_rebuild_lock';
147	}
148
149	/** @return bool */
150	private function isHitsTableRebuildLocked(): bool {
151	if (!function_exists('get_option')) {
152	return false;
153	}
154	$lockValue = get_option($this->getHitsTableRebuildLockOptionName(), false);
155	if ($lockValue === false \|\| $lockValue === null \|\| $lockValue === '') {
156	return false;
157	}
158	// Defensive: if lock is corrupted (non-numeric), clear it so rebuilds can resume.
159	if (!is_numeric($lockValue)) {
160	if (function_exists('delete_option')) {
161	delete_option($this->getHitsTableRebuildLockOptionName());
162	}
163	return false;
164	}
165	$lockTimestamp = (int)$lockValue;
166	if ($lockTimestamp > 0 && (time() - $lockTimestamp) > self::HITS_TABLE_REBUILD_LOCK_TTL_SECONDS) {
167	if (function_exists('delete_option')) {
168	delete_option($this->getHitsTableRebuildLockOptionName());
169	}
170	return false;
171	}
172	return true;
173	}
174
175	/** @return int\|null */
176	function getLogsHitsTableLastCheckedAt() {
177	$rawTsFlag = $this->getRuntimeFlag(self::HITS_TABLE_LAST_CHECKED_FLAG);
178	$ts = is_scalar($rawTsFlag) ? (int)$rawTsFlag : 0;
179	return $ts > 0 ? $ts : null;
180	}
181
182	/** @return int\|null */
183	function getLogsHitsTableLastScheduledAt() {
184	$rawTsFlag2 = $this->getRuntimeFlag(self::HITS_TABLE_LAST_SCHEDULED_FLAG);
185	$ts = is_scalar($rawTsFlag2) ? (int)$rawTsFlag2 : 0;
186	return $ts > 0 ? $ts : null;
187	}
188
189	/** @return string */
190	function getLogsHitsTableLastDecision(): string {
191	$v = $this->getRuntimeFlag(self::HITS_TABLE_LAST_DECISION_FLAG);
192	return is_string($v) ? $v : '';
193	}
194
195	/** @return bool */
196	private function acquireHitsTableRebuildLock(): bool {
197	if (!function_exists('add_option')) {
198	return true;
199	}
200	if ($this->isHitsTableRebuildLocked()) {
201	return false;
202	}
203	return (bool)add_option(
204	$this->getHitsTableRebuildLockOptionName(),
205	(string)time(),
206	'',
207	false
208	);
209	}
210
211	/** @return void */
212	private function releaseHitsTableRebuildLock(): void {
213	if (function_exists('delete_option')) {
214	delete_option($this->getHitsTableRebuildLockOptionName());
215	}
216	}
217
218	/** @return bool */
219	private function logsHitsTableExistsViaShowTables(): bool {
220	global $wpdb;
221	if (!isset($wpdb) \|\| !method_exists($wpdb, 'prepare')) {
222	return false;
223	}
224	$tableName = $this->doTableNameReplacements('{wp_abj404_logs_hits}');
225	/** @var wpdb $wpdb */
226	// DAO-bypass-approved: $wpdb->prepare is read-only string formatting; the resulting SQL is executed below via queryAndGetResults
227	$showTablesQuery = $wpdb->prepare("SHOW TABLES LIKE %s", $tableName);
228	if ($showTablesQuery === null) {
229	return false;
230	}
231	$fallback = $this->queryAndGetResults($showTablesQuery, array('log_errors' => false));
232	if (empty($fallback['rows'])) {
233	return false;
234	}
235	$fbRows = is_array($fallback['rows']) ? $fallback['rows'] : array();
236	$firstRow = isset($fbRows[0]) ? $fbRows[0] : null;
237	if (!is_array($firstRow)) {
238	return false;
239	}
240	$value = reset($firstRow);
241	return ((string)$value === (string)$tableName);
242	}
243
244	/**
245	* Check if the logs_hits table exists.
246	* Used to verify table was created before using it in queries.
247	* @return bool
248	*/
249	function logsHitsTableExists() {
250	$query = "SELECT 1 FROM information_schema.tables WHERE table_name = '{wp_abj404_logs_hits}' AND table_schema = DATABASE() LIMIT 1";
251	$query = $this->doTableNameReplacements($query);
252	$results = $this->queryAndGetResults($query);
253	if ($results['rows'] != null && !empty($results['rows'])) {
254	return true;
255	}
256	if (!empty($results['last_error'])) {
257	// Some hosts restrict information_schema access; fall back to SHOW TABLES.
258	return $this->logsHitsTableExistsViaShowTables();
259	}
260	return false;
261	}
262
263	/**
264	* Get the maximum log ID from the logs table.
265	*
266	* Used to detect if logs have changed since the hits table was last built.
267	* O(1) query using primary key index.
268	*
269	* @return int Maximum log ID, or 0 if table is empty
270	*/
271	function getMaxLogId() {
272	$query = "SELECT MAX(id) FROM {wp_abj404_logsv2}";
273	$query = $this->doTableNameReplacements($query);
274	$results = $this->queryAndGetResults($query);
275
276	$resultRows = is_array($results['rows']) ? $results['rows'] : array();
277	if (empty($resultRows)) {
278	return 0;
279	}
280
281	$row = $resultRows[0];
282	// Handle both object and array results
283	$maxId = is_array($row) ? array_values($row)[0] : (array_values((array)$row)[0] ?? 0);
284	return (int)($maxId ?? 0);
285	}
286
287	/** @return int */
288	function getMinLogId() {
289	$query = "SELECT MIN(id) FROM {wp_abj404_logsv2}";
290	$query = $this->doTableNameReplacements($query);
291	$results = $this->queryAndGetResults($query);
292
293	$resultRows = is_array($results['rows']) ? $results['rows'] : array();
294	if (empty($resultRows)) {
295	return 0;
296	}
297
298	$row = $resultRows[0];
299	$minId = is_array($row) ? array_values($row)[0] : (array_values((array)$row)[0] ?? 0);
300	return is_numeric($minId) ? (int)$minId : 0;
301	}
302
303	/**
304	* Get the stored max log ID from the hits table comment.
305	*
306	* Comment format: "elapsed_time\|max_log_id" (e.g., "0.35\|12345")
307	*
308	* @return int Stored max log ID, or 0 if not found
309	*/
310	function getStoredMaxLogId() {
311	$query = "SELECT table_comment FROM information_schema.tables WHERE table_name = '{wp_abj404_logs_hits}' AND table_schema = DATABASE()";
312	$query = $this->doTableNameReplacements($query);
313	$results = $this->queryAndGetResults($query);
314
315	$storedRows = is_array($results['rows']) ? $results['rows'] : array();
316	if (empty($storedRows)) {
317	if (!empty($results['last_error'])) {
318	$statusRow = $this->getLogsHitsTableStatusRow();
319	$commentFromStatus = $statusRow['comment'] ?? '';
320	if ($commentFromStatus !== '') {
321	$parts = explode('\|', is_string($commentFromStatus) ? $commentFromStatus : '');
322	if (count($parts) >= 2) {
323	return (int)$parts[1];
324	}
325	}
326	}
327	return 0;
328	}
329
330	$row = is_array($storedRows[0] ?? null) ? $storedRows[0] : array();
331	$row = array_change_key_case($row);
332	$comment = $row['table_comment'] ?? '';
333
334	// Parse comment format: "elapsed_time\|max_log_id"
335	$parts = explode('\|', is_string($comment) ? $comment : '');
336	if (count($parts) >= 2) {
337	return (int)$parts[1];
338	}
339
340	// Old format (just elapsed time) or empty - treat as needing rebuild
341	return 0;
342	}
343	}
344