404-solution / includes / database / upgrades / DatabaseUpgradeDailyMaintenance.php

DatabaseUpgradeDailyMaintenance.php in 404 Solution trunk, at includes/database/upgrades/DatabaseUpgradeDailyMaintenance.php

180 lines 9.0 KB

Raw Download Zip

1	<?php
2
3	if (!defined('ABSPATH')) {
4	exit;
5	}
6
7	/**
8	* Daily-cron entry point that fans out to every database-maintenance sub-task.
9	*
10	* Called by abj404_dailyMaintenanceCronJobListener() in 404-solution.php.
11	* Coordinates (in order): self-heal prologue, ngram cache sync/cleanup, expired
12	* transient cleanup, auto-redirect retention,
13	* canonical_url backfill (redirects + logsv2), redirects denorm backfill (Step
14	* 3a) + nightly full reconcile (Step 3d), internal-link scan, and an
15	* inline view_done snapshot refresh. The orchestrator owns the small in-house
16	* tasks that only the daily cron triggers (transient cleanup, view-done
17	* refresh); everything else routes through the coordinator delegate map.
18	*/
19	class ABJ_404_Solution_DatabaseUpgradeDailyMaintenance extends ABJ_404_Solution_DatabaseUpgradeComponent {
20
21	/**
22	* Run all database maintenance tasks.
23	*
24	* This is the main orchestrator method called by the daily maintenance cron job.
25	* It coordinates all database-related maintenance tasks in the proper order.
26	*
27	* Called by: abj404_dailyMaintenanceCronJobListener() in 404-solution.php
28	*
29	* @return void
30	*/
31	public function runDatabaseMaintenanceTasks() {
32	// Insurance: Verify tables exist (per-site or network-wide based on activation mode)
33	// This catches failed activations, database corruption, and edge cases.
34	// Routed through runSelfHealPrologue() so the SelfHealingPrologueReachabilityTest
35	// can confirm the daily cron reaches the canonical prologue token.
36	$this->upgrades()->selfHealUpgrade()->runSelfHealPrologue();
37
38	// Ngram cache maintenance: sync missing entries and cleanup orphaned ones
39	$this->upgrades()->nGramUpgrade()->syncMissingNGrams();
40	$this->upgrades()->nGramUpgrade()->cleanupOrphanedNGrams();
41
42	// Clean up expired rate limit transients to prevent wp_options bloat
43	$this->cleanupExpiredRateLimitTransients();
44
45	// Expire auto-created redirects that exceed the configured age threshold
46	abj_service('redirects_retention_service')->expireOldAutoRedirects();
47
48	// Backfill canonical_url on legacy redirect rows so the captured-page
49	// JOIN to logs_hits.requested_url stays index-friendly. Chunked + rate-
50	// limited so the daily cron continues progress without blocking large
51	// sites; converges on its own across successive runs.
52	$this->upgrades()->canonicalUrlBackfillUpgrade()->backfillRedirectsCanonicalUrl();
53
54	// Same idea for logsv2: legacy rows (pre-4.1.x) lack canonical_url, so
55	// the hits-rebuild JOIN falls back to CONCAT/TRIM and can't use
56	// idx_canonical_url. Chunked + rate-limited so even a multi-hundred-K
57	// logsv2 backlog converges across successive cron ticks. Tighter
58	// 15-second budget (vs redirects' 25) because this same function is
59	// also reachable from the browser-triggered lazy-backfill AJAX endpoint.
60	$this->upgrades()->canonicalUrlBackfillUpgrade()->backfillLogsv2CanonicalUrl();
61
62	// Denorm Step 3a (i459): populate the four derived columns (logshits,
63	// last_used, dest_for_view, published_status) on legacy redirect rows
64	// that pre-date the column add. Chunked + wall-clock-bounded so even a
65	// large redirects table converges across successive daily ticks without
66	// blocking activation. Runs after the canonical_url backfills so the
67	// hits rollup join matches on the freshly populated canonical_url column.
68	$this->upgrades()->redirectsDenormBackfillUpgrade()->backfillRedirectsDenormColumns();
69
70	// report5.md Finding 1: an install upgraded ACROSS the dest_sort_key
71	// column add already has dest_for_view populated, so the main backfill's
72	// dest_for_view IS NULL sentinel skips it and the converged-row live
73	// write-back never fires; the indexable Destination sort key would stay
74	// NULL until the nightly reconcile below happened to walk that row. This
75	// cheap narrow drain (no per-type joins) closes that window directly,
76	// keyed on the self-clearing dest_sort_key IS NULL AND dest_for_view IS
77	// NOT NULL sentinel. Runs after the main backfill so freshly-resolved rows
78	// (which already got their sort key) are skipped, and converges to a no-op.
79	$this->upgrades()->redirectsSortKeyBackfillUpgrade()->backfillRedirectsDestSortKey();
80
81	// report6.md: same one-time drain for url_sort_key (the indexable URL sort
82	// key, LEFT(url, 191)). New/edited rows get it in real time via the Step 3c
83	// recompute; this converges legacy rows that pre-date the column add so the
84	// URL sort is index-ordered on the captured tab too, without waiting for the
85	// nightly reconcile below.
86	$this->upgrades()->redirectsSortKeyBackfillUpgrade()->backfillRedirectsUrlSortKey();
87
88	// Denorm Step 3d (i462): nightly full reconcile of the same four derived
89	// columns for ALL redirect rows. The Tier-3 floor / backstop: a
90	// brute-force cursor-walked recompute that catches drift from raw-SQL
91	// writers that bypassed the Step 3c real-time hooks (no hook fired, no
92	// timestamp bumped). Chunked + wall-clock-bounded with a resumable id
93	// cursor so a large table converges across successive nightly ticks.
94	// Background-only: never on the read path, never blocks the table view;
95	// a stale value can only delay a refresh by one pass, never blank rows.
96	// Runs after the backfill so a fresh install's one-time population goes
97	// first and the reconcile then keeps the populated columns honest.
98	$this->upgrades()->redirectsDenormReconcileUpgrade()->reconcileRedirectsDenormColumns();
99
100	// Nightly internal-link scan: find broken internal links in published content.
101	if (class_exists('ABJ_404_Solution_InternalLinkScanner')) {
102	$scanner = new ABJ_404_Solution_InternalLinkScanner();
103	$scanner->runNightlyScan();
104	}
105	}
106
107	/**
108	* Clean up expired rate limit transients from wp_options table.
109	*
110	* WordPress transients are supposed to auto-delete when they expire, but in practice
111	* they can accumulate over time. This maintenance task removes expired rate limit
112	* transients to prevent wp_options table bloat.
113	*
114	* Called during daily maintenance cron job.
115	*
116	* @return array<string, mixed> Statistics: ['deleted' => int, 'errors' => int]
117	*/
118	public function cleanupExpiredRateLimitTransients() {
119	global $wpdb;
120
121	$this->logger->debugMessage("Cleaning up expired rate limit transients...");
122
123	$stats = ['deleted' => 0, 'errors' => 0];
124
125	// Delete expired rate limit transients
126	// WordPress stores transients as two rows: _transient_* and _transient_timeout_*
127	// The timeout row contains the expiration timestamp
128	// We delete both the value and timeout rows for expired transients
129
130	$currentTime = abj_clock()->now();
131
132	// Find all expired rate limit timeout keys
133	// DAO-bypass-approved: WP-core wp_options probe; $wpdb->prepare is read-only string formatting, executed via $wpdb->get_col below
134	$query = $wpdb->prepare(
135	"SELECT option_name FROM {$wpdb->options}
136	WHERE option_name LIKE %s
137	AND option_value < %d",
138	$wpdb->esc_like('_transient_timeout_abj404_rate_limit_') . '%',
139	$currentTime
140	);
141
142	// DAO-bypass-approved: Outside-plugin-tables wp_options cleanup probe (parallels DataAccessTrait_ViewQueries:478 transient clear)
143	$expiredTimeouts = $wpdb->get_col($query);
144
145	$lastError = (string)($wpdb->last_error ?? '');
146	if ($lastError !== '') {
147	if (!$this->dbCore->errorClassifier()->classifyAndHandleInfrastructureError($lastError)) {
148	$this->logger->errorMessage("Failed to query for expired rate limit transients: " . $lastError);
149	}
150	return ['deleted' => 0, 'errors' => 1, 'error' => $lastError];
151	}
152
153	if (!empty($expiredTimeouts)) {
154	$this->logger->debugMessage("Found " . count($expiredTimeouts) . " expired rate limit transients to delete.");
155
156	foreach ($expiredTimeouts as $timeoutKey) {
157	// Get the corresponding value key (remove '_timeout' from the name)
158	$valueKey = str_replace('_transient_timeout_', '_transient_', $timeoutKey);
159
160	// Delete both the timeout and value rows
161	$timeoutDeleted = delete_option($timeoutKey);
162	$valueDeleted = delete_option($valueKey);
163
164	if ($timeoutDeleted \|\| $valueDeleted) {
165	$stats['deleted']++;
166	} else {
167	$stats['errors']++;
168	}
169	}
170
171	$this->logger->debugMessage("Deleted {$stats['deleted']} expired rate limit transients, {$stats['errors']} errors.");
172	} else {
173	$this->logger->debugMessage("No expired rate limit transients found.");
174	}
175
176	return $stats;
177	}
178
179	}
180