404-solution / includes / redirects / RedirectDeadDestinationChecker.php

RedirectDeadDestinationChecker.php in 404 Solution trunk, at includes/redirects/RedirectDeadDestinationChecker.php

266 lines 10.0 KB

Raw Download Zip

1	<?php
2
3	if (!defined('ABSPATH')) {
4	exit;
5	}
6
7	/**
8	* Checks whether specific redirects point at a dead destination, i.e. a
9	* destination URL that shows recent failed hits in the logs_hits rollup.
10	*
11	* Unlike the former global precompute (which scanned every redirect into a
12	* transient), this checker is bounded by the caller's id list: it only ever
13	* inspects the redirect ids it is handed, so the read never loads more than
14	* the caller's working set.
15	*/
16	class ABJ_404_Solution_RedirectDeadDestinationChecker {
17
18	/** @var ABJ_404_Solution_DatabaseCore */
19	private $dbCore;
20
21	/** @var ABJ_404_Solution_Logging */
22	private $logger;
23
24	/**
25	* Per-request memo of rollup readiness (table present AND failed_hits column).
26	* Invariant within a request, so the two schema probes run at most once even
27	* when matching evaluates a redirect across several URL-normalization passes.
28	*
29	* @var bool\|null
30	*/
31	private $rollupReady = null;
32
33	/**
34	* Per-request memo of resolved ids: id => is its destination dead. Lets the
35	* repeated matching passes for the same redirect avoid re-querying.
36	*
37	* @var array<int, bool>
38	*/
39	private $deadByIdMemo = array();
40
41	/**
42	* @param ABJ_404_Solution_DatabaseCore $dbCore
43	* @param ABJ_404_Solution_Logging $logger
44	*/
45	public function __construct(
46	ABJ_404_Solution_DatabaseCore $dbCore,
47	ABJ_404_Solution_Logging $logger
48	) {
49	$this->dbCore = $dbCore;
50	$this->logger = $logger;
51	}
52
53	/**
54	* Returns the subset of the given redirect ids whose destination URL shows
55	* recent failed hits (i.e. the redirect points at a dead page). Bounded by
56	* the caller's id list: the Page Redirects tab passes the ids it is
57	* rendering (~25), and live redirect matching passes the single redirect
58	* being evaluated, so the read never loads more than the caller's working
59	* set. Fails open (returns empty) when the logs_hits rollup is absent or the
60	* query times out, so a redirect is never blocked by an infrastructure gap.
61	*
62	* @param array<int, int\|string> $redirectIds
63	* @return array<int, string> the subset of $redirectIds that are dead, as strings
64	*/
65	public function findDeadDestinationIds(array $redirectIds): array {
66	$ids = array();
67	foreach ($redirectIds as $candidate) {
68	if (is_scalar($candidate)) {
69	$intId = (int) $candidate;
70	if ($intId > 0) {
71	$ids[$intId] = $intId;
72	}
73	}
74	}
75	if (empty($ids)) {
76	return array();
77	}
78
79	// Serve ids already resolved this request from the memo; only query the rest.
80	$deadIds = array();
81	$toQuery = array();
82	foreach ($ids as $intId) {
83	if (array_key_exists($intId, $this->deadByIdMemo)) {
84	if ($this->deadByIdMemo[$intId]) {
85	$deadIds[] = (string) $intId;
86	}
87	} else {
88	$toQuery[$intId] = $intId;
89	}
90	}
91	if (empty($toQuery)) {
92	return $deadIds;
93	}
94
95	if (!$this->rollupReady()) {
96	// Rollup absent: cannot determine. Do NOT memo as "not dead" -- the
97	// rollup may be rebuilt later in the request lifecycle.
98	return $deadIds;
99	}
100
101	$freshDead = $this->queryDeadDestinationIds($toQuery);
102	if ($freshDead === null) {
103	// Fail open (timeout/error/throwable): never block a redirect, and do
104	// NOT memo -- the answer is unknown, not "not dead".
105	return $deadIds;
106	}
107
108	// Resolve every queried id (dead or not) so repeated matching passes in
109	// the same request never re-query.
110	$freshDeadSet = array();
111	foreach ($freshDead as $deadIdStr) {
112	$freshDeadSet[(int) $deadIdStr] = true;
113	}
114	foreach ($toQuery as $intId) {
115	$isDead = isset($freshDeadSet[$intId]);
116	$this->deadByIdMemo[$intId] = $isDead;
117	if ($isDead) {
118	$deadIds[] = (string) $intId;
119	}
120	}
121	return $deadIds;
122	}
123
124	/**
125	* Runs the bounded dead-destination read for the given (already-sanitized,
126	* positive-int) redirect ids. Returns the dead-id strings on success, or null
127	* when the read could not complete (timeout, DAO error, or a rethrown
128	* Throwable) -- this method moved from daily cron to request time, so a DB
129	* fault must fail open here and never propagate into frontend redirect
130	* matching. queryAndGetResults already logs the underlying error.
131	*
132	* @param array<int, int> $ids
133	* @return array<int, string>\|null
134	*/
135	private function queryDeadDestinationIds(array $ids): ?array {
136	$idList = implode(',', array_map('intval', array_values($ids)));
137	$logsHitsTable = $this->dbCore->doTableNameReplacements('{wp_abj404_logs_hits}');
138	$destinationUrl = "CONCAT('/', TRIM(BOTH '/' FROM r.final_dest))";
139	$comparableDestinationUrl = $this->dbCore->collationHelper()->coerceExpressionToColumnCollation(
140	$destinationUrl,
141	array('table' => $logsHitsTable, 'column' => 'requested_url')
142	);
143	// allow-unbounded-select: bounded by the caller's r.id IN (...) working set -- the Page Redirects tab passes the page's row ids, live matching passes the single matched redirect id
144	$sql = "SELECT DISTINCT r.id
145	FROM {wp_abj404_redirects} r
146	INNER JOIN {wp_abj404_logs_hits} h
147	ON h.requested_url = " . $comparableDestinationUrl . "
148	AND BINARY h.requested_url = BINARY CONCAT('/', TRIM(BOTH '/' FROM r.final_dest))
149	WHERE r.id IN (" . $idList . ")
150	AND h.last_used > %d
151	AND h.failed_hits > 0
152	AND r.disabled = 0
153	AND r.final_dest != ''
154	AND r.final_dest != '0'";
155	try {
156	$sql = $this->dbCore->doTableNameReplacements($sql);
157	$result = $this->dbCore->queryAndGetResults($sql, array(
158	'query_params' => array(abj_clock()->now() - 7 * 86400),
159	'timeout' => 5,
160	));
161	} catch (Throwable $e) {
162	$this->logger->debugMessage(__CLASS__ . '/' . __FUNCTION__
163	. ': dead-destination read failed open (' . $e->getMessage() . '); no redirect suspended.');
164	return null;
165	}
166
167	if (!empty($result['timed_out']) \|\| (isset($result['last_error']) && $result['last_error'] != '')) {
168	$this->logger->debugMessage(__CLASS__ . '/' . __FUNCTION__
169	. ': dead-destination read timed out or errored; failing open, no redirect suspended.');
170	return null;
171	}
172
173	$deadIds = array();
174	$rows = is_array($result['rows'] ?? null) ? $result['rows'] : array();
175	foreach ($rows as $row) {
176	$value = $this->extractIdValue($row);
177	if ($value !== null) {
178	$deadIds[] = $value;
179	}
180	}
181	return $deadIds;
182	}
183
184	/**
185	* Is the logs_hits rollup usable (table present AND failed_hits column)?
186	* Memoized for the request. When not ready, schedules a rebuild once (the
187	* rollup's own lifecycle in LogsHitsRollupService also schedules rebuilds
188	* from the logging and admin paths, so this is an opportunistic top-up, not
189	* the primary maintainer). The probes are wrapped so a schema-query throwable
190	* fails open rather than breaking a frontend redirect.
191	*
192	* @return bool
193	*/
194	private function rollupReady(): bool {
195	if ($this->rollupReady !== null) {
196	return $this->rollupReady;
197	}
198
199	try {
200	$hitsTable = $this->dbCore->doTableNameReplacements('{wp_abj404_logs_hits}');
201	$ready = $this->dbCore->tableNameResolver()->tableExists($hitsTable)
202	&& $this->logsHitsHasFailedHitsColumn();
203	} catch (Throwable $e) {
204	$this->logger->debugMessage(__CLASS__ . '/' . __FUNCTION__
205	. ': rollup readiness probe failed open (' . $e->getMessage() . ').');
206	$ready = false;
207	}
208
209	if (!$ready) {
210	// Scheduling a rebuild is opportunistic and must also fail open: a
211	// throwable from service resolution or scheduling must never escape
212	// into frontend redirect matching.
213	try {
214	/** @var ABJ_404_Solution_LogsRepository\|null $logsRepo */
215	$logsRepo = abj_service('logs_repository');
216	if ($logsRepo !== null) {
217	$logsRepo->scheduleHitsTableRebuild();
218	}
219	} catch (Throwable $e) {
220	$this->logger->debugMessage(__CLASS__ . '/' . __FUNCTION__
221	. ': rollup rebuild scheduling failed open (' . $e->getMessage() . ').');
222	}
223	}
224
225	$this->rollupReady = $ready;
226	return $ready;
227	}
228
229	/**
230	* @param mixed $row
231	* @return string\|null
232	*/
233	private function extractIdValue($row): ?string {
234	if (is_array($row)) {
235	$value = $row['id'] ?? reset($row);
236	} elseif (is_object($row)) {
237	$value = $row->id ?? null;
238	} else {
239	$value = $row;
240	}
241
242	if (is_scalar($value) && (string)$value !== '') {
243	return (string)$value;
244	}
245
246	return null;
247	}
248
249	private function logsHitsHasFailedHitsColumn(): bool {
250	$tableName = $this->dbCore->doTableNameReplacements('{wp_abj404_logs_hits}');
251	$sql = "SELECT 1 FROM information_schema.columns "
252	. "WHERE table_schema = DATABASE() "
253	. "AND table_name = %s "
254	. "AND column_name = 'failed_hits' LIMIT 1";
255	$result = $this->dbCore->queryAndGetResults($sql, array(
256	'query_params' => array($tableName),
257	'log_errors' => false,
258	));
259	if (!empty($result['last_error'])) {
260	return false;
261	}
262	$rows = is_array($result['rows'] ?? null) ? $result['rows'] : array();
263	return !empty($rows);
264	}
265	}
266