404-solution / includes / redirects / RedirectsBulkReader.php

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

229 lines 9.0 KB

Raw Download Zip

1	<?php
2
3	if (!defined('ABSPATH')) {
4	exit;
5	}
6
7	/**
8	* Specialized, non-paginated reads of the redirects table for callers that
9	* don't go through the staged admin-list pipeline.
10	*
11	* Owns:
12	* - doRedirectsExport: stream the redirects table to a CSV temp file
13	* - getRedirectsWithRegEx: regex redirects with a static request-scoped cache
14	* - getManualRedirectsWithRegexMetachars: manual redirects whose URL
15	* contains regex metacharacters (for the matcher's wildcard fallback)
16	* - getExtraDataToPermalinkSuggestions: post metadata for suggestion ids
17	*
18	* Extracted from ViewReadService in the i805 decomposition. The regex
19	* cache is still owned by RedirectsRepository (its static cache holds the
20	* per-request state); this reader just consults it.
21	*/
22	class ABJ_404_Solution_RedirectsBulkReader {
23
24	/** Number of rows fetched by each keyset page once caching is unsafe. */
25	const REGEX_READ_BATCH_SIZE = 250;
26
27	/** @var ABJ_404_Solution_DatabaseCore */
28	private $dbCore;
29
30	/** @var ABJ_404_Solution_ViewQueryBuilder */
31	private $queryBuilder;
32
33	/** @var ABJ_404_Solution_Functions */
34	private $f;
35
36	/**
37	* @param ABJ_404_Solution_DatabaseCore $dbCore
38	* @param ABJ_404_Solution_ViewQueryBuilder $queryBuilder
39	* @param ABJ_404_Solution_Functions $f
40	*/
41	public function __construct(
42	ABJ_404_Solution_DatabaseCore $dbCore,
43	ABJ_404_Solution_ViewQueryBuilder $queryBuilder,
44	$f
45	) {
46	$this->dbCore = $dbCore;
47	$this->queryBuilder = $queryBuilder;
48	$this->f = $f;
49	}
50
51	/**
52	* Stream the export query straight to a CSV temp file via mysqli to keep
53	* the row buffer bounded on large redirect tables.
54	*
55	* @param string $tempFile
56	* @return void
57	*/
58	public function doRedirectsExport(string $tempFile): void {
59	global $wpdb;
60
61	if (file_exists($tempFile)) {
62	ABJ_404_Solution_FileSystemService::safeUnlink($tempFile);
63	}
64
65	$query = ABJ_404_Solution_FileSystemService::readFileContents(__DIR__ . "/../sql/getRedirectsExport.sql");
66	$query = $this->dbCore->doTableNameReplacements($query);
67
68	$result = mysqli_query($wpdb->dbh, $query);
69	if ($result instanceof \mysqli_result) {
70	$fh = fopen($tempFile, 'w');
71	if ($fh === false) {
72	mysqli_free_result($result);
73	return;
74	}
75	// try/finally: mysqli's default error mode (MYSQLI_REPORT_ERROR \|
76	// MYSQLI_REPORT_STRICT since PHP 8.1) throws mysqli_sql_exception
77	// on a dropped connection mid-fetch. Without a guaranteed close
78	// here, that exception would skip fclose($fh) and leak the file
79	// handle. Same resource-lifecycle shape as
80	// includes/import/ImportService.php::doImportFile().
81	try {
82	fputcsv($fh, array('from_url', 'status', 'type', 'to_url', 'wp_type', 'engine', 'code'), ',', '"', '\\');
83
84	while (($row = mysqli_fetch_array($result, MYSQLI_ASSOC))) {
85	fputcsv($fh, array(
86	$row['from_url'],
87	$row['status'],
88	$row['type'],
89	$row['to_url'],
90	$row['type_wp'],
91	isset($row['engine']) ? $row['engine'] : '',
92	isset($row['code']) ? $row['code'] : '301'
93	), ',', '"', '\\');
94	}
95	} finally {
96	fclose($fh);
97	mysqli_free_result($result);
98	}
99	}
100	}
101
102	/** @return iterable<int, array<string, mixed>> */
103	public function getRedirectsWithRegEx(): iterable {
104	$cached = ABJ_404_Solution_RedirectsRepository::getRegexRedirectsCache();
105	$disabled = ABJ_404_Solution_RedirectsRepository::isRegexCacheDisabled();
106
107	if ($cached !== null && !$disabled) {
108	return $cached;
109	}
110
111	if ($disabled) {
112	return $this->iterateAllRegexRedirectsInBatches();
113	}
114
115	$results = $this->queryBuilder->queryRegexRedirects(ABJ_404_Solution_RedirectsRepository::REGEX_CACHE_MAX_COUNT + 1);
116
117	if (count($results) <= ABJ_404_Solution_RedirectsRepository::REGEX_CACHE_MAX_COUNT) {
118	ABJ_404_Solution_RedirectsRepository::setRegexRedirectsCache($results);
119	} else {
120	ABJ_404_Solution_RedirectsRepository::setRegexCacheDisabled(true);
121	return $this->iterateAllRegexRedirectsInBatches($results);
122	}
123
124	return $results;
125	}
126
127	/**
128	* Stream a complete regex redirect read without allowing either a SQL
129	* result set or the PHP row collection to grow without bound. The optional
130	* leading rows are the cache-threshold probe and are reused so the common
131	* 51+ path does not reread them.
132	*
133	* @param array<int, array<string, mixed>> $leadingRows
134	* @return iterable<int, array<string, mixed>>
135	*/
136	private function iterateAllRegexRedirectsInBatches(array $leadingRows = array()): iterable {
137	// DESIGN-AUDIT-OK(2026-08-21, owner): A total-work cap would silently make later valid regex rules unreachable, reproducing Troy's 51-rule defect.
138	// The admin-owned finite table is streamed in 250-row pages, holds bounded memory, and stops as soon as a caller finds a match.
139	$afterId = $this->greatestRedirectId($leadingRows);
140	foreach ($leadingRows as $row) {
141	yield $row;
142	}
143
144	do {
145	$page = $this->queryBuilder->queryRegexRedirectsPage(array(
146	'after_id' => $afterId,
147	'limit' => self::REGEX_READ_BATCH_SIZE,
148	));
149	if (empty($page)) {
150	break;
151	}
152
153	$nextAfterId = $this->greatestRedirectId($page);
154	if ($nextAfterId <= $afterId) {
155	throw new UnexpectedValueException(
156	'Regex redirect keyset page did not advance past id ' . $afterId . '.'
157	);
158	}
159
160	foreach ($page as $row) {
161	yield $row;
162	}
163	$afterId = $nextAfterId;
164	} while (count($page) === self::REGEX_READ_BATCH_SIZE);
165	}
166
167	/**
168	* @param array<int, array<string, mixed>> $rows
169	*/
170	private function greatestRedirectId(array $rows): int {
171	$greatestId = 0;
172	foreach ($rows as $row) {
173	$id = isset($row['id']) && is_scalar($row['id']) ? (int)$row['id'] : 0;
174	$greatestId = max($greatestId, $id);
175	}
176	return $greatestId;
177	}
178
179	/** @return array<int, array<string, mixed>> */
180	// DESIGN-AUDIT-OK(2026-06-19, owner): status=MANUAL AND disabled=0 seeks via
181	// idx_status_disabled to the few hand-made MANUAL rows first, so INSTR() runs only
182	// over that bounded subset, not a full table scan. A LIMIT would drop valid manual
183	// regex redirects (changes results), so no cap. Reviewed + accepted 2026-06-18.
184	public function getManualRedirectsWithRegexMetachars(): array {
185	$query = "select \n {wp_abj404_redirects}.id,\n {wp_abj404_redirects}.url,\n {wp_abj404_redirects}.status,\n"
186	. " {wp_abj404_redirects}.type,\n {wp_abj404_redirects}.final_dest,\n {wp_abj404_redirects}.code,\n"
187	. " {wp_abj404_redirects}.timestamp,\n {wp_posts}.id as wp_post_id\n ";
188	$query .= "from {wp_abj404_redirects}\n " .
189	" LEFT OUTER JOIN {wp_posts} \n " .
190	" on {wp_abj404_redirects}.final_dest = {wp_posts}.id \n ";
191
192	$query .= "where status = " . ABJ404_STATUS_MANUAL . " \n " .
193	" and disabled = 0 \n " .
194	" and (INSTR(`url`, '*') > 0 " .
195	" OR INSTR(`url`, '[') > 0 " .
196	" OR INSTR(`url`, ']') > 0 " .
197	" OR INSTR(`url`, '\|') > 0 " .
198	" OR INSTR(`url`, '^') > 0 " .
199	" OR INSTR(`url`, '\\\\') > 0 " .
200	" OR INSTR(`url`, '{') > 0 " .
201	" OR INSTR(`url`, '}') > 0)";
202	$results = $this->dbCore->queryAndGetResults($query);
203
204	/** @var array<int, array<string, mixed>> $rows */
205	$rows = is_array($results['rows']) ? $results['rows'] : array();
206	return $rows;
207	}
208
209	/**
210	* @param array<int, string> $postIDs
211	* @return array<int, mixed>
212	*/
213	public function getExtraDataToPermalinkSuggestions(array $postIDs): array {
214	$postIDs = array_map('absint', $postIDs);
215	$postIDJoined = implode(', ', $postIDs);
216
217	$query = ABJ_404_Solution_FileSystemService::readFileContents(__DIR__ . "/../sql/getAdditionalPostData.sql");
218	$query = $this->f->str_replace('{IDS_TO_INCLUDE}', $postIDJoined, $query);
219	$query = $this->dbCore->doTableNameReplacements($query);
220	$query = $this->f->doNormalReplacements($query);
221
222	$results = $this->dbCore->queryAndGetResults($query);
223
224	/** @var array<int, mixed> $rows */
225	$rows = is_array($results['rows']) ? $results['rows'] : array();
226	return $rows;
227	}
228	}
229