404-solution / includes / frontend / PublishedPostsProvider.php

PublishedPostsProvider.php in 404 Solution trunk, at includes/frontend/PublishedPostsProvider.php

173 lines 5.0 KB

Raw Download Zip

1	<?php
2
3
4	if (!defined('ABSPATH')) {
5	exit;
6	}
7
8	/** Return some of the published pages when requested. */
9	class ABJ_404_Solution_PublishedPostsProvider {
10
11	/** Track which rows to get from the database using limit.
12	* @var integer */
13	private $currentLowRowNumber = 0;
14
15	/** When not null then use this data instead of querying the database.
16	* @var array<int, mixed>\|null
17	*/
18	private $dataToUse = null;
19
20	/** Tracks whether we're using user supplied data or not.
21	* @var bool */
22	private $useDataMode = false;
23
24	/** When set, only return posts with these IDs (whitelist mode).
25	* @var array<int, int>\|null */
26	private $restrictedIds = null;
27
28	/** @var self\|null */
29	private static $instance = null;
30	/**
31	* Test seam: install or clear the cached singleton instance without
32	* private-field reflection. Pass null to reset between tests; pass a
33	* configured instance (or double) to install it. Mirrors the setInstance()
34	* contract on DataAccess / PluginLogic (M105 singleton-reset seam).
35	*
36	* @param self\|null $instance
37	* @return void
38	*/
39	public static function setInstance($instance) {
40	self::$instance = $instance;
41	}
42
43
44	/** @var ABJ_404_Solution_ContentRepository */
45	private $contentRepository;
46
47	/**
48	* @param ABJ_404_Solution_ContentRepository\|null $contentRepository Content repository
49	*/
50	public function __construct($contentRepository = null) {
51	$this->contentRepository = $contentRepository !== null ? $contentRepository : abj_service('content_repository');
52	}
53
54	public static function resetForTests(?self $replacement = null): void {
55	self::$instance = $replacement;
56	}
57
58	/** @return self */
59	public static function getInstance() {
60	if (self::$instance == null) {
61	self::$instance = new ABJ_404_Solution_PublishedPostsProvider();
62	}
63
64	return self::$instance;
65	}
66
67	/** Use this data instead of querying the database.
68	* @param array<int, mixed> $data
69	* @return void
70	*/
71	function useThisData(array $data): void {
72	$this->dataToUse = $data;
73	$this->useDataMode = true;
74	}
75
76	/**
77	* Restrict results to only posts with the given IDs (whitelist mode).
78	* This enables N-gram prefiltering to limit the dataset before expensive processing.
79	*
80	* @param array<int, int\|string> $ids Array of post IDs to include
81	* @return void
82	*/
83	function restrictToIds(array $ids): void {
84	if (!empty($ids) && is_array($ids)) {
85	$this->restrictedIds = array_map('intval', $ids);
86	}
87	}
88
89	/**
90	* Check if we're in restricted ID mode.
91	* @return bool
92	*/
93	function hasRestrictedIds() {
94	return !empty($this->restrictedIds);
95	}
96
97	/**
98	* @param int $permalinkLength Order by prioritizing things with a permalink
99	* close to this length.
100	* @param int $batchSize The number of results to return. e.g. 100.
101	* @param int\|null $maxAcceptableDistance
102	* @return array<int, mixed>
103	*/
104	function getNextBatch(int $permalinkLength, int $batchSize = 1000, $maxAcceptableDistance = null): array {
105	if ($this->useDataMode) {
106	return $this->getNextBatchFromLocalData($permalinkLength, $batchSize, $maxAcceptableDistance);
107	}
108
109	return $this->getNextBatchFromTheDatabase($permalinkLength, $batchSize, $maxAcceptableDistance);
110	}
111
112	/**
113	* @param int $permalinkLength
114	* @param int $batchSize
115	* @param int\|null $maxAcceptableDistance
116	* @return array<int, mixed>
117	*/
118	private function getNextBatchFromLocalData(int $permalinkLength, int $batchSize, $maxAcceptableDistance): array {
119	$data = $this->dataToUse ?? array();
120	// get the rows to return.
121	$rows = array_slice($data, 0, $batchSize);
122
123	// remove the rows we'll return.
124	$this->dataToUse = array_slice($data, $batchSize);
125
126	return $rows;
127	}
128
129	/**
130	* @param int $permalinkLength
131	* @param int $batchSize
132	* @param int\|null $maxAcceptableDistance
133	* @return array<int, mixed>
134	*/
135	private function getNextBatchFromTheDatabase(int $permalinkLength, int $batchSize, $maxAcceptableDistance): array {
136	$orderBy = "abs(plc.url_length - " . $permalinkLength . "), wp_posts.id";
137	$limit = $this->currentLowRowNumber . ", " . $batchSize;
138	$extraWhereClause = '';
139
140	if ($maxAcceptableDistance != null) {
141	$extraWhereClause = "and abs(plc.url_length - " . $permalinkLength .
142	") <= " . $maxAcceptableDistance;
143	}
144
145	// N-gram prefiltering: restrict to specific IDs if whitelist is set
146	if (!empty($this->restrictedIds)) {
147	$idList = implode(',', $this->restrictedIds);
148	$extraWhereClause .= " and wp_posts.id IN (" . $idList . ")";
149	}
150
151	$rows = $this->contentRepository->getPublishedPagesAndPostsIDs(array(
152	'limit_results' => $limit,
153	'order_results' => $orderBy,
154	'extra_where_clause' => $extraWhereClause,
155	));
156
157	$this->currentLowRowNumber += $batchSize;
158
159	return $rows;
160	}
161
162	/** Start over at 0 when getting the next batch.
163	* @return void
164	*/
165	function resetBatch(): void {
166	$this->currentLowRowNumber = 0;
167	$this->dataToUse = null;
168	$this->useDataMode = false;
169	$this->restrictedIds = null;
170	}
171
172	}
173