404-solution / includes / PublishedPostsProvider.php

PublishedPostsProvider.php in 404 Solution 4.2.0, at includes/PublishedPostsProvider.php

156 lines 4.5 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	/** @var ABJ_404_Solution_ContentRepository */
32	private $contentRepository;
33
34	/**
35	* @param ABJ_404_Solution_ContentRepository\|null $contentRepository Content repository
36	*/
37	public function __construct($contentRepository = null) {
38	$this->contentRepository = $contentRepository !== null ? $contentRepository : abj_service('content_repository');
39	}
40
41	public static function resetForTests(?self $replacement = null): void {
42	self::$instance = $replacement;
43	}
44
45	/** @return self */
46	public static function getInstance() {
47	if (self::$instance == null) {
48	self::$instance = new ABJ_404_Solution_PublishedPostsProvider();
49	}
50
51	return self::$instance;
52	}
53
54	/** Use this data instead of querying the database.
55	* @param array<int, mixed> $data
56	* @return void
57	*/
58	function useThisData(array $data): void {
59	$this->dataToUse = $data;
60	$this->useDataMode = true;
61	}
62
63	/**
64	* Restrict results to only posts with the given IDs (whitelist mode).
65	* This enables N-gram prefiltering to limit the dataset before expensive processing.
66	*
67	* @param array<int, int\|string> $ids Array of post IDs to include
68	* @return void
69	*/
70	function restrictToIds(array $ids): void {
71	if (!empty($ids) && is_array($ids)) {
72	$this->restrictedIds = array_map('intval', $ids);
73	}
74	}
75
76	/**
77	* Check if we're in restricted ID mode.
78	* @return bool
79	*/
80	function hasRestrictedIds() {
81	return !empty($this->restrictedIds);
82	}
83
84	/**
85	* @param int $permalinkLength Order by prioritizing things with a permalink
86	* close to this length.
87	* @param int $batchSize The number of results to return. e.g. 100.
88	* @param int\|null $maxAcceptableDistance
89	* @return array<int, mixed>
90	*/
91	function getNextBatch(int $permalinkLength, int $batchSize = 1000, $maxAcceptableDistance = null): array {
92	if ($this->useDataMode) {
93	return $this->getNextBatchFromLocalData($permalinkLength, $batchSize, $maxAcceptableDistance);
94	}
95
96	return $this->getNextBatchFromTheDatabase($permalinkLength, $batchSize, $maxAcceptableDistance);
97	}
98
99	/**
100	* @param int $permalinkLength
101	* @param int $batchSize
102	* @param int\|null $maxAcceptableDistance
103	* @return array<int, mixed>
104	*/
105	private function getNextBatchFromLocalData(int $permalinkLength, int $batchSize, $maxAcceptableDistance): array {
106	$data = $this->dataToUse ?? array();
107	// get the rows to return.
108	$rows = array_slice($data, 0, $batchSize);
109
110	// remove the rows we'll return.
111	$this->dataToUse = array_slice($data, $batchSize);
112
113	return $rows;
114	}
115
116	/**
117	* @param int $permalinkLength
118	* @param int $batchSize
119	* @param int\|null $maxAcceptableDistance
120	* @return array<int, mixed>
121	*/
122	private function getNextBatchFromTheDatabase(int $permalinkLength, int $batchSize, $maxAcceptableDistance): array {
123	$orderBy = "abs(plc.url_length - " . $permalinkLength . "), wp_posts.id";
124	$limit = $this->currentLowRowNumber . ", " . $batchSize;
125	$extraWhereClause = '';
126
127	if ($maxAcceptableDistance != null) {
128	$extraWhereClause = "and abs(plc.url_length - " . $permalinkLength .
129	") <= " . $maxAcceptableDistance;
130	}
131
132	// N-gram prefiltering: restrict to specific IDs if whitelist is set
133	if (!empty($this->restrictedIds)) {
134	$idList = implode(',', $this->restrictedIds);
135	$extraWhereClause .= " and wp_posts.id IN (" . $idList . ")";
136	}
137
138	$rows = $this->contentRepository->getPublishedPagesAndPostsIDs('', '', $limit, $orderBy, $extraWhereClause);
139
140	$this->currentLowRowNumber += $batchSize;
141
142	return $rows;
143	}
144
145	/** Start over at 0 when getting the next batch.
146	* @return void
147	*/
148	function resetBatch(): void {
149	$this->currentLowRowNumber = 0;
150	$this->dataToUse = null;
151	$this->useDataMode = false;
152	$this->restrictedIds = null;
153	}
154
155	}
156