404-solution / includes / PublishedPostsProvider.php

PublishedPostsProvider.php in 404 Solution 4.1.13, at includes/PublishedPostsProvider.php

144 lines 4.1 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 $instance = null;
30
31	/** @return self */
32	public static function getInstance() {
33	if (self::$instance == null) {
34	self::$instance = new ABJ_404_Solution_PublishedPostsProvider();
35	}
36
37	return self::$instance;
38	}
39
40	/** Use this data instead of querying the database.
41	* @param array<int, mixed> $data
42	* @return void
43	*/
44	function useThisData(array $data): void {
45	$this->dataToUse = $data;
46	$this->useDataMode = true;
47	}
48
49	/**
50	* Restrict results to only posts with the given IDs (whitelist mode).
51	* This enables N-gram prefiltering to limit the dataset before expensive processing.
52	*
53	* @param array<int, int\|string> $ids Array of post IDs to include
54	* @return void
55	*/
56	function restrictToIds(array $ids): void {
57	if (!empty($ids) && is_array($ids)) {
58	$this->restrictedIds = array_map('intval', $ids);
59	}
60	}
61
62	/**
63	* Check if we're in restricted ID mode.
64	* @return bool
65	*/
66	function hasRestrictedIds() {
67	return !empty($this->restrictedIds);
68	}
69
70	/**
71	* @param int $permalinkLength Order by prioritizing things with a permalink
72	* close to this length.
73	* @param int $batchSize The number of results to return. e.g. 100.
74	* @param int\|null $maxAcceptableDistance
75	* @return array<int, mixed>
76	*/
77	function getNextBatch(int $permalinkLength, int $batchSize = 1000, $maxAcceptableDistance = null): array {
78	if ($this->useDataMode) {
79	return $this->getNextBatchFromLocalData($permalinkLength, $batchSize, $maxAcceptableDistance);
80	}
81
82	return $this->getNextBatchFromTheDatabase($permalinkLength, $batchSize, $maxAcceptableDistance);
83	}
84
85	/**
86	* @param int $permalinkLength
87	* @param int $batchSize
88	* @param int\|null $maxAcceptableDistance
89	* @return array<int, mixed>
90	*/
91	private function getNextBatchFromLocalData(int $permalinkLength, int $batchSize, $maxAcceptableDistance): array {
92	$data = $this->dataToUse ?? array();
93	// get the rows to return.
94	$rows = array_slice($data, 0, $batchSize);
95
96	// remove the rows we'll return.
97	$this->dataToUse = array_slice($data, $batchSize);
98
99	return $rows;
100	}
101
102	/**
103	* @param int $permalinkLength
104	* @param int $batchSize
105	* @param int\|null $maxAcceptableDistance
106	* @return array<int, mixed>
107	*/
108	private function getNextBatchFromTheDatabase(int $permalinkLength, int $batchSize, $maxAcceptableDistance): array {
109	$abj404dao = abj_service('data_access');
110
111	$orderBy = "abs(plc.url_length - " . $permalinkLength . "), wp_posts.id";
112	$limit = $this->currentLowRowNumber . ", " . $batchSize;
113	$extraWhereClause = '';
114
115	if ($maxAcceptableDistance != null) {
116	$extraWhereClause = "and abs(plc.url_length - " . $permalinkLength .
117	") <= " . $maxAcceptableDistance;
118	}
119
120	// N-gram prefiltering: restrict to specific IDs if whitelist is set
121	if (!empty($this->restrictedIds)) {
122	$idList = implode(',', $this->restrictedIds);
123	$extraWhereClause .= " and wp_posts.id IN (" . $idList . ")";
124	}
125
126	$rows = $abj404dao->getPublishedPagesAndPostsIDs('', '', $limit, $orderBy, $extraWhereClause);
127
128	$this->currentLowRowNumber += $batchSize;
129
130	return $rows;
131	}
132
133	/** Start over at 0 when getting the next batch.
134	* @return void
135	*/
136	function resetBatch(): void {
137	$this->currentLowRowNumber = 0;
138	$this->dataToUse = null;
139	$this->useDataMode = false;
140	$this->restrictedIds = null;
141	}
142
143	}
144