404-solution / includes / engine / UrlFixEngine.php

UrlFixEngine.php in 404 Solution 4.3.0, at includes/engine/UrlFixEngine.php

156 lines 5.6 KB

Raw Download Zip

1	<?php
2
3	if (!defined('ABSPATH')) {
4	exit;
5	}
6
7	/**
8	* Matching engine that applies common URL fixes and checks if the corrected URL
9	* resolves to a real page.
10	*
11	* Handles structural URL issues that add too much edit distance for the Spelling
12	* engine's Levenshtein matching:
13	* - File extensions: /about.html → /about
14	* - Trailing punctuation: /about. → /about (from copy-paste in emails)
15	*
16	* Runs early in the pipeline (after exact Slug, before Title/Spelling) since each
17	* fix is an O(1) slug lookup — much cheaper than keyword or Levenshtein matching.
18	*/
19	class ABJ_404_Solution_UrlFixEngine implements ABJ_404_Solution_MatchingEngine {
20
21	/**
22	* Common file extensions to strip, ordered by frequency.
23	* @var array<int, string>
24	*/
25	private static $extensionsToStrip = [
26	'.html',
27	'.htm',
28	'.php',
29	'.asp',
30	'.aspx',
31	'.shtml',
32	'.jsp',
33	'.cfm',
34	];
35
36	/**
37	* Trailing punctuation characters commonly appended by email clients
38	* and word processors when URLs appear at the end of sentences.
39	* @var string
40	*/
41	private static $trailingPunctuation = '.,;:!?)>';
42
43	/** @var ABJ_404_Solution_SpellChecker */
44	private $spellChecker;
45
46	/** @var ABJ_404_Solution_Functions */
47	private $f;
48
49	/** @var ABJ_404_Solution_Logging */
50	private $logger;
51
52	/**
53	* @param ABJ_404_Solution_SpellChecker $spellChecker
54	* @param ABJ_404_Solution_Functions $f
55	* @param ABJ_404_Solution_Logging $logger
56	*/
57	public function __construct(
58	ABJ_404_Solution_SpellChecker $spellChecker,
59	ABJ_404_Solution_Functions $f,
60	ABJ_404_Solution_Logging $logger
61	) {
62	$this->spellChecker = $spellChecker;
63	$this->f = $f;
64	$this->logger = $logger;
65	}
66
67	/** @return string */
68	public function getName(): string {
69	return __('url fix', '404-solution');
70	}
71
72	/** @param ABJ_404_Solution_MatchRequest $request */
73	public function shouldRun(ABJ_404_Solution_MatchRequest $request): bool {
74	return $request->getUrlSlugOnly() !== '';
75	}
76
77	/** @param ABJ_404_Solution_MatchRequest $request */
78	public function match(ABJ_404_Solution_MatchRequest $request): ?ABJ_404_Solution_MatchResult {
79	$slug = $request->getUrlSlugOnly();
80	$candidates = $this->generateFixedCandidates($slug);
81
82	foreach ($candidates as $fixDescription => $fixedSlug) {
83	$permalink = $this->spellChecker->getPermalinkUsingSlug($fixedSlug);
84
85	if (!empty($permalink)) {
86	$id = isset($permalink['id']) && is_scalar($permalink['id']) ? (string)$permalink['id'] : '';
87	$type = isset($permalink['type']) && is_scalar($permalink['type']) ? (string)$permalink['type'] : '';
88	$link = isset($permalink['link']) && is_string($permalink['link']) ? $permalink['link'] : '';
89	$title = isset($permalink['title']) && is_string($permalink['title']) ? $permalink['title'] : '';
90	$score = isset($permalink['score']) && is_scalar($permalink['score']) ? (float)$permalink['score'] : 0.0;
91
92	$this->logger->debugMessage("URL fix engine: matched via " . $fixDescription .
93	" ('" . $slug . "' → '" . $fixedSlug . "')");
94
95	return new ABJ_404_Solution_MatchResult($id, $type, $link, $title, $score, $this->getName());
96	}
97	}
98
99	return null;
100	}
101
102	/**
103	* Generate fixed URL slug candidates by applying common transformations.
104	*
105	* Each candidate is keyed by a human-readable description of the fix applied.
106	* Only candidates that differ from the original slug are included.
107	*
108	* @param string $slug
109	* @return array<string, string> fix description => fixed slug
110	*/
111	private function generateFixedCandidates(string $slug): array {
112	$candidates = [];
113
114	// 1. Strip file extensions
115	$slugLower = $this->f->strtolower($slug);
116	foreach (self::$extensionsToStrip as $ext) {
117	$extLen = strlen($ext);
118	if (strlen($slugLower) > $extLen && substr($slugLower, -$extLen) === $ext) {
119	$fixed = substr($slug, 0, -$extLen);
120	if ($fixed !== '' && $fixed !== $slug) {
121	$candidates['strip extension ' . $ext] = $fixed;
122	}
123	break; // Only strip one extension
124	}
125	}
126
127	// 2. Strip trailing punctuation (one character at a time, up to 3 chars)
128	$stripped = $slug;
129	$puncCount = 0;
130	while ($stripped !== '' && $puncCount < 3 && strpos(self::$trailingPunctuation, substr($stripped, -1)) !== false) {
131	$stripped = substr($stripped, 0, -1);
132	$puncCount++;
133	}
134	if ($stripped !== '' && $stripped !== $slug) {
135	$candidates['strip trailing punctuation'] = $stripped;
136	}
137
138	// 3. Combined: extension + trailing punctuation (e.g., "/about.html.")
139	if (isset($candidates['strip trailing punctuation'])) {
140	$strippedLower = $this->f->strtolower($stripped);
141	foreach (self::$extensionsToStrip as $ext) {
142	$extLen = strlen($ext);
143	if (strlen($strippedLower) > $extLen && substr($strippedLower, -$extLen) === $ext) {
144	$combined = substr($stripped, 0, -$extLen);
145	if ($combined !== '' && $combined !== $slug && !in_array($combined, $candidates, true)) {
146	$candidates['strip punctuation + extension ' . $ext] = $combined;
147	}
148	break;
149	}
150	}
151	}
152
153	return $candidates;
154	}
155	}
156