404-solution / includes / matching / WordPressGuessFallback.php

WordPressGuessFallback.php in 404 Solution trunk, at includes/matching/WordPressGuessFallback.php

225 lines 9.0 KB

Raw Download Zip

1	<?php
2
3	if (!defined('ABSPATH')) {
4	exit;
5	}
6
7	/**
8	* Last-resort redirect attempt using WordPress's built-in 404 permalink guess.
9	*
10	* WordPress matches partial slugs via LIKE 'slug%', a complementary strategy
11	* to our Levenshtein-based spell checker (e.g. /redes matches /redes-social
12	* because the slug starts with "redes"). When a guess matches and is not
13	* excluded, this records the redirect and emits it, then exits. If no guess
14	* matches, the guess is a self-redirect, the destination is excluded, or the
15	* emit is blocked, it records a trace step and returns so the caller can
16	* fall through to the normal-post query / 404 page.
17	*
18	* Gated by the `abj404_wp_guess_fallback_enabled` filter and the engine
19	* profile resolver (per-URL engine enablement).
20	*/
21	class ABJ_404_Solution_WordPressGuessFallback {
22
23	/** @var ABJ_404_Solution_PluginLogicUrlNormalization */
24	private $urlNormalization;
25
26	/** @var ABJ_404_Solution_RedirectsRepository */
27	private $redirectsRepository;
28
29	/** @var ABJ_404_Solution_NotFoundResponseService */
30	private $notFoundResponse;
31
32	/** @var ABJ_404_Solution_RedirectExclusionPolicy */
33	private $exclusionPolicy;
34
35	/** @var mixed */
36	private $logsRepository;
37
38	/**
39	* @param ABJ_404_Solution_PluginLogicUrlNormalization $urlNormalization
40	* @param ABJ_404_Solution_RedirectsRepository $redirectsRepository
41	* @param ABJ_404_Solution_NotFoundResponseService $notFoundResponse
42	* @param ABJ_404_Solution_RedirectExclusionPolicy $exclusionPolicy
43	* @param mixed $logsRepository Object with logRedirectHit(); shape is duck-typed.
44	*/
45	function __construct($urlNormalization, $redirectsRepository, $notFoundResponse, $exclusionPolicy, $logsRepository) {
46	$this->urlNormalization = $urlNormalization;
47	$this->redirectsRepository = $redirectsRepository;
48	$this->notFoundResponse = $notFoundResponse;
49	$this->exclusionPolicy = $exclusionPolicy;
50	$this->logsRepository = $logsRepository;
51	}
52
53	/**
54	* @param bool $autoRedirectsAreOn Whether auto-redirect creation is enabled.
55	* @param string $requestedURL The normalized requested URL that 404'd.
56	* @param array<string, mixed> $options
57	* @param ABJ_404_Solution_FrontendPipelineTrace $trace
58	*/
59	function tryFallback(bool $autoRedirectsAreOn, string $requestedURL, array $options, ABJ_404_Solution_FrontendPipelineTrace $trace): void {
60	$wpGuessFallbackEnabled = $autoRedirectsAreOn && $this->shouldRunGuess($requestedURL);
61	if (!$wpGuessFallbackEnabled) {
62	$reason = !$autoRedirectsAreOn ? 'auto_redirects off' : 'engine profile/filter';
63	$trace->add('WordPress URL guess', 'Skipped: ' . $reason);
64	return;
65	}
66
67	// The original branch ran only when redirect_guess_404_permalink() existed and
68	// did nothing (no trace) when it was absent; preserve that silent no-op.
69	if (!function_exists('redirect_guess_404_permalink')) {
70	return;
71	}
72
73	$wpGuess = redirect_guess_404_permalink();
74	if ($wpGuess && is_string($wpGuess)) {
75	// attemptGuessRedirect() either exit()s on a successful redirect or returns
76	// so the shared 'No match' trace below is recorded for every fall-through.
77	$this->attemptGuessRedirect($wpGuess, $requestedURL, $options, $trace);
78	}
79	$trace->add('WordPress URL guess', 'No match');
80	}
81
82	/**
83	* Given a concrete WordPress permalink guess, decide whether to redirect to it and,
84	* if so, record the redirect and emit it (which exits). Each branch (self-redirect,
85	* excluded destination, blocked redirect) records a trace step and returns so the
86	* caller can fall through to the normal-post query / 404 page.
87	*
88	* @param string $wpGuess The candidate URL returned by redirect_guess_404_permalink().
89	* @param string $requestedURL The normalized requested URL that 404'd.
90	* @param array<string, mixed> $options
91	* @param ABJ_404_Solution_FrontendPipelineTrace $trace
92	*/
93	private function attemptGuessRedirect(string $wpGuess, string $requestedURL, array $options, ABJ_404_Solution_FrontendPipelineTrace $trace): void {
94	$wpGuessEngineName = __('wp guess', '404-solution');
95
96	$normalizedGuess = $this->normalizeGuessedUrlToRequestShape($wpGuess);
97	if ($normalizedGuess !== '' && $normalizedGuess === $requestedURL) {
98	$trace->add('WordPress URL guess', 'Ignored self-redirect guess', $wpGuess);
99	return;
100	}
101
102	$trace->add('WordPress URL guess', 'Matched candidate', $wpGuess);
103	$defaultRedirect = isset($options['default_redirect']) && is_scalar($options['default_redirect'])
104	? (string)$options['default_redirect'] : '301';
105	$wpGuessType = (string)$this->typePost();
106	$wpGuessPostId = $this->resolveGuessPostId($wpGuess);
107
108	$wpGuessResult = new ABJ_404_Solution_MatchResult(
109	$wpGuessPostId !== '' ? $wpGuessPostId : '0',
110	$wpGuessType,
111	$wpGuess,
112	'',
113	0.0,
114	$wpGuessEngineName
115	);
116	if ($this->exclusionPolicy->isExcluded($wpGuessResult, $options)) {
117	$trace->add('WordPress URL guess', 'Excluded destination: skipped', $wpGuess);
118	return;
119	}
120
121	$trace->add('WordPress URL guess', 'Matched: redirecting', $wpGuess);
122	$this->redirectsRepository->setupRedirect(ABJ_404_Solution_RedirectSpec::create(
123	$requestedURL, (string)ABJ404_STATUS_AUTO,
124	$wpGuessType, $wpGuessPostId, $defaultRedirect, 0, $wpGuessEngineName
125	));
126	$this->writeHit($requestedURL, $wpGuess, $wpGuessEngineName, null, $trace->getSteps());
127	$redirectSent = $this->notFoundResponse->forceRedirect(esc_url($wpGuess), (int)$defaultRedirect);
128	if ($redirectSent !== false) {
129	exit;
130	}
131	$trace->add('WordPress URL guess', 'Redirect blocked: continued', $wpGuess);
132	}
133
134	/**
135	* Resolve a guessed URL to its WordPress post id as a string, or '' when the URL
136	* maps to no post (or url_to_postid() is unavailable).
137	*
138	* @param string $wpGuess
139	* @return string
140	*/
141	private function resolveGuessPostId(string $wpGuess): string {
142	if (!function_exists('url_to_postid')) {
143	return '';
144	}
145	$postId = url_to_postid($wpGuess);
146	return $postId > 0 ? (string)$postId : '';
147	}
148
149	/**
150	* @param string $requestedURL
151	* @return bool
152	*/
153	private function shouldRunGuess(string $requestedURL): bool {
154	$enabled = true;
155	if (function_exists('apply_filters')) {
156	$enabled = (bool) apply_filters(
157	'abj404_wp_guess_fallback_enabled',
158	true,
159	$requestedURL
160	);
161	}
162	if (!$enabled) {
163	return false;
164	}
165
166	return ABJ_404_Solution_EngineProfileResolver::getInstance()
167	->isEngineEnabledForUrl($requestedURL, 'ABJ_404_Solution_WordPressUrlGuessEngine');
168	}
169
170	/**
171	* Normalize a guessed URL into the same request-shape used by $requestedURL:
172	* path (relative to WP home directory) plus sorted query string.
173	*
174	* @param string $guessedUrl
175	* @return string
176	*/
177	private function normalizeGuessedUrlToRequestShape(string $guessedUrl): string {
178	$normalized = abj_service('sanitizer')->normalizeUrlString($guessedUrl);
179	if ($normalized === '') {
180	return '';
181	}
182
183	$parts = parse_url($normalized);
184	if (!is_array($parts)) {
185	return '';
186	}
187
188	$path = isset($parts['path']) ? $parts['path'] : '/';
189	if ($path === '') {
190	$path = '/';
191	}
192	$path = $this->urlNormalization->removeHomeDirectory($path);
193	if ($path === '') {
194	$path = '/';
195	}
196	if ($path[0] !== '/') {
197	$path = '/' . $path;
198	}
199
200	/** @var array<string, string> $urlPartsStr */
201	$urlPartsStr = array_map('strval', $parts);
202	$sortedQuery = abj_service('query_string_helper')->sortQueryString($urlPartsStr);
203	return $path . $sortedQuery;
204	}
205
206	/** @return int */
207	private function typePost(): int {
208	return (int)ABJ404_TYPE_POST;
209	}
210
211	/**
212	* @param string $requestedUrl
213	* @param string $action
214	* @param string $matchReason
215	* @param string\|null $requestedUrlDetail
216	* @param list<array{step: string, outcome: string, detail: string}>\|null $pipelineTrace
217	*/
218	private function writeHit(string $requestedUrl, string $action, string $matchReason, ?string $requestedUrlDetail = null, ?array $pipelineTrace = null): void {
219	if (!is_object($this->logsRepository) \|\| !is_callable(array($this->logsRepository, 'logRedirectHit'))) {
220	return;
221	}
222	call_user_func(array($this->logsRepository, 'logRedirectHit'), ABJ_404_Solution_RedirectHitLogEntry::create($requestedUrl, $action, $matchReason, $requestedUrlDetail, $pipelineTrace));
223	}
224	}
225