404-solution / includes / redirects / RegexAutoPromote.php

RegexAutoPromote.php in 404 Solution trunk, at includes/redirects/RegexAutoPromote.php

191 lines 8.5 KB

Raw Download Zip

1	<?php
2
3	if (!defined('ABSPATH')) {
4	exit;
5	}
6
7	/**
8	* Server-side regex auto-promotion helpers.
9	*
10	* The JS in includes/ajax/redirect_to_ajax.js already auto-ticks the
11	* "Treat as regex" checkbox when an admin types a URL that looks like a
12	* regex pattern. This class is the server-side counterpart: it lets the
13	* admin-save, import, and runtime-matching paths reach the same conclusion
14	* when the JS detection didn't run (paste-and-submit with JS disabled, CSV
15	* import, existing MANUAL row written by an older version, etc.).
16	*
17	* The character set used here (`* [ ] \| ^ \ { }`) is intentionally narrower
18	* than the JS detector: every character is reserved/unsafe per RFC 3986, so
19	* browsers URL-encode them. Their RAW presence in a stored from_url is
20	* near-certain regex intent. False positives would require a user to
21	* paste an already URL-encoded query string verbatim, which is not a real
22	* workflow this plugin needs to support.
23	*
24	* Explicitly NOT included: `. ? + ( ) & = #`. These appear in real URLs
25	* constantly (file extensions, query strings, fragments) and a sniff that
26	* promoted them would catch many real MANUAL redirects by mistake.
27	*/
28	class ABJ_404_Solution_RegexAutoPromote {
29
30	/**
31	* Decide whether a raw from_url contains regex metachars unambiguous
32	* enough to auto-promote status=MANUAL to status=REGEX.
33	*
34	* @param string $url The raw from_url as stored / as posted.
35	* @return bool True when the URL contains at least one unambiguous regex
36	* metacharacter from the safe set described in the class
37	* docblock.
38	*/
39	public static function looksLikeUnambiguousRegex($url) {
40	if (!is_string($url) \|\| $url === '') {
41	return false;
42	}
43
44	// Each char in the bracket below is RFC 3986-reserved or non-URL-safe.
45	// A browser submitting these via the address bar would URL-encode
46	// them (`%5B`, `%7C`, etc.), so a raw presence is strong regex
47	// intent. Backslash is included because a plain URL never contains
48	// one. `\d`, `\w`, `\s` are PCRE shorthand classes only.
49	return preg_match('/[\*\[\]\\|\^\\\\\{\}]/', $url) === 1;
50	}
51
52	/**
53	* Apply minimal glob-to-regex normalization on the auto-promote path
54	* only. This handles the "user typed `/sales/` and meant `/sales/.`"
55	* case: shells use bare `*` for "zero or more of anything", but PCRE's
56	* `*` is a quantifier that needs something in front of it. Without this
57	* fix, `/sales/*` compiles as the literal string `/sales/` followed by
58	* the regex error "nothing to repeat". Every 404 hit logs a PHP warning
59	* and the redirect never fires.
60	*
61	* Rules (deliberately narrow). The user can always uncheck the
62	* "Treat as regex" box if we guess wrong, and the admin notice points
63	* out the rewrite so it is visible:
64	*
65	* 1. Pattern contains `` AND nowhere contains `.`: replace every
66	* bare `` with `.`. Covers `/sales/` and `-old.html` both.
67	* 2. Pattern does not contain `*`: no change.
68	* 3. Pattern already contains `.` somewhere: leave the bare ``s
69	* alone. The user (or upstream) clearly knows the difference.
70	*
71	* Explicitly NOT applied:
72	* - Auto-escaping `.` (would break legitimate `.*` users).
73	* - Auto-anchoring with `^` / `$` (some users want substring match).
74	* - Quoting other metachars. We trust the pattern shape from here on
75	* and the import-time validator (validateRegexPattern) catches the
76	* leftover compile failures.
77	*
78	* This MUST be called only on the auto-promote path. When the user
79	* explicitly checks "Treat as regex" on a weird pattern like `(foo)*`,
80	* they meant exactly what they wrote and we don't second-guess them.
81	*
82	* @param string $url Raw from_url to consider for normalization.
83	* @return array{url: string, changed: bool}
84	*/
85	public static function applyGlobFixup($url) {
86	if (!is_string($url)) {
87	return array('url' => '', 'changed' => false);
88	}
89	if ($url === '') {
90	return array('url' => '', 'changed' => false);
91	}
92	if (strpos($url, '*') === false) {
93	return array('url' => $url, 'changed' => false);
94	}
95	if (strpos($url, '.*') !== false) {
96	return array('url' => $url, 'changed' => false);
97	}
98	// Replace each bare `` with `.`. preg_replace with a negative
99	// lookbehind for `.` would also work, but a straight str_replace
100	// suffices since we already proved no `.*` exists anywhere.
101	$rewritten = str_replace('', '.', $url);
102	return array('url' => $rewritten, 'changed' => $rewritten !== $url);
103	}
104
105	/**
106	* Single-tenant transient key holding the most recent auto-promote
107	* event. Sites with concurrent admins editing redirects will see
108	* latest-wins behavior on this notice; the underlying redirect rows
109	* are still per-row so no data is mixed up, just the dismissable
110	* UI banner. This is the canonical "small UX feature, big
111	* complexity if multi-keyed" tradeoff: a per-user key would force
112	* every test that renders the admin header to stub
113	* get_current_user_id even after Brain Monkey teardown, because
114	* Patchwork keeps the function registered globally.
115	*/
116	const NOTICE_TRANSIENT_KEY = 'abj404_regex_autopromote_notice';
117
118	/**
119	* Persist an auto-promote event. The payload is small (~5 fields)
120	* and one-shot per save; the next admin page render reads it back
121	* and shows a notice with [Edit] and [Undo] links.
122	*
123	* Kept on the helper class (not PluginLogic) so the view layer can
124	* read the notice without going through a typed PluginLogic method
125	* call. Mockery's strict-call behavior would otherwise force every
126	* existing PluginLogic mock to pre-declare an expectation on the
127	* read method.
128	*
129	* @param int $redirectId The redirect row id that was just saved.
130	* @param string $originalURL The from_url posted by the admin (pre-rewrite).
131	* @param string $newURL The from_url that ended up stored.
132	* @param bool $urlRewritten True when the glob fixup mutated the URL.
133	* @return void
134	*/
135	public static function saveNotice($redirectId, $originalURL, $newURL, $urlRewritten) {
136	if ($redirectId <= 0 \|\| !function_exists('set_transient')) {
137	return;
138	}
139	$payload = array(
140	'redirect_id' => (int)$redirectId,
141	'original_url' => (string)$originalURL,
142	'new_url' => (string)$newURL,
143	'url_rewritten' => (bool)$urlRewritten,
144	'created_at' => abj_clock()->now(),
145	);
146	$ttl = defined('HOUR_IN_SECONDS') ? HOUR_IN_SECONDS : 3600;
147	// allow-cache-empty: $payload built locally from method args, never a failure-derived empty/null.
148	set_transient(self::NOTICE_TRANSIENT_KEY, $payload, $ttl);
149	}
150
151	/**
152	* Read the pending notice payload, or null when no notice is
153	* pending. Leaves the transient in place so the Undo handler can
154	* still find the original from_url after the notice has been
155	* rendered. The transient self-expires via its TTL.
156	*
157	* @return array{redirect_id: int, original_url: string, new_url: string, url_rewritten: bool, created_at: int}\|null
158	*/
159	public static function readNotice() {
160	if (!function_exists('get_transient')) {
161	return null;
162	}
163	$data = get_transient(self::NOTICE_TRANSIENT_KEY);
164	if (!is_array($data) \|\| !isset($data['redirect_id'])) {
165	return null;
166	}
167	$redirectId = isset($data['redirect_id']) && is_numeric($data['redirect_id']) ? (int)$data['redirect_id'] : 0;
168	$createdAt = isset($data['created_at']) && is_numeric($data['created_at']) ? (int)$data['created_at'] : 0;
169	return array(
170	'redirect_id' => $redirectId,
171	'original_url' => isset($data['original_url']) && is_string($data['original_url']) ? $data['original_url'] : '',
172	'new_url' => isset($data['new_url']) && is_string($data['new_url']) ? $data['new_url'] : '',
173	'url_rewritten' => !empty($data['url_rewritten']),
174	'created_at' => $createdAt,
175	);
176	}
177
178	/**
179	* Delete the pending notice (called after a successful Undo or an
180	* explicit dismissal).
181	*
182	* @return void
183	*/
184	public static function clearNotice() {
185	if (!function_exists('delete_transient')) {
186	return;
187	}
188	delete_transient(self::NOTICE_TRANSIENT_KEY);
189	}
190	}
191