activitypub / includes / functions-url.php

functions-url.php in ActivityPub trunk, at includes/functions-url.php

210 lines 5.6 KB

Raw Download Zip

1	<?php
2	/**
3	* URL functions.
4	*
5	* Functions for URL manipulation.
6	*
7	* @package Activitypub
8	*/
9
10	namespace Activitypub;
11
12	/**
13	* Get the REST URL relative to this plugin's namespace.
14	*
15	* @param string $path Optional. REST route path. Default ''.
16	*
17	* @return string REST URL relative to this plugin's namespace.
18	*/
19	function get_rest_url_by_path( $path = '' ) {
20	// We'll handle the leading slash.
21	$path = \ltrim( $path, '/' );
22	$namespaced_path = \sprintf( '/%s/%s', ACTIVITYPUB_REST_NAMESPACE, $path );
23	return \get_rest_url( null, $namespaced_path );
24	}
25
26	/**
27	* Normalize a URL.
28	*
29	* @param string $url The URL.
30	*
31	* @return string The normalized URL.
32	*/
33	function normalize_url( $url ) {
34	// Remove ActivityPub-specific query parameters.
35	$url = \remove_query_arg( array( 'activitypub', 'preview' ), $url );
36	$url = \untrailingslashit( $url );
37	$url = \preg_replace( '/^https?:\/\/(www\.)?/', '', $url );
38
39	return $url;
40	}
41
42	/**
43	* Normalize a host.
44	*
45	* @param string $host The host.
46	*
47	* @return string The normalized host.
48	*/
49	function normalize_host( $host ) {
50	return \preg_replace( '/^www\./', '', $host );
51	}
52
53	/**
54	* Fold a host to its canonical spelling for comparison.
55	*
56	* Hosts are case-insensitive, IPv6 literals are bracketed in a URI but not in a stored host,
57	* and a trailing dot is the same fully qualified name. Two spellings that differ only in those
58	* ways address the same host, so anything comparing hosts has to fold them first.
59	*
60	* Unlike {@see normalize_host()} this is lossless: it does not strip `www.`, which is a
61	* different host rather than a different spelling of one.
62	*
63	* @since 9.3.0
64	*
65	* @param string $host The host.
66	*
67	* @return string The folded host.
68	*/
69	function fold_host( $host ) {
70	return \rtrim( \trim( \strtolower( (string) $host ), '[]' ), '.' );
71	}
72
73	/**
74	* Fold an authority to the host it names.
75	*
76	* An authority is what a `Host` header carries: a host, plus a port whenever it is not the
77	* default one, with an IPv6 literal in brackets. {@see fold_host()} expects a bare host and
78	* strips brackets without understanding a port, so `[::1]:443` would fold to `::1]:443` and
79	* `example.org:8080` would never match the `example.org` that {@see home_host()} returns.
80	* Use this wherever an authority has to be compared against a host.
81	*
82	* @since 9.3.1
83	*
84	* @param string $authority A host, optionally bracketed and optionally with a port.
85	*
86	* @return string The folded host, or an empty string when the value names none.
87	*/
88	function fold_authority( $authority ) {
89	// Parsed as a scheme-relative URL, which is the shape an authority already has.
90	$host = \wp_parse_url( '//' . \ltrim( (string) $authority, '/' ), \PHP_URL_HOST );
91
92	return fold_host( (string) $host );
93	}
94
95	/**
96	* Check if a URL is from the same domain as the site.
97	*
98	* @param string $url The URL to check.
99	*
100	* @return boolean True if the URL is from the same domain, false otherwise.
101	*/
102	function is_same_domain( $url ) {
103	$remote = \wp_parse_url( $url, PHP_URL_HOST );
104
105	if ( ! $remote ) {
106	return false;
107	}
108
109	$remote = normalize_host( $remote );
110	$self = normalize_host( home_host() );
111
112	return $remote === $self;
113	}
114
115	/**
116	* Retrieves the Host for the current site where the front end is accessible.
117	*
118	* @return string The host for the current site.
119	*/
120	function home_host() {
121	return \wp_parse_url( \home_url(), PHP_URL_HOST );
122	}
123
124	/**
125	* Get the base URL for uploads.
126	*
127	* @return string The upload base URL.
128	*/
129	function get_upload_baseurl() {
130	/**
131	* Early filter to allow plugins to set the upload base URL.
132	*
133	* @param string\|false $maybe_upload_dir The upload base URL or false if not set.
134	*/
135	$maybe_upload_dir = \apply_filters( 'pre_activitypub_get_upload_baseurl', false );
136	if ( false !== $maybe_upload_dir ) {
137	return $maybe_upload_dir;
138	}
139
140	$upload_dir = \wp_get_upload_dir();
141
142	/**
143	* Filters the upload base URL.
144	*
145	* @param string $upload_dir The upload base URL. Default \wp_get_upload_dir()['baseurl']
146	*/
147	return \apply_filters( 'activitypub_get_upload_baseurl', $upload_dir['baseurl'] );
148	}
149
150	/**
151	* Get the authority (scheme + host) from a URL.
152	*
153	* @param string $url The URL to parse.
154	*
155	* @return string\|false The authority, or false on failure.
156	*/
157	function get_url_authority( $url ) {
158	$parsed = \wp_parse_url( $url );
159
160	if ( ! $parsed \|\| empty( $parsed['scheme'] ) \|\| empty( $parsed['host'] ) ) {
161	return false;
162	}
163
164	return $parsed['scheme'] . '://' . $parsed['host'];
165	}
166
167	/**
168	* Infer a shortname from the Actor ID or URL. Used only for fallbacks,
169	* we will try to use what's supplied.
170	*
171	* @param string $uri The URI.
172	*
173	* @return string Hopefully the name of the Follower.
174	*/
175	function extract_name_from_uri( $uri ) {
176	$name = $uri;
177
178	if ( \filter_var( $name, FILTER_VALIDATE_URL ) ) {
179	$name = \rtrim( $name, '/' );
180	$path = \wp_parse_url( $name, PHP_URL_PATH );
181	if ( $path && '/' !== $path ) {
182	if ( \strpos( $name, '@' ) !== false ) {
183	// Expected: https://example.com/@user (default URL pattern).
184	$name = \preg_replace( '\|^/@?\|', '', $path );
185	} else {
186	// Expected: https://example.com/users/user (default ID pattern).
187	$parts = \explode( '/', $path );
188	$name = \array_pop( $parts );
189	}
190	} else {
191	$name = \wp_parse_url( $name, PHP_URL_HOST );
192	$name = \str_replace( 'www.', '', $name );
193	}
194	} elseif (
195	\is_email( $name ) \|\|
196	\strpos( $name, 'acct' ) === 0 \|\|
197	\strpos( $name, '@' ) === 0
198	) {
199	// Expected: user@example.com or acct:user@example (WebFinger).
200	$name = \ltrim( $name, '@' );
201	if ( \str_starts_with( $name, 'acct:' ) ) {
202	$name = \substr( $name, 5 );
203	}
204	$parts = \explode( '@', $name );
205	$name = $parts[0];
206	}
207
208	return $name;
209	}
210