friends / feed-parsers / class-feed-parser.php

class-feed-parser.php in Friends 4.3.1, at feed-parsers/class-feed-parser.php

252 lines 7.1 KB

Raw Download Zip

1	<?php
2	/**
3	* Friends Feed Parser
4	*
5	* This contains the reference implementation for a parser.
6	*
7	* @package Friends
8	*/
9
10	namespace Friends;
11
12	/**
13	* This class describes a friends feed parser.
14	*/
15	abstract class Feed_Parser {
16	/**
17	* Get the badge to display for this parser type.
18	*
19	* Return an array with badge info:
20	* array(
21	* 'label' => 'RSS', // Short text to display
22	* 'color' => '#ee802f', // Background color
23	* 'title' => 'RSS/Atom Feed' // Tooltip text
24	* )
25	*
26	* @return array\|null Badge info array, or null for no badge.
27	*/
28	public function get_badge() {
29	return null;
30	}
31
32	/**
33	* Determines if this is a supported feed and to what degree we feel it's supported.
34	*
35	* @param string $url The url.
36	* @param string $mime_type The mime type.
37	* @param string $title The title.
38	* @param string\|null $content The content, it can't be assumed that it's always available.
39	*
40	* @return int Return 0 if unsupported, a positive value representing the confidence for the feed, use 10 if you're reasonably confident.
41	*/
42	public function feed_support_confidence( $url, $mime_type, $title, $content = null ) { // phpcs:ignore Generic.CodeAnalysis.UnusedFunctionParameter.FoundAfterLastUsed
43	return 0;
44	}
45
46	/**
47	* Allow augmenting or modifying the details of a feed.
48	*
49	* The incoming $feed_details array looks like this:
50	*
51	* $feed_details = array(
52	* 'url' => 'https://url.of/the/feed',
53	* 'title' => 'Title from the <link> tag if any',
54	* 'mime-type' => 'mime-type from the <link> tag if any',
55	* // You can add these fields in the response:
56	* 'autoselect' => true\|false,
57	* 'post-format' => 'standard', // or 'aside', etc. see get_post_format_strings() of WordPress core
58	* );
59	*
60	* @param array $feed_details The feed details.
61	*
62	* @return array The (potentially) modified feed details.
63	*/
64	public function update_feed_details( $feed_details ) {
65	return $feed_details;
66	}
67
68	/**
69	* Discover the feeds available at the URL specified.
70	*
71	* The content for the URL has already been fetched for you which can be analyzed.
72	*
73	* Return an array of supported feeds in the format of the $feed_details above:
74	*
75	* return array(
76	* array(
77	* 'url' => 'https://url.of/the/feed',
78	* 'title' => 'Title for the feed',
79	* 'mime-type' => 'mime-type for the feed',
80	* 'rel' => 'e.g. alternate',
81	* ),
82	* );
83	*
84	* @param string $content The content for the URL is already provided here.
85	* @param string $url The url to search.
86	*
87	* @return array A list of supported feeds at the URL.
88	*/
89	public function discover_available_feeds( $content, $url ) { // phpcs:ignore Generic.CodeAnalysis.UnusedFunctionParameter.FoundAfterLastUsed
90	return array();
91	}
92
93	/**
94	* Strip tracking parameters from a URL.
95	*
96	* @param string $url The URL to clean.
97	*
98	* @return string The URL without tracking parameters.
99	*/
100	public static function strip_tracking_parameters( $url ) {
101	$parsed = wp_parse_url( $url );
102	if ( ! $parsed \|\| empty( $parsed['query'] ) ) {
103	return $url;
104	}
105
106	parse_str( $parsed['query'], $params );
107
108	/**
109	* Filter the list of tracking parameter prefixes to strip from URLs.
110	*
111	* @param array $prefixes List of parameter prefixes to strip (e.g., 'utm_' matches utm_source, utm_medium, etc.).
112	*/
113	$tracking_prefixes = apply_filters(
114	'friends_tracking_parameter_prefixes',
115	array(
116	'utm_', // Google Analytics.
117	'mtm_', // Matomo.
118	'pk_', // Piwik/Matomo.
119	)
120	);
121
122	/**
123	* Filter the list of exact tracking parameters to strip from URLs.
124	*
125	* @param array $params List of exact parameter names to strip.
126	*/
127	$tracking_params = apply_filters(
128	'friends_tracking_parameters',
129	array(
130	'fbclid', // Facebook.
131	'gclid', // Google Ads.
132	'gclsrc', // Google Ads.
133	'dclid', // Google Display Network.
134	'gbraid', // Google Ads (iOS).
135	'wbraid', // Google Ads (web-to-app).
136	'msclkid', // Microsoft Ads.
137	'twclid', // Twitter.
138	'igshid', // Instagram.
139	'mc_eid', // Mailchimp.
140	'mc_cid', // Mailchimp.
141	'oly_enc_id', // Omeda.
142	'oly_anon_id', // Omeda.
143	'vero_id', // Vero.
144	'_hsenc', // HubSpot.
145	'_hsmi', // HubSpot.
146	'hsCtaTracking', // HubSpot.
147	'ref', // Generic referrer.
148	'ref_src', // Generic referrer.
149	'ref_url', // Generic referrer.
150	)
151	);
152
153	$cleaned_params = array();
154	foreach ( $params as $key => $value ) {
155	$is_tracking = false;
156
157	// Check exact matches.
158	if ( in_array( $key, $tracking_params, true ) ) {
159	$is_tracking = true;
160	}
161
162	// Check prefix matches.
163	if ( ! $is_tracking ) {
164	foreach ( $tracking_prefixes as $prefix ) {
165	if ( str_starts_with( $key, $prefix ) ) {
166	$is_tracking = true;
167	break;
168	}
169	}
170	}
171
172	if ( ! $is_tracking ) {
173	$cleaned_params[ $key ] = $value;
174	}
175	}
176
177	// Rebuild the URL.
178	$clean_url = '';
179	if ( ! empty( $parsed['scheme'] ) ) {
180	$clean_url .= $parsed['scheme'] . '://';
181	}
182	if ( ! empty( $parsed['host'] ) ) {
183	if ( ! empty( $parsed['user'] ) ) {
184	$clean_url .= $parsed['user'];
185	if ( ! empty( $parsed['pass'] ) ) {
186	$clean_url .= ':' . $parsed['pass'];
187	}
188	$clean_url .= '@';
189	}
190	$clean_url .= $parsed['host'];
191	if ( ! empty( $parsed['port'] ) ) {
192	$clean_url .= ':' . $parsed['port'];
193	}
194	}
195	if ( ! empty( $parsed['path'] ) ) {
196	$clean_url .= $parsed['path'];
197	}
198	if ( ! empty( $cleaned_params ) ) {
199	$clean_url .= '?' . http_build_query( $cleaned_params );
200	}
201	if ( ! empty( $parsed['fragment'] ) ) {
202	$clean_url .= '#' . $parsed['fragment'];
203	}
204
205	return $clean_url;
206	}
207
208	/**
209	* Convert relative URLs to absolute ones in incoming content.
210	*
211	* @param string $html The html.
212	* @param string $permalink The permalink of the feed.
213	*
214	* @return string The HTML with URLs replaced to their absolute represenation.
215	*/
216	public function convert_relative_urls_to_absolute_urls( $html, $permalink ) {
217	if ( ! $html ) {
218	$html = '';
219	}
220
221	// Strip off the hash.
222	$permalink = strtok( $permalink, '#' );
223
224	// For now this only converts links and image srcs.
225	return preg_replace_callback(
226	'~(src\|href)=(?:"([^"]+)\|\'([^\']+))~i',
227	function ( $m ) use ( $permalink ) {
228	// Don't update hash-only links.
229	if ( str_starts_with( $m[2], '#' ) ) {
230	return $m[0];
231	}
232
233	// Remove absolute URL from hashes so that it can become relative.
234	if ( str_starts_with( $m[2], $permalink . '#' ) ) {
235	return str_replace( $permalink, '', $m[0] );
236	}
237
238	// Don't convert content URLs like data:image/png;base64, etc.
239	if ( str_starts_with( $m[2], 'data:' ) ) {
240	return $m[0];
241	}
242
243	// Convert relative URLs to absolute ones and strip tracking parameters.
244	$absolute_url = Mf2\resolveUrl( $permalink, $m[2] );
245	$clean_url = self::strip_tracking_parameters( $absolute_url );
246	return str_replace( $m[2], $clean_url, $m[0] );
247	},
248	$html
249	);
250	}
251	}
252