wordpress-seo / inc / sitemaps / class-sitemaps-renderer.php

class-sitemaps-renderer.php in Yoast SEO – Advanced SEO with real-time guidance and built-in AI trunk, at inc/sitemaps/class-sitemaps-renderer.php

356 lines 9.4 KB

Raw Download Zip

1	<?php
2	/**
3	* WPSEO plugin file.
4	*
5	* @package WPSEO\XML_Sitemaps
6	*/
7
8	/**
9	* Renders XML output for sitemaps.
10	*/
11	class WPSEO_Sitemaps_Renderer {
12
13	/**
14	* XSL stylesheet for styling a sitemap for web browsers.
15	*
16	* @var string
17	*/
18	protected $stylesheet = '';
19
20	/**
21	* Holds the get_bloginfo( 'charset' ) value to reuse for performance.
22	*
23	* @var string
24	*/
25	protected $charset = 'UTF-8';
26
27	/**
28	* Holds charset of output, might be converted.
29	*
30	* @var string
31	*/
32	protected $output_charset = 'UTF-8';
33
34	/**
35	* If data encoding needs to be converted for output.
36	*
37	* @var bool
38	*/
39	protected $needs_conversion = false;
40
41	/**
42	* Set up object properties.
43	*/
44	public function __construct() {
45	$stylesheet_url = preg_replace( '/(^http[s]?:)/', '', $this->get_xsl_url() );
46	$this->stylesheet = '<?xml-stylesheet type="text/xsl" href="' . esc_url( $stylesheet_url ) . '"?>';
47	$this->charset = get_bloginfo( 'charset' );
48	$this->output_charset = $this->charset;
49
50	if (
51	$this->charset !== 'UTF-8'
52	&& function_exists( 'mb_list_encodings' )
53	&& in_array( $this->charset, mb_list_encodings(), true )
54	) {
55	$this->output_charset = 'UTF-8';
56	}
57
58	$this->needs_conversion = $this->output_charset !== $this->charset;
59	}
60
61	/**
62	* Builds the sitemap index.
63	*
64	* @param array<string> $links Set of sitemaps index links.
65	*
66	* @return string
67	*/
68	public function get_index( $links ) {
69
70	$xml = '<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n";
71
72	foreach ( $links as $link ) {
73	$xml .= $this->sitemap_index_url( $link );
74	}
75
76	/**
77	* Filter to append sitemaps to the index.
78	*
79	* @param string $index String to append to sitemaps index, defaults to empty.
80	*/
81	$xml .= apply_filters( 'wpseo_sitemap_index', '' );
82	$xml .= '</sitemapindex>';
83
84	return $xml;
85	}
86
87	/**
88	* Builds the sitemap.
89	*
90	* @param array<string> $links Set of sitemap links.
91	* @param string $type Sitemap type.
92	* @param int $current_page Current sitemap page number.
93	*
94	* @return string
95	*/
96	public function get_sitemap( $links, $type, $current_page ) {
97
98	$urlset = '<urlset xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:image="http://www.google.com/schemas/sitemap-image/1.1" '
99	. 'xsi:schemaLocation="http://www.sitemaps.org/schemas/sitemap/0.9 http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd '
100	. 'http://www.google.com/schemas/sitemap-image/1.1 http://www.google.com/schemas/sitemap-image/1.1/sitemap-image.xsd" '
101	. 'xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n";
102
103	/**
104	* Filters the `urlset` for all sitemaps.
105	*
106	* @param string $urlset The output for the sitemap's `urlset`.
107	*/
108	$urlset = apply_filters( 'wpseo_sitemap_urlset', $urlset );
109
110	/**
111	* Filters the `urlset` for a sitemap by type.
112	*
113	* @param string $urlset The output for the sitemap's `urlset`.
114	*/
115	$xml = apply_filters( "wpseo_sitemap_{$type}_urlset", $urlset );
116
117	foreach ( $links as $url ) {
118	$xml .= $this->sitemap_url( $url );
119	}
120
121	/**
122	* Filter to add extra URLs to the XML sitemap by type.
123	*
124	* Only runs for the first page, not on all.
125	*
126	* @param string $content String content to add, defaults to empty.
127	*/
128	if ( $current_page === 1 ) {
129	$xml .= apply_filters( "wpseo_sitemap_{$type}_content", '' );
130	}
131
132	$xml .= '</urlset>';
133
134	return $xml;
135	}
136
137	/**
138	* Produce final XML output with debug information.
139	*
140	* @param string $sitemap Sitemap XML.
141	*
142	* @return string
143	*/
144	public function get_output( $sitemap ) {
145
146	$output = '<?xml version="1.0" encoding="' . esc_attr( $this->output_charset ) . '"?>';
147
148	if ( $this->stylesheet ) {
149	/**
150	* Filter the stylesheet URL for the XML sitemap.
151	*
152	* @param string $stylesheet Stylesheet URL.
153	*/
154	$output .= apply_filters( 'wpseo_stylesheet_url', $this->stylesheet ) . "\n";
155	}
156
157	$output .= $sitemap;
158	$output .= "\n<!-- XML Sitemap generated by Yoast SEO -->";
159
160	return $output;
161	}
162
163	/**
164	* Get charset for the output.
165	*
166	* @return string
167	*/
168	public function get_output_charset() {
169	return $this->output_charset;
170	}
171
172	/**
173	* Set a custom stylesheet for this sitemap. Set to empty to just remove the default stylesheet.
174	*
175	* @param string $stylesheet Full XML-stylesheet declaration.
176	*
177	* @return void
178	*/
179	public function set_stylesheet( $stylesheet ) {
180	$this->stylesheet = $stylesheet;
181	}
182
183	/**
184	* Build the `<sitemap>` tag for a given URL.
185	*
186	* @param array<string> $url Array of parts that make up this entry.
187	*
188	* @return string
189	*/
190	protected function sitemap_index_url( $url ) {
191
192	$date = null;
193
194	if ( ! empty( $url['lastmod'] ) ) {
195	$date = YoastSEO()->helpers->date->format( $url['lastmod'] );
196	}
197
198	$url['loc'] = htmlspecialchars( $url['loc'], ENT_COMPAT, $this->output_charset, false );
199
200	$output = "\t<sitemap>\n";
201	$output .= "\t\t<loc>" . $url['loc'] . "</loc>\n";
202	$output .= empty( $date ) ? '' : "\t\t<lastmod>" . htmlspecialchars( $date, ENT_COMPAT, $this->output_charset, false ) . "</lastmod>\n";
203	$output .= "\t</sitemap>\n";
204
205	return $output;
206	}
207
208	/**
209	* Build the `<url>` tag for a given URL.
210	*
211	* Public access for backwards compatibility reasons.
212	*
213	* @param array<string> $url Array of parts that make up this entry.
214	*
215	* @return string
216	*/
217	public function sitemap_url( $url ) {
218
219	$date = null;
220
221	if ( ! empty( $url['mod'] ) ) {
222	// Create a DateTime object date in the correct timezone.
223	$date = YoastSEO()->helpers->date->format( $url['mod'] );
224	}
225
226	$output = "\t<url>\n";
227	$output .= "\t\t<loc>" . $this->encode_and_escape( $url['loc'] ) . "</loc>\n";
228	$output .= empty( $date ) ? '' : "\t\t<lastmod>" . htmlspecialchars( $date, ENT_COMPAT, $this->output_charset, false ) . "</lastmod>\n";
229
230	if ( empty( $url['images'] ) ) {
231	$url['images'] = [];
232	}
233
234	foreach ( $url['images'] as $img ) {
235
236	if ( empty( $img['src'] ) ) {
237	continue;
238	}
239
240	$output .= "\t\t<image:image>\n";
241	$output .= "\t\t\t<image:loc>" . $this->encode_and_escape( $img['src'] ) . "</image:loc>\n";
242	$output .= "\t\t</image:image>\n";
243	}
244	unset( $img );
245
246	$output .= "\t</url>\n";
247
248	/**
249	* Filters the output for the sitemap URL tag.
250	*
251	* @param string $output The output for the sitemap url tag.
252	* @param array $url The sitemap URL array on which the output is based.
253	*/
254	return apply_filters( 'wpseo_sitemap_url', $output, $url );
255	}
256
257	/**
258	* Ensure the URL is encoded per RFC3986 and correctly escaped for use in an XML sitemap.
259	*
260	* This method works around a two quirks in esc_url():
261	* 1. `esc_url()` leaves schema-relative URLs alone, while according to the sitemap specs,
262	* the URL must always begin with a protocol.
263	* 2. `esc_url()` escapes ampersands as `&` instead of the more common `&`.
264	* According to the specs, `&` should be used, and even though this shouldn't
265	* really make a difference in practice, to quote Jono: "I'd be nervous about &
266	* given how many weird and wonderful things eat sitemaps", so better safe than sorry.
267	*
268	* @link https://www.sitemaps.org/protocol.html#xmlTagDefinitions
269	* @link https://www.sitemaps.org/protocol.html#escaping
270	* @link https://developer.wordpress.org/reference/functions/esc_url/
271	*
272	* @param string $url URL to encode and escape.
273	*
274	* @return string
275	*/
276	protected function encode_and_escape( $url ) {
277	$url = $this->encode_url_rfc3986( $url );
278	$url = esc_url( $url );
279	$url = str_replace( '&', '&', $url );
280	$url = str_replace( ''', ''', $url );
281
282	if ( strpos( $url, '//' ) === 0 ) {
283	// Schema-relative URL for which esc_url() does not add a scheme.
284	$url = 'http:' . $url;
285	}
286
287	return $url;
288	}
289
290	/**
291	* Apply some best effort conversion to comply with RFC3986.
292	*
293	* @param string $url URL to encode.
294	*
295	* @return string
296	*/
297	protected function encode_url_rfc3986( $url ) {
298
299	if ( filter_var( $url, FILTER_VALIDATE_URL ) ) {
300	return $url;
301	}
302
303	$path = wp_parse_url( $url, PHP_URL_PATH );
304
305	if ( ! empty( $path ) && $path !== '/' ) {
306	$encoded_path = explode( '/', $path );
307
308	// First decode the path, to prevent double encoding.
309	$encoded_path = array_map( 'rawurldecode', $encoded_path );
310
311	$encoded_path = array_map( 'rawurlencode', $encoded_path );
312	$encoded_path = implode( '/', $encoded_path );
313
314	$url = str_replace( $path, $encoded_path, $url );
315	}
316
317	$query = wp_parse_url( $url, PHP_URL_QUERY );
318
319	if ( ! empty( $query ) ) {
320
321	parse_str( $query, $parsed_query );
322
323	$parsed_query = http_build_query( $parsed_query, '', '&', PHP_QUERY_RFC3986 );
324
325	$url = str_replace( $query, $parsed_query, $url );
326	}
327
328	return $url;
329	}
330
331	/**
332	* Retrieves the XSL URL that should be used in the current environment
333	*
334	* When home_url and site_url are not the same, the home_url should be used.
335	* This is because the XSL needs to be served from the same domain, protocol and port
336	* as the XML file that is loading it.
337	*
338	* @return string The XSL URL that needs to be used.
339	*/
340	protected function get_xsl_url() {
341	if ( home_url() !== site_url() ) {
342	return apply_filters( 'wpseo_sitemap_public_url', home_url( 'main-sitemap.xsl' ) );
343	}
344
345	/*
346	* Fallback to circumvent a cross-domain security problem when the XLS file is
347	* loaded from a different (sub)domain.
348	*/
349	if ( strpos( plugins_url(), home_url() ) !== 0 ) {
350	return home_url( 'main-sitemap.xsl' );
351	}
352
353	return plugin_dir_url( WPSEO_FILE ) . 'css/main-sitemap.xsl';
354	}
355	}
356