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 18.2, at inc/sitemaps/class-sitemaps-renderer.php

343 lines 8.6 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 $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 $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 a sitemap by type.
105	*
106	* @api string $urlset The output for the sitemap's `urlset`.
107	*/
108	$xml = apply_filters( "wpseo_sitemap_{$type}_urlset", $urlset );
109
110	foreach ( $links as $url ) {
111	$xml .= $this->sitemap_url( $url );
112	}
113
114	/**
115	* Filter to add extra URLs to the XML sitemap by type.
116	*
117	* Only runs for the first page, not on all.
118	*
119	* @param string $content String content to add, defaults to empty.
120	*/
121	if ( $current_page === 1 ) {
122	$xml .= apply_filters( "wpseo_sitemap_{$type}_content", '' );
123	}
124
125	$xml .= '</urlset>';
126
127	return $xml;
128	}
129
130	/**
131	* Produce final XML output with debug information.
132	*
133	* @param string $sitemap Sitemap XML.
134	*
135	* @return string
136	*/
137	public function get_output( $sitemap ) {
138
139	$output = '<?xml version="1.0" encoding="' . esc_attr( $this->output_charset ) . '"?>';
140
141	if ( $this->stylesheet ) {
142	/**
143	* Filter the stylesheet URL for the XML sitemap.
144	*
145	* @param string $stylesheet Stylesheet URL.
146	*/
147	$output .= apply_filters( 'wpseo_stylesheet_url', $this->stylesheet ) . "\n";
148	}
149
150	$output .= $sitemap;
151	$output .= "\n<!-- XML Sitemap generated by Yoast SEO -->";
152
153	return $output;
154	}
155
156	/**
157	* Get charset for the output.
158	*
159	* @return string
160	*/
161	public function get_output_charset() {
162	return $this->output_charset;
163	}
164
165	/**
166	* Set a custom stylesheet for this sitemap. Set to empty to just remove the default stylesheet.
167	*
168	* @param string $stylesheet Full XML-stylesheet declaration.
169	*/
170	public function set_stylesheet( $stylesheet ) {
171	$this->stylesheet = $stylesheet;
172	}
173
174	/**
175	* Build the `<sitemap>` tag for a given URL.
176	*
177	* @param array $url Array of parts that make up this entry.
178	*
179	* @return string
180	*/
181	protected function sitemap_index_url( $url ) {
182
183	$date = null;
184
185	if ( ! empty( $url['lastmod'] ) ) {
186	$date = YoastSEO()->helpers->date->format( $url['lastmod'] );
187	}
188
189	$url['loc'] = htmlspecialchars( $url['loc'], ENT_COMPAT, $this->output_charset, false );
190
191	$output = "\t<sitemap>\n";
192	$output .= "\t\t<loc>" . $url['loc'] . "</loc>\n";
193	$output .= empty( $date ) ? '' : "\t\t<lastmod>" . htmlspecialchars( $date, ENT_COMPAT, $this->output_charset, false ) . "</lastmod>\n";
194	$output .= "\t</sitemap>\n";
195
196	return $output;
197	}
198
199	/**
200	* Build the `<url>` tag for a given URL.
201	*
202	* Public access for backwards compatibility reasons.
203	*
204	* @param array $url Array of parts that make up this entry.
205	*
206	* @return string
207	*/
208	public function sitemap_url( $url ) {
209
210	$date = null;
211
212
213	if ( ! empty( $url['mod'] ) ) {
214	// Create a DateTime object date in the correct timezone.
215	$date = YoastSEO()->helpers->date->format( $url['mod'] );
216	}
217
218	$url['loc'] = htmlspecialchars( $url['loc'], ENT_COMPAT, $this->output_charset, false );
219
220	$output = "\t<url>\n";
221	$output .= "\t\t<loc>" . $this->encode_url_rfc3986( $url['loc'] ) . "</loc>\n";
222	$output .= empty( $date ) ? '' : "\t\t<lastmod>" . htmlspecialchars( $date, ENT_COMPAT, $this->output_charset, false ) . "</lastmod>\n";
223
224	if ( empty( $url['images'] ) ) {
225	$url['images'] = [];
226	}
227
228	foreach ( $url['images'] as $img ) {
229
230	if ( empty( $img['src'] ) ) {
231	continue;
232	}
233
234	$output .= "\t\t<image:image>\n";
235	$output .= "\t\t\t<image:loc>" . esc_html( $this->encode_url_rfc3986( $img['src'] ) ) . "</image:loc>\n";
236
237	if ( ! empty( $img['title'] ) ) {
238
239	$title = $img['title'];
240
241	if ( $this->needs_conversion ) {
242	$title = mb_convert_encoding( $title, $this->output_charset, $this->charset );
243	}
244
245	$title = _wp_specialchars( html_entity_decode( $title, ENT_QUOTES, $this->output_charset ) );
246	$output .= "\t\t\t<image:title><![CDATA[{$title}]]></image:title>\n";
247	}
248
249	if ( ! empty( $img['alt'] ) ) {
250
251	$alt = $img['alt'];
252
253	if ( $this->needs_conversion ) {
254	$alt = mb_convert_encoding( $alt, $this->output_charset, $this->charset );
255	}
256
257	$alt = _wp_specialchars( html_entity_decode( $alt, ENT_QUOTES, $this->output_charset ) );
258	$output .= "\t\t\t<image:caption><![CDATA[{$alt}]]></image:caption>\n";
259	}
260
261	$output .= "\t\t</image:image>\n";
262	}
263	unset( $img, $title, $alt );
264
265	$output .= "\t</url>\n";
266
267	/**
268	* Filters the output for the sitemap URL tag.
269	*
270	* @api string $output The output for the sitemap url tag.
271	*
272	* @param array $url The sitemap URL array on which the output is based.
273	*/
274	return apply_filters( 'wpseo_sitemap_url', $output, $url );
275	}
276
277	/**
278	* Apply some best effort conversion to comply with RFC3986.
279	*
280	* @param string $url URL to encode.
281	*
282	* @return string
283	*/
284	protected function encode_url_rfc3986( $url ) {
285
286	if ( filter_var( $url, FILTER_VALIDATE_URL ) ) {
287	return $url;
288	}
289
290	$path = wp_parse_url( $url, PHP_URL_PATH );
291
292	if ( ! empty( $path ) && $path !== '/' ) {
293	$encoded_path = explode( '/', $path );
294
295	// First decode the path, to prevent double encoding.
296	$encoded_path = array_map( 'rawurldecode', $encoded_path );
297
298	$encoded_path = array_map( 'rawurlencode', $encoded_path );
299	$encoded_path = implode( '/', $encoded_path );
300
301	$url = str_replace( $path, $encoded_path, $url );
302	}
303
304	$query = wp_parse_url( $url, PHP_URL_QUERY );
305
306	if ( ! empty( $query ) ) {
307
308	parse_str( $query, $parsed_query );
309
310	$parsed_query = http_build_query( $parsed_query, '', '&', PHP_QUERY_RFC3986 );
311
312	$url = str_replace( $query, $parsed_query, $url );
313	}
314
315	return $url;
316	}
317
318	/**
319	* Retrieves the XSL URL that should be used in the current environment
320	*
321	* When home_url and site_url are not the same, the home_url should be used.
322	* This is because the XSL needs to be served from the same domain, protocol and port
323	* as the XML file that is loading it.
324	*
325	* @return string The XSL URL that needs to be used.
326	*/
327	protected function get_xsl_url() {
328	if ( home_url() !== site_url() ) {
329	return home_url( 'main-sitemap.xsl' );
330	}
331
332	/*
333	* Fallback to circumvent a cross-domain security problem when the XLS file is
334	* loaded from a different (sub)domain.
335	*/
336	if ( strpos( plugins_url(), home_url() ) !== 0 ) {
337	return home_url( 'main-sitemap.xsl' );
338	}
339
340	return plugin_dir_url( WPSEO_FILE ) . 'css/main-sitemap.xsl';
341	}
342	}
343