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

372 lines 9.7 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	if ( ! empty( $url['mod'] ) ) {
213	// Create a DateTime object date in the correct timezone.
214	$date = YoastSEO()->helpers->date->format( $url['mod'] );
215	}
216
217	$output = "\t<url>\n";
218	$output .= "\t\t<loc>" . $this->encode_and_escape( $url['loc'] ) . "</loc>\n";
219	$output .= empty( $date ) ? '' : "\t\t<lastmod>" . htmlspecialchars( $date, ENT_COMPAT, $this->output_charset, false ) . "</lastmod>\n";
220
221	if ( empty( $url['images'] ) ) {
222	$url['images'] = [];
223	}
224
225	foreach ( $url['images'] as $img ) {
226
227	if ( empty( $img['src'] ) ) {
228	continue;
229	}
230
231	$output .= "\t\t<image:image>\n";
232	$output .= "\t\t\t<image:loc>" . $this->encode_and_escape( $img['src'] ) . "</image:loc>\n";
233
234	if ( ! empty( $img['title'] ) ) {
235
236	$title = $img['title'];
237
238	if ( $this->needs_conversion ) {
239	$title = mb_convert_encoding( $title, $this->output_charset, $this->charset );
240	}
241
242	$title = _wp_specialchars( html_entity_decode( $title, ENT_QUOTES, $this->output_charset ) );
243	$output .= "\t\t\t<image:title><![CDATA[{$title}]]></image:title>\n";
244	}
245
246	if ( ! empty( $img['alt'] ) ) {
247
248	$alt = $img['alt'];
249
250	if ( $this->needs_conversion ) {
251	$alt = mb_convert_encoding( $alt, $this->output_charset, $this->charset );
252	}
253
254	$alt = _wp_specialchars( html_entity_decode( $alt, ENT_QUOTES, $this->output_charset ) );
255	$output .= "\t\t\t<image:caption><![CDATA[{$alt}]]></image:caption>\n";
256	}
257
258	$output .= "\t\t</image:image>\n";
259	}
260	unset( $img, $title, $alt );
261
262	$output .= "\t</url>\n";
263
264	/**
265	* Filters the output for the sitemap URL tag.
266	*
267	* @api string $output The output for the sitemap url tag.
268	*
269	* @param array $url The sitemap URL array on which the output is based.
270	*/
271	return apply_filters( 'wpseo_sitemap_url', $output, $url );
272	}
273
274	/**
275	* Ensure the URL is encoded per RFC3986 and correctly escaped for use in an XML sitemap.
276	*
277	* This method works around a two quirks in esc_url():
278	* 1. `esc_url()` leaves schema-relative URLs alone, while according to the sitemap specs,
279	* the URL must always begin with a protocol.
280	* 2. `esc_url()` escapes ampersands as `&` instead of the more common `&`.
281	* According to the specs, `&` should be used, and even though this shouldn't
282	* really make a difference in practice, to quote Jono: "I'd be nervous about &
283	* given how many weird and wonderful things eat sitemaps", so better safe than sorry.
284	*
285	* @link https://www.sitemaps.org/protocol.html#xmlTagDefinitions
286	* @link https://www.sitemaps.org/protocol.html#escaping
287	* @link https://developer.wordpress.org/reference/functions/esc_url/
288	*
289	* @param string $url URL to encode and escape.
290	*
291	* @return string
292	*/
293	protected function encode_and_escape( $url ) {
294	$url = $this->encode_url_rfc3986( $url );
295	$url = esc_url( $url );
296	$url = str_replace( '&', '&', $url );
297
298	if ( strpos( $url, '//' ) === 0 ) {
299	// Schema-relative URL for which esc_url() does not add a scheme.
300	$url = 'http:' . $url;
301	}
302
303	return $url;
304	}
305
306	/**
307	* Apply some best effort conversion to comply with RFC3986.
308	*
309	* @param string $url URL to encode.
310	*
311	* @return string
312	*/
313	protected function encode_url_rfc3986( $url ) {
314
315	if ( filter_var( $url, FILTER_VALIDATE_URL ) ) {
316	return $url;
317	}
318
319	$path = wp_parse_url( $url, PHP_URL_PATH );
320
321	if ( ! empty( $path ) && $path !== '/' ) {
322	$encoded_path = explode( '/', $path );
323
324	// First decode the path, to prevent double encoding.
325	$encoded_path = array_map( 'rawurldecode', $encoded_path );
326
327	$encoded_path = array_map( 'rawurlencode', $encoded_path );
328	$encoded_path = implode( '/', $encoded_path );
329
330	$url = str_replace( $path, $encoded_path, $url );
331	}
332
333	$query = wp_parse_url( $url, PHP_URL_QUERY );
334
335	if ( ! empty( $query ) ) {
336
337	parse_str( $query, $parsed_query );
338
339	$parsed_query = http_build_query( $parsed_query, '', '&', PHP_QUERY_RFC3986 );
340
341	$url = str_replace( $query, $parsed_query, $url );
342	}
343
344	return $url;
345	}
346
347	/**
348	* Retrieves the XSL URL that should be used in the current environment
349	*
350	* When home_url and site_url are not the same, the home_url should be used.
351	* This is because the XSL needs to be served from the same domain, protocol and port
352	* as the XML file that is loading it.
353	*
354	* @return string The XSL URL that needs to be used.
355	*/
356	protected function get_xsl_url() {
357	if ( home_url() !== site_url() ) {
358	return home_url( 'main-sitemap.xsl' );
359	}
360
361	/*
362	* Fallback to circumvent a cross-domain security problem when the XLS file is
363	* loaded from a different (sub)domain.
364	*/
365	if ( strpos( plugins_url(), home_url() ) !== 0 ) {
366	return home_url( 'main-sitemap.xsl' );
367	}
368
369	return plugin_dir_url( WPSEO_FILE ) . 'css/main-sitemap.xsl';
370	}
371	}
372