/wp-content/plugins/wordpress-seo/inc/sitemaps/class-sitemaps-renderer.php
PHP | 344 lines | 159 code | 76 blank | 109 comment | 26 complexity | bdded78a9631cf24a51c92cc461a09fd MD5 | raw file
1<?php 2/** 3 * WPSEO plugin file. 4 * 5 * @package WPSEO\XML_Sitemaps 6 */ 7 8/** 9 * Renders XML output for sitemaps. 10 */ 11class WPSEO_Sitemaps_Renderer { 12 13 /** @var string $stylesheet XSL stylesheet for styling a sitemap for web browsers. */ 14 protected $stylesheet = ''; 15 16 /** @var string $charset Holds the get_bloginfo( 'charset' ) value to reuse for performance. */ 17 protected $charset = 'UTF-8'; 18 19 /** @var string $output_charset Holds charset of output, might be converted. */ 20 protected $output_charset = 'UTF-8'; 21 22 /** @var bool $needs_conversion If data encoding needs to be converted for output. */ 23 protected $needs_conversion = false; 24 25 /** @var WPSEO_Sitemap_Timezone $timezone */ 26 protected $timezone; 27 28 /** 29 * Set up object properties. 30 */ 31 public function __construct() { 32 $stylesheet_url = preg_replace( '/(^http[s]?:)/', '', $this->get_xsl_url() ); 33 $this->stylesheet = '<?xml-stylesheet type="text/xsl" href="' . esc_url( $stylesheet_url ) . '"?>'; 34 $this->charset = get_bloginfo( 'charset' ); 35 $this->output_charset = $this->charset; 36 $this->timezone = new WPSEO_Sitemap_Timezone(); 37 38 if ( 39 'UTF-8' !== $this->charset 40 && function_exists( 'mb_list_encodings' ) 41 && in_array( $this->charset, mb_list_encodings(), true ) 42 ) { 43 $this->output_charset = 'UTF-8'; 44 } 45 46 $this->needs_conversion = $this->output_charset !== $this->charset; 47 } 48 49 /** 50 * @param array $links Set of sitemaps index links. 51 * 52 * @return string 53 */ 54 public function get_index( $links ) { 55 56 $xml = '<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n"; 57 58 foreach ( $links as $link ) { 59 $xml .= $this->sitemap_index_url( $link ); 60 } 61 62 /** 63 * Filter to append sitemaps to the index. 64 * 65 * @param string $index String to append to sitemaps index, defaults to empty. 66 */ 67 $xml .= apply_filters( 'wpseo_sitemap_index', '' ); 68 $xml .= '</sitemapindex>'; 69 70 return $xml; 71 } 72 73 /** 74 * @param array $links Set of sitemap links. 75 * @param string $type Sitemap type. 76 * @param int $current_page Current sitemap page number. 77 * 78 * @return string 79 */ 80 public function get_sitemap( $links, $type, $current_page ) { 81 82 $urlset = '<urlset xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:image="http://www.google.com/schemas/sitemap-image/1.1" ' 83 . 'xsi:schemaLocation="http://www.sitemaps.org/schemas/sitemap/0.9 http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd ' 84 . 'http://www.google.com/schemas/sitemap-image/1.1 http://www.google.com/schemas/sitemap-image/1.1/sitemap-image.xsd" ' 85 . 'xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n"; 86 87 /** 88 * Filters the `urlset` for a sitemap by type. 89 * 90 * @api string $urlset The output for the sitemap's `urlset`. 91 */ 92 $xml = apply_filters( "wpseo_sitemap_{$type}_urlset", $urlset ); 93 94 foreach ( $links as $url ) { 95 $xml .= $this->sitemap_url( $url ); 96 } 97 98 /** 99 * Filter to add extra URLs to the XML sitemap by type. 100 * 101 * Only runs for the first page, not on all. 102 * 103 * @param string $content String content to add, defaults to empty. 104 */ 105 if ( $current_page === 1 ) { 106 $xml .= apply_filters( "wpseo_sitemap_{$type}_content", '' ); 107 } 108 109 $xml .= '</urlset>'; 110 111 return $xml; 112 } 113 114 /** 115 * Produce final XML output with debug information. 116 * 117 * @param string $sitemap Sitemap XML. 118 * @param boolean $transient Transient cache flag. 119 * 120 * @return string 121 */ 122 public function get_output( $sitemap, $transient ) { 123 124 $output = '<?xml version="1.0" encoding="' . esc_attr( $this->output_charset ) . '"?>'; 125 126 if ( $this->stylesheet ) { 127 /** 128 * Filter the stylesheet URL for the XML sitemap. 129 * 130 * @param string $stylesheet Stylesheet URL. 131 */ 132 $output .= apply_filters( 'wpseo_stylesheet_url', $this->stylesheet ) . "\n"; 133 } 134 135 $output .= $sitemap; 136 $output .= "\n<!-- XML Sitemap generated by Yoast SEO -->"; 137 138 $debug = WP_DEBUG || ( defined( 'WPSEO_DEBUG' ) && true === WPSEO_DEBUG ); 139 140 if ( ! WP_DEBUG_DISPLAY || ! $debug ) { 141 return $output; 142 } 143 144 $memory_used = number_format( ( memory_get_peak_usage() / 1048576 ), 2 ); 145 $queries_run = ( $transient ) ? 'Served from transient cache' : 'Queries executed ' . absint( $GLOBALS['wpdb']->num_queries ); 146 147 $output .= "\n<!-- {$memory_used}MB | {$queries_run} -->"; 148 149 if ( defined( 'SAVEQUERIES' ) && SAVEQUERIES ) { 150 151 $queries = print_r( $GLOBALS['wpdb']->queries, true ); 152 $output .= "\n<!-- {$queries} -->"; 153 } 154 155 return $output; 156 } 157 158 /** 159 * Get charset for the output. 160 * 161 * @return string 162 */ 163 public function get_output_charset() { 164 return $this->output_charset; 165 } 166 167 /** 168 * Set a custom stylesheet for this sitemap. Set to empty to just remove the default stylesheet. 169 * 170 * @param string $stylesheet Full xml-stylesheet declaration. 171 */ 172 public function set_stylesheet( $stylesheet ) { 173 $this->stylesheet = $stylesheet; 174 } 175 176 /** 177 * Build the `<sitemap>` tag for a given URL. 178 * 179 * @param array $url Array of parts that make up this entry. 180 * 181 * @return string 182 */ 183 protected function sitemap_index_url( $url ) { 184 185 $date = null; 186 187 if ( ! empty( $url['lastmod'] ) ) { 188 $date = $this->timezone->format_date( $url['lastmod'] ); 189 } 190 191 $url['loc'] = htmlspecialchars( $url['loc'] ); 192 193 $output = "\t<sitemap>\n"; 194 $output .= "\t\t<loc>" . $url['loc'] . "</loc>\n"; 195 $output .= empty( $date ) ? '' : "\t\t<lastmod>" . htmlspecialchars( $date ) . "</lastmod>\n"; 196 $output .= "\t</sitemap>\n"; 197 198 return $output; 199 } 200 201 /** 202 * Build the `<url>` tag for a given URL. 203 * 204 * Public access for backwards compatibility reasons. 205 * 206 * @param array $url Array of parts that make up this entry. 207 * 208 * @return string 209 */ 210 public function sitemap_url( $url ) { 211 212 $date = null; 213 214 215 if ( ! empty( $url['mod'] ) ) { 216 // Create a DateTime object date in the correct timezone. 217 $date = $this->timezone->format_date( $url['mod'] ); 218 } 219 220 $url['loc'] = htmlspecialchars( $url['loc'] ); 221 222 $output = "\t<url>\n"; 223 $output .= "\t\t<loc>" . $this->encode_url_rfc3986( $url['loc'] ) . "</loc>\n"; 224 $output .= empty( $date ) ? '' : "\t\t<lastmod>" . htmlspecialchars( $date ) . "</lastmod>\n"; 225 226 if ( empty( $url['images'] ) ) { 227 $url['images'] = array(); 228 } 229 230 foreach ( $url['images'] as $img ) { 231 232 if ( empty( $img['src'] ) ) { 233 continue; 234 } 235 236 $output .= "\t\t<image:image>\n"; 237 $output .= "\t\t\t<image:loc>" . esc_html( $this->encode_url_rfc3986( $img['src'] ) ) . "</image:loc>\n"; 238 239 if ( ! empty( $img['title'] ) ) { 240 241 $title = $img['title']; 242 243 if ( $this->needs_conversion ) { 244 $title = mb_convert_encoding( $title, $this->output_charset, $this->charset ); 245 } 246 247 $title = _wp_specialchars( html_entity_decode( $title, ENT_QUOTES, $this->output_charset ) ); 248 $output .= "\t\t\t<image:title><![CDATA[{$title}]]></image:title>\n"; 249 } 250 251 if ( ! empty( $img['alt'] ) ) { 252 253 $alt = $img['alt']; 254 255 if ( $this->needs_conversion ) { 256 $alt = mb_convert_encoding( $alt, $this->output_charset, $this->charset ); 257 } 258 259 $alt = _wp_specialchars( html_entity_decode( $alt, ENT_QUOTES, $this->output_charset ) ); 260 $output .= "\t\t\t<image:caption><![CDATA[{$alt}]]></image:caption>\n"; 261 } 262 263 $output .= "\t\t</image:image>\n"; 264 } 265 unset( $img, $title, $alt ); 266 267 $output .= "\t</url>\n"; 268 269 /** 270 * Filters the output for the sitemap url tag. 271 * 272 * @api string $output The output for the sitemap url tag. 273 * 274 * @param array $url The sitemap url array on which the output is based. 275 */ 276 return apply_filters( 'wpseo_sitemap_url', $output, $url ); 277 } 278 279 /** 280 * Apply some best effort conversion to comply with RFC3986. 281 * 282 * @param string $url URL to encode. 283 * 284 * @return string 285 */ 286 protected function encode_url_rfc3986( $url ) { 287 288 if ( filter_var( $url, FILTER_VALIDATE_URL ) ) { 289 return $url; 290 } 291 292 $path = wp_parse_url( $url, PHP_URL_PATH ); 293 294 if ( ! empty( $path ) && '/' !== $path ) { 295 $encoded_path = explode( '/', $path ); 296 297 // First decode the path, to prevent double encoding. 298 $encoded_path = array_map( 'rawurldecode', $encoded_path ); 299 300 $encoded_path = array_map( 'rawurlencode', $encoded_path ); 301 $encoded_path = implode( '/', $encoded_path ); 302 $encoded_path = str_replace( '%7E', '~', $encoded_path ); // PHP <5.3. 303 304 $url = str_replace( $path, $encoded_path, $url ); 305 } 306 307 $query = wp_parse_url( $url, PHP_URL_QUERY ); 308 309 if ( ! empty( $query ) ) { 310 311 parse_str( $query, $parsed_query ); 312 313 if ( defined( 'PHP_QUERY_RFC3986' ) ) { // PHP 5.4+. 314 $parsed_query = http_build_query( $parsed_query, null, '&', PHP_QUERY_RFC3986 ); 315 } 316 else { 317 $parsed_query = http_build_query( $parsed_query, null, '&' ); 318 $parsed_query = str_replace( '+', '%20', $parsed_query ); 319 $parsed_query = str_replace( '%7E', '~', $parsed_query ); 320 } 321 322 $url = str_replace( $query, $parsed_query, $url ); 323 } 324 325 return $url; 326 } 327 328 /** 329 * Retrieves the XSL URL that should be used in the current environment 330 * 331 * When home_url and site_url are not the same, the home_url should be used. 332 * This is because the XSL needs to be served from the same domain, protocol and port 333 * as the XML file that is loading it. 334 * 335 * @return string The XSL URL that needs to be used. 336 */ 337 protected function get_xsl_url() { 338 if ( home_url() !== site_url() ) { 339 return home_url( 'main-sitemap.xsl' ); 340 } 341 342 return plugin_dir_url( WPSEO_FILE ) . 'css/main-sitemap.xsl'; 343 } 344}