| @@ -1,0 +1,313 @@ | ||
| 1 | +<?php // phpcs:ignore WordPress.Files.FileName.InvalidClassFileName | |
| 2 | +/** | |
| 3 | + * Sitemaps (per the protocol) are essentially lists of XML fragments; | |
| 4 | + * lists which are subject to size constraints. The Jetpack_Sitemap_Buffer | |
| 5 | + * class abstracts the details of constructing these lists while | |
| 6 | + * maintaining the constraints. | |
| 7 | + * | |
| 8 | + * @since 4.8.0 | |
| 9 | + * @package automattic/jetpack | |
| 10 | + */ | |
| 11 | + | |
| 12 | +if ( ! defined( 'ABSPATH' ) ) { | |
| 13 | + exit( 0 ); | |
| 14 | +} | |
| 15 | + | |
| 16 | +/** | |
| 17 | + * A buffer for constructing sitemap xml files. | |
| 18 | + * | |
| 19 | + * Models a list of strings such that | |
| 20 | + * | |
| 21 | + * 1. the list must have a bounded number of entries, | |
| 22 | + * 2. the concatenation of the strings must have bounded | |
| 23 | + * length (including some header and footer strings), and | |
| 24 | + * 3. each item has a timestamp, and we need to keep track | |
| 25 | + * of the most recent timestamp of the items in the list. | |
| 26 | + * | |
| 27 | + * @since 4.8.0 | |
| 28 | + */ | |
| 29 | +abstract class Jetpack_Sitemap_Buffer { | |
| 30 | + | |
| 31 | + /** | |
| 32 | + * Largest number of items the buffer can hold. | |
| 33 | + * | |
| 34 | + * @access protected | |
| 35 | + * @since 4.8.0 | |
| 36 | + * @var int $item_capacity The item capacity. | |
| 37 | + */ | |
| 38 | + protected $item_capacity; | |
| 39 | + | |
| 40 | + /** | |
| 41 | + * Largest number of bytes the buffer can hold. | |
| 42 | + * | |
| 43 | + * @access protected | |
| 44 | + * @since 4.8.0 | |
| 45 | + * @var int $byte_capacity The byte capacity. | |
| 46 | + */ | |
| 47 | + protected $byte_capacity; | |
| 48 | + | |
| 49 | + /** | |
| 50 | + * Flag which detects when the buffer is full. | |
| 51 | + * | |
| 52 | + * @access protected | |
| 53 | + * @since 4.8.0 | |
| 54 | + * @var bool $is_full_flag The flag value. This flag is set to false on construction and only flipped to true if we've tried to add something and failed. | |
| 55 | + */ | |
| 56 | + protected $is_full_flag; | |
| 57 | + | |
| 58 | + /** | |
| 59 | + * Flag which detects when the buffer is empty. | |
| 60 | + * | |
| 61 | + * @access protected | |
| 62 | + * @since 4.8.0 | |
| 63 | + * @var bool $is_empty_flag The flag value. This flag is set to true on construction and only flipped to false if we've tried to add something and succeeded. | |
| 64 | + */ | |
| 65 | + protected $is_empty_flag; | |
| 66 | + | |
| 67 | + /** | |
| 68 | + * The most recent timestamp seen by the buffer. | |
| 69 | + * | |
| 70 | + * @access protected | |
| 71 | + * @since 4.8.0 | |
| 72 | + * @var string $timestamp Must be in 'YYYY-MM-DD hh:mm:ss' format. | |
| 73 | + */ | |
| 74 | + protected $timestamp; | |
| 75 | + | |
| 76 | + /** | |
| 77 | + * The DOM document object that is currently being used to construct the XML doc. | |
| 78 | + * | |
| 79 | + * @access protected | |
| 80 | + * @since 5.3.0 | |
| 81 | + * @var DOMDocument $doc | |
| 82 | + */ | |
| 83 | + protected $doc = null; | |
| 84 | + | |
| 85 | + /** | |
| 86 | + * The root DOM element object that holds everything inside. Do not use directly, call | |
| 87 | + * the get_root_element getter method instead. | |
| 88 | + * | |
| 89 | + * @access protected | |
| 90 | + * @since 5.3.0 | |
| 91 | + * @var DOMElement $doc | |
| 92 | + */ | |
| 93 | + protected $root = null; | |
| 94 | + | |
| 95 | + /** | |
| 96 | + * Helper class to construct sitemap paths. | |
| 97 | + * | |
| 98 | + * @since 5.3.0 | |
| 99 | + * @protected | |
| 100 | + * @var Jetpack_Sitemap_Finder | |
| 101 | + */ | |
| 102 | + protected $finder; | |
| 103 | + | |
| 104 | + /** | |
| 105 | + * Construct a new Jetpack_Sitemap_Buffer. | |
| 106 | + * | |
| 107 | + * @since 4.8.0 | |
| 108 | + * | |
| 109 | + * @param int $item_limit The maximum size of the buffer in items. | |
| 110 | + * @param int $byte_limit The maximum size of the buffer in bytes. | |
| 111 | + * @param string $time The initial datetime of the buffer. Must be in 'YYYY-MM-DD hh:mm:ss' format. | |
| 112 | + */ | |
| 113 | + public function __construct( $item_limit, $byte_limit, $time ) { | |
| 114 | + $this->is_full_flag = false; | |
| 115 | + $this->timestamp = $time; | |
| 116 | + | |
| 117 | + $this->finder = new Jetpack_Sitemap_Finder(); | |
| 118 | + $this->doc = new DOMDocument( '1.0', 'UTF-8' ); | |
| 119 | + $this->doc->formatOutput = true; | |
| 120 | + $this->doc->preserveWhiteSpace = false; | |
| 121 | + | |
| 122 | + $this->item_capacity = max( 1, (int) $item_limit ); | |
| 123 | + $this->byte_capacity = max( 1, (int) $byte_limit ) - strlen( $this->doc->saveXML() ); | |
| 124 | + } | |
| 125 | + | |
| 126 | + /** | |
| 127 | + * Returns a DOM element that contains all sitemap elements. | |
| 128 | + * | |
| 129 | + * @access protected | |
| 130 | + * @since 5.3.0 | |
| 131 | + * @return DOMElement $root | |
| 132 | + */ | |
| 133 | + abstract protected function get_root_element(); | |
| 134 | + | |
| 135 | + /** | |
| 136 | + * Append an item to the buffer, if there is room for it, | |
| 137 | + * and set is_empty_flag to false. If there is no room, | |
| 138 | + * we set is_full_flag to true. If $item is null, | |
| 139 | + * don't do anything and report success. | |
| 140 | + * | |
| 141 | + * @since 5.3.0 | |
| 142 | + * | |
| 143 | + * @param array $array The item to be added. | |
| 144 | + * | |
| 145 | + * @return bool True if the append succeeded, False if not. | |
| 146 | + */ | |
| 147 | + public function append( $array ) { | |
| 148 | + if ( $array === null ) { | |
| 149 | + return true; | |
| 150 | + } | |
| 151 | + | |
| 152 | + if ( $this->is_full_flag ) { | |
| 153 | + return false; | |
| 154 | + } | |
| 155 | + | |
| 156 | + if ( 0 >= $this->item_capacity || 0 >= $this->byte_capacity ) { | |
| 157 | + $this->is_full_flag = true; | |
| 158 | + return false; | |
| 159 | + } else { | |
| 160 | + $this->item_capacity -= 1; | |
| 161 | + $added_element = $this->array_to_xml_string( $array, $this->get_root_element(), $this->doc ); | |
| 162 | + | |
| 163 | + $this->byte_capacity -= strlen( $this->doc->saveXML( $added_element ) ); | |
| 164 | + | |
| 165 | + return true; | |
| 166 | + } | |
| 167 | + } | |
| 168 | + | |
| 169 | + /** | |
| 170 | + * Retrieve the contents of the buffer. | |
| 171 | + * | |
| 172 | + * @since 4.8.0 | |
| 173 | + * | |
| 174 | + * @return string The contents of the buffer (with the footer included). | |
| 175 | + */ | |
| 176 | + public function contents() { | |
| 177 | + if ( $this->is_empty() ) { | |
| 178 | + // The sitemap should have at least the root element added to the DOM. | |
| 179 | + $this->get_root_element(); | |
| 180 | + } | |
| 181 | + return $this->doc->saveXML(); | |
| 182 | + } | |
| 183 | + | |
| 184 | + /** | |
| 185 | + * Retrieve the document object. | |
| 186 | + * | |
| 187 | + * @since 5.3.0 | |
| 188 | + * @return DOMDocument $doc | |
| 189 | + */ | |
| 190 | + public function get_document() { | |
| 191 | + return $this->doc; | |
| 192 | + } | |
| 193 | + | |
| 194 | + /** | |
| 195 | + * Detect whether the buffer is full. | |
| 196 | + * | |
| 197 | + * @since 4.8.0 | |
| 198 | + * | |
| 199 | + * @return bool True if the buffer is full, false otherwise. | |
| 200 | + */ | |
| 201 | + public function is_full() { | |
| 202 | + return $this->is_full_flag; | |
| 203 | + } | |
| 204 | + | |
| 205 | + /** | |
| 206 | + * Detect whether the buffer is empty. | |
| 207 | + * | |
| 208 | + * @since 4.8.0 | |
| 209 | + * | |
| 210 | + * @return bool True if the buffer is empty, false otherwise. | |
| 211 | + */ | |
| 212 | + public function is_empty() { | |
| 213 | + return ( | |
| 214 | + ! isset( $this->root ) | |
| 215 | + || ! $this->root->hasChildNodes() | |
| 216 | + ); | |
| 217 | + } | |
| 218 | + | |
| 219 | + /** | |
| 220 | + * Update the timestamp of the buffer. | |
| 221 | + * | |
| 222 | + * @since 4.8.0 | |
| 223 | + * | |
| 224 | + * @param string $new_time A datetime string in 'YYYY-MM-DD hh:mm:ss' format. | |
| 225 | + */ | |
| 226 | + public function view_time( $new_time ) { | |
| 227 | + $this->timestamp = max( $this->timestamp, $new_time ); | |
| 228 | + } | |
| 229 | + | |
| 230 | + /** | |
| 231 | + * Retrieve the timestamp of the buffer. | |
| 232 | + * | |
| 233 | + * @since 4.8.0 | |
| 234 | + * | |
| 235 | + * @return string A datetime string in 'YYYY-MM-DD hh:mm:ss' format. | |
| 236 | + */ | |
| 237 | + public function last_modified() { | |
| 238 | + return $this->timestamp; | |
| 239 | + } | |
| 240 | + | |
| 241 | + /** | |
| 242 | + * Render an associative array as an XML string. This is needed because | |
| 243 | + * SimpleXMLElement only handles valid XML, but we sometimes want to | |
| 244 | + * pass around (possibly invalid) fragments. Note that 'null' values make | |
| 245 | + * a tag self-closing; this is only sometimes correct (depending on the | |
| 246 | + * version of HTML/XML); see the list of 'void tags'. | |
| 247 | + * | |
| 248 | + * Example: | |
| 249 | + * | |
| 250 | + * array( | |
| 251 | + * 'html' => array( |<html xmlns="foo"> | |
| 252 | + * 'head' => array( | <head> | |
| 253 | + * 'title' => 'Woo!', | <title>Woo!</title> | |
| 254 | + * ), | </head> | |
| 255 | + * 'body' => array( ==> | <body> | |
| 256 | + * 'h2' => 'Some thing', | <h2>Some thing</h2> | |
| 257 | + * 'p' => 'it's all up ons', | <p>it's all up ons</p> | |
| 258 | + * 'br' => null, | <br /> | |
| 259 | + * ), | </body> | |
| 260 | + * ), |</html> | |
| 261 | + * ) | |
| 262 | + * | |
| 263 | + * @access protected | |
| 264 | + * @since 3.9.0 | |
| 265 | + * @since 4.8.0 Rename, add $depth parameter, and change return type. | |
| 266 | + * @since 5.3.0 Refactor, remove $depth parameter, add $parent and $root, make access protected. | |
| 267 | + * | |
| 268 | + * @param array $array A recursive associative array of tag/child relationships. | |
| 269 | + * @param DOMElement $parent (optional) an element to which new children should be added. | |
| 270 | + * @param DOMDocument $root (optional) the parent document. | |
| 271 | + * | |
| 272 | + * @return string|DOMDocument The rendered XML string or an object if root element is specified. | |
| 273 | + */ | |
| 274 | + protected function array_to_xml_string( $array, $parent = null, $root = null ) { | |
| 275 | + $element = null; | |
| 276 | + $return_string = false; | |
| 277 | + | |
| 278 | + if ( null === $parent ) { | |
| 279 | + $return_string = true; | |
| 280 | + $root = new DOMDocument(); | |
| 281 | + $parent = $root; | |
| 282 | + } | |
| 283 | + | |
| 284 | + if ( is_array( $array ) ) { | |
| 285 | + | |
| 286 | + foreach ( $array as $key => $value ) { | |
| 287 | + $element = $root->createElement( $key ); | |
| 288 | + $parent->appendChild( $element ); | |
| 289 | + | |
| 290 | + if ( is_array( $value ) ) { | |
| 291 | + foreach ( $value as $child_key => $child_value ) { | |
| 292 | + $child = $root->createElement( $child_key ); | |
| 293 | + $element->appendChild( $child ); | |
| 294 | + $child->appendChild( self::array_to_xml_string( $child_value, $child, $root ) ); | |
| 295 | + } | |
| 296 | + } else { | |
| 297 | + $element->appendChild( | |
| 298 | + $root->createTextNode( $value ) | |
| 299 | + ); | |
| 300 | + } | |
| 301 | + } | |
| 302 | + } else { | |
| 303 | + $element = $root->createTextNode( $array ); | |
| 304 | + $parent->appendChild( $element ); | |
| 305 | + } | |
| 306 | + | |
| 307 | + if ( $return_string ) { | |
| 308 | + return $root->saveHTML(); | |
| 309 | + } else { | |
| 310 | + return $element; | |
| 311 | + } | |
| 312 | + } | |
| 313 | +} | |