Helper.php
3 months ago
Importer_Alterator.php
3 months ago
Quiet_Skin.php
5 years ago
Quiet_Skin_Legacy.php
5 years ago
Slug_Mapping.php
3 months ago
Slug_Mapping.php
458 lines
| 1 | <?php |
| 2 | /** |
| 3 | * Slug mapping helper. |
| 4 | * |
| 5 | * Keeps the old/new slug map and source URLs for the current import and |
| 6 | * provides URL rewrite helpers that use this state. |
| 7 | * |
| 8 | * @package templates-patterns-collection |
| 9 | */ |
| 10 | |
| 11 | namespace TIOB\Importers\Helpers; |
| 12 | |
| 13 | /** |
| 14 | * Class Slug_Mapping |
| 15 | */ |
| 16 | class Slug_Mapping { |
| 17 | |
| 18 | /** |
| 19 | * Transient key used for the import slug mapping state. |
| 20 | */ |
| 21 | const TRANSIENT_KEY = 'ti_tpc_import_slug_mapping'; |
| 22 | |
| 23 | /** |
| 24 | * Map array key. |
| 25 | */ |
| 26 | const MAP_KEY = 'slug_map'; |
| 27 | |
| 28 | /** |
| 29 | * Source URLs array key. |
| 30 | */ |
| 31 | const SOURCES_KEY = 'source_urls'; |
| 32 | |
| 33 | /** |
| 34 | * Persisted state expiration. |
| 35 | */ |
| 36 | const EXPIRATION = 12 * HOUR_IN_SECONDS; |
| 37 | |
| 38 | /** |
| 39 | * Clear stored state. |
| 40 | * |
| 41 | * @return void |
| 42 | */ |
| 43 | public static function clear() { |
| 44 | delete_transient( self::TRANSIENT_KEY ); |
| 45 | } |
| 46 | |
| 47 | /** |
| 48 | * Set slug map. |
| 49 | * |
| 50 | * @param array $slug_map old slug => new slug map. |
| 51 | * |
| 52 | * @return void |
| 53 | */ |
| 54 | public static function set_slug_map( $slug_map ) { |
| 55 | if ( ! is_array( $slug_map ) ) { |
| 56 | return; |
| 57 | } |
| 58 | |
| 59 | $state = self::get_state(); |
| 60 | $map = array(); |
| 61 | |
| 62 | foreach ( $slug_map as $old_slug => $new_slug ) { |
| 63 | if ( ! is_string( $old_slug ) || ! is_string( $new_slug ) ) { |
| 64 | continue; |
| 65 | } |
| 66 | |
| 67 | $old_slug = trim( $old_slug ); |
| 68 | $new_slug = trim( $new_slug ); |
| 69 | if ( $old_slug === '' || $new_slug === '' ) { |
| 70 | continue; |
| 71 | } |
| 72 | |
| 73 | $map[ $old_slug ] = $new_slug; |
| 74 | } |
| 75 | |
| 76 | $state[ self::MAP_KEY ] = $map; |
| 77 | self::save_state( $state ); |
| 78 | } |
| 79 | |
| 80 | /** |
| 81 | * Return the stored slug map. |
| 82 | * |
| 83 | * @return array |
| 84 | */ |
| 85 | public static function get_slug_map() { |
| 86 | $state = self::get_state(); |
| 87 | return is_array( $state[ self::MAP_KEY ] ) ? $state[ self::MAP_KEY ] : array(); |
| 88 | } |
| 89 | |
| 90 | /** |
| 91 | * Resolve a slug through the map. |
| 92 | * |
| 93 | * @param string $slug old slug. |
| 94 | * |
| 95 | * @return string |
| 96 | */ |
| 97 | public static function resolve_slug( $slug ) { |
| 98 | $map = self::get_slug_map(); |
| 99 | if ( isset( $map[ $slug ] ) ) { |
| 100 | return $map[ $slug ]; |
| 101 | } |
| 102 | |
| 103 | return $slug; |
| 104 | } |
| 105 | |
| 106 | /** |
| 107 | * Register an import source URL. |
| 108 | * |
| 109 | * @param string $source_url source URL. |
| 110 | * |
| 111 | * @return void |
| 112 | */ |
| 113 | public static function register_source_url( $source_url ) { |
| 114 | if ( ! is_string( $source_url ) || empty( $source_url ) ) { |
| 115 | return; |
| 116 | } |
| 117 | |
| 118 | $source_url = esc_url_raw( trim( $source_url ) ); |
| 119 | if ( empty( $source_url ) ) { |
| 120 | return; |
| 121 | } |
| 122 | |
| 123 | $state = self::get_state(); |
| 124 | $source_urls = isset( $state[ self::SOURCES_KEY ] ) && is_array( $state[ self::SOURCES_KEY ] ) ? $state[ self::SOURCES_KEY ] : array(); |
| 125 | |
| 126 | if ( ! in_array( $source_url, $source_urls, true ) ) { |
| 127 | $source_urls[] = $source_url; |
| 128 | } |
| 129 | |
| 130 | $state[ self::SOURCES_KEY ] = $source_urls; |
| 131 | self::save_state( $state ); |
| 132 | } |
| 133 | |
| 134 | /** |
| 135 | * Get all registered source URLs. |
| 136 | * |
| 137 | * @return array |
| 138 | */ |
| 139 | public static function get_source_urls() { |
| 140 | $state = self::get_state(); |
| 141 | return isset( $state[ self::SOURCES_KEY ] ) && is_array( $state[ self::SOURCES_KEY ] ) ? $state[ self::SOURCES_KEY ] : array(); |
| 142 | } |
| 143 | |
| 144 | /** |
| 145 | * Recursively rewrite internal links in a value. |
| 146 | * |
| 147 | * @param mixed $value value to rewrite. |
| 148 | * |
| 149 | * @return mixed |
| 150 | */ |
| 151 | public static function rewrite_value( $value ) { |
| 152 | if ( is_array( $value ) ) { |
| 153 | foreach ( $value as $key => $item ) { |
| 154 | $value[ $key ] = self::rewrite_value( $item ); |
| 155 | } |
| 156 | |
| 157 | return $value; |
| 158 | } |
| 159 | |
| 160 | if ( is_string( $value ) ) { |
| 161 | return self::rewrite_string_value( $value ); |
| 162 | } |
| 163 | |
| 164 | return $value; |
| 165 | } |
| 166 | |
| 167 | /** |
| 168 | * Rewrite a single URL if it belongs to an import source. |
| 169 | * |
| 170 | * @param string $url URL. |
| 171 | * |
| 172 | * @return string |
| 173 | */ |
| 174 | public static function rewrite_url( $url ) { |
| 175 | if ( ! is_string( $url ) || empty( $url ) ) { |
| 176 | return $url; |
| 177 | } |
| 178 | |
| 179 | $source_url = self::get_matching_source_url( $url ); |
| 180 | if ( empty( $source_url ) ) { |
| 181 | return $url; |
| 182 | } |
| 183 | |
| 184 | $url_parts = wp_parse_url( $url ); |
| 185 | $source_parts = wp_parse_url( $source_url ); |
| 186 | $home_parts = wp_parse_url( get_home_url() ); |
| 187 | |
| 188 | if ( ! is_array( $url_parts ) || ! is_array( $source_parts ) || ! is_array( $home_parts ) ) { |
| 189 | return $url; |
| 190 | } |
| 191 | |
| 192 | $target_path = isset( $url_parts['path'] ) ? $url_parts['path'] : '/'; |
| 193 | $source_path = isset( $source_parts['path'] ) ? rtrim( $source_parts['path'], '/' ) : ''; |
| 194 | $home_path = isset( $home_parts['path'] ) ? rtrim( $home_parts['path'], '/' ) : ''; |
| 195 | $relative = $target_path; |
| 196 | |
| 197 | if ( ! empty( $source_path ) && $source_path !== '/' ) { |
| 198 | if ( $target_path === $source_path ) { |
| 199 | $relative = ''; |
| 200 | } elseif ( strpos( $target_path, $source_path . '/' ) === 0 ) { |
| 201 | $relative = substr( $target_path, strlen( $source_path ) ); |
| 202 | } else { |
| 203 | return $url; |
| 204 | } |
| 205 | } |
| 206 | |
| 207 | $rewritten_relative = $relative; |
| 208 | if ( $rewritten_relative !== '' ) { |
| 209 | $rewritten_relative = self::replace_path_slugs( $rewritten_relative ); |
| 210 | } |
| 211 | |
| 212 | $new_path = $home_path . $rewritten_relative; |
| 213 | if ( $new_path === '' ) { |
| 214 | $new_path = '/'; |
| 215 | } |
| 216 | if ( strpos( $new_path, '/' ) !== 0 ) { |
| 217 | $new_path = '/' . $new_path; |
| 218 | } |
| 219 | |
| 220 | $url_parts['scheme'] = isset( $home_parts['scheme'] ) ? $home_parts['scheme'] : ( isset( $url_parts['scheme'] ) ? $url_parts['scheme'] : 'http' ); |
| 221 | $url_parts['host'] = isset( $home_parts['host'] ) ? $home_parts['host'] : $url_parts['host']; |
| 222 | $url_parts['path'] = $new_path; |
| 223 | |
| 224 | if ( isset( $home_parts['port'] ) ) { |
| 225 | $url_parts['port'] = $home_parts['port']; |
| 226 | } else { |
| 227 | unset( $url_parts['port'] ); |
| 228 | } |
| 229 | |
| 230 | if ( isset( $home_parts['user'] ) ) { |
| 231 | $url_parts['user'] = $home_parts['user']; |
| 232 | } else { |
| 233 | unset( $url_parts['user'] ); |
| 234 | } |
| 235 | |
| 236 | if ( isset( $home_parts['pass'] ) ) { |
| 237 | $url_parts['pass'] = $home_parts['pass']; |
| 238 | } else { |
| 239 | unset( $url_parts['pass'] ); |
| 240 | } |
| 241 | |
| 242 | return self::build_url( $url_parts ); |
| 243 | } |
| 244 | |
| 245 | /** |
| 246 | * Rewrite string values and support serialized content. |
| 247 | * |
| 248 | * @param string $value string value. |
| 249 | * |
| 250 | * @return string |
| 251 | */ |
| 252 | private static function rewrite_string_value( $value ) { |
| 253 | if ( $value === '' ) { |
| 254 | return $value; |
| 255 | } |
| 256 | |
| 257 | if ( is_serialized( $value ) ) { |
| 258 | $unserialized = maybe_unserialize( $value ); |
| 259 | $rewritten = self::rewrite_value( $unserialized ); |
| 260 | |
| 261 | return maybe_serialize( $rewritten ); |
| 262 | } |
| 263 | |
| 264 | $value = self::rewrite_plain_urls_in_string( $value ); |
| 265 | $value = self::rewrite_escaped_urls_in_string( $value ); |
| 266 | |
| 267 | return $value; |
| 268 | } |
| 269 | |
| 270 | /** |
| 271 | * Rewrite plain URLs found in string. |
| 272 | * |
| 273 | * @param string $value value. |
| 274 | * |
| 275 | * @return string |
| 276 | */ |
| 277 | private static function rewrite_plain_urls_in_string( $value ) { |
| 278 | return preg_replace_callback( |
| 279 | '#https?://[^\s"\'<>()]+#i', |
| 280 | function ( $matches ) { |
| 281 | return self::rewrite_url( $matches[0] ); |
| 282 | }, |
| 283 | $value |
| 284 | ); |
| 285 | } |
| 286 | |
| 287 | /** |
| 288 | * Rewrite escaped URLs found in string. |
| 289 | * |
| 290 | * @param string $value value. |
| 291 | * |
| 292 | * @return string |
| 293 | */ |
| 294 | private static function rewrite_escaped_urls_in_string( $value ) { |
| 295 | return preg_replace_callback( |
| 296 | '#https?:\\\\/\\\\/[^\s"\'<>()]+#i', |
| 297 | function ( $matches ) { |
| 298 | $escaped_url = $matches[0]; |
| 299 | $decoded_url = str_replace( '\\/', '/', $escaped_url ); |
| 300 | $rewritten = self::rewrite_url( $decoded_url ); |
| 301 | $rewritten = str_replace( '/', '\\/', $rewritten ); |
| 302 | |
| 303 | return $rewritten; |
| 304 | }, |
| 305 | $value |
| 306 | ); |
| 307 | } |
| 308 | |
| 309 | /** |
| 310 | * Replace path segments using slug map. |
| 311 | * |
| 312 | * @param string $path path. |
| 313 | * |
| 314 | * @return string |
| 315 | */ |
| 316 | private static function replace_path_slugs( $path ) { |
| 317 | $map = self::get_slug_map(); |
| 318 | if ( empty( $map ) || ! is_string( $path ) || $path === '' ) { |
| 319 | return $path; |
| 320 | } |
| 321 | |
| 322 | if ( $path === '/' ) { |
| 323 | return $path; |
| 324 | } |
| 325 | |
| 326 | $starts_with_slash = strpos( $path, '/' ) === 0; |
| 327 | $ends_with_slash = substr( $path, -1 ) === '/'; |
| 328 | $segments = explode( '/', trim( $path, '/' ) ); |
| 329 | |
| 330 | foreach ( $segments as $index => $segment ) { |
| 331 | $decoded = rawurldecode( $segment ); |
| 332 | if ( isset( $map[ $decoded ] ) ) { |
| 333 | $segments[ $index ] = rawurlencode( $map[ $decoded ] ); |
| 334 | } |
| 335 | } |
| 336 | |
| 337 | $new_path = implode( '/', $segments ); |
| 338 | if ( $starts_with_slash ) { |
| 339 | $new_path = '/' . $new_path; |
| 340 | } |
| 341 | if ( $ends_with_slash && $new_path !== '/' ) { |
| 342 | $new_path .= '/'; |
| 343 | } |
| 344 | if ( $new_path === '' ) { |
| 345 | $new_path = '/'; |
| 346 | } |
| 347 | |
| 348 | return $new_path; |
| 349 | } |
| 350 | |
| 351 | /** |
| 352 | * Return the matching source URL for a URL. |
| 353 | * |
| 354 | * @param string $url URL. |
| 355 | * |
| 356 | * @return string |
| 357 | */ |
| 358 | private static function get_matching_source_url( $url ) { |
| 359 | $url_parts = wp_parse_url( $url ); |
| 360 | if ( ! is_array( $url_parts ) || ! isset( $url_parts['host'] ) ) { |
| 361 | return ''; |
| 362 | } |
| 363 | |
| 364 | $target_host = strtolower( $url_parts['host'] ); |
| 365 | $target_path = isset( $url_parts['path'] ) ? $url_parts['path'] : '/'; |
| 366 | $best_match = ''; |
| 367 | $best_length = -1; |
| 368 | |
| 369 | foreach ( self::get_source_urls() as $source_url ) { |
| 370 | $source_parts = wp_parse_url( $source_url ); |
| 371 | if ( ! is_array( $source_parts ) || ! isset( $source_parts['host'] ) ) { |
| 372 | continue; |
| 373 | } |
| 374 | |
| 375 | if ( strtolower( $source_parts['host'] ) !== $target_host ) { |
| 376 | continue; |
| 377 | } |
| 378 | |
| 379 | if ( isset( $source_parts['port'], $url_parts['port'] ) && (int) $source_parts['port'] !== (int) $url_parts['port'] ) { |
| 380 | continue; |
| 381 | } |
| 382 | |
| 383 | $source_path = isset( $source_parts['path'] ) ? rtrim( $source_parts['path'], '/' ) : ''; |
| 384 | if ( empty( $source_path ) || $source_path === '/' ) { |
| 385 | $path_length = 0; |
| 386 | if ( $path_length > $best_length ) { |
| 387 | $best_match = $source_url; |
| 388 | $best_length = $path_length; |
| 389 | } |
| 390 | continue; |
| 391 | } |
| 392 | |
| 393 | if ( $target_path === $source_path || strpos( $target_path, $source_path . '/' ) === 0 ) { |
| 394 | $path_length = strlen( $source_path ); |
| 395 | if ( $path_length > $best_length ) { |
| 396 | $best_match = $source_url; |
| 397 | $best_length = $path_length; |
| 398 | } |
| 399 | } |
| 400 | } |
| 401 | |
| 402 | return $best_match; |
| 403 | } |
| 404 | |
| 405 | /** |
| 406 | * Build URL from parsed URL parts. |
| 407 | * |
| 408 | * @param array $parts parsed URL parts. |
| 409 | * |
| 410 | * @return string |
| 411 | */ |
| 412 | private static function build_url( $parts ) { |
| 413 | $scheme = isset( $parts['scheme'] ) ? $parts['scheme'] . '://' : ''; |
| 414 | $user = isset( $parts['user'] ) ? $parts['user'] : ''; |
| 415 | $pass = isset( $parts['pass'] ) ? ':' . $parts['pass'] : ''; |
| 416 | $auth = ( $user || $pass ) ? "{$user}{$pass}@" : ''; |
| 417 | $host = isset( $parts['host'] ) ? $parts['host'] : ''; |
| 418 | $port = isset( $parts['port'] ) ? ':' . $parts['port'] : ''; |
| 419 | $path = isset( $parts['path'] ) ? $parts['path'] : ''; |
| 420 | $query = isset( $parts['query'] ) ? '?' . $parts['query'] : ''; |
| 421 | $fragment = isset( $parts['fragment'] ) ? '#' . $parts['fragment'] : ''; |
| 422 | |
| 423 | return "{$scheme}{$auth}{$host}{$port}{$path}{$query}{$fragment}"; |
| 424 | } |
| 425 | |
| 426 | /** |
| 427 | * Get state from transient. |
| 428 | * |
| 429 | * @return array |
| 430 | */ |
| 431 | private static function get_state() { |
| 432 | $state = get_transient( self::TRANSIENT_KEY ); |
| 433 | if ( ! is_array( $state ) ) { |
| 434 | $state = array(); |
| 435 | } |
| 436 | |
| 437 | if ( ! isset( $state[ self::MAP_KEY ] ) || ! is_array( $state[ self::MAP_KEY ] ) ) { |
| 438 | $state[ self::MAP_KEY ] = array(); |
| 439 | } |
| 440 | if ( ! isset( $state[ self::SOURCES_KEY ] ) || ! is_array( $state[ self::SOURCES_KEY ] ) ) { |
| 441 | $state[ self::SOURCES_KEY ] = array(); |
| 442 | } |
| 443 | |
| 444 | return $state; |
| 445 | } |
| 446 | |
| 447 | /** |
| 448 | * Save state. |
| 449 | * |
| 450 | * @param array $state state. |
| 451 | * |
| 452 | * @return void |
| 453 | */ |
| 454 | private static function save_state( $state ) { |
| 455 | set_transient( self::TRANSIENT_KEY, $state, self::EXPIRATION ); |
| 456 | } |
| 457 | } |
| 458 |