| 1 |
<?php |
| 2 |
/** |
| 3 |
* Indexable abstract class. |
| 4 |
* |
| 5 |
* An indexable is a type of "data" in WP e.g. post type, term, user, etc. |
| 6 |
* |
| 7 |
* @since 3.0 |
| 8 |
* @package elasticpress |
| 9 |
*/ |
| 10 |
|
| 11 |
namespace ElasticPress; |
| 12 |
|
| 13 |
use ElasticPress\Elasticsearch; |
| 14 |
use ElasticPress\SyncManager; |
| 15 |
|
| 16 |
if ( ! defined( 'ABSPATH' ) ) { |
| 17 |
exit; // Exit if accessed directly. |
| 18 |
} |
| 19 |
|
| 20 |
/** |
| 21 |
* An indexable is essentially a document type that can be indexed |
| 22 |
* and queried against |
| 23 |
* |
| 24 |
* @since 3.0 |
| 25 |
*/ |
| 26 |
abstract class Indexable { |
| 27 |
|
| 28 |
/** |
| 29 |
* Declaring an Indexable global means it won't have an index for each blog in |
| 30 |
* the network. Instead it will just have one index. There will also be no |
| 31 |
* network alias. |
| 32 |
* |
| 33 |
* @var boolean |
| 34 |
* @since 3.0 |
| 35 |
*/ |
| 36 |
public $global = false; |
| 37 |
|
| 38 |
/** |
| 39 |
* Instance of SyncManager. This should handle automated syncing of indexable |
| 40 |
* objects. |
| 41 |
* |
| 42 |
* @var SyncManager |
| 43 |
* @since 3.0 |
| 44 |
*/ |
| 45 |
public $sync_manager; |
| 46 |
|
| 47 |
/** |
| 48 |
* Instance of QueryIntegration. This should handle integrating with a default |
| 49 |
* WP query. |
| 50 |
* |
| 51 |
* @var object |
| 52 |
* @since 3.0 |
| 53 |
*/ |
| 54 |
public $query_integration; |
| 55 |
|
| 56 |
/** |
| 57 |
* Flag to indicate if the indexable has support for |
| 58 |
* `id_range` pagination method during a sync. |
| 59 |
* |
| 60 |
* @var boolean |
| 61 |
* @since 4.1.0 |
| 62 |
*/ |
| 63 |
public $support_indexing_advanced_pagination = false; |
| 64 |
|
| 65 |
/** |
| 66 |
* Indexable slug |
| 67 |
* |
| 68 |
* @since 4.5.0 |
| 69 |
* @var string |
| 70 |
*/ |
| 71 |
public $slug = ''; |
| 72 |
|
| 73 |
/** |
| 74 |
* Indexable labels |
| 75 |
* |
| 76 |
* @since 4.5.0 |
| 77 |
* @var array |
| 78 |
*/ |
| 79 |
public $labels = [ |
| 80 |
'plural' => '', |
| 81 |
'singular' => '', |
| 82 |
]; |
| 83 |
|
| 84 |
/** |
| 85 |
* Get number of bulk items to index per page |
| 86 |
* |
| 87 |
* @since 3.0 |
| 88 |
* @return int |
| 89 |
*/ |
| 90 |
public function get_bulk_items_per_page() { |
| 91 |
/** |
| 92 |
* Filter bulk items to sync per batch |
| 93 |
* |
| 94 |
* @hook ep_bulk_items_per_page |
| 95 |
* @param {int} $number Number of items per batch |
| 96 |
* @param {Indexable} $indexable Current indexable |
| 97 |
* @return {int} New number of items |
| 98 |
* @since 3.0 |
| 99 |
*/ |
| 100 |
return apply_filters( 'ep_bulk_items_per_page', 350, $this ); |
| 101 |
} |
| 102 |
|
| 103 |
/** |
| 104 |
* Get the name of the index. Each indexable needs a unique index name |
| 105 |
* |
| 106 |
* @param int $blog_id `null` means current blog. |
| 107 |
* @since 3.0 |
| 108 |
* @return string |
| 109 |
*/ |
| 110 |
public function get_index_name( $blog_id = null ) { |
| 111 |
if ( $this->global ) { |
| 112 |
$site_url = network_site_url(); |
| 113 |
|
| 114 |
if ( ! empty( $site_url ) ) { |
| 115 |
$index_name = preg_replace( '#https?://(www\.)?#i', '', $site_url ); |
| 116 |
$index_name = preg_replace( '#[^\w]#', '', $index_name ) . '-' . $this->slug; |
| 117 |
} else { |
| 118 |
$index_name = false; |
| 119 |
} |
| 120 |
} else { |
| 121 |
if ( ! $blog_id ) { |
| 122 |
$blog_id = get_current_blog_id(); |
| 123 |
} |
| 124 |
|
| 125 |
$site_url = get_site_url( $blog_id ); |
| 126 |
|
| 127 |
if ( ! empty( $site_url ) ) { |
| 128 |
$index_name = preg_replace( '#https?://(www\.)?#i', '', $site_url ); |
| 129 |
$index_name = preg_replace( '#[^\w]#', '', $index_name ) . '-' . $this->slug . '-' . $blog_id; |
| 130 |
} else { |
| 131 |
$index_name = false; |
| 132 |
} |
| 133 |
} |
| 134 |
|
| 135 |
$prefix = Utils\get_index_prefix(); |
| 136 |
|
| 137 |
if ( ! empty( $prefix ) ) { |
| 138 |
$index_name = $prefix . '-' . $index_name; |
| 139 |
} |
| 140 |
|
| 141 |
$index_name = strtolower( $index_name ); |
| 142 |
|
| 143 |
/** |
| 144 |
* Filter index name |
| 145 |
* |
| 146 |
* @hook ep_index_name |
| 147 |
* @param {string} $index_name Name of index |
| 148 |
* @param {int} $blog_id Blog ID |
| 149 |
* @param {Indexable} $indexable Current indexable |
| 150 |
* @return {string} Index name |
| 151 |
* @since 3.0 |
| 152 |
*/ |
| 153 |
return apply_filters( 'ep_index_name', $index_name, $blog_id, $this ); |
| 154 |
} |
| 155 |
|
| 156 |
/** |
| 157 |
* Get unique indexable network alias |
| 158 |
* |
| 159 |
* @since 3.0 |
| 160 |
* @return string |
| 161 |
*/ |
| 162 |
public function get_network_alias() { |
| 163 |
$url = network_site_url(); |
| 164 |
$slug = preg_replace( '#https?://(www\.)?#i', '', $url ); |
| 165 |
$slug = preg_replace( '#[^\w]#', '', $slug ); |
| 166 |
|
| 167 |
$alias = $slug . '-' . $this->slug . '-global'; |
| 168 |
|
| 169 |
$prefix = Utils\get_index_prefix(); |
| 170 |
|
| 171 |
if ( ! empty( $prefix ) ) { |
| 172 |
$alias = $prefix . '-' . $alias; |
| 173 |
} |
| 174 |
|
| 175 |
/** |
| 176 |
* Filter global/network Elasticsearch alias |
| 177 |
* |
| 178 |
* @hook ep_global_alias |
| 179 |
* @param {string} $number Current alias |
| 180 |
* @return {string} New alias |
| 181 |
*/ |
| 182 |
return apply_filters( 'ep_global_alias', $alias ); |
| 183 |
} |
| 184 |
|
| 185 |
/** |
| 186 |
* Delete unique indexable network alias |
| 187 |
* |
| 188 |
* @since 3.0 |
| 189 |
* @return boolean |
| 190 |
*/ |
| 191 |
public function delete_network_alias() { |
| 192 |
return Elasticsearch::factory()->delete_network_alias( $this->get_network_alias() ); |
| 193 |
} |
| 194 |
|
| 195 |
/** |
| 196 |
* Create unique indexable network alias |
| 197 |
* |
| 198 |
* @param array $indexes Array of indexes. |
| 199 |
* @since 3.0 |
| 200 |
* @return boolean |
| 201 |
*/ |
| 202 |
public function create_network_alias( $indexes ) { |
| 203 |
return Elasticsearch::factory()->create_network_alias( $indexes, $this->get_network_alias() ); |
| 204 |
} |
| 205 |
|
| 206 |
/** |
| 207 |
* Delete an object within the indexable |
| 208 |
* |
| 209 |
* @param int $object_id Object to delete. |
| 210 |
* @param boolean $blocking Whether to issue blocking HTTP request or not. |
| 211 |
* @since 3.0 |
| 212 |
* @return boolean |
| 213 |
*/ |
| 214 |
public function delete( $object_id, $blocking = true ) { |
| 215 |
/** |
| 216 |
* Fires before object deletion |
| 217 |
* |
| 218 |
* @hook ep_delete_{indexable_slug} |
| 219 |
* @param {int} $object_id ID of object being deleted |
| 220 |
* @param {string} $indexable_slug The slug of the indexable type that is being deleted |
| 221 |
*/ |
| 222 |
do_action( 'ep_delete_' . $this->slug, $object_id, $this->slug ); |
| 223 |
|
| 224 |
return Elasticsearch::factory()->delete_document( $this->get_index_name(), $this->slug, $object_id, $blocking ); |
| 225 |
} |
| 226 |
|
| 227 |
/** |
| 228 |
* Get an object within the indexable |
| 229 |
* |
| 230 |
* @param int $object_id Object to get. |
| 231 |
* @since 3.0 |
| 232 |
* @return boolean|array |
| 233 |
*/ |
| 234 |
public function get( $object_id ) { |
| 235 |
return Elasticsearch::factory()->get_document( $this->get_index_name(), $this->slug, $object_id ); |
| 236 |
} |
| 237 |
|
| 238 |
/** |
| 239 |
* Get objects within the indexable |
| 240 |
* |
| 241 |
* @param int $object_ids Array of object ids to get. |
| 242 |
* @since 3.6.0 |
| 243 |
* @return boolean|array |
| 244 |
*/ |
| 245 |
public function multi_get( $object_ids ) { |
| 246 |
return Elasticsearch::factory()->get_documents( $this->get_index_name(), $this->slug, $object_ids ); |
| 247 |
} |
| 248 |
|
| 249 |
/** |
| 250 |
* Delete an index within the indexable |
| 251 |
* |
| 252 |
* @param int $blog_id `null` means current blog. |
| 253 |
* @since 3.0 |
| 254 |
* @return boolean |
| 255 |
*/ |
| 256 |
public function delete_index( $blog_id = null ) { |
| 257 |
return Elasticsearch::factory()->delete_index( $this->get_index_name( $blog_id ) ); |
| 258 |
} |
| 259 |
|
| 260 |
/** |
| 261 |
* Index an object within the indexable. This calls prepare_document |
| 262 |
* |
| 263 |
* @param int $object_id Object to index. |
| 264 |
* @param boolean $blocking Blocking HTTP request or not. |
| 265 |
* @since 3.0 |
| 266 |
* @return object|boolean |
| 267 |
*/ |
| 268 |
public function index( $object_id, $blocking = false ) { |
| 269 |
try { |
| 270 |
$document = $this->prepare_document( $object_id ); |
| 271 |
} catch ( \Throwable $th ) { |
| 272 |
return false; |
| 273 |
} |
| 274 |
|
| 275 |
if ( false === $document ) { |
| 276 |
return false; |
| 277 |
} |
| 278 |
|
| 279 |
/** |
| 280 |
* Conditionally kill indexing on a specific object |
| 281 |
* |
| 282 |
* @deprecated 5.3.3 Use ep_{indexable_slug}_sync_kill instead |
| 283 |
* @hook ep_{indexable_slug}_index_kill |
| 284 |
* @param {bool} $kill True to not index |
| 285 |
* @param {int} $object_id Id of object to index |
| 286 |
* @since 3.0 |
| 287 |
* @return {bool} New kill value |
| 288 |
*/ |
| 289 |
if ( apply_filters_deprecated( 'ep_' . $this->slug . '_index_kill', [ false, $object_id ], 'ElasticPress 5.3.3', 'ep_' . $this->slug . '_sync_kill' ) ) { |
| 290 |
return false; |
| 291 |
} |
| 292 |
|
| 293 |
/** |
| 294 |
* Conditionally kill indexing for an object. |
| 295 |
* |
| 296 |
* @hook ep_{$this->slug}_sync_kill |
| 297 |
* @param {bool} $kill True means dont sync |
| 298 |
* @param {int} $object_id Object ID |
| 299 |
* @return {bool} New value |
| 300 |
*/ |
| 301 |
$ep_indexable_sync_kill = apply_filters( 'ep_' . $this->slug . '_sync_kill', false, $object_id ); |
| 302 |
if ( $ep_indexable_sync_kill ) { |
| 303 |
return false; |
| 304 |
} |
| 305 |
|
| 306 |
/** |
| 307 |
* Filter document before index |
| 308 |
* |
| 309 |
* @hook ep_pre_index_{indexable_slug} |
| 310 |
* @param {array} $document Document to index |
| 311 |
* @return {array} New document |
| 312 |
* @since 3.0 |
| 313 |
*/ |
| 314 |
$document = apply_filters( 'ep_pre_index_' . $this->slug, $document ); |
| 315 |
|
| 316 |
$return = Elasticsearch::factory()->index_document( $this->get_index_name(), $this->slug, $document, $blocking ); |
| 317 |
|
| 318 |
/** |
| 319 |
* Fires after document is indexed |
| 320 |
* |
| 321 |
* @hook ep_after_index_{indexable_slug} |
| 322 |
* @param {array} $document Document to index |
| 323 |
* @param {object|boolean} $return ES response on success, false on failure |
| 324 |
* @since 3.0 |
| 325 |
*/ |
| 326 |
do_action( 'ep_after_index_' . $this->slug, $document, $return ); |
| 327 |
|
| 328 |
return $return; |
| 329 |
} |
| 330 |
|
| 331 |
/** |
| 332 |
* Determine if indexable index exists |
| 333 |
* |
| 334 |
* @param int $blog_id Blog to check index for. |
| 335 |
* @since 3.0 |
| 336 |
* @return boolean |
| 337 |
*/ |
| 338 |
public function index_exists( $blog_id = null ) { |
| 339 |
return Elasticsearch::factory()->index_exists( $this->get_index_name( $blog_id ) ); |
| 340 |
} |
| 341 |
|
| 342 |
/** |
| 343 |
* Bulk index objects. This calls prepare_document on each object |
| 344 |
* |
| 345 |
* @param array $object_ids Array of object IDs. |
| 346 |
* @since 3.0 |
| 347 |
* @return WP_Error|array |
| 348 |
*/ |
| 349 |
public function bulk_index( $object_ids ) { |
| 350 |
$body = ''; |
| 351 |
|
| 352 |
$non_es_errors = []; |
| 353 |
|
| 354 |
foreach ( $object_ids as $object_id ) { |
| 355 |
$action_args = array( |
| 356 |
'index' => array( |
| 357 |
'_id' => absint( $object_id ), |
| 358 |
), |
| 359 |
); |
| 360 |
|
| 361 |
try { |
| 362 |
$document = $this->prepare_document( $object_id ); |
| 363 |
} catch ( \Throwable $th ) { |
| 364 |
$non_es_errors[] = [ |
| 365 |
'index' => [ |
| 366 |
'_id' => absint( $object_id ), |
| 367 |
'error' => [ |
| 368 |
'type' => 'prepare_document_error', |
| 369 |
'reason' => $th->getMessage(), |
| 370 |
], |
| 371 |
], |
| 372 |
]; |
| 373 |
continue; |
| 374 |
} |
| 375 |
|
| 376 |
/** |
| 377 |
* Conditionally kill indexing on a specific object |
| 378 |
* |
| 379 |
* @hook ep_bulk_index_action_args |
| 380 |
* @param {array} $action_args Bulk action arguments |
| 381 |
* @param {array} $document Document to index |
| 382 |
* @since 3.0 |
| 383 |
* @return {array} New action args |
| 384 |
*/ |
| 385 |
$body .= wp_json_encode( apply_filters( 'ep_bulk_index_action_args', $action_args, $document ) ) . "\n"; |
| 386 |
$body .= addcslashes( wp_json_encode( $document ), "\n" ); |
| 387 |
|
| 388 |
$body .= "\n\n"; |
| 389 |
} |
| 390 |
|
| 391 |
$result = Elasticsearch::factory()->bulk_index( $this->get_index_name(), $this->slug, $body ); |
| 392 |
|
| 393 |
if ( ! empty( $non_es_errors ) ) { |
| 394 |
$result['errors'] = true; |
| 395 |
$result['items'] = isset( $result['items'] ) ? array_merge( $result['items'], $non_es_errors ) : $non_es_errors; |
| 396 |
} |
| 397 |
|
| 398 |
/** |
| 399 |
* Perform actions after a bulk indexing is completed |
| 400 |
* |
| 401 |
* @hook ep_after_bulk_index |
| 402 |
* @param {array} $object_ids List of object ids attempted to be indexed |
| 403 |
* @param {string} $slug Current indexable slug |
| 404 |
* @param {array|bool} $result Result of the Elasticsearch query. False on error. |
| 405 |
*/ |
| 406 |
do_action( 'ep_after_bulk_index', $object_ids, $this->slug, $result ); |
| 407 |
|
| 408 |
return $result; |
| 409 |
} |
| 410 |
|
| 411 |
/** |
| 412 |
* Bulk index objects but with a dynamic size of queue. |
| 413 |
* |
| 414 |
* @since 4.0.0 |
| 415 |
* @param array $object_ids Array of object IDs. |
| 416 |
* @return array[WP_Error|array] The return of each request made. |
| 417 |
*/ |
| 418 |
public function bulk_index_dynamically( $object_ids ) { |
| 419 |
$documents = []; |
| 420 |
|
| 421 |
$non_es_errors = []; |
| 422 |
|
| 423 |
foreach ( $object_ids as $object_id ) { |
| 424 |
$action_args = array( |
| 425 |
'index' => array( |
| 426 |
'_id' => absint( $object_id ), |
| 427 |
), |
| 428 |
); |
| 429 |
|
| 430 |
try { |
| 431 |
$document = $this->prepare_document( $object_id ); |
| 432 |
} catch ( \Throwable $th ) { |
| 433 |
$non_es_errors[] = [ |
| 434 |
'index' => [ |
| 435 |
'_id' => absint( $object_id ), |
| 436 |
'error' => [ |
| 437 |
'type' => 'prepare_document_error', |
| 438 |
'reason' => $th->getMessage(), |
| 439 |
], |
| 440 |
], |
| 441 |
]; |
| 442 |
continue; |
| 443 |
} |
| 444 |
|
| 445 |
if ( empty( $document ) ) { |
| 446 |
continue; |
| 447 |
} |
| 448 |
|
| 449 |
/** |
| 450 |
* Conditionally kill indexing on a specific object |
| 451 |
* |
| 452 |
* @hook ep_bulk_index_action_args |
| 453 |
* @param {array} $action_args Bulk action arguments |
| 454 |
* @param {array} $document Document to index |
| 455 |
* @since 3.0 |
| 456 |
* @return {array} New action args |
| 457 |
*/ |
| 458 |
$document_str = wp_json_encode( apply_filters( 'ep_bulk_index_action_args', $action_args, $document ) ) . "\n"; |
| 459 |
$document_str .= addcslashes( wp_json_encode( $document ), "\n" ); |
| 460 |
$document_str .= "\n\n"; |
| 461 |
|
| 462 |
$documents[] = $document_str; |
| 463 |
} |
| 464 |
|
| 465 |
if ( empty( $documents ) ) { |
| 466 |
return ( ! empty( $non_es_errors ) ? [ |
| 467 |
[ |
| 468 |
'errors' => true, |
| 469 |
'items' => $non_es_errors, |
| 470 |
], |
| 471 |
] : [ |
| 472 |
new \WP_Error( |
| 473 |
'ep_bulk_index_no_documents', |
| 474 |
esc_html__( 'It was not possible to create a body request with the document IDs provided.', 'elasticpress' ), |
| 475 |
$object_ids |
| 476 |
), |
| 477 |
] ); |
| 478 |
} |
| 479 |
|
| 480 |
$results = $this->send_bulk_index_request( $documents ); |
| 481 |
|
| 482 |
if ( ! empty( $non_es_errors ) ) { |
| 483 |
$results = [ |
| 484 |
[ |
| 485 |
'errors' => true, |
| 486 |
'items' => $non_es_errors, |
| 487 |
], |
| 488 |
...$results, |
| 489 |
]; |
| 490 |
} |
| 491 |
|
| 492 |
/** |
| 493 |
* Perform actions after a dynamic bulk indexing is completed |
| 494 |
* |
| 495 |
* @hook ep_after_bulk_index_dynamically |
| 496 |
* @since 4.0.0 |
| 497 |
* @param {array} $object_ids List of object ids attempted to be indexed |
| 498 |
* @param {string} $slug Current indexable slug |
| 499 |
* @param {array|bool} $result Result of the Elasticsearch query. False on error. |
| 500 |
*/ |
| 501 |
do_action( 'ep_after_bulk_index_dynamically', $object_ids, $this->slug, $results ); |
| 502 |
|
| 503 |
return $results; |
| 504 |
} |
| 505 |
|
| 506 |
/** |
| 507 |
* Bulk index documents through several requests with dynamic size. |
| 508 |
* |
| 509 |
* @param array $documents The documents to be sent to Elasticsearch (already formatted.) |
| 510 |
* @return array[WP_Error|array] |
| 511 |
*/ |
| 512 |
protected function send_bulk_index_request( $documents ) { |
| 513 |
static $min_buffer_size, $max_buffer_size, $current_buffer_size, $incremental_step; |
| 514 |
|
| 515 |
if ( ! $min_buffer_size ) { |
| 516 |
/** |
| 517 |
* Filter the minimum buffer size for dynamic bulk index requests. |
| 518 |
* |
| 519 |
* @hook ep_dynamic_bulk_min_buffer_size |
| 520 |
* @since 4.0.0 |
| 521 |
* @param {int} $min_buffer_size Min buffer size for dynamic bulk index (in bytes.) |
| 522 |
* @return {int} New size. |
| 523 |
*/ |
| 524 |
$min_buffer_size = apply_filters( 'ep_dynamic_bulk_min_buffer_size', MB_IN_BYTES / 2 ); |
| 525 |
} |
| 526 |
|
| 527 |
if ( ! $max_buffer_size ) { |
| 528 |
/** |
| 529 |
* Filter the max buffer size for dynamic bulk index requests. |
| 530 |
* |
| 531 |
* @hook ep_dynamic_bulk_max_buffer_size |
| 532 |
* @since 4.0.0 |
| 533 |
* @param {int} $max_buffer_size Max buffer size for dynamic bulk index (in bytes.) |
| 534 |
* @return {int} New size. |
| 535 |
*/ |
| 536 |
$max_buffer_size = apply_filters( 'ep_dynamic_bulk_max_buffer_size', 150 * MB_IN_BYTES ); |
| 537 |
} |
| 538 |
|
| 539 |
if ( ! $incremental_step ) { |
| 540 |
/** |
| 541 |
* Filter the number of bytes the current buffer size should be incremented in case of success. |
| 542 |
* |
| 543 |
* @hook ep_dynamic_bulk_incremental_step |
| 544 |
* @since 4.0.0 |
| 545 |
* @param {int} $incremental_step Number of bytes to add to the current buffer size. |
| 546 |
* @return {int} New incremental step. |
| 547 |
*/ |
| 548 |
$incremental_step = apply_filters( 'ep_dynamic_bulk_incremental_step', MB_IN_BYTES / 2 ); |
| 549 |
} |
| 550 |
|
| 551 |
/** |
| 552 |
* Perform actions before a new batch of documents is processed. |
| 553 |
* |
| 554 |
* @hook ep_before_send_dynamic_bulk_requests |
| 555 |
* @since 4.0.0 |
| 556 |
* @param {array} $documents Array of documents to be sent to Elasticsearch. |
| 557 |
*/ |
| 558 |
do_action( 'ep_before_send_dynamic_bulk_requests', $documents ); |
| 559 |
|
| 560 |
if ( ! $current_buffer_size ) { |
| 561 |
$current_buffer_size = $min_buffer_size; |
| 562 |
} |
| 563 |
|
| 564 |
$results = []; |
| 565 |
|
| 566 |
$body = []; |
| 567 |
$current_body_size = 0; |
| 568 |
$requests = 0; |
| 569 |
|
| 570 |
/* |
| 571 |
* This script will use two main arrays: $body and $documents, being $body the |
| 572 |
* documents to be sent in the next request and $documents the list of docs to be indexed. |
| 573 |
* The do-while loop will stop if all documents are sent or if a request fails even sending |
| 574 |
* a buffer as small as possible. |
| 575 |
*/ |
| 576 |
do { |
| 577 |
$next_document = array_shift( $documents ); |
| 578 |
$next_document_size = mb_strlen( $next_document ); |
| 579 |
$has_buffered_documents = count( $body ) > 0; |
| 580 |
|
| 581 |
// If the next document alone takes the entire current buffer size, |
| 582 |
// let's add it back to the pipe and send what we have first |
| 583 |
if ( $next_document_size > $current_buffer_size && $has_buffered_documents ) { |
| 584 |
array_unshift( $documents, $next_document ); |
| 585 |
} else { |
| 586 |
if ( $next_document_size > $max_buffer_size ) { |
| 587 |
/** |
| 588 |
* Perform actions when a post is bigger than the max buffer size. |
| 589 |
* |
| 590 |
* @hook ep_dynamic_bulk_post_too_big |
| 591 |
* @since 4.0.0 |
| 592 |
* @param {string} $document JSON string of the post detected as too big. |
| 593 |
*/ |
| 594 |
do_action( 'ep_dynamic_bulk_post_too_big', $next_document ); |
| 595 |
$results[] = new \WP_Error( 'ep_too_big_request_skipped', 'Indexable too big. Request not sent.' ); |
| 596 |
continue; |
| 597 |
} |
| 598 |
|
| 599 |
$body[] = $next_document; |
| 600 |
$current_body_size += $next_document_size; |
| 601 |
|
| 602 |
$can_add_more_documents = ( $current_body_size < $current_buffer_size && ! empty( $documents ) ); |
| 603 |
if ( $can_add_more_documents ) { |
| 604 |
continue; |
| 605 |
} |
| 606 |
|
| 607 |
if ( $current_body_size > $max_buffer_size ) { |
| 608 |
// The last document added to body made it too big, so let's give it back. |
| 609 |
$removed_document = array_pop( $body ); |
| 610 |
$current_body_size -= mb_strlen( $removed_document ); |
| 611 |
array_unshift( $documents, $removed_document ); |
| 612 |
} |
| 613 |
} |
| 614 |
|
| 615 |
// Try the request. |
| 616 |
timer_start(); |
| 617 |
$result = Elasticsearch::factory()->bulk_index( $this->get_index_name(), $this->slug, implode( '', $body ) ); |
| 618 |
$request_time = timer_stop(); |
| 619 |
++$requests; |
| 620 |
|
| 621 |
/** |
| 622 |
* Perform actions before a new batch of documents is processed. |
| 623 |
* |
| 624 |
* @hook ep_after_send_dynamic_bulk_request |
| 625 |
* @since 4.0.0 |
| 626 |
* @param {WP_Error|array} $result Result of the request. |
| 627 |
* @param {array} $body Array of documents sent to Elasticsearch. |
| 628 |
* @param {array} $documents Array of documents to be sent to Elasticsearch. |
| 629 |
* @param {int} $min_buffer_size Min buffer size for dynamic bulk index (in bytes.) |
| 630 |
* @param {int} $max_buffer_size Max buffer size for dynamic bulk index (in bytes.) |
| 631 |
* @param {int} $current_buffer_size Current buffer size for dynamic bulk index (in bytes.) |
| 632 |
* @param {int} $request_time Total time of the request. |
| 633 |
*/ |
| 634 |
do_action( 'ep_after_send_dynamic_bulk_request', $result, $body, $documents, $min_buffer_size, $max_buffer_size, $current_buffer_size, $request_time ); |
| 635 |
|
| 636 |
// It failed, possibly adjust the buffer size and try again. |
| 637 |
if ( is_wp_error( $result ) ) { |
| 638 |
$error_code = $result->get_error_code(); |
| 639 |
|
| 640 |
// Too many requests, wait and try again. |
| 641 |
if ( 429 === $error_code ) { |
| 642 |
sleep( 2 ); |
| 643 |
} |
| 644 |
|
| 645 |
// If the error is not a "Request too big" then we really fail this batch of documents. |
| 646 |
if ( 413 !== $error_code ) { |
| 647 |
$results[] = $result; |
| 648 |
continue; |
| 649 |
} |
| 650 |
|
| 651 |
if ( count( $body ) === 1 ) { |
| 652 |
$max_buffer_size = min( $max_buffer_size, $current_body_size ); |
| 653 |
$results[] = $result; |
| 654 |
$body = []; |
| 655 |
$current_body_size = 0; |
| 656 |
continue; |
| 657 |
} |
| 658 |
|
| 659 |
// As the buffer is as small as possible, return the error. |
| 660 |
if ( $current_body_size === $min_buffer_size ) { |
| 661 |
$results[] = $result; |
| 662 |
continue; |
| 663 |
} |
| 664 |
|
| 665 |
// We have a too big buffer. Remove one doc from the body, and set both max and current as its size. |
| 666 |
$removed_document = array_pop( $body ); |
| 667 |
$current_body_size -= mb_strlen( $removed_document ); |
| 668 |
array_unshift( $documents, $removed_document ); |
| 669 |
|
| 670 |
$max_buffer_size = count( $body ) ? |
| 671 |
max( $min_buffer_size, $current_body_size ) : |
| 672 |
$min_buffer_size; |
| 673 |
|
| 674 |
$current_buffer_size = $max_buffer_size; |
| 675 |
continue; |
| 676 |
} |
| 677 |
|
| 678 |
// Things worked so we can try to bump the buffer size. |
| 679 |
if ( $current_buffer_size < $max_buffer_size && $current_body_size > $current_buffer_size ) { |
| 680 |
$current_buffer_size = min( ( $current_buffer_size + $incremental_step ), $max_buffer_size ); |
| 681 |
} |
| 682 |
|
| 683 |
$results[] = $result; |
| 684 |
|
| 685 |
$body = []; |
| 686 |
$current_body_size = 0; |
| 687 |
} while ( ! empty( $documents ) ); |
| 688 |
|
| 689 |
/** |
| 690 |
* Perform actions after a batch of documents was processed. |
| 691 |
* |
| 692 |
* @hook ep_after_send_dynamic_bulk_requests |
| 693 |
* @since 4.0.0 |
| 694 |
* @param {array} $results Array of results sent. |
| 695 |
* @param {int} $requests Number of all requests sent. |
| 696 |
*/ |
| 697 |
do_action( 'ep_after_send_dynamic_bulk_requests', $results, $requests ); |
| 698 |
|
| 699 |
return $results; |
| 700 |
} |
| 701 |
|
| 702 |
/** |
| 703 |
* Query Elasticsearch for documents |
| 704 |
* |
| 705 |
* @param array $formatted_args Formatted es query arguments. |
| 706 |
* @param array $query_args WP_Query args. |
| 707 |
* @param string $index Index(es) to query. Comma separate for multiple. Defaults to current. |
| 708 |
* @param mixed $query_object Could be WP_Query, WP_User_Query, etc. |
| 709 |
* @since 3.0 |
| 710 |
* @return array |
| 711 |
*/ |
| 712 |
public function query_es( $formatted_args, $query_args, $index = null, $query_object = null ) { |
| 713 |
if ( null === $index ) { |
| 714 |
$index = $this->get_index_name(); |
| 715 |
} |
| 716 |
|
| 717 |
return Elasticsearch::factory()->query( $index, $this->slug, $formatted_args, $query_args, $query_object ); |
| 718 |
} |
| 719 |
|
| 720 |
/** |
| 721 |
* Check to see if we should allow elasticpress to override this query |
| 722 |
* |
| 723 |
* @param \WP_Query|\WP_User_Query|\WP_Term_Query $query WP_Query or WP_User_Query or WP_Term_Query instance |
| 724 |
* @return bool |
| 725 |
* @since 3.0 |
| 726 |
*/ |
| 727 |
public function elasticpress_enabled( $query ) { |
| 728 |
$enabled = false; |
| 729 |
|
| 730 |
if ( ! empty( $query->query_vars['ep_integrate'] ) ) { |
| 731 |
$enabled = true; |
| 732 |
} |
| 733 |
|
| 734 |
/** |
| 735 |
* Determine if ElasticPress should integrate with a query |
| 736 |
* |
| 737 |
* @hook ep_elasticpress_enabled |
| 738 |
* @param {bool} $enabled Whether to integrate with Elasticsearch or not |
| 739 |
* @param {WP_Query} $query WP_Query to evaluate |
| 740 |
* @return {bool} Enabled value |
| 741 |
*/ |
| 742 |
$enabled = apply_filters( 'ep_elasticpress_enabled', $enabled, $query ); |
| 743 |
|
| 744 |
if ( isset( $query->query_vars['ep_integrate'] ) && ! filter_var( $query->query_vars['ep_integrate'], FILTER_VALIDATE_BOOLEAN ) ) { |
| 745 |
$enabled = false; |
| 746 |
} |
| 747 |
|
| 748 |
return $enabled; |
| 749 |
} |
| 750 |
|
| 751 |
/** |
| 752 |
* Prepare meta type values to send to ES |
| 753 |
* |
| 754 |
* @param array $meta Array of meta. |
| 755 |
* @since 3.0 |
| 756 |
* @return array |
| 757 |
*/ |
| 758 |
public function prepare_meta_types( $meta ) { |
| 759 |
|
| 760 |
$prepared_meta = []; |
| 761 |
|
| 762 |
foreach ( $meta as $meta_key => $meta_values ) { |
| 763 |
if ( ! is_array( $meta_values ) ) { |
| 764 |
$meta_values = array( $meta_values ); |
| 765 |
} |
| 766 |
|
| 767 |
$prepared_meta[ $meta_key ] = array_map( array( $this, 'prepare_meta_value_types' ), $meta_values ); |
| 768 |
} |
| 769 |
|
| 770 |
return $prepared_meta; |
| 771 |
} |
| 772 |
|
| 773 |
/** |
| 774 |
* Prepare meta types for meta value |
| 775 |
* |
| 776 |
* @param mixed $meta_value Meta value to prepare. |
| 777 |
* @since 3.0 |
| 778 |
* @return array |
| 779 |
*/ |
| 780 |
public function prepare_meta_value_types( $meta_value ) { |
| 781 |
|
| 782 |
$max_java_int_value = PHP_INT_MAX; |
| 783 |
|
| 784 |
$meta_types = []; |
| 785 |
|
| 786 |
if ( is_array( $meta_value ) || is_object( $meta_value ) ) { |
| 787 |
$meta_value = serialize( $meta_value ); // phpcs:ignore WordPress.PHP.DiscouragedPHPFunctions.serialize_serialize |
| 788 |
} |
| 789 |
|
| 790 |
$meta_types['value'] = $meta_value; |
| 791 |
$meta_types['raw'] = $meta_value; |
| 792 |
|
| 793 |
if ( is_numeric( $meta_value ) ) { |
| 794 |
$long = intval( $meta_value ); |
| 795 |
|
| 796 |
if ( $max_java_int_value < $long ) { |
| 797 |
$long = $max_java_int_value; |
| 798 |
} |
| 799 |
|
| 800 |
$double = floatval( $meta_value ); |
| 801 |
|
| 802 |
if ( ! is_finite( $double ) ) { |
| 803 |
$double = 0; |
| 804 |
} |
| 805 |
|
| 806 |
$meta_types['long'] = $long; |
| 807 |
$meta_types['double'] = $double; |
| 808 |
} |
| 809 |
|
| 810 |
$meta_types['boolean'] = filter_var( $meta_value, FILTER_VALIDATE_BOOLEAN ); |
| 811 |
|
| 812 |
$meta_types = $this->prepare_date_meta_values( $meta_types, $meta_value ); |
| 813 |
|
| 814 |
return $meta_types; |
| 815 |
} |
| 816 |
|
| 817 |
/** |
| 818 |
* Checks if a meta_value is a valid date and prepare extra meta-data. |
| 819 |
* |
| 820 |
* @param array $meta_types Array of currently prepared data |
| 821 |
* @param string $meta_value Meta value to prepare. |
| 822 |
* |
| 823 |
* @return array |
| 824 |
*/ |
| 825 |
public function prepare_date_meta_values( $meta_types, $meta_value ) { |
| 826 |
|
| 827 |
if ( empty( $meta_value ) || ! is_string( $meta_value ) ) { |
| 828 |
return $meta_types; |
| 829 |
} |
| 830 |
|
| 831 |
$meta_types['date'] = '1970-01-01'; |
| 832 |
$meta_types['datetime'] = '1970-01-01 00:00:01'; |
| 833 |
$meta_types['time'] = '00:00:01'; |
| 834 |
|
| 835 |
// is this is a recognizable date format? |
| 836 |
$new_date = date_create( $meta_value, \wp_timezone() ); |
| 837 |
if ( $new_date ) { |
| 838 |
$timestamp = $new_date->getTimestamp(); |
| 839 |
|
| 840 |
/** |
| 841 |
* Filter the maximum year limit for date conversion. |
| 842 |
* |
| 843 |
* Use default date if year is greater than max limit. EP has limitation that doesn't allow to have year greater than 2099. |
| 844 |
* |
| 845 |
* @see https://github.com/10up/ElasticPress/issues/2769 |
| 846 |
* |
| 847 |
* @hook ep_max_year_limit |
| 848 |
* @param {int} $year Maximum year limit. |
| 849 |
* @return {int} Maximum year limit. |
| 850 |
* @since 4.2.1 |
| 851 |
*/ |
| 852 |
$max_year = apply_filters( 'ep_max_year_limit', 2099 ); |
| 853 |
|
| 854 |
// PHP allows DateTime to build dates with the non-existing year 0000, and this causes |
| 855 |
// issues when integrating into stricter systems. This is by design: |
| 856 |
// https://bugs.php.net/bug.php?id=60288 |
| 857 |
if ( false !== $timestamp && '0000' !== $new_date->format( 'Y' ) && $new_date->format( 'Y' ) <= $max_year ) { |
| 858 |
$meta_types['date'] = $new_date->format( 'Y-m-d' ); |
| 859 |
$meta_types['datetime'] = $new_date->format( 'Y-m-d H:i:s' ); |
| 860 |
$meta_types['time'] = $new_date->format( 'H:i:s' ); |
| 861 |
} |
| 862 |
} |
| 863 |
|
| 864 |
return $meta_types; |
| 865 |
} |
| 866 |
|
| 867 |
/** |
| 868 |
* Build Elasticsearch filter query for WP meta_query |
| 869 |
* |
| 870 |
* @since 2.2 |
| 871 |
* @param array $meta_queries Array of queries |
| 872 |
* @return array |
| 873 |
*/ |
| 874 |
public function build_meta_query( $meta_queries ) { |
| 875 |
$meta_filter = []; |
| 876 |
|
| 877 |
$outer_relation = 'must'; |
| 878 |
if ( ! empty( $meta_queries['relation'] ) && 'or' === strtolower( $meta_queries['relation'] ) ) { |
| 879 |
$outer_relation = 'should'; |
| 880 |
} |
| 881 |
|
| 882 |
$meta_query_type_mapping = [ |
| 883 |
'numeric' => 'long', |
| 884 |
'binary' => 'raw', |
| 885 |
'char' => 'raw', |
| 886 |
'date' => 'date', |
| 887 |
'datetime' => 'datetime', |
| 888 |
'decimal' => 'double', |
| 889 |
'signed' => 'long', |
| 890 |
'time' => 'time', |
| 891 |
'unsigned' => 'long', |
| 892 |
]; |
| 893 |
|
| 894 |
foreach ( $meta_queries as $single_meta_query ) { |
| 895 |
if ( ! empty( $single_meta_query['key'] ) ) { |
| 896 |
|
| 897 |
$terms_obj = false; |
| 898 |
|
| 899 |
$compare = '='; |
| 900 |
if ( ! empty( $single_meta_query['compare'] ) ) { |
| 901 |
$compare = strtolower( $single_meta_query['compare'] ); |
| 902 |
} elseif ( ! isset( $single_meta_query['value'] ) ) { |
| 903 |
$compare = 'exists'; |
| 904 |
} |
| 905 |
|
| 906 |
$type = null; |
| 907 |
if ( ! empty( $single_meta_query['type'] ) ) { |
| 908 |
$type = strtolower( $single_meta_query['type'] ); |
| 909 |
} |
| 910 |
|
| 911 |
// Comparisons need to look at different paths |
| 912 |
if ( in_array( $compare, array( 'exists', 'not exists' ), true ) ) { |
| 913 |
$meta_key_path = 'meta.' . $single_meta_query['key']; |
| 914 |
} elseif ( in_array( $compare, array( '=', '!=' ), true ) && ! $type ) { |
| 915 |
$meta_key_path = 'meta.' . $single_meta_query['key'] . '.raw'; |
| 916 |
} elseif ( in_array( $compare, array( 'like', 'not like' ), true ) ) { |
| 917 |
$meta_key_path = 'meta.' . $single_meta_query['key'] . '.value'; |
| 918 |
} elseif ( $type && isset( $meta_query_type_mapping[ $type ] ) ) { |
| 919 |
// Map specific meta field types to different Elasticsearch core types |
| 920 |
$meta_key_path = 'meta.' . $single_meta_query['key'] . '.' . $meta_query_type_mapping[ $type ]; |
| 921 |
} elseif ( in_array( $compare, array( '>=', '<=', '>', '<', 'between', 'not between' ), true ) ) { |
| 922 |
$meta_key_path = 'meta.' . $single_meta_query['key'] . '.double'; |
| 923 |
} else { |
| 924 |
$meta_key_path = 'meta.' . $single_meta_query['key'] . '.raw'; |
| 925 |
} |
| 926 |
|
| 927 |
switch ( $compare ) { |
| 928 |
case 'not in': |
| 929 |
case '!=': |
| 930 |
if ( isset( $single_meta_query['value'] ) ) { |
| 931 |
$terms_obj = array( |
| 932 |
'bool' => array( |
| 933 |
'must_not' => array( |
| 934 |
array( |
| 935 |
'terms' => array( |
| 936 |
$meta_key_path => (array) $single_meta_query['value'], |
| 937 |
), |
| 938 |
), |
| 939 |
), |
| 940 |
), |
| 941 |
); |
| 942 |
} |
| 943 |
|
| 944 |
break; |
| 945 |
case 'exists': |
| 946 |
$terms_obj = array( |
| 947 |
'exists' => array( |
| 948 |
'field' => $meta_key_path, |
| 949 |
), |
| 950 |
); |
| 951 |
|
| 952 |
break; |
| 953 |
case 'not exists': |
| 954 |
$terms_obj = array( |
| 955 |
'bool' => array( |
| 956 |
'must_not' => array( |
| 957 |
array( |
| 958 |
'exists' => array( |
| 959 |
'field' => $meta_key_path, |
| 960 |
), |
| 961 |
), |
| 962 |
), |
| 963 |
), |
| 964 |
); |
| 965 |
|
| 966 |
break; |
| 967 |
case '>=': |
| 968 |
if ( isset( $single_meta_query['value'] ) ) { |
| 969 |
$terms_obj = array( |
| 970 |
'bool' => array( |
| 971 |
'must' => array( |
| 972 |
array( |
| 973 |
'range' => array( |
| 974 |
$meta_key_path => array( |
| 975 |
'gte' => $single_meta_query['value'], |
| 976 |
), |
| 977 |
), |
| 978 |
), |
| 979 |
), |
| 980 |
), |
| 981 |
); |
| 982 |
} |
| 983 |
|
| 984 |
break; |
| 985 |
case 'between': |
| 986 |
if ( isset( $single_meta_query['value'] ) && is_array( $single_meta_query['value'] ) && 2 === count( $single_meta_query['value'] ) ) { |
| 987 |
$terms_obj = array( |
| 988 |
'bool' => array( |
| 989 |
'must' => array( |
| 990 |
array( |
| 991 |
'range' => array( |
| 992 |
$meta_key_path => array( |
| 993 |
'gte' => $single_meta_query['value'][0], |
| 994 |
), |
| 995 |
), |
| 996 |
), |
| 997 |
array( |
| 998 |
'range' => array( |
| 999 |
$meta_key_path => array( |
| 1000 |
'lte' => $single_meta_query['value'][1], |
| 1001 |
), |
| 1002 |
), |
| 1003 |
), |
| 1004 |
), |
| 1005 |
), |
| 1006 |
); |
| 1007 |
} |
| 1008 |
|
| 1009 |
break; |
| 1010 |
case 'not between': |
| 1011 |
if ( isset( $single_meta_query['value'] ) && is_array( $single_meta_query['value'] ) && 2 === count( $single_meta_query['value'] ) ) { |
| 1012 |
$terms_obj = array( |
| 1013 |
'bool' => array( |
| 1014 |
'should' => array( |
| 1015 |
array( |
| 1016 |
'range' => array( |
| 1017 |
$meta_key_path => array( |
| 1018 |
'lte' => $single_meta_query['value'][0], |
| 1019 |
), |
| 1020 |
), |
| 1021 |
), |
| 1022 |
array( |
| 1023 |
'range' => array( |
| 1024 |
$meta_key_path => array( |
| 1025 |
'gte' => $single_meta_query['value'][1], |
| 1026 |
), |
| 1027 |
), |
| 1028 |
), |
| 1029 |
), |
| 1030 |
), |
| 1031 |
); |
| 1032 |
} |
| 1033 |
|
| 1034 |
break; |
| 1035 |
case '<=': |
| 1036 |
if ( isset( $single_meta_query['value'] ) ) { |
| 1037 |
$terms_obj = array( |
| 1038 |
'bool' => array( |
| 1039 |
'must' => array( |
| 1040 |
array( |
| 1041 |
'range' => array( |
| 1042 |
$meta_key_path => array( |
| 1043 |
'lte' => $single_meta_query['value'], |
| 1044 |
), |
| 1045 |
), |
| 1046 |
), |
| 1047 |
), |
| 1048 |
), |
| 1049 |
); |
| 1050 |
} |
| 1051 |
|
| 1052 |
break; |
| 1053 |
case '>': |
| 1054 |
if ( isset( $single_meta_query['value'] ) ) { |
| 1055 |
$terms_obj = array( |
| 1056 |
'bool' => array( |
| 1057 |
'must' => array( |
| 1058 |
array( |
| 1059 |
'range' => array( |
| 1060 |
$meta_key_path => array( |
| 1061 |
'gt' => $single_meta_query['value'], |
| 1062 |
), |
| 1063 |
), |
| 1064 |
), |
| 1065 |
), |
| 1066 |
), |
| 1067 |
); |
| 1068 |
} |
| 1069 |
|
| 1070 |
break; |
| 1071 |
case '<': |
| 1072 |
if ( isset( $single_meta_query['value'] ) ) { |
| 1073 |
$terms_obj = array( |
| 1074 |
'bool' => array( |
| 1075 |
'must' => array( |
| 1076 |
array( |
| 1077 |
'range' => array( |
| 1078 |
$meta_key_path => array( |
| 1079 |
'lt' => $single_meta_query['value'], |
| 1080 |
), |
| 1081 |
), |
| 1082 |
), |
| 1083 |
), |
| 1084 |
), |
| 1085 |
); |
| 1086 |
} |
| 1087 |
|
| 1088 |
break; |
| 1089 |
case 'like': |
| 1090 |
if ( isset( $single_meta_query['value'] ) ) { |
| 1091 |
$terms_obj = array( |
| 1092 |
'match_phrase' => array( |
| 1093 |
$meta_key_path => $single_meta_query['value'], |
| 1094 |
), |
| 1095 |
); |
| 1096 |
} |
| 1097 |
break; |
| 1098 |
case 'not like': |
| 1099 |
if ( isset( $single_meta_query['value'] ) ) { |
| 1100 |
$terms_obj = array( |
| 1101 |
'bool' => array( |
| 1102 |
'must_not' => array( |
| 1103 |
array( |
| 1104 |
'match_phrase' => array( |
| 1105 |
$meta_key_path => $single_meta_query['value'], |
| 1106 |
), |
| 1107 |
), |
| 1108 |
), |
| 1109 |
), |
| 1110 |
); |
| 1111 |
} |
| 1112 |
break; |
| 1113 |
case '=': |
| 1114 |
default: |
| 1115 |
if ( isset( $single_meta_query['value'] ) ) { |
| 1116 |
$terms_obj = array( |
| 1117 |
'terms' => array( |
| 1118 |
$meta_key_path => (array) $single_meta_query['value'], |
| 1119 |
), |
| 1120 |
); |
| 1121 |
} |
| 1122 |
|
| 1123 |
break; |
| 1124 |
} |
| 1125 |
|
| 1126 |
// Add the meta query filter |
| 1127 |
if ( false !== $terms_obj ) { |
| 1128 |
$meta_filter[] = $terms_obj; |
| 1129 |
} |
| 1130 |
} elseif ( is_array( $single_meta_query ) ) { |
| 1131 |
/** |
| 1132 |
* Handle multidimensional array. Something like: |
| 1133 |
* |
| 1134 |
* 'meta_query' => array( |
| 1135 |
* 'relation' => 'AND', |
| 1136 |
* array( |
| 1137 |
* 'key' => 'meta_key_1', |
| 1138 |
* 'value' => '1', |
| 1139 |
* ), |
| 1140 |
* array( |
| 1141 |
* 'relation' => 'OR', |
| 1142 |
* array( |
| 1143 |
* 'key' => 'meta_key_2', |
| 1144 |
* 'value' => '2', |
| 1145 |
* ), |
| 1146 |
* array( |
| 1147 |
* 'key' => 'meta_key_3', |
| 1148 |
* 'value' => '4', |
| 1149 |
* ), |
| 1150 |
* ), |
| 1151 |
* ), |
| 1152 |
*/ |
| 1153 |
$inner_relation = 'must'; |
| 1154 |
if ( ! empty( $single_meta_query['relation'] ) && 'or' === strtolower( $single_meta_query['relation'] ) ) { |
| 1155 |
$inner_relation = 'should'; |
| 1156 |
} |
| 1157 |
|
| 1158 |
$meta_filter[] = array( |
| 1159 |
'bool' => array( |
| 1160 |
$inner_relation => $this->build_meta_query( $single_meta_query ), |
| 1161 |
), |
| 1162 |
); |
| 1163 |
} |
| 1164 |
} |
| 1165 |
|
| 1166 |
if ( ! empty( $meta_filter ) ) { |
| 1167 |
return [ |
| 1168 |
'bool' => [ |
| 1169 |
$outer_relation => $meta_filter, |
| 1170 |
], |
| 1171 |
]; |
| 1172 |
} else { |
| 1173 |
return false; |
| 1174 |
} |
| 1175 |
} |
| 1176 |
|
| 1177 |
/** |
| 1178 |
* Get the indexable mapping. |
| 1179 |
* |
| 1180 |
* @since 3.6.0 |
| 1181 |
* @return boolean|array |
| 1182 |
*/ |
| 1183 |
public function get_mapping() { |
| 1184 |
return Elasticsearch::factory()->get_mapping( $this->get_index_name() ); |
| 1185 |
} |
| 1186 |
|
| 1187 |
/** |
| 1188 |
* Compare the mapping generated by the plugin and the mapping stored in Elasticsearch. |
| 1189 |
* |
| 1190 |
* @todo properly implement the check. |
| 1191 |
* |
| 1192 |
* @since 3.6.0 |
| 1193 |
* @return bool|WP_Error |
| 1194 |
*/ |
| 1195 |
public function compare_mappings() { |
| 1196 |
if ( ! method_exists( $this, 'generate_mapping' ) ) { |
| 1197 |
return new \WP_Error( 'ep_generate_mapping_not_implemented' ); |
| 1198 |
} |
| 1199 |
|
| 1200 |
$new_mapping = $this->generate_mapping(); |
| 1201 |
$stored_mapping = $this->get_mapping(); |
| 1202 |
|
| 1203 |
return ( (string) $new_mapping['settings']['index.number_of_shards'] === $stored_mapping[ $this->get_index_name() ]['settings']['index']['number_of_shards'] ); |
| 1204 |
} |
| 1205 |
|
| 1206 |
/** |
| 1207 |
* Utility function to check if the indexable is being fully reindexed, i.e., |
| 1208 |
* the index was deleted, a new mapping was sent and content is being reindexed. |
| 1209 |
* |
| 1210 |
* @param int|null $blog_id Blog ID |
| 1211 |
* @return boolean |
| 1212 |
*/ |
| 1213 |
public function is_full_reindexing( $blog_id = null ) { |
| 1214 |
if ( $this->global ) { |
| 1215 |
$blog_id = null; |
| 1216 |
} elseif ( ! $blog_id ) { |
| 1217 |
$blog_id = get_current_blog_id(); |
| 1218 |
} |
| 1219 |
|
| 1220 |
return \ElasticPress\IndexHelper::factory()->is_full_reindexing( $this->slug, $blog_id ); |
| 1221 |
} |
| 1222 |
|
| 1223 |
/** |
| 1224 |
* Send mapping to Elasticsearch |
| 1225 |
* |
| 1226 |
* @param string $return_type Desired return type. Can be either 'bool' or 'raw' |
| 1227 |
* @return bool|WP_Error |
| 1228 |
*/ |
| 1229 |
public function put_mapping( $return_type = 'bool' ) { |
| 1230 |
$mapping = $this->generate_mapping(); |
| 1231 |
|
| 1232 |
return Elasticsearch::factory()->put_mapping( $this->get_index_name(), $mapping, $return_type ); |
| 1233 |
} |
| 1234 |
|
| 1235 |
/** |
| 1236 |
* Must implement a method that given an object ID, returns a formatted Elasticsearch |
| 1237 |
* document |
| 1238 |
* |
| 1239 |
* @param int $object_id Object to prepare. |
| 1240 |
* @return array |
| 1241 |
*/ |
| 1242 |
abstract public function prepare_document( $object_id ); |
| 1243 |
|
| 1244 |
/** |
| 1245 |
* Must implement a method that queries MySQL for objects and returns them |
| 1246 |
* in a standardized format. This is necessary so we can genericize the index |
| 1247 |
* process across indexables. |
| 1248 |
* |
| 1249 |
* @param array $args Array to query DB against. |
| 1250 |
* @return array |
| 1251 |
*/ |
| 1252 |
abstract public function query_db( $args ); |
| 1253 |
|
| 1254 |
/** |
| 1255 |
* Shim function for backwards-compatibility on custom Indexables. |
| 1256 |
* |
| 1257 |
* @since 4.1.0 |
| 1258 |
* @return array |
| 1259 |
*/ |
| 1260 |
public function generate_mapping() { |
| 1261 |
_doing_it_wrong( __METHOD__, 'The Indexable class should not call generate_mapping() directly.', 'ElasticPress 4.0' ); |
| 1262 |
|
| 1263 |
return []; |
| 1264 |
} |
| 1265 |
|
| 1266 |
/** |
| 1267 |
* Get the search algorithm that should be used. |
| 1268 |
* |
| 1269 |
* @since 4.3.0 |
| 1270 |
* @param string $search_text Search term(s) |
| 1271 |
* @param array $search_fields Search fields |
| 1272 |
* @param array $query_vars Query vars |
| 1273 |
* @return SearchAlgorithm Instance of search algorithm to be used |
| 1274 |
*/ |
| 1275 |
public function get_search_algorithm( string $search_text, array $search_fields, array $query_vars ): \ElasticPress\SearchAlgorithm { |
| 1276 |
/** |
| 1277 |
* Filter the search algorithm to be used |
| 1278 |
* |
| 1279 |
* @hook ep_{$indexable_slug}_search_algorithm |
| 1280 |
* @since 4.3.0 |
| 1281 |
* @param {string} $search_algorithm Slug of the search algorithm used as fallback |
| 1282 |
* @param {string} $search_term Search term |
| 1283 |
* @param {array} $search_fields Fields to be searched |
| 1284 |
* @param {array} $query_vars Query variables |
| 1285 |
* @return {string} New search algorithm slug |
| 1286 |
*/ |
| 1287 |
$search_algorithm = apply_filters( "ep_{$this->slug}_search_algorithm", 'basic', $search_text, $search_fields, $query_vars ); |
| 1288 |
|
| 1289 |
return \ElasticPress\SearchAlgorithms::factory()->get( $search_algorithm ); |
| 1290 |
} |
| 1291 |
|
| 1292 |
/** |
| 1293 |
* Get all distinct meta field keys. |
| 1294 |
* |
| 1295 |
* @since 4.3.0 |
| 1296 |
* @param null|int $blog_id (Optional) The blog ID. Sending `null` will use the current blog ID. |
| 1297 |
* @return array |
| 1298 |
* @throws \Exception An exception if meta fields are not available. |
| 1299 |
*/ |
| 1300 |
public function get_distinct_meta_field_keys( $blog_id = null ) { |
| 1301 |
$mapping = $this->get_mapping(); |
| 1302 |
|
| 1303 |
try { |
| 1304 |
if ( version_compare( (string) Elasticsearch::factory()->get_elasticsearch_version(), '7.0', '<' ) ) { |
| 1305 |
$meta_fields = (array) $mapping[ $this->get_index_name( $blog_id ) ]['mappings']['post']['properties']['meta']['properties']; |
| 1306 |
} else { |
| 1307 |
$meta_fields = (array) $mapping[ $this->get_index_name( $blog_id ) ]['mappings']['properties']['meta']['properties']; |
| 1308 |
} |
| 1309 |
$meta_keys = array_values( array_keys( $meta_fields ) ); |
| 1310 |
sort( $meta_keys ); |
| 1311 |
} catch ( \Throwable $th ) { |
| 1312 |
throw new \Exception( 'Meta fields not available.', 0 ); |
| 1313 |
} |
| 1314 |
|
| 1315 |
return $meta_keys; |
| 1316 |
} |
| 1317 |
|
| 1318 |
/** |
| 1319 |
* Get all distinct values for a given field. |
| 1320 |
* |
| 1321 |
* @since 4.3.0 |
| 1322 |
* @param string $field Field full name. For example: `meta.name.raw` |
| 1323 |
* @param int $count (Optional) Max number of different distinct values to be returned |
| 1324 |
* @param int $blog_id (Optional) The blog ID. Sending `null` will use the current blog ID. |
| 1325 |
* @return array |
| 1326 |
*/ |
| 1327 |
public function get_all_distinct_values( $field, $count = 10000, $blog_id = null ) { |
| 1328 |
$aggregation_name = 'distinct_values'; |
| 1329 |
|
| 1330 |
$es_query = [ |
| 1331 |
'_source' => false, |
| 1332 |
'size' => 0, |
| 1333 |
'aggs' => [ |
| 1334 |
$aggregation_name => [ |
| 1335 |
'terms' => [ |
| 1336 |
/** |
| 1337 |
* Filter the max. number of different distinct values to be returned by Elasticsearch. |
| 1338 |
* |
| 1339 |
* @since 4.3.0 |
| 1340 |
* @hook ep_{$indexable_slug}_all_distinct_values |
| 1341 |
* @param {int} $size The number of different values. Default: 10000 |
| 1342 |
* @param {string} $field The meta field |
| 1343 |
* @return {string} The new number of different values |
| 1344 |
*/ |
| 1345 |
'size' => apply_filters( 'ep_' . $this->slug . '_all_distinct_values', $count, $field ), |
| 1346 |
'field' => $field, |
| 1347 |
], |
| 1348 |
], |
| 1349 |
], |
| 1350 |
]; |
| 1351 |
|
| 1352 |
$response = Elasticsearch::factory()->query( $this->get_index_name( $blog_id ), $this->slug, $es_query, [] ); |
| 1353 |
|
| 1354 |
if ( ! $response || empty( $response['aggregations'] ) || empty( $response['aggregations'][ $aggregation_name ] ) || empty( $response['aggregations'][ $aggregation_name ]['buckets'] ) ) { |
| 1355 |
return []; |
| 1356 |
} |
| 1357 |
|
| 1358 |
$values = []; |
| 1359 |
foreach ( $response['aggregations'][ $aggregation_name ]['buckets'] as $es_bucket ) { |
| 1360 |
$values[] = $es_bucket['key']; |
| 1361 |
} |
| 1362 |
|
| 1363 |
return $values; |
| 1364 |
} |
| 1365 |
|
| 1366 |
/** |
| 1367 |
* Should instantiate the indexable SyncManager and QueryIntegration, the main responsibles for the WP integration. |
| 1368 |
* |
| 1369 |
* @since 4.5.0 |
| 1370 |
*/ |
| 1371 |
public function setup() {} |
| 1372 |
|
| 1373 |
/** |
| 1374 |
* Given a mapping, add the ngram analyzer to it |
| 1375 |
* |
| 1376 |
* @since 4.5.0 |
| 1377 |
* @param array $mapping The mapping |
| 1378 |
* @return array |
| 1379 |
*/ |
| 1380 |
public function add_ngram_analyzer( array $mapping ): array { |
| 1381 |
$mapping['settings']['analysis']['analyzer']['edge_ngram_analyzer'] = array( |
| 1382 |
'type' => 'custom', |
| 1383 |
'tokenizer' => 'standard', |
| 1384 |
'filter' => array( |
| 1385 |
'lowercase', |
| 1386 |
'edge_ngram', |
| 1387 |
), |
| 1388 |
); |
| 1389 |
|
| 1390 |
return $mapping; |
| 1391 |
} |
| 1392 |
} |
| 1393 |
|
