class-query-builder.php
438 lines
| 1 | <?php |
| 2 | /** |
| 3 | * Provides an interface for easily building a complex search query that |
| 4 | * combines multiple ranking signals. |
| 5 | * |
| 6 | * $bldr = new Query_Builder(); |
| 7 | * $bldr->add_filter( ... ); |
| 8 | * $bldr->add_filter( ... ); |
| 9 | * $bldr->add_query( ... ); |
| 10 | * $es_query = $bldr->build_query(); |
| 11 | * |
| 12 | * |
| 13 | * All ES queries take a standard form with main query (with some filters), |
| 14 | * wrapped in a function_score |
| 15 | * |
| 16 | * Most functions are chainable, e.g. $bldr->add_filter( ... )->add_query( ... )->build_query(); |
| 17 | * |
| 18 | * Bucketed queries use an aggregation to diversify results. eg a bunch |
| 19 | * of separate filters where to get different sets of results. |
| 20 | * |
| 21 | * @package automattic/jetpack-search |
| 22 | */ |
| 23 | |
| 24 | // Disables comment checks. |
| 25 | // phpcs:disable Squiz.Commenting |
| 26 | |
| 27 | namespace Automattic\Jetpack\Search\WPES; |
| 28 | |
| 29 | /** |
| 30 | * Query builder class. |
| 31 | */ |
| 32 | class Query_Builder { |
| 33 | /** |
| 34 | * ElasticSerach filters. |
| 35 | * |
| 36 | * @var array |
| 37 | */ |
| 38 | protected $es_filters = array(); |
| 39 | |
| 40 | // |
| 41 | // Variables for handling custom boosting with function_score. |
| 42 | // |
| 43 | protected $functions = array(); |
| 44 | protected $weighting_functions = array(); |
| 45 | protected $decays = array(); |
| 46 | protected $scripts = array(); |
| 47 | protected $functions_max_boost = 2.0; |
| 48 | protected $functions_score_mode = 'multiply'; |
| 49 | protected $functions_boost_mode = 'multiply'; |
| 50 | protected $query_bool_boost = null; |
| 51 | |
| 52 | // |
| 53 | // General aggregations for buckets and metrics. |
| 54 | // |
| 55 | protected $aggs_query = false; |
| 56 | protected $aggs = array(); |
| 57 | |
| 58 | // |
| 59 | // The set of top level text queries to combine. |
| 60 | // |
| 61 | protected $must_queries = array(); |
| 62 | protected $should_queries = array(); |
| 63 | protected $dis_max_queries = array(); |
| 64 | |
| 65 | protected $diverse_buckets_query = false; |
| 66 | protected $bucket_filters = array(); |
| 67 | protected $bucket_sub_aggs = array(); |
| 68 | |
| 69 | /** |
| 70 | * Language codes. |
| 71 | * |
| 72 | * @var array |
| 73 | */ |
| 74 | private $langs; |
| 75 | |
| 76 | public function get_langs() { |
| 77 | if ( isset( $this->langs ) ) { |
| 78 | return $this->langs; |
| 79 | } |
| 80 | return false; |
| 81 | } |
| 82 | |
| 83 | // |
| 84 | // Methods for building a query. |
| 85 | // |
| 86 | public function add_filter( $filter ) { |
| 87 | $this->es_filters[] = $filter; |
| 88 | |
| 89 | return $this; |
| 90 | } |
| 91 | |
| 92 | public function add_query( $query, $type = 'must' ) { |
| 93 | switch ( $type ) { |
| 94 | case 'dis_max': |
| 95 | $this->dis_max_queries[] = $query; |
| 96 | break; |
| 97 | |
| 98 | case 'should': |
| 99 | $this->should_queries[] = $query; |
| 100 | break; |
| 101 | |
| 102 | case 'must': |
| 103 | default: |
| 104 | $this->must_queries[] = $query; |
| 105 | break; |
| 106 | } |
| 107 | |
| 108 | return $this; |
| 109 | } |
| 110 | |
| 111 | /** |
| 112 | * Add any weighting function to the query |
| 113 | * |
| 114 | * @see https://www.elastic.co/guide/en/elasticsearch/reference/current/query-dsl-function-score-query.html |
| 115 | * |
| 116 | * @param $function array A function structure to apply to the query |
| 117 | * |
| 118 | * @return void |
| 119 | */ |
| 120 | public function add_weighting_function( $function ) { |
| 121 | // check for danger. |
| 122 | if ( isset( $function['random_score'] ) ) { |
| 123 | return $this; |
| 124 | } |
| 125 | if ( isset( $function['script_score'] ) ) { |
| 126 | return $this; |
| 127 | } |
| 128 | |
| 129 | $this->weighting_functions[] = $function; |
| 130 | |
| 131 | return $this; |
| 132 | } |
| 133 | |
| 134 | /** |
| 135 | * Add a scoring function to the query |
| 136 | * |
| 137 | * NOTE: For decays (linear, exp, or gauss), use Query_Builder::add_decay() instead |
| 138 | * |
| 139 | * @see https://www.elastic.co/guide/en/elasticsearch/reference/current/query-dsl-function-score-query.html |
| 140 | * |
| 141 | * @param $function string name of the function |
| 142 | * @param $params array functions parameters |
| 143 | * |
| 144 | * @return void |
| 145 | */ |
| 146 | public function add_function( $function, $params ) { |
| 147 | $this->functions[ $function ][] = $params; |
| 148 | |
| 149 | return $this; |
| 150 | } |
| 151 | |
| 152 | /** |
| 153 | * Add a decay function to score results |
| 154 | * |
| 155 | * This method should be used instead of Query_Builder::add_function() for decays, as the internal ES structure |
| 156 | * is slightly different for them. |
| 157 | * |
| 158 | * @see https://www.elastic.co/guide/en/elasticsearch/guide/current/decay-functions.html |
| 159 | * |
| 160 | * @param $function string name of the decay function - linear, exp, or gauss |
| 161 | * @param $params array The decay functions parameters, passed to ES directly |
| 162 | * |
| 163 | * @return void |
| 164 | */ |
| 165 | public function add_decay( $function, $params ) { |
| 166 | $this->decays[ $function ][] = $params; |
| 167 | |
| 168 | return $this; |
| 169 | } |
| 170 | |
| 171 | /** |
| 172 | * Add a scoring mode to the query |
| 173 | * |
| 174 | * @see https://www.elastic.co/guide/en/elasticsearch/reference/current/query-dsl-function-score-query.html |
| 175 | * |
| 176 | * @param $mode string name of how to score |
| 177 | * |
| 178 | * @return void |
| 179 | */ |
| 180 | public function add_score_mode_to_functions( $mode = 'multiply' ) { |
| 181 | $this->functions_score_mode = $mode; |
| 182 | |
| 183 | return $this; |
| 184 | } |
| 185 | |
| 186 | public function add_boost_mode_to_functions( $mode = 'multiply' ) { |
| 187 | $this->functions_boost_mode = $mode; |
| 188 | |
| 189 | return $this; |
| 190 | } |
| 191 | |
| 192 | public function add_max_boost_to_functions( $boost ) { |
| 193 | $this->functions_max_boost = $boost; |
| 194 | |
| 195 | return $this; |
| 196 | } |
| 197 | |
| 198 | public function add_boost_to_query_bool( $boost ) { |
| 199 | $this->query_bool_boost = $boost; |
| 200 | |
| 201 | return $this; |
| 202 | } |
| 203 | |
| 204 | public function add_aggs( $aggs_name, $aggs ) { |
| 205 | $this->aggs_query = true; |
| 206 | $this->aggs[ $aggs_name ] = $aggs; |
| 207 | |
| 208 | return $this; |
| 209 | } |
| 210 | |
| 211 | public function set_all_aggs( $aggs ) { |
| 212 | $this->aggs_query = true; |
| 213 | $this->aggs = $aggs; |
| 214 | |
| 215 | return $this; |
| 216 | } |
| 217 | |
| 218 | public function add_aggs_sub_aggs( $aggs_name, $sub_aggs ) { |
| 219 | if ( ! array_key_exists( 'aggs', $this->aggs[ $aggs_name ] ) ) { |
| 220 | $this->aggs[ $aggs_name ]['aggs'] = array(); |
| 221 | } |
| 222 | $this->aggs[ $aggs_name ]['aggs'] = $sub_aggs; |
| 223 | |
| 224 | return $this; |
| 225 | } |
| 226 | |
| 227 | public function add_bucketed_query( $name, $query ) { |
| 228 | $this->add_bucket_filter( $name, $query ); |
| 229 | |
| 230 | $this->add_query( $query, 'dis_max' ); |
| 231 | |
| 232 | return $this; |
| 233 | } |
| 234 | |
| 235 | public function add_bucketed_terms( $name, $field, $terms, $boost = 1 ) { |
| 236 | if ( ! is_array( $terms ) ) { |
| 237 | $terms = array( $terms ); |
| 238 | } |
| 239 | |
| 240 | $this->add_bucket_filter( |
| 241 | $name, |
| 242 | array( |
| 243 | 'terms' => array( |
| 244 | $field => $terms, |
| 245 | ), |
| 246 | ) |
| 247 | ); |
| 248 | |
| 249 | $this->add_query( |
| 250 | array( |
| 251 | 'constant_score' => array( |
| 252 | 'filter' => array( |
| 253 | 'terms' => array( |
| 254 | $field => $terms, |
| 255 | ), |
| 256 | ), |
| 257 | 'boost' => $boost, |
| 258 | ), |
| 259 | ), |
| 260 | 'dis_max' |
| 261 | ); |
| 262 | |
| 263 | return $this; |
| 264 | } |
| 265 | |
| 266 | public function add_bucket_sub_aggs( $agg ) { |
| 267 | $this->bucket_sub_aggs = array_merge( $this->bucket_sub_aggs, $agg ); |
| 268 | |
| 269 | return $this; |
| 270 | } |
| 271 | |
| 272 | protected function add_bucket_filter( $name, $filter ) { |
| 273 | $this->diverse_buckets_query = true; |
| 274 | $this->bucket_filters[ $name ] = $filter; |
| 275 | } |
| 276 | |
| 277 | //////////////////////////////////// |
| 278 | // Building Final Query |
| 279 | |
| 280 | /** |
| 281 | * Combine all the queries, functions, decays, scripts, and max_boost into an ES query |
| 282 | * |
| 283 | * @return array Array representation of the built ES query |
| 284 | */ |
| 285 | public function build_query() { |
| 286 | $query = array(); |
| 287 | |
| 288 | //dis_max queries just become a single must query |
| 289 | if ( ! empty( $this->dis_max_queries ) ) { |
| 290 | $this->must_queries[] = array( |
| 291 | 'dis_max' => array( |
| 292 | 'queries' => $this->dis_max_queries, |
| 293 | ), |
| 294 | ); |
| 295 | } |
| 296 | |
| 297 | if ( empty( $this->must_queries ) ) { |
| 298 | $this->must_queries = array( |
| 299 | array( |
| 300 | 'match_all' => array(), |
| 301 | ), |
| 302 | ); |
| 303 | } |
| 304 | |
| 305 | if ( empty( $this->should_queries ) ) { |
| 306 | $query = array( |
| 307 | 'bool' => array( |
| 308 | 'must' => $this->must_queries, |
| 309 | ), |
| 310 | ); |
| 311 | } else { |
| 312 | $query = array( |
| 313 | 'bool' => array( |
| 314 | 'must' => $this->must_queries, |
| 315 | 'should' => $this->should_queries, |
| 316 | ), |
| 317 | ); |
| 318 | } |
| 319 | |
| 320 | if ( $this->query_bool_boost !== null ) { |
| 321 | $query['bool']['boost'] = $this->query_bool_boost; |
| 322 | } |
| 323 | |
| 324 | // If there are any function score adjustments, then combine those |
| 325 | if ( $this->functions || $this->decays || $this->scripts || $this->weighting_functions ) { |
| 326 | $weighting_functions = $this->weighting_functions; |
| 327 | |
| 328 | if ( $this->functions ) { |
| 329 | foreach ( $this->functions as $function_type => $configs ) { |
| 330 | foreach ( $configs as $config ) { |
| 331 | foreach ( $config as $field => $params ) { |
| 332 | $func_arr = $params; |
| 333 | |
| 334 | $func_arr['field'] = $field; |
| 335 | |
| 336 | $weighting_functions[] = array( |
| 337 | $function_type => $func_arr, |
| 338 | ); |
| 339 | } |
| 340 | } |
| 341 | } |
| 342 | } |
| 343 | |
| 344 | if ( $this->decays ) { |
| 345 | foreach ( $this->decays as $decay_type => $configs ) { |
| 346 | foreach ( $configs as $config ) { |
| 347 | foreach ( $config as $field => $params ) { |
| 348 | $weighting_functions[] = array( |
| 349 | $decay_type => array( |
| 350 | $field => $params, |
| 351 | ), |
| 352 | ); |
| 353 | } |
| 354 | } |
| 355 | } |
| 356 | } |
| 357 | |
| 358 | if ( $this->scripts ) { |
| 359 | foreach ( $this->scripts as $script ) { |
| 360 | $weighting_functions[] = array( |
| 361 | 'script_score' => array( |
| 362 | 'script' => $script, |
| 363 | ), |
| 364 | ); |
| 365 | } |
| 366 | } |
| 367 | |
| 368 | $query = array( |
| 369 | 'function_score' => array( |
| 370 | 'query' => $query, |
| 371 | 'functions' => $weighting_functions, |
| 372 | 'max_boost' => $this->functions_max_boost, |
| 373 | 'score_mode' => $this->functions_score_mode, |
| 374 | 'boost_mode' => $this->functions_boost_mode, |
| 375 | ), |
| 376 | ); |
| 377 | } |
| 378 | |
| 379 | return $query; |
| 380 | } |
| 381 | |
| 382 | /** |
| 383 | * Assemble the 'filter' portion of an ES query, from all registered filters |
| 384 | * |
| 385 | * @return array|null Combined ES filters, or null if none have been defined |
| 386 | */ |
| 387 | public function build_filter() { |
| 388 | if ( empty( $this->es_filters ) ) { |
| 389 | $filter = null; |
| 390 | } elseif ( 1 === count( $this->es_filters ) ) { |
| 391 | $filter = $this->es_filters[0]; |
| 392 | } else { |
| 393 | $filter = array( |
| 394 | 'and' => $this->es_filters, |
| 395 | ); |
| 396 | } |
| 397 | |
| 398 | return $filter; |
| 399 | } |
| 400 | |
| 401 | /** |
| 402 | * Assemble the 'aggregation' portion of an ES query, from all general aggregations. |
| 403 | * |
| 404 | * @return array An aggregation query as an array of topics, filters, and bucket names |
| 405 | */ |
| 406 | public function build_aggregation() { |
| 407 | if ( empty( $this->bucket_sub_aggs ) && empty( $this->aggs_query ) ) { |
| 408 | return array(); |
| 409 | } |
| 410 | |
| 411 | if ( ! $this->diverse_buckets_query && empty( $this->aggs_query ) ) { |
| 412 | return $this->bucket_sub_aggs; |
| 413 | } |
| 414 | |
| 415 | $aggregations = array( |
| 416 | 'topics' => array( |
| 417 | 'filters' => array( |
| 418 | 'filters' => array(), |
| 419 | ), |
| 420 | ), |
| 421 | ); |
| 422 | |
| 423 | if ( ! empty( $this->bucket_sub_aggs ) ) { |
| 424 | $aggregations['topics']['aggs'] = $this->bucket_sub_aggs; |
| 425 | } |
| 426 | |
| 427 | foreach ( $this->bucket_filters as $bucket_name => $filter ) { |
| 428 | $aggregations['topics']['filters']['filters'][ $bucket_name ] = $filter; |
| 429 | } |
| 430 | |
| 431 | if ( ! empty( $this->aggs_query ) ) { |
| 432 | $aggregations = $this->aggs; |
| 433 | } |
| 434 | |
| 435 | return $aggregations; |
| 436 | } |
| 437 | } |
| 438 |