Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
97.70% |
85 / 87 |
|
75.00% |
6 / 8 |
CRAP | |
0.00% |
0 / 1 |
| CustomMatchFeature | |
97.70% |
85 / 87 |
|
75.00% |
6 / 8 |
23 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getKeywords | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| doApply | |
93.75% |
15 / 16 |
|
0.00% |
0 / 1 |
3.00 | |||
| applyFunctionScore | |
92.86% |
13 / 14 |
|
0.00% |
0 / 1 |
3.00 | |||
| combineQueries | |
100.00% |
10 / 10 |
|
100.00% |
1 / 1 |
2 | |||
| parseValue | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
2 | |||
| parseSearchString | |
100.00% |
21 / 21 |
|
100.00% |
1 / 1 |
7 | |||
| generateParameterizedQueries | |
100.00% |
19 / 19 |
|
100.00% |
1 / 1 |
4 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace Wikibase\MediaInfo\Search\Feature; |
| 4 | |
| 5 | use CirrusSearch\Query\SimpleKeywordFeature; |
| 6 | use CirrusSearch\Search\SearchContext; |
| 7 | use CirrusSearch\WarningCollector; |
| 8 | use Elastica\Query\AbstractQuery; |
| 9 | use Elastica\Query\BoolQuery; |
| 10 | use Elastica\Query\FunctionScore; |
| 11 | use Elastica\Query\MatchQuery; |
| 12 | use Elastica\Script\Script; |
| 13 | use RuntimeException; |
| 14 | |
| 15 | /** |
| 16 | * Handles the search keyword 'custommatch:' |
| 17 | * |
| 18 | * Allows the user to search using custom-configured Match queries. The user supplies a name for |
| 19 | * the custom profile plus a search term, and a Bool query is created comprising Matches on the |
| 20 | * fields specified in the config for the custom profile |
| 21 | * |
| 22 | * The custom profiles are specified in config like this |
| 23 | * $wgCirrusSearchCustomMatchFeature = [ |
| 24 | * string $profileName => [ |
| 25 | * 'fields' => [ |
| 26 | * string $fieldName => [ |
| 27 | * [ 'prefix' => string $prefix, 'boost' => float $boost ], |
| 28 | * ... |
| 29 | * ], |
| 30 | * ... |
| 31 | * ], |
| 32 | * ] |
| 33 | * ... |
| 34 | * ]; |
| 35 | * |
| 36 | * So for example if we search using `custommatch:depicts_or_linked_from=Q999` and the config |
| 37 | * looks like this: |
| 38 | * [ |
| 39 | * 'depicts_or_linked_from' => [ |
| 40 | * 'fields' => [ |
| 41 | * 'statement_keywords' => [ |
| 42 | * [ 'prefix' => 'P180=', 'boost' => 10 ], |
| 43 | * ], |
| 44 | * 'weighted_tags' => [ |
| 45 | * [ 'prefix' => 'image.linked.from.wikidata.p18/', 'boost' => 9 ], |
| 46 | * [ 'prefix' => 'image.linked.from.wikidata.p373/', 'boost' => 8 ], |
| 47 | * ] |
| 48 | * ] |
| 49 | * ] |
| 50 | * ]; |
| 51 | * |
| 52 | * Then the resulting query will look like this: |
| 53 | * { |
| 54 | * "query": { |
| 55 | * "bool": { |
| 56 | * "should": [ |
| 57 | * { |
| 58 | * "match": { |
| 59 | * "statement_keywords": { |
| 60 | * "query": "P180=Q999", |
| 61 | * "boost": 10 |
| 62 | * } |
| 63 | * } |
| 64 | * }, |
| 65 | * { |
| 66 | * "match": { |
| 67 | * "weighted_tags": { |
| 68 | * "query": "image.linked.from.wikidata.p18\/Q999", |
| 69 | * "boost": 9 |
| 70 | * } |
| 71 | * } |
| 72 | * }, |
| 73 | * { |
| 74 | * "match": { |
| 75 | * "weighted_tags": { |
| 76 | * "query": "image.linked.from.wikidata.p373\/Q999", |
| 77 | * "boost": 8 |
| 78 | * } |
| 79 | * } |
| 80 | * }, |
| 81 | * ] |
| 82 | * } |
| 83 | * } |
| 84 | * } |
| 85 | * |
| 86 | * A profile can also have a function score associated with it. Configuration looks like this: |
| 87 | * string $profileName => [ |
| 88 | * 'fields' => [ ... ], |
| 89 | * 'functionScore' => [ |
| 90 | * 'scriptCode' => string $code, |
| 91 | * 'params' => [], |
| 92 | * ] |
| 93 | * ] |
| 94 | * |
| 95 | * In our previous example if we have the following config for the profile 'depicts_or_linked_from': |
| 96 | * [ |
| 97 | * 'depicts_or_linked_from' => [ |
| 98 | * 'fields' => [ (as above) ], |
| 99 | * 'functionScore' => [ |
| 100 | * 'scriptCode' => '100 / ( 1 + exp( -1 * ( _score + intercept ) ) )', |
| 101 | * 'params' => [ 'intercept' => -1.11111 ] |
| 102 | * ] |
| 103 | * ] |
| 104 | * ]; |
| 105 | * |
| 106 | * ... then the resulting query will look like this: |
| 107 | * { |
| 108 | * "function_score": { |
| 109 | * "query": { (as above) }, |
| 110 | * "functions": [ |
| 111 | * { |
| 112 | * "script_score": { |
| 113 | * "script": { |
| 114 | * "source": "100 \/ ( 1 + exp( -1 * ( _score + intercept ) ) )", |
| 115 | * "params": { |
| 116 | * "intercept": -1.11111 |
| 117 | * }, |
| 118 | * "lang": "expression" |
| 119 | * } |
| 120 | * } |
| 121 | * } |
| 122 | * ] |
| 123 | * } |
| 124 | * } |
| 125 | * |
| 126 | * @uses CirrusSearch |
| 127 | * @see https://phabricator.wikimedia.org/T296309 |
| 128 | */ |
| 129 | class CustomMatchFeature extends SimpleKeywordFeature { |
| 130 | |
| 131 | /** @var array */ |
| 132 | private $featureConfig; |
| 133 | |
| 134 | public function __construct( array $featureConfig ) { |
| 135 | $this->featureConfig = $featureConfig; |
| 136 | } |
| 137 | |
| 138 | /** |
| 139 | * @return string[] |
| 140 | */ |
| 141 | protected function getKeywords() { |
| 142 | return [ 'custommatch' ]; |
| 143 | } |
| 144 | |
| 145 | /** |
| 146 | * @param SearchContext $context |
| 147 | * @param string $key The keyword |
| 148 | * @param string $value The value attached to the keyword with quotes stripped |
| 149 | * @param string $quotedValue The original value in the search string, including quotes if used |
| 150 | * @param bool $negated Is the search negated? Not used to generate the returned AbstractQuery, |
| 151 | * that will be negated as necessary. Used for any other building/context necessary. |
| 152 | * @return array Two element array, first an AbstractQuery or null to apply to the |
| 153 | * query. Second a boolean indicating if the quotedValue should be kept in the search |
| 154 | * string. |
| 155 | */ |
| 156 | protected function doApply( SearchContext $context, $key, $value, $quotedValue, $negated ) { |
| 157 | $queries = $this->parseValue( |
| 158 | $key, |
| 159 | $value, |
| 160 | $quotedValue, |
| 161 | '', |
| 162 | '', |
| 163 | $context |
| 164 | ); |
| 165 | if ( count( $queries ) === 0 ) { |
| 166 | $context->setResultsPossible( false ); |
| 167 | return [ null, false ]; |
| 168 | } |
| 169 | $query = $this->combineQueries( $queries ); |
| 170 | if ( !$negated ) { |
| 171 | $context->addNonTextQuery( $query ); |
| 172 | return [ null, false ]; |
| 173 | } else { |
| 174 | return [ $query, false ]; |
| 175 | } |
| 176 | } |
| 177 | |
| 178 | /** |
| 179 | * Wraps query in a FunctionScore |
| 180 | * |
| 181 | * @param string $profileName |
| 182 | * @param AbstractQuery $query |
| 183 | * @return AbstractQuery |
| 184 | */ |
| 185 | private function applyFunctionScore( string $profileName, AbstractQuery $query |
| 186 | ): AbstractQuery { |
| 187 | $config = $this->featureConfig[$profileName]['functionScore'] ?? null; |
| 188 | if ( $config === null ) { |
| 189 | return $query; |
| 190 | } |
| 191 | if ( !isset( $config['scriptCode'] ) ) { |
| 192 | return $query; |
| 193 | } |
| 194 | |
| 195 | return ( new FunctionScore() ) |
| 196 | ->setQuery( $query ) |
| 197 | ->addScriptScoreFunction( |
| 198 | new Script( |
| 199 | $config['scriptCode'], |
| 200 | $config['params'] ?? [], |
| 201 | 'expression' |
| 202 | ) |
| 203 | ); |
| 204 | } |
| 205 | |
| 206 | /** |
| 207 | * Combines parameterized queries into single query containing MatchQuery objects |
| 208 | * |
| 209 | * @param string[][] $queries queries to combine. See generateParameterizedQueries() for fields. |
| 210 | * @return AbstractQuery |
| 211 | */ |
| 212 | private function combineQueries( array $queries ): AbstractQuery { |
| 213 | $profileName = ''; |
| 214 | $return = new BoolQuery(); |
| 215 | $return->setMinimumShouldMatch( 1 ); |
| 216 | foreach ( $queries as $query ) { |
| 217 | $return->addShould( new MatchQuery( |
| 218 | $query['field'], |
| 219 | [ 'query' => $query['string'], 'boost' => $query['boost'] ] |
| 220 | ) ); |
| 221 | $profileName = $query['profileName']; |
| 222 | } |
| 223 | return $this->applyFunctionScore( $profileName, $return ); |
| 224 | } |
| 225 | |
| 226 | /** |
| 227 | * @param string $key |
| 228 | * @param string $value |
| 229 | * @param string $quotedValue |
| 230 | * @param string $valueDelimiter |
| 231 | * @param string $suffix |
| 232 | * @param WarningCollector $warningCollector |
| 233 | * @return array [ |
| 234 | * [ |
| 235 | * 'field' => document field to run the query against, |
| 236 | * 'string' => string to search for, |
| 237 | * 'weight' => the boost for the query |
| 238 | * ], |
| 239 | * ... |
| 240 | * ] |
| 241 | */ |
| 242 | public function parseValue( |
| 243 | $key, |
| 244 | $value, |
| 245 | $quotedValue, |
| 246 | $valueDelimiter, |
| 247 | $suffix, |
| 248 | WarningCollector $warningCollector |
| 249 | ) { |
| 250 | $parsedSearchString = $this->parseSearchString( $value, $key, $warningCollector ); |
| 251 | if ( $parsedSearchString === null ) { |
| 252 | return []; |
| 253 | } |
| 254 | return $this->generateParameterizedQueries( |
| 255 | $parsedSearchString['profileName'], $parsedSearchString['searchTerm'] ); |
| 256 | } |
| 257 | |
| 258 | /** |
| 259 | * We expect the search string to be in the form <profile name>=<search term>. This function |
| 260 | * checks the format and if it's ok returns an array with the profile name and search |
| 261 | * terms separated |
| 262 | * |
| 263 | * @param string $searchString |
| 264 | * @param string $keyword |
| 265 | * @param WarningCollector|null $warningCollector |
| 266 | * @return array|null |
| 267 | */ |
| 268 | private function parseSearchString( string $searchString, string $keyword = '', |
| 269 | WarningCollector $warningCollector = null |
| 270 | ): ?array { |
| 271 | if ( !preg_match( '/^(\w+)=(.+)$/i', $searchString, $matches ) ) { |
| 272 | if ( $warningCollector !== null ) { |
| 273 | $warningCollector->addWarning( |
| 274 | 'wikibasemediainfo-custommatch-feature-invalid-term', |
| 275 | $keyword |
| 276 | ); |
| 277 | } |
| 278 | return null; |
| 279 | } |
| 280 | $profileName = $matches[1]; |
| 281 | $searchTerm = $matches[2]; |
| 282 | if ( !isset( $this->featureConfig[ $profileName ] ) ) { |
| 283 | if ( $warningCollector !== null ) { |
| 284 | $warningCollector->addWarning( 'wikibasemediainfo-custommatch-feature-no-profile', |
| 285 | $profileName ); |
| 286 | } |
| 287 | return null; |
| 288 | } |
| 289 | if ( !isset( $this->featureConfig[ $profileName ][ 'fields' ] ) || |
| 290 | !is_array( $this->featureConfig[ $profileName ][ 'fields' ] ) |
| 291 | ) { |
| 292 | throw new RuntimeException( 'The CustomMatch cirrussearch feature is misconfigured' ); |
| 293 | } |
| 294 | return [ |
| 295 | 'profileName' => $profileName, |
| 296 | 'searchTerm' => $searchTerm, |
| 297 | ]; |
| 298 | } |
| 299 | |
| 300 | private function generateParameterizedQueries( string $profileName, string $searchTerm |
| 301 | ): array { |
| 302 | $queries = []; |
| 303 | foreach ( $this->featureConfig[ $profileName ][ 'fields' ] as $field => $config ) { |
| 304 | if ( is_array( $config ) ) { |
| 305 | foreach ( $config as $configRow ) { |
| 306 | $prefix = $configRow['prefix'] ?? ''; |
| 307 | $boost = $configRow['boost'] ?? 1; |
| 308 | $queries[] = [ |
| 309 | 'field' => $field, |
| 310 | 'string' => $prefix . $searchTerm, |
| 311 | 'boost' => $boost, |
| 312 | 'profileName' => $profileName, |
| 313 | ]; |
| 314 | } |
| 315 | } else { |
| 316 | $queries[] = [ |
| 317 | 'field' => $config, |
| 318 | 'string' => $searchTerm, |
| 319 | 'boost' => 1, |
| 320 | 'profileName' => $profileName, |
| 321 | ]; |
| 322 | } |
| 323 | } |
| 324 | return $queries; |
| 325 | } |
| 326 | } |