Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
100.00% |
93 / 93 |
|
100.00% |
9 / 9 |
CRAP | |
100.00% |
1 / 1 |
| HasWbStatementFeature | |
100.00% |
93 / 93 |
|
100.00% |
9 / 9 |
24 | |
100.00% |
1 / 1 |
| getKeywords | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| doApply | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
2 | |||
| combineQueries | |
100.00% |
18 / 18 |
|
100.00% |
1 / 1 |
5 | |||
| parseValue | |
100.00% |
36 / 36 |
|
100.00% |
1 / 1 |
7 | |||
| isStatementStringValid | |
100.00% |
15 / 15 |
|
100.00% |
1 / 1 |
2 | |||
| statementContainsPropertyOnly | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| statementEndsWithWildcard | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| statementContainsOnlyWildcard | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getFilterQuery | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace Wikibase\Search\Elastic\Query; |
| 4 | |
| 5 | use CirrusSearch\Parser\AST\KeywordFeatureNode; |
| 6 | use CirrusSearch\Query\Builder\QueryBuildingContext; |
| 7 | use CirrusSearch\Query\FilterQueryFeature; |
| 8 | use CirrusSearch\Query\SimpleKeywordFeature; |
| 9 | use CirrusSearch\Search\SearchContext; |
| 10 | use CirrusSearch\WarningCollector; |
| 11 | use Elastica\Query\AbstractQuery; |
| 12 | use Elastica\Query\BoolQuery; |
| 13 | use Elastica\Query\Exists; |
| 14 | use Elastica\Query\MatchQuery; |
| 15 | use Elastica\Query\Prefix; |
| 16 | use Wikibase\DataModel\Entity\NumericPropertyId; |
| 17 | use Wikibase\Search\Elastic\Fields\StatementsField; |
| 18 | |
| 19 | /** |
| 20 | * Handles the search keyword 'haswbstatement:' |
| 21 | * |
| 22 | * Allows the user to search for pages/items that have wikibase properties or statements associated |
| 23 | * with them. |
| 24 | * |
| 25 | * If a file page has ANY statement about property 'P180' ('depicts') then it can be found |
| 26 | * by including 'haswbstatement:P180' in the search query. |
| 27 | * |
| 28 | * If a file page has the statement 'P180=Q527' (meaning 'depicts sky') associated |
| 29 | * with it then it can be found by including 'haswbstatement:P180=Q527' in the |
| 30 | * search query. |
| 31 | * |
| 32 | * If a file page has the statement 'P2014=79802' (meaning 'MoMA artwork id 79802') |
| 33 | * associated with it then it can be found by including 'haswbstatement:P2014=79802' in the |
| 34 | * search query. |
| 35 | * |
| 36 | * A '*' at the end of a 'haswbstatement' string triggers a prefix search. If different file pages |
| 37 | * have the statements: |
| 38 | * - P180=Q146[P462=Q23445] ('depicts cat, color black') |
| 39 | * - P180=Q146[P462=Q23444] ('depicts cat, color white') |
| 40 | * ... then both those pages will be found if 'P180=Q146[P462=*' is |
| 41 | * included in the search query. |
| 42 | * |
| 43 | * A '*' as the statement triggers existence search. Any page containing any statement will |
| 44 | * be returned. |
| 45 | * |
| 46 | * Statements can be combined using logical OR by separating them with a | character in a single |
| 47 | * haswbstatement query e.g. 'haswbstatement:P999=Q888|P999=Q777' |
| 48 | * |
| 49 | * Statements can be combined using logical AND by using two separate haswbstatement queries e.g. |
| 50 | * 'haswbstatement:P999=Q888 haswbstatement:P999=Q777' |
| 51 | * |
| 52 | * |
| 53 | * Note that NOT ALL STATEMENTS ARE INDEXED. Searching for a statement about a property that has |
| 54 | * not been indexed will give an empty result set. |
| 55 | * |
| 56 | * @uses CirrusSearch |
| 57 | * @see https://phabricator.wikimedia.org/T190022 |
| 58 | */ |
| 59 | class HasWbStatementFeature extends SimpleKeywordFeature implements FilterQueryFeature { |
| 60 | |
| 61 | /** |
| 62 | * @return string[] |
| 63 | */ |
| 64 | protected function getKeywords() { |
| 65 | return [ 'haswbstatement' ]; |
| 66 | } |
| 67 | |
| 68 | /** |
| 69 | * @param SearchContext $context |
| 70 | * @param string $key The keyword |
| 71 | * @param string $value The value attached to the keyword with quotes stripped |
| 72 | * @param string $quotedValue The original value in the search string, including quotes if used |
| 73 | * @param bool $negated Is the search negated? Not used to generate the returned AbstractQuery, |
| 74 | * that will be negated as necessary. Used for any other building/context necessary. |
| 75 | * @return array Two element array, first an AbstractQuery or null to apply to the |
| 76 | * query. Second a boolean indicating if the quotedValue should be kept in the search |
| 77 | * string. |
| 78 | */ |
| 79 | protected function doApply( SearchContext $context, $key, $value, $quotedValue, $negated ) { |
| 80 | $queries = $this->parseValue( |
| 81 | $key, |
| 82 | $value, |
| 83 | $quotedValue, |
| 84 | '', |
| 85 | '', |
| 86 | $context |
| 87 | ); |
| 88 | if ( count( $queries ) == 0 ) { |
| 89 | $context->setResultsPossible( false ); |
| 90 | return [ null, false ]; |
| 91 | } |
| 92 | |
| 93 | return [ $this->combineQueries( $queries ), false ]; |
| 94 | } |
| 95 | |
| 96 | /** |
| 97 | * Builds an OR between many statements about the wikibase item |
| 98 | * |
| 99 | * @param string[][] $queries queries to combine. See parseValue() for fields. |
| 100 | * @return \Elastica\Query\AbstractQuery |
| 101 | */ |
| 102 | private function combineQueries( array $queries ) { |
| 103 | $return = new BoolQuery(); |
| 104 | foreach ( $queries as $query ) { |
| 105 | if ( $query['class'] === Prefix::class ) { |
| 106 | $return->addShould( new Prefix( [ |
| 107 | $query['field'] => |
| 108 | [ |
| 109 | 'value' => $query['string'], |
| 110 | 'rewrite' => 'top_terms_1024' |
| 111 | ] |
| 112 | ] ) ); |
| 113 | } elseif ( $query['class'] === MatchQuery::class ) { |
| 114 | $return->addShould( new MatchQuery( |
| 115 | $query['field'], |
| 116 | [ 'query' => $query['string'] ] |
| 117 | ) ); |
| 118 | } elseif ( $query['class'] === Exists::class ) { |
| 119 | // In a boolean 'OR' having an existence check negates the |
| 120 | // need for the remaining queries. |
| 121 | return new Exists( $query['field'] ); |
| 122 | } |
| 123 | } |
| 124 | return $return; |
| 125 | } |
| 126 | |
| 127 | /** |
| 128 | * @param string $key |
| 129 | * @param string $value |
| 130 | * @param string $quotedValue |
| 131 | * @param string $valueDelimiter |
| 132 | * @param string $suffix |
| 133 | * @param WarningCollector $warningCollector |
| 134 | * @return array [ |
| 135 | * [ |
| 136 | * 'class' => \Elastica\Query class name to be used to construct the query, |
| 137 | * 'field' => document field to run the query against, |
| 138 | * 'string' => string to search for |
| 139 | * ], |
| 140 | * ... |
| 141 | * ] |
| 142 | */ |
| 143 | public function parseValue( |
| 144 | $key, |
| 145 | $value, |
| 146 | $quotedValue, |
| 147 | $valueDelimiter, |
| 148 | $suffix, |
| 149 | WarningCollector $warningCollector |
| 150 | ) { |
| 151 | $queries = []; |
| 152 | $statementStrings = explode( '|', $value ); |
| 153 | foreach ( $statementStrings as $statementString ) { |
| 154 | if ( !$this->isStatementStringValid( $statementString ) ) { |
| 155 | // TODO: Add warning to avoid unexpected behaviour |
| 156 | continue; |
| 157 | } |
| 158 | if ( $this->statementContainsOnlyWildcard( $statementString ) ) { |
| 159 | $queries[] = [ |
| 160 | 'class' => Exists::class, |
| 161 | 'field' => StatementsField::NAME |
| 162 | ]; |
| 163 | continue; |
| 164 | } |
| 165 | if ( $this->statementContainsPropertyOnly( $statementString ) ) { |
| 166 | $queries[] = [ |
| 167 | 'class' => MatchQuery::class, |
| 168 | 'field' => StatementsField::NAME . '.property', |
| 169 | 'string' => $statementString, |
| 170 | ]; |
| 171 | continue; |
| 172 | } |
| 173 | if ( $this->statementEndsWithWildcard( $statementString ) ) { |
| 174 | $queries[] = [ |
| 175 | 'class' => Prefix::class, |
| 176 | 'field' => StatementsField::NAME, |
| 177 | 'string' => substr( $statementString, 0, strlen( $statementString ) - 1 ), |
| 178 | ]; |
| 179 | continue; |
| 180 | } |
| 181 | $queries[] = [ |
| 182 | 'class' => MatchQuery::class, |
| 183 | 'field' => StatementsField::NAME, |
| 184 | 'string' => $statementString, |
| 185 | ]; |
| 186 | } |
| 187 | if ( count( $queries ) == 0 ) { |
| 188 | $warningCollector->addWarning( |
| 189 | 'wikibasecirrus-haswbstatement-feature-no-valid-statements', |
| 190 | $key |
| 191 | ); |
| 192 | } |
| 193 | return $queries; |
| 194 | } |
| 195 | |
| 196 | /** |
| 197 | * Check that a statement string is valid. A valid string is a P-id |
| 198 | * optionally suffixed with an equals sign. |
| 199 | * |
| 200 | * The following strings are valid: |
| 201 | * P2014=79802 |
| 202 | * P999 |
| 203 | * |
| 204 | * The following strings are invalid: |
| 205 | * PrefixedId:P123=Q537 |
| 206 | * PA=Q888 |
| 207 | * PF=1234567 |
| 208 | * |
| 209 | * @param string $statementString |
| 210 | * @return bool |
| 211 | */ |
| 212 | private function isStatementStringValid( $statementString ) { |
| 213 | if ( $this->statementContainsOnlyWildcard( $statementString ) ) { |
| 214 | // Simpler than integrating into basically unrelated regex |
| 215 | return true; |
| 216 | } |
| 217 | |
| 218 | //Strip delimiters, anchors and pattern modifiers from NumericPropertyId::PATTERN |
| 219 | $propertyIdPattern = preg_replace( |
| 220 | '/([^\sa-zA-Z0-9\\\])(\^|\\\A)?(.*?)(\$|\\\z|\\\Z)?\\1[a-zA-Z]*/', |
| 221 | '$3', |
| 222 | NumericPropertyId::PATTERN |
| 223 | ); |
| 224 | $validStatementStringPattern = '/^' . |
| 225 | $propertyIdPattern . |
| 226 | '(' . StatementsField::STATEMENT_SEPARATOR . '|$)' . |
| 227 | '/i'; |
| 228 | |
| 229 | return (bool)preg_match( |
| 230 | $validStatementStringPattern, |
| 231 | $statementString |
| 232 | ); |
| 233 | } |
| 234 | |
| 235 | private function statementContainsPropertyOnly( $statementString ) { |
| 236 | if ( strpos( $statementString, '=' ) === false ) { |
| 237 | return true; |
| 238 | } |
| 239 | return false; |
| 240 | } |
| 241 | |
| 242 | private function statementEndsWithWildcard( $statementString ) { |
| 243 | if ( substr( $statementString, -1 ) == '*' ) { |
| 244 | return true; |
| 245 | } |
| 246 | return false; |
| 247 | } |
| 248 | |
| 249 | private function statementContainsOnlyWildcard( $statementString ) { |
| 250 | return $statementString === '*'; |
| 251 | } |
| 252 | |
| 253 | /** |
| 254 | * @param KeywordFeatureNode $node |
| 255 | * @param QueryBuildingContext $context |
| 256 | * @return AbstractQuery|null |
| 257 | */ |
| 258 | public function getFilterQuery( KeywordFeatureNode $node, QueryBuildingContext $context ) { |
| 259 | $statements = $node->getParsedValue(); |
| 260 | if ( $statements === [] ) { |
| 261 | return null; |
| 262 | } |
| 263 | return $this->combineQueries( $statements ); |
| 264 | } |
| 265 | |
| 266 | } |