Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
75.00% |
24 / 32 |
|
33.33% |
4 / 12 |
CRAP | |
0.00% |
0 / 1 |
| MediaSearchASTEntitiesExtractor | |
75.00% |
24 / 32 |
|
33.33% |
4 / 12 |
20.00 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getEntities | |
90.00% |
9 / 10 |
|
0.00% |
0 / 1 |
4.02 | |||
| getTerms | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
2 | |||
| getTerm | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
1 | |||
| visitWordsQueryNode | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| visitPhraseQueryNode | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| visitPhrasePrefixNode | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| visitFuzzyNode | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| visitPrefixNode | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| visitWildcardNode | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| visitEmptyQueryNode | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| visitKeywordFeatureNode | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace Wikibase\MediaInfo\Search; |
| 4 | |
| 5 | use CirrusSearch\Parser\AST\EmptyQueryNode; |
| 6 | use CirrusSearch\Parser\AST\FuzzyNode; |
| 7 | use CirrusSearch\Parser\AST\KeywordFeatureNode; |
| 8 | use CirrusSearch\Parser\AST\ParsedNode; |
| 9 | use CirrusSearch\Parser\AST\ParsedQuery; |
| 10 | use CirrusSearch\Parser\AST\PhrasePrefixNode; |
| 11 | use CirrusSearch\Parser\AST\PhraseQueryNode; |
| 12 | use CirrusSearch\Parser\AST\PrefixNode; |
| 13 | use CirrusSearch\Parser\AST\Visitor\LeafVisitor; |
| 14 | use CirrusSearch\Parser\AST\WildcardNode; |
| 15 | use CirrusSearch\Parser\AST\WordsQueryNode; |
| 16 | use SplObjectStorage; |
| 17 | |
| 18 | /** |
| 19 | * This visitor simply extracts all individual search terms out of |
| 20 | * a query that might contain multiple clauses (AND, OR, ...) so that |
| 21 | * we can look them all up individually to fetch relevant entities. |
| 22 | */ |
| 23 | class MediaSearchASTEntitiesExtractor extends LeafVisitor { |
| 24 | /** @var ParsedQuery */ |
| 25 | private $parsedQuery; |
| 26 | |
| 27 | /** @var MediaSearchEntitiesFetcher */ |
| 28 | private $entitiesFetcher; |
| 29 | |
| 30 | /** @var SplObjectStorage */ |
| 31 | private $terms; |
| 32 | |
| 33 | /** @var SplObjectStorage */ |
| 34 | private $entities; |
| 35 | |
| 36 | public function __construct( MediaSearchEntitiesFetcher $entitiesFetcher ) { |
| 37 | parent::__construct(); |
| 38 | $this->entitiesFetcher = $entitiesFetcher; |
| 39 | } |
| 40 | |
| 41 | public function getEntities( ParsedQuery $parsedQuery, ParsedNode $parsedNode ): array { |
| 42 | if ( !isset( $this->entities[$parsedQuery] ) ) { |
| 43 | // walk the AST and extract all terms |
| 44 | $terms = $this->getTerms( $parsedQuery ); |
| 45 | // then fetch entities for all of them |
| 46 | // re-init entities map; runtime code will usually only have only 1 query, |
| 47 | // but tests might execute plenty & we don't want to keep results for an |
| 48 | // entire suite in memory |
| 49 | $this->entities = new SplObjectStorage(); |
| 50 | $this->entities[$parsedQuery] = $this->entitiesFetcher->get( $terms ); |
| 51 | } |
| 52 | |
| 53 | if ( !isset( $this->terms[$parsedNode] ) ) { |
| 54 | return []; |
| 55 | } |
| 56 | |
| 57 | $term = $this->terms[$parsedNode]; |
| 58 | if ( !isset( $this->entities[$parsedQuery][$term] ) ) { |
| 59 | return []; |
| 60 | } |
| 61 | |
| 62 | return $this->entities[$parsedQuery][$term]; |
| 63 | } |
| 64 | |
| 65 | private function getTerms( ParsedQuery $parsedQuery ): array { |
| 66 | $this->terms = new SplObjectStorage(); |
| 67 | $this->parsedQuery = $parsedQuery; |
| 68 | $this->parsedQuery->getRoot()->accept( $this ); |
| 69 | |
| 70 | $terms = []; |
| 71 | foreach ( $this->terms as $node ) { |
| 72 | $terms[] = $this->terms[$node]; |
| 73 | } |
| 74 | |
| 75 | return $terms; |
| 76 | } |
| 77 | |
| 78 | private function getTerm( ParsedNode $node ): string { |
| 79 | return substr( |
| 80 | $this->parsedQuery->getQuery(), |
| 81 | $node->getStartOffset(), |
| 82 | $node->getEndOffset() - $node->getStartOffset() |
| 83 | ); |
| 84 | } |
| 85 | |
| 86 | /** |
| 87 | * @inheritDoc |
| 88 | */ |
| 89 | public function visitWordsQueryNode( WordsQueryNode $node ) { |
| 90 | $this->terms[$node] = $this->getTerm( $node ); |
| 91 | } |
| 92 | |
| 93 | /** |
| 94 | * @inheritDoc |
| 95 | */ |
| 96 | public function visitPhraseQueryNode( PhraseQueryNode $node ) { |
| 97 | $this->terms[$node] = $this->getTerm( $node ); |
| 98 | } |
| 99 | |
| 100 | /** |
| 101 | * @inheritDoc |
| 102 | */ |
| 103 | public function visitPhrasePrefixNode( PhrasePrefixNode $node ) { |
| 104 | // @todo implement this |
| 105 | // $this->terms[$node] = $this->getTerm( $node ); |
| 106 | } |
| 107 | |
| 108 | /** |
| 109 | * @inheritDoc |
| 110 | */ |
| 111 | public function visitFuzzyNode( FuzzyNode $node ) { |
| 112 | // @todo implement this |
| 113 | // $this->terms[$node] = $this->getTerm( $node ); |
| 114 | } |
| 115 | |
| 116 | /** |
| 117 | * @inheritDoc |
| 118 | */ |
| 119 | public function visitPrefixNode( PrefixNode $node ) { |
| 120 | // @todo implement this |
| 121 | // $this->terms[$node] = $this->getTerm( $node ); |
| 122 | } |
| 123 | |
| 124 | /** |
| 125 | * @inheritDoc |
| 126 | */ |
| 127 | public function visitWildcardNode( WildcardNode $node ) { |
| 128 | // @todo implement this |
| 129 | // $this->terms[$node] = $this->getTerm( $node ); |
| 130 | } |
| 131 | |
| 132 | /** |
| 133 | * @inheritDoc |
| 134 | */ |
| 135 | public function visitEmptyQueryNode( EmptyQueryNode $node ) { |
| 136 | // @todo implement this |
| 137 | // $this->terms[$node] = $this->getTerm( $node ); |
| 138 | } |
| 139 | |
| 140 | /** |
| 141 | * @inheritDoc |
| 142 | */ |
| 143 | public function visitKeywordFeatureNode( KeywordFeatureNode $node ) { |
| 144 | // not relevant here |
| 145 | } |
| 146 | } |