Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
96.39% |
80 / 83 |
|
62.50% |
5 / 8 |
CRAP | |
0.00% |
0 / 1 |
| FullTextCirrusSearchResultBuilder | |
96.39% |
80 / 83 |
|
62.50% |
5 / 8 |
33 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| newBuilder | |
75.00% |
3 / 4 |
|
0.00% |
0 / 1 |
2.06 | |||
| build | |
100.00% |
20 / 20 |
|
100.00% |
1 / 1 |
4 | |||
| getTitleHelper | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| doTitleSnippet | |
100.00% |
23 / 23 |
|
100.00% |
1 / 1 |
11 | |||
| doMainSnippet | |
100.00% |
15 / 15 |
|
100.00% |
1 / 1 |
6 | |||
| doHeadings | |
88.89% |
8 / 9 |
|
0.00% |
0 / 1 |
4.02 | |||
| doCategory | |
87.50% |
7 / 8 |
|
0.00% |
0 / 1 |
4.03 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace CirrusSearch\Search; |
| 4 | |
| 5 | use CirrusSearch\Search\Fetch\HighlightedField; |
| 6 | use CirrusSearch\Search\Fetch\HighlightingTrait; |
| 7 | use MediaWiki\Title\Title; |
| 8 | use MWTimestamp; |
| 9 | |
| 10 | class FullTextCirrusSearchResultBuilder { |
| 11 | use HighlightingTrait; |
| 12 | |
| 13 | /** @var CirrusSearchResultBuilder|null */ |
| 14 | private $builder; |
| 15 | |
| 16 | /** @var TitleHelper */ |
| 17 | private $titleHelper; |
| 18 | |
| 19 | /** @var HighlightedField[][] indexed per target and ordered by priority */ |
| 20 | private $highligtedFields; |
| 21 | |
| 22 | /** @var string[] */ |
| 23 | private $extraFields; |
| 24 | |
| 25 | /** |
| 26 | * @param TitleHelper $titleHelper |
| 27 | * @param HighlightedField[][] $hlFieldsPerTarget list of highlighted field indexed per target and sorted by priority |
| 28 | * @param string[] $extraFields list of extra fields to extract from the source doc |
| 29 | */ |
| 30 | public function __construct( TitleHelper $titleHelper, array $hlFieldsPerTarget, array $extraFields = [] ) { |
| 31 | $this->titleHelper = $titleHelper; |
| 32 | $this->highligtedFields = $hlFieldsPerTarget; |
| 33 | $this->extraFields = $extraFields; |
| 34 | } |
| 35 | |
| 36 | /** |
| 37 | * @param Title $title |
| 38 | * @param string $docId |
| 39 | * @return CirrusSearchResultBuilder |
| 40 | */ |
| 41 | private function newBuilder( Title $title, $docId ): CirrusSearchResultBuilder { |
| 42 | if ( $this->builder === null ) { |
| 43 | $this->builder = new CirrusSearchResultBuilder( $title, $docId ); |
| 44 | } else { |
| 45 | $this->builder->reset( $title, $docId ); |
| 46 | } |
| 47 | return $this->builder; |
| 48 | } |
| 49 | |
| 50 | /** |
| 51 | * @param \Elastica\Result $result |
| 52 | * @return CirrusSearchResult |
| 53 | */ |
| 54 | public function build( \Elastica\Result $result ): CirrusSearchResult { |
| 55 | $title = $this->getTitleHelper()->makeTitle( $result ); |
| 56 | $fields = $result->getFields(); |
| 57 | $builder = $this->newBuilder( $title, $result->getId() ) |
| 58 | ->wordCount( $fields['text.word_count'][0] ?? 0 ) |
| 59 | ->byteSize( $result->text_bytes ?? 0 ) |
| 60 | ->timestamp( new MWTimestamp( $result->timestamp ) ) |
| 61 | ->score( $result->getScore() ) |
| 62 | ->explanation( $result->getExplanation() ); |
| 63 | |
| 64 | if ( isset( $result->namespace_text ) ) { |
| 65 | $builder->interwikiNamespaceText( $result->namespace_text ); |
| 66 | } |
| 67 | |
| 68 | $highlights = $result->getHighlights(); |
| 69 | $this->doTitleSnippet( $title, $result, $highlights ); |
| 70 | $this->doMainSnippet( $highlights ); |
| 71 | $this->doHeadings( $title, $highlights ); |
| 72 | $this->doCategory( $highlights ); |
| 73 | $source = $result->getData(); |
| 74 | |
| 75 | foreach ( $this->extraFields as $field ) { |
| 76 | if ( isset( $source[$field] ) ) { |
| 77 | $builder->addExtraField( $field, $source[$field] ); |
| 78 | } |
| 79 | } |
| 80 | |
| 81 | return $builder->build(); |
| 82 | } |
| 83 | |
| 84 | /** |
| 85 | * @return TitleHelper |
| 86 | */ |
| 87 | protected function getTitleHelper(): TitleHelper { |
| 88 | return $this->titleHelper; |
| 89 | } |
| 90 | |
| 91 | private function doTitleSnippet( Title $title, \Elastica\Result $result, array $highlights ) { |
| 92 | $matched = false; |
| 93 | foreach ( $this->highligtedFields[ArrayCirrusSearchResult::TITLE_SNIPPET] as $hlField ) { |
| 94 | if ( isset( $highlights[$hlField->getFieldName()] ) ) { |
| 95 | $nstext = $title->getNamespace() === 0 ? '' : $this->titleHelper->getNamespaceText( $title ) . ':'; |
| 96 | $snippet = $nstext . $this->escapeHighlightedText( $highlights[ $hlField->getFieldName() ][ 0 ] ); |
| 97 | $this->builder |
| 98 | ->titleSnippet( $snippet ) |
| 99 | ->titleSnippetField( $hlField->getFieldName() ); |
| 100 | $matched = true; |
| 101 | break; |
| 102 | } |
| 103 | } |
| 104 | if ( !$matched && $title->isExternal() ) { |
| 105 | // Interwiki searches are weird. They won't have title highlights by design, but |
| 106 | // if we don't return a title snippet we'll get weird display results. |
| 107 | $this->builder->titleSnippet( $title->getText() ); |
| 108 | } |
| 109 | |
| 110 | if ( !$matched && isset( $this->highligtedFields[ArrayCirrusSearchResult::REDIRECT_SNIPPET] ) ) { |
| 111 | foreach ( $this->highligtedFields[ArrayCirrusSearchResult::REDIRECT_SNIPPET] as $hlField ) { |
| 112 | // Make sure to find the redirect title before escaping because escaping breaks it.... |
| 113 | if ( !isset( $highlights[$hlField->getFieldName()][0] ) ) { |
| 114 | continue; |
| 115 | } |
| 116 | $redirTitle = $this->findRedirectTitle( $result, $highlights[$hlField->getFieldName()][0] ); |
| 117 | if ( $redirTitle !== null ) { |
| 118 | $this->builder |
| 119 | ->redirectTitle( $redirTitle ) |
| 120 | ->redirectSnippet( $this->escapeHighlightedText( $highlights[$hlField->getFieldName()][0] ) ) |
| 121 | ->redirectSnippetField( $hlField->getFieldName() ); |
| 122 | break; |
| 123 | } |
| 124 | } |
| 125 | } |
| 126 | } |
| 127 | |
| 128 | private function doMainSnippet( $highlights ) { |
| 129 | $hasTextSnippet = false; |
| 130 | foreach ( $this->highligtedFields[ArrayCirrusSearchResult::TEXT_SNIPPET] as $hlField ) { |
| 131 | if ( isset( $highlights[$hlField->getFieldName()][0] ) ) { |
| 132 | $snippet = $highlights[$hlField->getFieldName()][0]; |
| 133 | if ( $this->containsMatches( $snippet ) ) { |
| 134 | $this->builder |
| 135 | ->textSnippet( $this->escapeHighlightedText( $snippet ) ) |
| 136 | ->textSnippetField( $hlField->getFieldName() ) |
| 137 | ->fileMatch( $hlField->getFieldName() === 'file_text' ); |
| 138 | $hasTextSnippet = true; |
| 139 | break; |
| 140 | } |
| 141 | } |
| 142 | } |
| 143 | |
| 144 | // Hardcode the fallback to the "text" highlight, it generally contains the beginning of the |
| 145 | // text content if nothing has matched. |
| 146 | if ( !$hasTextSnippet && isset( $highlights['text'][0] ) ) { |
| 147 | $this->builder |
| 148 | ->textSnippet( $this->escapeHighlightedText( $highlights['text'][0] ) ) |
| 149 | ->textSnippetField( 'text' ); |
| 150 | } |
| 151 | } |
| 152 | |
| 153 | private function doHeadings( Title $title, $highlights ) { |
| 154 | if ( !isset( $this->highligtedFields[ArrayCirrusSearchResult::SECTION_SNIPPET] ) ) { |
| 155 | return; |
| 156 | } |
| 157 | foreach ( $this->highligtedFields[ArrayCirrusSearchResult::SECTION_SNIPPET] as $hlField ) { |
| 158 | if ( isset( $highlights[$hlField->getFieldName()] ) ) { |
| 159 | $this->builder |
| 160 | ->sectionSnippet( $this->escapeHighlightedText( $highlights[$hlField->getFieldName()][0] ) ) |
| 161 | ->sectionSnippetField( $hlField->getFieldName() ) |
| 162 | ->sectionTitle( $this->findSectionTitle( $highlights[$hlField->getFieldName()][0], $title ) ); |
| 163 | break; |
| 164 | } |
| 165 | } |
| 166 | } |
| 167 | |
| 168 | private function doCategory( $highlights ) { |
| 169 | if ( !isset( $this->highligtedFields[ArrayCirrusSearchResult::CATEGORY_SNIPPET] ) ) { |
| 170 | return; |
| 171 | } |
| 172 | foreach ( $this->highligtedFields[ArrayCirrusSearchResult::CATEGORY_SNIPPET] as $hlField ) { |
| 173 | if ( isset( $highlights[$hlField->getFieldName()] ) ) { |
| 174 | $this->builder |
| 175 | ->categorySnippet( $this->escapeHighlightedText( $highlights[$hlField->getFieldName()][0] ) ) |
| 176 | ->categorySnippetField( $hlField->getFieldName() ); |
| 177 | } |
| 178 | break; |
| 179 | } |
| 180 | } |
| 181 | } |