Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
98.59% |
70 / 71 |
|
92.31% |
12 / 13 |
CRAP | |
0.00% |
0 / 1 |
| FetchPhaseConfigBuilder | |
98.59% |
70 / 71 |
|
92.31% |
12 / 13 |
25 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| newHighlightField | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
5 | |||
| addNewRegexHLField | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| supportsRegexFields | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| newRegexField | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| addHLField | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| getHLField | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| buildHLConfig | |
100.00% |
14 / 14 |
|
100.00% |
1 / 1 |
4 | |||
| withConfig | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getHLFieldsPerTargetAndPriority | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
2 | |||
| suppressRedirectTitleHighlight | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| configureDefaultFullTextFields | |
100.00% |
15 / 15 |
|
100.00% |
1 / 1 |
2 | |||
| clearSkipIfLastMatched | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace CirrusSearch\Search\Fetch; |
| 4 | |
| 5 | use CirrusSearch\CirrusConfigNames; |
| 6 | use CirrusSearch\SearchConfig; |
| 7 | use CirrusSearch\Searcher; |
| 8 | use Elastica\Query\AbstractQuery; |
| 9 | use Wikimedia\Assert\Assert; |
| 10 | |
| 11 | /** |
| 12 | * Class holding the building state of the fetch phase elements of |
| 13 | * an elasticsearch query. |
| 14 | * Currently only supports the highlight section but can be extended to support |
| 15 | * source filtering and stored field. |
| 16 | */ |
| 17 | class FetchPhaseConfigBuilder implements HighlightFieldGenerator { |
| 18 | |
| 19 | /** @var HighlightedField[] */ |
| 20 | private $highlightedFields = []; |
| 21 | |
| 22 | /** @var SearchConfig */ |
| 23 | private $config; |
| 24 | |
| 25 | /** |
| 26 | * @var string |
| 27 | */ |
| 28 | private $factoryGroup; |
| 29 | |
| 30 | /** |
| 31 | * @var bool |
| 32 | */ |
| 33 | private $provideAllSnippets; |
| 34 | |
| 35 | /** |
| 36 | * @var bool whether to drop the redirect.title highlight from the default |
| 37 | * full text fields. Set by the query side when the search is in redirect |
| 38 | * scope, where the matching redirect is its own result. |
| 39 | */ |
| 40 | private bool $suppressRedirectTitleHighlight = false; |
| 41 | |
| 42 | /** |
| 43 | * @param SearchConfig $config |
| 44 | * @param string|null $factoryGroup |
| 45 | * @param bool $provideAllSnippets |
| 46 | */ |
| 47 | public function __construct( |
| 48 | SearchConfig $config, |
| 49 | $factoryGroup = null, |
| 50 | bool $provideAllSnippets = false |
| 51 | ) { |
| 52 | $this->config = $config; |
| 53 | $this->factoryGroup = $factoryGroup; |
| 54 | $this->provideAllSnippets = $provideAllSnippets; |
| 55 | } |
| 56 | |
| 57 | /** |
| 58 | * @inheritDoc |
| 59 | */ |
| 60 | public function newHighlightField( |
| 61 | $name, |
| 62 | $target, |
| 63 | $priority = HighlightedField::DEFAULT_TARGET_PRIORITY |
| 64 | ): BaseHighlightedField { |
| 65 | $useExp = $this->config->get( CirrusConfigNames::UseExperimentalHighlighter ); |
| 66 | if ( $useExp ) { |
| 67 | $factories = ExperimentalHighlightedFieldBuilder::getFactories(); |
| 68 | } else { |
| 69 | $factories = BaseHighlightedField::getFactories(); |
| 70 | } |
| 71 | if ( $this->factoryGroup !== null && isset( $factories[$this->factoryGroup][$name] ) ) { |
| 72 | return ( $factories[$this->factoryGroup][$name] )( $this->config, $name, $target, $priority ); |
| 73 | } |
| 74 | if ( $useExp ) { |
| 75 | return new ExperimentalHighlightedFieldBuilder( $name, $target, $priority ); |
| 76 | } else { |
| 77 | return new BaseHighlightedField( $name, BaseHighlightedField::FVH_HL_TYPE, $target, $priority ); |
| 78 | } |
| 79 | } |
| 80 | |
| 81 | /** |
| 82 | * @param string $name |
| 83 | * @param string $target |
| 84 | * @param string $pattern |
| 85 | * @param bool $caseInsensitive |
| 86 | * @param int $priority |
| 87 | */ |
| 88 | public function addNewRegexHLField( |
| 89 | $name, |
| 90 | $target, |
| 91 | $pattern, |
| 92 | $caseInsensitive, |
| 93 | $priority = HighlightedField::COSTLY_EXPERT_SYNTAX_PRIORITY |
| 94 | ) { |
| 95 | if ( !$this->supportsRegexFields() ) { |
| 96 | return; |
| 97 | } |
| 98 | $this->addHLField( $this->newRegexField( $name, $target, $pattern, $caseInsensitive, $priority ) ); |
| 99 | } |
| 100 | |
| 101 | /** |
| 102 | * Whether this builder can generate regex fields |
| 103 | * @return bool |
| 104 | */ |
| 105 | public function supportsRegexFields() { |
| 106 | return (bool)$this->config->get( CirrusConfigNames::UseExperimentalHighlighter ); |
| 107 | } |
| 108 | |
| 109 | /** |
| 110 | * @inheritDoc |
| 111 | */ |
| 112 | public function newRegexField( |
| 113 | $name, |
| 114 | $target, |
| 115 | $pattern, |
| 116 | $caseInsensitive, |
| 117 | $priority = HighlightedField::COSTLY_EXPERT_SYNTAX_PRIORITY, |
| 118 | $regexFlavor = 'lucene' |
| 119 | ): BaseHighlightedField { |
| 120 | Assert::precondition( $this->supportsRegexFields(), 'Regex fields not supported' ); |
| 121 | return ExperimentalHighlightedFieldBuilder::newRegexField( |
| 122 | $this->config, $name, $target, $pattern, $caseInsensitive, $priority, $regexFlavor ); |
| 123 | } |
| 124 | |
| 125 | public function addHLField( HighlightedField $field ) { |
| 126 | $prev = $this->highlightedFields[$field->getFieldName()] ?? null; |
| 127 | if ( $prev === null ) { |
| 128 | $this->highlightedFields[$field->getFieldName()] = $field; |
| 129 | } else { |
| 130 | $this->highlightedFields[$field->getFieldName()] = $prev->merge( $field ); |
| 131 | } |
| 132 | } |
| 133 | |
| 134 | /** |
| 135 | * @param string $field |
| 136 | * @return HighlightedField|null |
| 137 | */ |
| 138 | public function getHLField( $field ) { |
| 139 | return $this->highlightedFields[$field] ?? null; |
| 140 | } |
| 141 | |
| 142 | /** |
| 143 | * @param AbstractQuery|null $mainHLQuery |
| 144 | * @return array |
| 145 | */ |
| 146 | public function buildHLConfig( ?AbstractQuery $mainHLQuery = null ): array { |
| 147 | $fields = []; |
| 148 | foreach ( $this->highlightedFields as $field ) { |
| 149 | $arr = $field->toArray(); |
| 150 | if ( $this->provideAllSnippets ) { |
| 151 | $arr = $this->clearSkipIfLastMatched( $arr ); |
| 152 | } |
| 153 | $fields[$field->getFieldName()] = $arr; |
| 154 | } |
| 155 | $config = [ |
| 156 | 'pre_tags' => [ Searcher::HIGHLIGHT_PRE_MARKER ], |
| 157 | 'post_tags' => [ Searcher::HIGHLIGHT_POST_MARKER ], |
| 158 | 'fields' => $fields, |
| 159 | ]; |
| 160 | |
| 161 | if ( $mainHLQuery !== null ) { |
| 162 | $config['highlight_query'] = $mainHLQuery->toArray(); |
| 163 | } |
| 164 | |
| 165 | return $config; |
| 166 | } |
| 167 | |
| 168 | public function withConfig( SearchConfig $config ): self { |
| 169 | return new self( $config, $this->factoryGroup ); |
| 170 | } |
| 171 | |
| 172 | /** |
| 173 | * Return the list of highlighted fields indexed per target |
| 174 | * and ordered by priority (reverse natural order) |
| 175 | * @return HighlightedField[][] |
| 176 | */ |
| 177 | public function getHLFieldsPerTargetAndPriority(): array { |
| 178 | $fields = []; |
| 179 | foreach ( $this->highlightedFields as $f ) { |
| 180 | $fields[$f->getTarget()][] = $f; |
| 181 | } |
| 182 | return array_map( |
| 183 | static function ( array $v ) { |
| 184 | usort( $v, static function ( HighlightedField $g1, HighlightedField $g2 ) { |
| 185 | return $g2->getPriority() <=> $g1->getPriority(); |
| 186 | } ); |
| 187 | return $v; |
| 188 | }, |
| 189 | $fields |
| 190 | ); |
| 191 | } |
| 192 | |
| 193 | /** |
| 194 | * Drop the redirect.title highlight from the default full text fields. |
| 195 | * |
| 196 | * Used when the search is in redirect scope: the matching redirect is its |
| 197 | * own result, so highlighting redirect.title would produce a redundant and |
| 198 | * misleading "redirected from" snippet. |
| 199 | */ |
| 200 | public function suppressRedirectTitleHighlight(): void { |
| 201 | $this->suppressRedirectTitleHighlight = true; |
| 202 | } |
| 203 | |
| 204 | public function configureDefaultFullTextFields() { |
| 205 | // TODO: find a better place for this |
| 206 | // Title/redir/category/template |
| 207 | $field = $this->newHighlightField( 'title', HighlightedField::TARGET_TITLE_SNIPPET ); |
| 208 | $this->addHLField( $field ); |
| 209 | if ( !$this->suppressRedirectTitleHighlight ) { |
| 210 | $field = $this->newHighlightField( 'redirect.title', HighlightedField::TARGET_REDIRECT_SNIPPET ); |
| 211 | $this->addHLField( $field->skipIfLastMatched() ); |
| 212 | } |
| 213 | $field = $this->newHighlightField( 'category', HighlightedField::TARGET_CATEGORY_SNIPPET ); |
| 214 | $this->addHLField( $field->skipIfLastMatched() ); |
| 215 | |
| 216 | $field = $this->newHighlightField( 'heading', HighlightedField::TARGET_SECTION_SNIPPET ); |
| 217 | $this->addHLField( $field->skipIfLastMatched() ); |
| 218 | |
| 219 | // content |
| 220 | $field = $this->newHighlightField( 'text', HighlightedField::TARGET_MAIN_SNIPPET ); |
| 221 | $this->addHLField( $field ); |
| 222 | |
| 223 | $field = $this->newHighlightField( 'auxiliary_text', HighlightedField::TARGET_MAIN_SNIPPET ); |
| 224 | $this->addHLField( $field->skipIfLastMatched() ); |
| 225 | |
| 226 | $field = $this->newHighlightField( 'file_text', HighlightedField::TARGET_MAIN_SNIPPET ); |
| 227 | $this->addHLField( $field->skipIfLastMatched() ); |
| 228 | } |
| 229 | |
| 230 | private function clearSkipIfLastMatched( array $arr ): array { |
| 231 | unset( $arr['options']['skip_if_last_matched'] ); |
| 232 | if ( empty( $arr['options'] ) ) { |
| 233 | unset( $arr['options'] ); |
| 234 | } |
| 235 | return $arr; |
| 236 | } |
| 237 | } |