Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 103 |
|
0.00% |
0 / 7 |
CRAP | |
0.00% |
0 / 1 |
| TextIndexField | |
0.00% |
0 / 103 |
|
0.00% |
0 / 7 |
1332 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
12 | |||
| setTextOptions | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| getTextOptions | |
0.00% |
0 / 10 |
|
0.00% |
0 / 1 |
42 | |||
| getMapping | |
0.00% |
0 / 64 |
|
0.00% |
0 / 1 |
156 | |||
| configureHighlighting | |
0.00% |
0 / 12 |
|
0.00% |
0 / 1 |
90 | |||
| initFlags | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| getSimilarity | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
20 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace CirrusSearch\Search; |
| 4 | |
| 5 | use CirrusSearch\CirrusSearch; |
| 6 | use CirrusSearch\Maintenance\MappingConfigBuilder; |
| 7 | use CirrusSearch\Profile\SearchProfileService; |
| 8 | use CirrusSearch\SearchConfig; |
| 9 | use SearchEngine; |
| 10 | use SearchIndexField; |
| 11 | |
| 12 | /** |
| 13 | * Index field representing keyword. |
| 14 | * Keywords use special analyzer. |
| 15 | * @package CirrusSearch |
| 16 | */ |
| 17 | class TextIndexField extends CirrusIndexField { |
| 18 | /** |
| 19 | * Distance that lucene places between multiple values of the same field. |
| 20 | * Set pretty high to prevent accidental phrase queries between those values. |
| 21 | */ |
| 22 | public const POSITION_INCREMENT_GAP = 10; |
| 23 | |
| 24 | /* Bit field parameters for string fields. |
| 25 | * ENABLE_NORMS: Enable norms on the field. Good for text you search against but useless |
| 26 | * for fields that don't get involved in the score. |
| 27 | * COPY_TO_SUGGEST: Copy the contents of this field to the suggest field for "Did you mean". |
| 28 | * SPEED_UP_HIGHLIGHTING: Store extra data in the field to speed up highlighting. This is important for |
| 29 | * long strings or fields with many values. |
| 30 | * SUPPORT_REGEX: If the wikimedia-extra plugin is available add a trigram |
| 31 | * index to speed up search. |
| 32 | */ |
| 33 | public const ENABLE_NORMS = 0x1000000; |
| 34 | // FIXME: when exactly we want to disable norms for text fields? |
| 35 | public const COPY_TO_SUGGEST = 0x2000000; |
| 36 | public const SPEED_UP_HIGHLIGHTING = 0x4000000; |
| 37 | public const SUPPORT_REGEX = 0x8000000; |
| 38 | public const STRING_FIELD_MASK = 0xFFFFFF; |
| 39 | |
| 40 | /** |
| 41 | * Extra definitions. |
| 42 | * @var array |
| 43 | */ |
| 44 | protected $extra; |
| 45 | /** |
| 46 | * Text options for this field |
| 47 | * @var int |
| 48 | */ |
| 49 | private $textOptions; |
| 50 | |
| 51 | /** |
| 52 | * Name of the type in Elastic |
| 53 | * @var string |
| 54 | */ |
| 55 | protected $typeName = 'text'; |
| 56 | |
| 57 | /** |
| 58 | * Are trigrams useful? |
| 59 | * @var bool |
| 60 | */ |
| 61 | protected $allowTrigrams = false; |
| 62 | |
| 63 | public function __construct( $name, $type, SearchConfig $config, $extra = [] ) { |
| 64 | parent::__construct( $name, $type, $config ); |
| 65 | |
| 66 | $this->extra = $extra; |
| 67 | |
| 68 | if ( $config->getElement( 'CirrusSearchWikimediaExtraPlugin', 'regex' ) && |
| 69 | in_array( 'build', $config->getElement( 'CirrusSearchWikimediaExtraPlugin', 'regex' ) ) |
| 70 | ) { |
| 71 | $this->allowTrigrams = true; |
| 72 | } |
| 73 | } |
| 74 | |
| 75 | /** |
| 76 | * Set text options for this field if non-default |
| 77 | * @param int $options |
| 78 | * @return self |
| 79 | */ |
| 80 | public function setTextOptions( $options ) { |
| 81 | $this->textOptions = $options; |
| 82 | return $this; |
| 83 | } |
| 84 | |
| 85 | /** |
| 86 | * Get text options for this field |
| 87 | * @param int $mappingFlags |
| 88 | * @return int |
| 89 | */ |
| 90 | protected function getTextOptions( $mappingFlags ) { |
| 91 | if ( $this->textOptions !== null ) { |
| 92 | return $this->textOptions; |
| 93 | } |
| 94 | $options = self::ENABLE_NORMS | self::SPEED_UP_HIGHLIGHTING; |
| 95 | if ( $this->config->get( 'CirrusSearchEnablePhraseSuggest' ) && |
| 96 | $mappingFlags & MappingConfigBuilder::PHRASE_SUGGEST_USE_TEXT && |
| 97 | !$this->checkFlag( SearchIndexField::FLAG_SCORING ) |
| 98 | ) { |
| 99 | // SCORING fields are not copied since this info is already in other fields |
| 100 | $options |= self::COPY_TO_SUGGEST; |
| 101 | } |
| 102 | if ( $this->checkFlag( SearchIndexField::FLAG_NO_HIGHLIGHT ) ) { |
| 103 | // Disable highlighting is asked to |
| 104 | $options &= ~self::SPEED_UP_HIGHLIGHTING; |
| 105 | } |
| 106 | return $options; |
| 107 | } |
| 108 | |
| 109 | /** |
| 110 | * @param SearchEngine $engine |
| 111 | * @return array |
| 112 | */ |
| 113 | public function getMapping( SearchEngine $engine ) { |
| 114 | if ( !( $engine instanceof CirrusSearch ) ) { |
| 115 | throw new \LogicException( "Cannot map CirrusSearch fields for another engine." ); |
| 116 | } |
| 117 | $this->initFlags(); |
| 118 | /** |
| 119 | * @var CirrusSearch $engine |
| 120 | */ |
| 121 | $field = parent::getMapping( $engine ); |
| 122 | |
| 123 | if ( $this->config->get( 'CirrusSearchEnablePhraseSuggest' ) && |
| 124 | $this->checkFlag( self::COPY_TO_SUGGEST ) |
| 125 | ) { |
| 126 | $field[ 'copy_to' ] = [ 'suggest' ]; |
| 127 | } |
| 128 | |
| 129 | if ( $this->checkFlag( self::FLAG_NO_INDEX ) ) { |
| 130 | // no need to configure further a not-indexed field |
| 131 | return $field; |
| 132 | } |
| 133 | |
| 134 | $extra = $this->extra; |
| 135 | |
| 136 | if ( $this->mappingFlags & MappingConfigBuilder::PREFIX_START_WITH_ANY ) { |
| 137 | $extra[] = [ |
| 138 | 'analyzer' => 'word_prefix', |
| 139 | 'search_analyzer' => 'plain_search', |
| 140 | 'index_options' => 'docs' |
| 141 | ]; |
| 142 | } |
| 143 | if ( $this->checkFlag( SearchIndexField::FLAG_CASEFOLD ) ) { |
| 144 | $extra[] = [ |
| 145 | 'analyzer' => 'lowercase_keyword', |
| 146 | 'norms' => false, |
| 147 | 'index_options' => 'docs', |
| 148 | // TODO: Re-enable in ES 5.2 with keyword type and s/analyzer/normalizer/ |
| 149 | // 'ignore_above' => KeywordIndexField::KEYWORD_IGNORE_ABOVE, |
| 150 | ]; |
| 151 | } |
| 152 | |
| 153 | if ( $this->allowTrigrams && $this->checkFlag( self::SUPPORT_REGEX ) ) { |
| 154 | $extra[] = [ |
| 155 | 'norms' => false, |
| 156 | 'type' => 'text', |
| 157 | 'analyzer' => 'trigram', |
| 158 | 'index_options' => 'docs', |
| 159 | ]; |
| 160 | } |
| 161 | |
| 162 | // multi_field is dead in 1.0 so we do this which actually looks less gnarly. |
| 163 | $field += [ |
| 164 | 'analyzer' => 'text', |
| 165 | 'search_analyzer' => 'text_search', |
| 166 | 'position_increment_gap' => self::POSITION_INCREMENT_GAP, |
| 167 | 'similarity' => self::getSimilarity( $this->config, $this->name ), |
| 168 | 'fields' => [ |
| 169 | 'plain' => [ |
| 170 | 'type' => 'text', |
| 171 | 'analyzer' => 'plain', |
| 172 | 'search_analyzer' => 'plain_search', |
| 173 | 'position_increment_gap' => self::POSITION_INCREMENT_GAP, |
| 174 | 'similarity' => self::getSimilarity( $this->config, $this->name, 'plain' ), |
| 175 | ], |
| 176 | ] |
| 177 | ]; |
| 178 | $disableNorms = !$this->checkFlag( self::ENABLE_NORMS ); |
| 179 | if ( $disableNorms ) { |
| 180 | $disableNorms = [ 'norms' => false ]; |
| 181 | $field = array_merge( $field, $disableNorms ); |
| 182 | $field[ 'fields' ][ 'plain' ] = array_merge( $field[ 'fields' ][ 'plain' ], $disableNorms ); |
| 183 | } |
| 184 | foreach ( $extra as $extraField ) { |
| 185 | $extraName = $extraField[ 'fieldName' ] ?? $extraField[ 'analyzer' ]; |
| 186 | unset( $extraField[ 'fieldName' ] ); |
| 187 | |
| 188 | $field[ 'fields' ][ $extraName ] = array_merge( [ |
| 189 | 'similarity' => self::getSimilarity( $this->config, $this->name, $extraName ), |
| 190 | 'type' => 'text', |
| 191 | ], $extraField ); |
| 192 | |
| 193 | if ( $disableNorms ) { |
| 194 | $field[ 'fields' ][ $extraName ] = array_merge( |
| 195 | $field[ 'fields' ][ $extraName ], $disableNorms ); |
| 196 | } |
| 197 | } |
| 198 | $this->configureHighlighting( $field, |
| 199 | [ 'plain', 'prefix', 'prefix_asciifolding', 'near_match', 'near_match_asciifolding' ] ); |
| 200 | return $field; |
| 201 | } |
| 202 | |
| 203 | /** |
| 204 | * Adapt the field options according to the highlighter used |
| 205 | * @param mixed[] &$field the mapping options being built |
| 206 | * @param string[] $subFields list of subfields to configure |
| 207 | * @param bool $rootField configure the root field (defaults to true) |
| 208 | */ |
| 209 | protected function configureHighlighting( array &$field, array $subFields, $rootField = true ) { |
| 210 | if ( $this->mappingFlags & MappingConfigBuilder::OPTIMIZE_FOR_EXPERIMENTAL_HIGHLIGHTER ) { |
| 211 | if ( $this->checkFlag( self::SPEED_UP_HIGHLIGHTING ) ) { |
| 212 | if ( $rootField ) { |
| 213 | $field[ 'index_options' ] = 'offsets'; |
| 214 | } |
| 215 | foreach ( $subFields as $fieldName ) { |
| 216 | if ( isset( $field[ 'fields' ][ $fieldName ] ) ) { |
| 217 | $field[ 'fields' ][ $fieldName ][ 'index_options' ] = 'offsets'; |
| 218 | } |
| 219 | } |
| 220 | } |
| 221 | } else { |
| 222 | // We use the FVH on all fields so turn on term vectors |
| 223 | if ( $rootField ) { |
| 224 | $field[ 'term_vector' ] = 'with_positions_offsets'; |
| 225 | } |
| 226 | foreach ( $subFields as $fieldName ) { |
| 227 | if ( isset( $field[ 'fields' ][ $fieldName ] ) ) { |
| 228 | $field[ 'fields' ][ $fieldName ][ 'term_vector' ] = 'with_positions_offsets'; |
| 229 | } |
| 230 | } |
| 231 | } |
| 232 | } |
| 233 | |
| 234 | /** |
| 235 | * Init the field flags |
| 236 | */ |
| 237 | protected function initFlags() { |
| 238 | $this->flags = |
| 239 | ( $this->flags & self::STRING_FIELD_MASK ) | $this->getTextOptions( $this->mappingFlags ); |
| 240 | } |
| 241 | |
| 242 | /** |
| 243 | * Get the field similarity |
| 244 | * @param SearchConfig $config |
| 245 | * @param string $field |
| 246 | * @param string|null $analyzer |
| 247 | * @return string |
| 248 | */ |
| 249 | public static function getSimilarity( SearchConfig $config, $field, $analyzer = null ) { |
| 250 | $similarity = $config->getProfileService()->loadProfile( SearchProfileService::SIMILARITY ); |
| 251 | $fieldSimilarity = $similarity['fields'][$field] ?? $similarity['fields']['__default__'] ?? null; |
| 252 | if ( $analyzer !== null && isset( $similarity['fields']["$field.$analyzer"] ) ) { |
| 253 | $fieldSimilarity = $similarity['fields']["$field.$analyzer"]; |
| 254 | } |
| 255 | if ( $fieldSimilarity === null ) { |
| 256 | throw new \RuntimeException( "Invalid similarity profile, unable to infer the similarity for " . |
| 257 | "the field $field, (defining a __default__ field might solve the issue" ); |
| 258 | } |
| 259 | return $fieldSimilarity; |
| 260 | } |
| 261 | } |