Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
100.00% |
75 / 75 |
|
100.00% |
19 / 19 |
CRAP | |
100.00% |
1 / 1 |
| GlobalCustomFilter | |
100.00% |
75 / 75 |
|
100.00% |
19 / 19 |
46 | |
100.00% |
1 / 1 |
| __construct | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setLanguageDenyList | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| setLanguageAllowList | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| setRequiredPlugins | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| setFallbackFilter | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| setApplyToAnalyzers | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getApplyToAnalyzers | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setRequiredTokenizer | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| setDisallowedTokenFilters | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| setDisallowedCharFilters | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| setMustFollowFilters | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| enableGlobalCustomFilters | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
7 | |||
| languageCheck | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
4 | |||
| pluginsAvailable | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
3 | |||
| requiredTokenizerUsed | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
4 | |||
| disallowedTokenFiltersPresent | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
3 | |||
| disallowedCharFiltersPresent | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
3 | |||
| analyzerCheck | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
7 | |||
| insertGlobalCustomFilter | |
100.00% |
10 / 10 |
|
100.00% |
1 / 1 |
4 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace CirrusSearch\Maintenance; |
| 4 | |
| 5 | class GlobalCustomFilter { |
| 6 | /** @var string filter type, probably 'filter' or 'char_filter'; 'filter' by default */ |
| 7 | private $type; |
| 8 | |
| 9 | /** @var string[] languages where this filter should not be used, by language codes */ |
| 10 | private $languageDenyList = []; |
| 11 | |
| 12 | /** @var string[] only languages where this filter should be used, by language codes */ |
| 13 | private $languageAllowList = []; |
| 14 | |
| 15 | /** @var string[] plugins that must be present to use the filter */ |
| 16 | private $requiredPlugins = []; |
| 17 | |
| 18 | /** @var string local filter to use instead if requiredPlugins are not available */ |
| 19 | private $fallbackFilter = ''; |
| 20 | |
| 21 | /** @var string[] which analyzers to apply to; 'text' and 'text_search' by default */ |
| 22 | private $applyToAnalyzers = [ 'text', 'text_search' ]; |
| 23 | |
| 24 | /** @var string tokenizer that must be present to use the filter */ |
| 25 | private $requiredTokenizer = ''; |
| 26 | |
| 27 | /** @var string[] token filters with which the filter is not allowed/needed */ |
| 28 | private $disallowedTokenFilters = []; |
| 29 | |
| 30 | /** @var string[] character filters with which the filter is not allowed/needed */ |
| 31 | private $disallowedCharFilters = []; |
| 32 | |
| 33 | /** @var string[] filters this one must come after. see T268730 */ |
| 34 | private $mustFollowFilters = []; |
| 35 | |
| 36 | public function __construct( string $type = 'filter' ) { |
| 37 | $this->type = $type; |
| 38 | } |
| 39 | |
| 40 | /** |
| 41 | * @param string[] $languageDenyList |
| 42 | * @return self |
| 43 | */ |
| 44 | public function setLanguageDenyList( array $languageDenyList ): self { |
| 45 | $this->languageDenyList = $languageDenyList; |
| 46 | return $this; |
| 47 | } |
| 48 | |
| 49 | /** |
| 50 | * @param string[] $languageAllowList |
| 51 | * @return self |
| 52 | */ |
| 53 | public function setLanguageAllowList( array $languageAllowList ): self { |
| 54 | $this->languageAllowList = $languageAllowList; |
| 55 | return $this; |
| 56 | } |
| 57 | |
| 58 | /** |
| 59 | * @param string[] $requiredPlugins |
| 60 | * @return self |
| 61 | */ |
| 62 | public function setRequiredPlugins( array $requiredPlugins ): self { |
| 63 | $this->requiredPlugins = $requiredPlugins; |
| 64 | return $this; |
| 65 | } |
| 66 | |
| 67 | /** |
| 68 | * @param string $fallbackFilter |
| 69 | * @return self |
| 70 | */ |
| 71 | public function setFallbackFilter( string $fallbackFilter ): self { |
| 72 | $this->fallbackFilter = $fallbackFilter; |
| 73 | return $this; |
| 74 | } |
| 75 | |
| 76 | /** |
| 77 | * @param string[] $applyToAnalyzers |
| 78 | * @return self |
| 79 | */ |
| 80 | public function setApplyToAnalyzers( array $applyToAnalyzers ): self { |
| 81 | $this->applyToAnalyzers = $applyToAnalyzers; |
| 82 | return $this; |
| 83 | } |
| 84 | |
| 85 | public function getApplyToAnalyzers() { |
| 86 | return $this->applyToAnalyzers; |
| 87 | } |
| 88 | |
| 89 | /** |
| 90 | * @param string $requiredTokenizer |
| 91 | * @return self |
| 92 | */ |
| 93 | public function setRequiredTokenizer( string $requiredTokenizer ): self { |
| 94 | $this->requiredTokenizer = $requiredTokenizer; |
| 95 | return $this; |
| 96 | } |
| 97 | |
| 98 | /** |
| 99 | * @param string[] $disallowedTokenFilters |
| 100 | * @return self |
| 101 | */ |
| 102 | public function setDisallowedTokenFilters( array $disallowedTokenFilters ): self { |
| 103 | $this->disallowedTokenFilters = $disallowedTokenFilters; |
| 104 | return $this; |
| 105 | } |
| 106 | |
| 107 | /** |
| 108 | * @param string[] $disallowedCharFilters |
| 109 | * @return self |
| 110 | */ |
| 111 | public function setDisallowedCharFilters( array $disallowedCharFilters ): self { |
| 112 | $this->disallowedCharFilters = $disallowedCharFilters; |
| 113 | return $this; |
| 114 | } |
| 115 | |
| 116 | /** |
| 117 | * @param string[] $mustFollowFilters |
| 118 | * @return self |
| 119 | */ |
| 120 | public function setMustFollowFilters( array $mustFollowFilters ): self { |
| 121 | $this->mustFollowFilters = $mustFollowFilters; |
| 122 | return $this; |
| 123 | } |
| 124 | |
| 125 | /** |
| 126 | * update languages with global custom filters (e.g., homoglyph & nnbsp filters) |
| 127 | * |
| 128 | * @param mixed[] $config |
| 129 | * @param string $language |
| 130 | * @param GlobalCustomFilter[] $customFilters list of filters and info |
| 131 | * @param string[] $installedPlugins |
| 132 | * @return mixed[] updated config |
| 133 | */ |
| 134 | public static function enableGlobalCustomFilters( array $config, string $language, |
| 135 | array $customFilters, array $installedPlugins ) { |
| 136 | foreach ( $customFilters as $filterName => $gcfInfo ) { |
| 137 | if ( !$gcfInfo->languageCheck( $language ) ) { |
| 138 | continue; |
| 139 | } |
| 140 | |
| 141 | if ( !$gcfInfo->pluginsAvailable( $installedPlugins ) ) { |
| 142 | if ( $gcfInfo->fallbackFilter ) { |
| 143 | $filterName = $gcfInfo->fallbackFilter; |
| 144 | } else { |
| 145 | continue; |
| 146 | } |
| 147 | } |
| 148 | |
| 149 | foreach ( $gcfInfo->getApplyToAnalyzers() as $analyzer ) { |
| 150 | if ( $gcfInfo->analyzerCheck( $config, $analyzer, $filterName ) ) { |
| 151 | $config = $gcfInfo->insertGlobalCustomFilter( $config, $analyzer, |
| 152 | $filterName ); |
| 153 | } |
| 154 | } |
| 155 | } |
| 156 | |
| 157 | return $config; |
| 158 | } |
| 159 | |
| 160 | /** |
| 161 | * check language deny and allow lists to see if this filter is allowed in this |
| 162 | * analyzer |
| 163 | * |
| 164 | * @param string $language |
| 165 | * @return bool |
| 166 | */ |
| 167 | private function languageCheck( string $language ): bool { |
| 168 | if ( in_array( $language, $this->languageDenyList ) |
| 169 | || ( $this->languageAllowList && |
| 170 | !in_array( $language, $this->languageAllowList ) ) |
| 171 | ) { |
| 172 | return false; |
| 173 | } |
| 174 | return true; |
| 175 | } |
| 176 | |
| 177 | /** |
| 178 | * check to see if the filter is compatible with the set of installed plugins |
| 179 | * |
| 180 | * @param string[] $installedPlugins |
| 181 | * @return bool |
| 182 | */ |
| 183 | private function pluginsAvailable( array $installedPlugins ): bool { |
| 184 | foreach ( $this->requiredPlugins as $reqPlugin ) { |
| 185 | if ( !in_array( $reqPlugin, $installedPlugins ) ) { |
| 186 | return false; |
| 187 | } |
| 188 | } |
| 189 | return true; |
| 190 | } |
| 191 | |
| 192 | /** |
| 193 | * check to see if the filter is compatible with the configured tokenizer |
| 194 | * |
| 195 | * @param mixed[] $analyzerConfig |
| 196 | * @return bool |
| 197 | */ |
| 198 | private function requiredTokenizerUsed( array $analyzerConfig ): bool { |
| 199 | if ( $this->requiredTokenizer ) { |
| 200 | if ( !array_key_exists( 'tokenizer', $analyzerConfig ) || |
| 201 | $analyzerConfig[ 'tokenizer' ] != $this->requiredTokenizer ) { |
| 202 | return false; |
| 203 | } |
| 204 | } |
| 205 | return true; |
| 206 | } |
| 207 | |
| 208 | /** |
| 209 | * check if any disqualifying token filters are already present |
| 210 | * |
| 211 | * @param mixed[] $config |
| 212 | * @param string $analyzer |
| 213 | * @return bool |
| 214 | */ |
| 215 | private function disallowedTokenFiltersPresent( array $config, string $analyzer ): bool { |
| 216 | $filters = $config['analyzer'][$analyzer]['filter'] ?? []; |
| 217 | foreach ( $this->disallowedTokenFilters as $df ) { |
| 218 | if ( in_array( $df, $filters ) ) { |
| 219 | return true; |
| 220 | } |
| 221 | } |
| 222 | return false; |
| 223 | } |
| 224 | |
| 225 | /** |
| 226 | * check if any disqualifying character filters are already present |
| 227 | * |
| 228 | * @param mixed[] $config |
| 229 | * @param string $analyzer |
| 230 | * @return bool |
| 231 | */ |
| 232 | private function disallowedCharFiltersPresent( array $config, string $analyzer ): bool { |
| 233 | $filters = $config['analyzer'][$analyzer]['char_filter'] ?? []; |
| 234 | |
| 235 | foreach ( $this->disallowedCharFilters as $df ) { |
| 236 | if ( in_array( $df, $filters ) ) { |
| 237 | return true; |
| 238 | } |
| 239 | } |
| 240 | return false; |
| 241 | } |
| 242 | |
| 243 | /** |
| 244 | * check that the analyzer checks all the boxes to insert this filter |
| 245 | * |
| 246 | * @param mixed[] $config |
| 247 | * @param string $analyzer |
| 248 | * @param string $filterName filter we want to add |
| 249 | * @return bool |
| 250 | */ |
| 251 | private function analyzerCheck( array $config, string $analyzer, |
| 252 | string $filterName ): bool { |
| 253 | $filters = $config['analyzer'][$analyzer][$this->type] ?? []; |
| 254 | |
| 255 | if ( !array_key_exists( $analyzer, $config['analyzer'] ) // array exists |
| 256 | || $config['analyzer'][$analyzer]['type'] != 'custom' // array is custom |
| 257 | || !$this->requiredTokenizerUsed( $config['analyzer'][$analyzer] ) |
| 258 | || $this->disallowedTokenFiltersPresent( $config, $analyzer ) |
| 259 | || $this->disallowedCharFiltersPresent( $config, $analyzer ) |
| 260 | || in_array( $filterName, $filters ) // not a duplicate |
| 261 | ) { |
| 262 | return false; |
| 263 | } |
| 264 | |
| 265 | return true; |
| 266 | } |
| 267 | |
| 268 | /** |
| 269 | * insert one of the global custom filters into the right spot in the analysis chain |
| 270 | * |
| 271 | * @param mixed[] $config the analysis config we are modifying |
| 272 | * @param string $analyzer the specifc analyzer we are modifying |
| 273 | * @param string $filterName filter to add |
| 274 | * @return mixed[] updated config |
| 275 | */ |
| 276 | private function insertGlobalCustomFilter( array $config, string $analyzer, |
| 277 | string $filterName ) { |
| 278 | $filters = $config['analyzer'][$analyzer][$this->type] ?? []; |
| 279 | |
| 280 | $lastMustFollow = -1; |
| 281 | foreach ( $this->mustFollowFilters as $mustFollow ) { |
| 282 | $mustFollowIdx = array_keys( $filters, $mustFollow ); |
| 283 | $mustFollowIdx = end( $mustFollowIdx ); |
| 284 | if ( $mustFollowIdx !== false && $mustFollowIdx > $lastMustFollow ) { |
| 285 | $lastMustFollow = $mustFollowIdx; |
| 286 | } |
| 287 | } |
| 288 | array_splice( $filters, $lastMustFollow + 1, 0, $filterName ); |
| 289 | |
| 290 | $config['analyzer'][$analyzer][$this->type] = $filters; |
| 291 | |
| 292 | return $config; |
| 293 | } |
| 294 | |
| 295 | } |