Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
25.84% |
23 / 89 |
|
87.50% |
7 / 8 |
CRAP | |
0.00% |
0 / 1 |
| PhraseSuggesterProfileRepoWrapper | |
25.84% |
23 / 89 |
|
87.50% |
7 / 8 |
1111.72 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| fromFile | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| fromConfig | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| repositoryType | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| repositoryName | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getProfile | |
18.52% |
15 / 81 |
|
0.00% |
0 / 1 |
1091.33 | |||
| hasProfile | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| listExposedProfiles | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace CirrusSearch\Profile; |
| 4 | |
| 5 | use CirrusSearch\Util; |
| 6 | use MediaWiki\Config\Config; |
| 7 | use Wikimedia\ObjectCache\BagOStuff; |
| 8 | |
| 9 | /** |
| 10 | * Wrapper to augment the phrase suggester profile settings |
| 11 | * with customization on-wiki using system messages. |
| 12 | */ |
| 13 | class PhraseSuggesterProfileRepoWrapper implements SearchProfileRepository { |
| 14 | |
| 15 | private const MAX_ERRORS_HARD_LIMIT = 2; |
| 16 | private const MAX_TERM_FREQ_HARD_LIMIT = 0.6; |
| 17 | private const PREFIX_LENGTH_HARD_LIMIT = 2; |
| 18 | public const CIRRUSSEARCH_DIDYOUMEAN_SETTINGS = 'cirrussearch-didyoumean-settings'; |
| 19 | |
| 20 | /** |
| 21 | * @var string[] |
| 22 | */ |
| 23 | private static $ALLOWED_MODE = [ 'missing', 'popular', 'always' ]; |
| 24 | |
| 25 | /** |
| 26 | * @var SearchProfileRepository |
| 27 | */ |
| 28 | private $wrapped; |
| 29 | |
| 30 | /** |
| 31 | * @var BagOStuff |
| 32 | */ |
| 33 | private $bagOStuff; |
| 34 | |
| 35 | public function __construct( SearchProfileRepository $wrapped, BagOStuff $bagOStuff ) { |
| 36 | $this->wrapped = $wrapped; |
| 37 | $this->bagOStuff = $bagOStuff; |
| 38 | } |
| 39 | |
| 40 | /** |
| 41 | * @param string $type |
| 42 | * @param string $name |
| 43 | * @param string $phpFile |
| 44 | * @param BagOStuff $cache |
| 45 | * @return SearchProfileRepository |
| 46 | */ |
| 47 | public static function fromFile( $type, $name, $phpFile, BagOStuff $cache ) { |
| 48 | return new self( ArrayProfileRepository::fromFile( $type, $name, $phpFile ), $cache ); |
| 49 | } |
| 50 | |
| 51 | /** |
| 52 | * @param string $type |
| 53 | * @param string $name |
| 54 | * @param string $configEntry |
| 55 | * @param Config $config |
| 56 | * @param BagOStuff $cache |
| 57 | * @return self |
| 58 | */ |
| 59 | public static function fromConfig( $type, $name, $configEntry, Config $config, BagOStuff $cache ): self { |
| 60 | return new self( new ConfigProfileRepository( $type, $name, $configEntry, $config ), $cache ); |
| 61 | } |
| 62 | |
| 63 | /** |
| 64 | * The repository type |
| 65 | * @return string |
| 66 | */ |
| 67 | public function repositoryType() { |
| 68 | return $this->wrapped->repositoryType(); |
| 69 | } |
| 70 | |
| 71 | /** |
| 72 | * The repository name |
| 73 | * @return string |
| 74 | */ |
| 75 | public function repositoryName() { |
| 76 | return $this->wrapped->repositoryName(); |
| 77 | } |
| 78 | |
| 79 | /** |
| 80 | * Load a profile named $name |
| 81 | * @param string $name |
| 82 | * @return array|null the profile data or null if not found |
| 83 | */ |
| 84 | public function getProfile( $name ) { |
| 85 | $settings = $this->wrapped->getProfile( $name ); |
| 86 | if ( $settings === null ) { |
| 87 | return null; |
| 88 | } |
| 89 | $lines = $this->bagOStuff->getWithSetCallback( |
| 90 | $this->bagOStuff->makeKey( self::CIRRUSSEARCH_DIDYOUMEAN_SETTINGS ), |
| 91 | 600, |
| 92 | static function () { |
| 93 | $source = wfMessage( 'cirrussearch-didyoumean-settings' )->inContentLanguage(); |
| 94 | if ( $source->isDisabled() ) { |
| 95 | return []; |
| 96 | } |
| 97 | return Util::parseSettingsInMessage( $source->plain() ); |
| 98 | } |
| 99 | ); |
| 100 | |
| 101 | $laplaceAlpha = null; |
| 102 | $stupidBackoffDiscount = null; |
| 103 | foreach ( $lines as $line ) { |
| 104 | $linePieces = explode( ':', $line, 2 ); |
| 105 | if ( count( $linePieces ) !== 2 ) { |
| 106 | // Skip improperly formatted lines without a key:value |
| 107 | continue; |
| 108 | } |
| 109 | [ $k, $v ] = $linePieces; |
| 110 | |
| 111 | switch ( $k ) { |
| 112 | case 'max_errors': |
| 113 | if ( is_numeric( $v ) && $v >= 1 && $v <= self::MAX_ERRORS_HARD_LIMIT ) { |
| 114 | $settings['max_errors'] = floatval( $v ); |
| 115 | } |
| 116 | break; |
| 117 | case 'confidence': |
| 118 | if ( is_numeric( $v ) && $v >= 0 ) { |
| 119 | $settings['confidence'] = floatval( $v ); |
| 120 | } |
| 121 | break; |
| 122 | case 'max_term_freq': |
| 123 | if ( is_numeric( $v ) && $v >= 0 && $v <= self::MAX_TERM_FREQ_HARD_LIMIT ) { |
| 124 | $settings['max_term_freq'] = floatval( $v ); |
| 125 | } |
| 126 | break; |
| 127 | case 'min_doc_freq': |
| 128 | if ( is_numeric( $v ) && $v >= 0 && $v < 1 ) { |
| 129 | $settings['min_doc_freq'] = floatval( $v ); |
| 130 | } |
| 131 | break; |
| 132 | case 'prefix_length': |
| 133 | if ( is_numeric( $v ) && $v >= 0 && $v <= self::PREFIX_LENGTH_HARD_LIMIT ) { |
| 134 | $settings['prefix_length'] = intval( $v ); |
| 135 | } |
| 136 | break; |
| 137 | case 'suggest_mode': |
| 138 | if ( in_array( $v, self::$ALLOWED_MODE ) ) { |
| 139 | $settings['mode'] = $v; |
| 140 | } |
| 141 | break; |
| 142 | case 'collate': |
| 143 | if ( $v === 'true' ) { |
| 144 | $settings['collate'] = true; |
| 145 | } elseif ( $v === 'false' ) { |
| 146 | $settings['collate'] = false; |
| 147 | } |
| 148 | break; |
| 149 | case 'smoothing': |
| 150 | if ( $v === 'laplace' ) { |
| 151 | $settings['smoothing_model'] = [ |
| 152 | 'laplace' => [ |
| 153 | 'alpha' => 0.5 |
| 154 | ] |
| 155 | ]; |
| 156 | } elseif ( $v === 'stupid_backoff' ) { |
| 157 | $settings['smoothing_model'] = [ |
| 158 | 'stupid_backoff' => [ |
| 159 | 'discount' => 0.4 |
| 160 | ] |
| 161 | ]; |
| 162 | } |
| 163 | break; |
| 164 | case 'laplace_alpha': |
| 165 | if ( is_numeric( $v ) && $v >= 0 && $v <= 1 ) { |
| 166 | $laplaceAlpha = floatval( $v ); |
| 167 | } |
| 168 | break; |
| 169 | case 'stupid_backoff_discount': |
| 170 | if ( is_numeric( $v ) && $v >= 0 && $v <= 1 ) { |
| 171 | $stupidBackoffDiscount = floatval( $v ); |
| 172 | } |
| 173 | break; |
| 174 | } |
| 175 | } |
| 176 | |
| 177 | // Apply smoothing model options, if none provided we'll use elasticsearch defaults |
| 178 | if ( isset( $settings['smoothing_model']['laplace'] ) && $laplaceAlpha !== null ) { |
| 179 | $settings['smoothing_model']['laplace'] = [ |
| 180 | 'alpha' => $laplaceAlpha |
| 181 | ]; |
| 182 | } |
| 183 | if ( isset( $settings['smoothing_model']['stupid_backoff'] ) && $stupidBackoffDiscount !== null ) { |
| 184 | $settings['smoothing_model']['stupid_backoff'] = [ |
| 185 | 'discount' => $stupidBackoffDiscount |
| 186 | ]; |
| 187 | } |
| 188 | return $settings; |
| 189 | } |
| 190 | |
| 191 | /** |
| 192 | * Check if a profile named $name exists in this repository |
| 193 | * @param string $name |
| 194 | * @return bool |
| 195 | */ |
| 196 | public function hasProfile( $name ) { |
| 197 | return $this->wrapped->hasProfile( $name ); |
| 198 | } |
| 199 | |
| 200 | /** |
| 201 | * Get the list of profiles that we want to expose to the user. |
| 202 | * |
| 203 | * @return array[] list of profiles index by name |
| 204 | */ |
| 205 | public function listExposedProfiles() { |
| 206 | return $this->wrapped->listExposedProfiles(); |
| 207 | } |
| 208 | } |