Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 122 |
|
0.00% |
0 / 9 |
CRAP | |
0.00% |
0 / 1 |
| TranslationStatsDataProvider | |
0.00% |
0 / 122 |
|
0.00% |
0 / 9 |
1980 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| getGraphSpecifications | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getGraphTypes | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getGraphData | |
0.00% |
0 / 72 |
|
0.00% |
0 / 1 |
420 | |||
| getStatsProvider | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
2 | |||
| roundTimestampToCutoff | |
0.00% |
0 / 18 |
|
0.00% |
0 / 1 |
110 | |||
| roundingAddition | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| getIncrement | |
0.00% |
0 / 10 |
|
0.00% |
0 / 1 |
30 | |||
| makeTimeCondition | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
12 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace MediaWiki\Extension\Translate\Statistics; |
| 5 | |
| 6 | use MediaWiki\Config\ServiceOptions; |
| 7 | use MediaWiki\Extension\Translate\MessageGroupProcessing\MessageGroups; |
| 8 | use MediaWiki\Extension\Translate\Utilities\Utilities; |
| 9 | use MediaWiki\Language\Language; |
| 10 | use Wikimedia\ObjectFactory\ObjectFactory; |
| 11 | use Wikimedia\Rdbms\IConnectionProvider; |
| 12 | use Wikimedia\Rdbms\IReadableDatabase; |
| 13 | use const TS_MW; |
| 14 | |
| 15 | /** |
| 16 | * Provides translation stats data |
| 17 | * @author Abijeet Patro |
| 18 | * @license GPL-2.0-or-later |
| 19 | * @since 2020.09 |
| 20 | */ |
| 21 | class TranslationStatsDataProvider { |
| 22 | public const CONSTRUCTOR_OPTIONS = [ |
| 23 | 'TranslateStatsProviders' |
| 24 | ]; |
| 25 | |
| 26 | private ObjectFactory $objectFactory; |
| 27 | private ServiceOptions $options; |
| 28 | private IConnectionProvider $dbProvider; |
| 29 | |
| 30 | public function __construct( |
| 31 | ServiceOptions $options, |
| 32 | ObjectFactory $objectFactory, |
| 33 | IConnectionProvider $dbProvider |
| 34 | ) { |
| 35 | $this->options = $options; |
| 36 | $this->objectFactory = $objectFactory; |
| 37 | $this->dbProvider = $dbProvider; |
| 38 | } |
| 39 | |
| 40 | private function getGraphSpecifications(): array { |
| 41 | return array_filter( $this->options->get( 'TranslateStatsProviders' ) ); |
| 42 | } |
| 43 | |
| 44 | public function getGraphTypes(): array { |
| 45 | return array_keys( $this->getGraphSpecifications() ); |
| 46 | } |
| 47 | |
| 48 | /** |
| 49 | * Fetches and preprocesses graph data that can be fed to graph drawer. |
| 50 | * @param TranslationStatsGraphOptions $opts |
| 51 | * @param Language $language |
| 52 | * @return array ( string => array ) Data indexed by their date labels. |
| 53 | */ |
| 54 | public function getGraphData( TranslationStatsGraphOptions $opts, Language $language ): array { |
| 55 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 56 | |
| 57 | $so = $this->getStatsProvider( $opts->getValue( 'count' ), $opts ); |
| 58 | |
| 59 | $fixedStart = $opts->getValue( 'start' ) !== ''; |
| 60 | |
| 61 | $now = time(); |
| 62 | $period = 3600 * 24 * $opts->getValue( 'days' ); |
| 63 | |
| 64 | if ( $fixedStart ) { |
| 65 | $cutoff = (int)wfTimestamp( TS_UNIX, $opts->getValue( 'start' ) ); |
| 66 | } else { |
| 67 | $cutoff = $now - $period; |
| 68 | } |
| 69 | $cutoff = self::roundTimestampToCutoff( $opts->getValue( 'scale' ), $cutoff, 'earlier' ); |
| 70 | |
| 71 | $start = $cutoff; |
| 72 | |
| 73 | if ( $fixedStart ) { |
| 74 | $end = self::roundTimestampToCutoff( $opts->getValue( 'scale' ), $start + $period, 'later' ) - 1; |
| 75 | } else { |
| 76 | $end = null; |
| 77 | } |
| 78 | |
| 79 | $timestampColumn = $so->getTimestampColumn(); |
| 80 | |
| 81 | $selectQueryBuilder = $so->createQueryBuilder( $dbr, __METHOD__ ); |
| 82 | $selectQueryBuilder->andWhere( |
| 83 | $this->makeTimeCondition( |
| 84 | $dbr, |
| 85 | $timestampColumn, |
| 86 | wfTimestamp( TS_MW, $start ), |
| 87 | wfTimestampOrNull( TS_MW, $end ) |
| 88 | ) |
| 89 | ); |
| 90 | |
| 91 | $res = $selectQueryBuilder->fetchResultSet(); |
| 92 | wfDebug( __METHOD__ . "-queryend\n" ); |
| 93 | |
| 94 | // Start processing the data |
| 95 | $dateFormat = $so->getDateFormat(); |
| 96 | $increment = self::getIncrement( $opts->getValue( 'scale' ) ); |
| 97 | |
| 98 | $labels = $so->labels(); |
| 99 | $keys = array_keys( $labels ); |
| 100 | $values = array_pad( [], count( $labels ), 0 ); |
| 101 | $defaults = array_combine( $keys, $values ); |
| 102 | |
| 103 | $data = []; |
| 104 | // Allow 10 seconds in the future for processing time |
| 105 | $lastValue = $end ?? $now + 10; |
| 106 | while ( $cutoff <= $lastValue ) { |
| 107 | $date = $language->sprintfDate( $dateFormat, wfTimestamp( TS_MW, $cutoff ) ); |
| 108 | $cutoff += $increment; |
| 109 | $data[$date] = $defaults; |
| 110 | } |
| 111 | // Ensure $lastValue is within range, in case the loop above jumped over it |
| 112 | $data[$language->sprintfDate( $dateFormat, wfTimestamp( TS_MW, $lastValue ) )] = $defaults; |
| 113 | |
| 114 | // Processing |
| 115 | $labelToIndex = array_flip( $labels ); |
| 116 | |
| 117 | foreach ( $res as $row ) { |
| 118 | $indexLabels = $so->indexOf( $row ); |
| 119 | if ( $indexLabels === null ) { |
| 120 | continue; |
| 121 | } |
| 122 | |
| 123 | foreach ( $indexLabels as $i ) { |
| 124 | if ( !isset( $labelToIndex[$i] ) ) { |
| 125 | continue; |
| 126 | } |
| 127 | $date = $language->sprintfDate( $dateFormat, $row->$timestampColumn ); |
| 128 | // Ignore values outside range |
| 129 | if ( !isset( $data[$date] ) ) { |
| 130 | continue; |
| 131 | } |
| 132 | |
| 133 | $data[$date][$labelToIndex[$i]]++; |
| 134 | } |
| 135 | } |
| 136 | |
| 137 | // Don't display dummy label |
| 138 | if ( count( $labels ) === 1 && $labels[0] === 'all' ) { |
| 139 | $labels = []; |
| 140 | } |
| 141 | |
| 142 | foreach ( $labels as &$label ) { |
| 143 | if ( !str_contains( $label, '@' ) ) { |
| 144 | continue; |
| 145 | } |
| 146 | [ $groupId, $code ] = explode( '@', $label, 2 ); |
| 147 | if ( $code && $groupId ) { |
| 148 | $code = Utilities::getLanguageName( $code, $language->getCode() ) . " ($code)"; |
| 149 | $group = MessageGroups::getGroup( $groupId ); |
| 150 | $group = $group ? $group->getLabel() : $groupId; |
| 151 | $label = "$group @ $code"; |
| 152 | } elseif ( $code ) { |
| 153 | $label = Utilities::getLanguageName( $code, $language->getCode() ) . " ($code)"; |
| 154 | } elseif ( $groupId ) { |
| 155 | $group = MessageGroups::getGroup( $groupId ); |
| 156 | $label = $group ? $group->getLabel() : $groupId; |
| 157 | } |
| 158 | } |
| 159 | |
| 160 | // Indicator that the last value is not full |
| 161 | if ( $end === null ) { |
| 162 | // Warning: do not user array_splice, which does not preserve numerical keys |
| 163 | $last = end( $data ); |
| 164 | $key = key( $data ); |
| 165 | unset( $data[$key] ); |
| 166 | $data[ "$key*" ] = $last; |
| 167 | } |
| 168 | |
| 169 | return [ $labels, $data ]; |
| 170 | } |
| 171 | |
| 172 | /** @noinspection PhpIncompatibleReturnTypeInspection */ |
| 173 | private function getStatsProvider( string $type, TranslationStatsGraphOptions $opts ): TranslationStatsInterface { |
| 174 | $specs = $this->getGraphSpecifications(); |
| 175 | return $this->objectFactory->createObject( |
| 176 | $specs[$type], |
| 177 | [ |
| 178 | 'allowClassName' => true, |
| 179 | 'extraArgs' => [ $opts ], |
| 180 | ] |
| 181 | ); |
| 182 | } |
| 183 | |
| 184 | /** |
| 185 | * Gets the closest earliest timestamp that corresponds to start of a |
| 186 | * period in given scale, like, midnight, monday or first day of the month. |
| 187 | */ |
| 188 | private static function roundTimestampToCutoff( |
| 189 | string $scale, int $cutoff, string $direction = 'earlier' |
| 190 | ): int { |
| 191 | $dir = $direction === 'earlier' ? -1 : 1; |
| 192 | |
| 193 | /* Ensure that the first item in the graph has full data even |
| 194 | * if it doesn't align with the given 'days' boundary */ |
| 195 | if ( $scale === 'hours' ) { |
| 196 | $cutoff += self::roundingAddition( $cutoff, 3600, $dir ); |
| 197 | } elseif ( $scale === 'days' ) { |
| 198 | $cutoff += self::roundingAddition( $cutoff, 86400, $dir ); |
| 199 | } elseif ( $scale === 'weeks' ) { |
| 200 | /* Here we assume that week starts on monday, which does not |
| 201 | * always hold true. Go Xwards day by day until we are on monday */ |
| 202 | while ( date( 'D', $cutoff ) !== 'Mon' ) { |
| 203 | $cutoff += $dir * 86400; |
| 204 | } |
| 205 | // Round to nearest day |
| 206 | $cutoff -= ( $cutoff % 86400 ); |
| 207 | } elseif ( $scale === 'months' ) { |
| 208 | // Go Xwards/ day by day until we are on the first day of the month |
| 209 | while ( date( 'j', $cutoff ) !== '1' ) { |
| 210 | $cutoff += $dir * 86400; |
| 211 | } |
| 212 | // Round to nearest day |
| 213 | $cutoff -= ( $cutoff % 86400 ); |
| 214 | } elseif ( $scale === 'years' ) { |
| 215 | // Go Xwards/ day by day until we are on the first day of the year |
| 216 | while ( date( 'z', $cutoff ) !== '0' ) { |
| 217 | $cutoff += $dir * 86400; |
| 218 | } |
| 219 | // Round to nearest day |
| 220 | $cutoff -= ( $cutoff % 86400 ); |
| 221 | } |
| 222 | |
| 223 | return $cutoff; |
| 224 | } |
| 225 | |
| 226 | private static function roundingAddition( int $ts, int $amount, int $dir ): int { |
| 227 | if ( $dir === -1 ) { |
| 228 | return -1 * ( $ts % $amount ); |
| 229 | } else { |
| 230 | return $amount - ( $ts % $amount ); |
| 231 | } |
| 232 | } |
| 233 | |
| 234 | /** |
| 235 | * Returns an increment in seconds for a given scale. |
| 236 | * The increment must be small enough that we will hit every item in the |
| 237 | * scale when using different multiples of the increment. It should be |
| 238 | * large enough to avoid hitting the same item multiple times. |
| 239 | */ |
| 240 | private static function getIncrement( string $scale ): int { |
| 241 | $increment = 3600 * 24; |
| 242 | if ( $scale === 'years' ) { |
| 243 | $increment = 3600 * 24 * 350; |
| 244 | } elseif ( $scale === 'months' ) { |
| 245 | /* We use increment to fill up the values. Use number small enough |
| 246 | * to ensure we hit each month */ |
| 247 | $increment = 3600 * 24 * 15; |
| 248 | } elseif ( $scale === 'weeks' ) { |
| 249 | $increment = 3600 * 24 * 7; |
| 250 | } elseif ( $scale === 'hours' ) { |
| 251 | $increment = 3600; |
| 252 | } |
| 253 | |
| 254 | return $increment; |
| 255 | } |
| 256 | |
| 257 | /** @return string[] */ |
| 258 | private function makeTimeCondition( |
| 259 | IReadableDatabase $database, |
| 260 | string $field, |
| 261 | ?string $start, |
| 262 | ?string $end |
| 263 | ): array { |
| 264 | $conditions = []; |
| 265 | if ( $start !== null ) { |
| 266 | $conditions[] = "$field >= " . $database->addQuotes( $database->timestamp( $start ) ); |
| 267 | } |
| 268 | if ( $end !== null ) { |
| 269 | $conditions[] = "$field <= " . $database->addQuotes( $database->timestamp( $end ) ); |
| 270 | } |
| 271 | |
| 272 | return $conditions; |
| 273 | } |
| 274 | } |