Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 52 |
|
0.00% |
0 / 12 |
CRAP | |
0.00% |
0 / 1 |
| TranslationUnitDTO | |
0.00% |
0 / 52 |
|
0.00% |
0 / 12 |
600 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
2 | |||
| setBlobForType | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
20 | |||
| toArray | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
2 | |||
| getUserBlob | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getMtBlob | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| hasUserBlob | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| hasSourceBlob | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| toCustomArray | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
20 | |||
| toCorporaDumpArray | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
20 | |||
| getRevision | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| getMwSectionNumber | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| getBaseSectionId | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace ContentTranslation\DTO; |
| 5 | |
| 6 | use InvalidArgumentException; |
| 7 | use MediaWiki\Parser\Sanitizer; |
| 8 | use UnexpectedValueException; |
| 9 | |
| 10 | class TranslationUnitDTO { |
| 11 | public const TYPE_SOURCE = 'source'; |
| 12 | public const TYPE_MT = 'mt'; |
| 13 | public const TYPE_USER = 'user'; |
| 14 | |
| 15 | public const VALID_BLOB_TYPES = [ |
| 16 | self::TYPE_MT, |
| 17 | self::TYPE_SOURCE, |
| 18 | self::TYPE_USER |
| 19 | ]; |
| 20 | |
| 21 | private ?int $sequenceId; |
| 22 | /** |
| 23 | * Fields: [ 'engine' => null, 'content' => string, timestamp: ISO string ] |
| 24 | */ |
| 25 | private ?array $source; |
| 26 | /** |
| 27 | * Fields: [ 'engine' => string (e.g. "Google"), 'content' => string, timestamp: ISO string ] |
| 28 | */ |
| 29 | private ?array $mt; |
| 30 | /** |
| 31 | * Fields: [ 'engine' => null, 'content' => string, timestamp: ISO string ] |
| 32 | */ |
| 33 | private ?array $user; |
| 34 | private string $sectionId; |
| 35 | |
| 36 | public function __construct( |
| 37 | string $sectionId, |
| 38 | ?int $sequenceId, |
| 39 | ?array $source = null, |
| 40 | ?array $mt = null, |
| 41 | ?array $user = null |
| 42 | ) { |
| 43 | $this->sectionId = $sectionId; |
| 44 | $this->sequenceId = $sequenceId; |
| 45 | $this->source = $source; |
| 46 | $this->mt = $mt; |
| 47 | $this->user = $user; |
| 48 | } |
| 49 | |
| 50 | /** |
| 51 | * @param string $type |
| 52 | * @param array|null $blob |
| 53 | */ |
| 54 | public function setBlobForType( string $type, ?array $blob ): void { |
| 55 | if ( !in_array( $type, self::VALID_BLOB_TYPES ) ) { |
| 56 | throw new InvalidArgumentException( '[CX] Invalid blob type during translation unit restoration' ); |
| 57 | } |
| 58 | |
| 59 | $existingBlob = $this->$type; |
| 60 | // It's possible we have a "conflict", since we don't enforce uniqueness |
| 61 | // in the database. In this case, the one with the latest timestamp is used. |
| 62 | // Note: TS_ISO_8601 is suitable for string comparison if timezone is Z. |
| 63 | /** @phan-suppress-next-line PhanTypeArraySuspiciousNullable */ |
| 64 | if ( $existingBlob === null || $blob['timestamp'] > $existingBlob['timestamp'] ) { |
| 65 | $this->$type = $blob; |
| 66 | } |
| 67 | } |
| 68 | |
| 69 | /** |
| 70 | * Returned fields: {sequenceid: int, mt: array, user: array, source: array} |
| 71 | * @return array |
| 72 | */ |
| 73 | public function toArray() { |
| 74 | return [ |
| 75 | 'sequenceid' => $this->sequenceId, |
| 76 | 'mt' => $this->mt, |
| 77 | 'user' => $this->user, |
| 78 | 'source' => $this->source, |
| 79 | ]; |
| 80 | } |
| 81 | |
| 82 | public function getUserBlob(): ?array { |
| 83 | return $this->user; |
| 84 | } |
| 85 | |
| 86 | public function getMtBlob(): ?array { |
| 87 | return $this->mt; |
| 88 | } |
| 89 | |
| 90 | /** |
| 91 | * This method returns a boolean indicating whether the translation unit has user provided input. |
| 92 | * @return bool |
| 93 | */ |
| 94 | public function hasUserBlob(): bool { |
| 95 | return $this->user !== null; |
| 96 | } |
| 97 | |
| 98 | /** |
| 99 | * This method returns a boolean indicating whether the translation unit has source content |
| 100 | * @return bool |
| 101 | */ |
| 102 | public function hasSourceBlob(): bool { |
| 103 | return $this->source !== null; |
| 104 | } |
| 105 | |
| 106 | /** |
| 107 | * This method returns an array representation of the translation unit, suitable to be used |
| 108 | * as payload for each (sub)section for the "contenttranslationcorpora" endpoint response. |
| 109 | * |
| 110 | * @param string[] $types The types that are used for this translation unit. e.g. ['source', 'user'] |
| 111 | * @param bool $sanitize |
| 112 | * @return array |
| 113 | */ |
| 114 | public function toCustomArray( array $types, bool $sanitize ): array { |
| 115 | $unit = [ 'sequenceid' => $this->sequenceId ]; |
| 116 | |
| 117 | $usedTypes = array_intersect( self::VALID_BLOB_TYPES, $types ); |
| 118 | |
| 119 | // filter out unused blob type fields |
| 120 | foreach ( $usedTypes as $usedType ) { |
| 121 | $unit[$usedType] = $this->$usedType; |
| 122 | if ( !isset( $unit[$usedType] ) ) { |
| 123 | continue; |
| 124 | } |
| 125 | |
| 126 | if ( $sanitize ) { |
| 127 | // @phan-suppress-next-line PhanTypeArraySuspiciousNullable |
| 128 | $unit[$usedType]['content'] = Sanitizer::stripAllTags( $unit[$usedType]['content'] ); |
| 129 | } |
| 130 | } |
| 131 | return $unit; |
| 132 | } |
| 133 | |
| 134 | /** |
| 135 | * Returned fields: {sequenceid: int, mt: array, user: array, source: array} |
| 136 | * |
| 137 | * @param bool $sanitize |
| 138 | * @return array |
| 139 | */ |
| 140 | public function toCorporaDumpArray( bool $sanitize ): array { |
| 141 | $unit = [ 'sequenceid' => $this->sequenceId ]; |
| 142 | |
| 143 | $unit['mt'] = $this->mt; |
| 144 | $unit['user'] = [ |
| 145 | // @phan-suppress-next-line PhanTypeArraySuspiciousNullable |
| 146 | 'content' => $this->user['content'], |
| 147 | // @phan-suppress-next-line PhanTypeArraySuspiciousNullable |
| 148 | 'timestamp' => $this->user['timestamp'] |
| 149 | ]; |
| 150 | $unit['source'] = [ |
| 151 | // @phan-suppress-next-line PhanTypeArraySuspiciousNullable |
| 152 | 'content' => $this->source['content'], |
| 153 | // @phan-suppress-next-line PhanTypeArraySuspiciousNullable |
| 154 | 'timestamp' => $this->source['timestamp'] |
| 155 | ]; |
| 156 | |
| 157 | if ( $sanitize ) { |
| 158 | // filter out unused blob type fields |
| 159 | foreach ( self::VALID_BLOB_TYPES as $type ) { |
| 160 | if ( isset( $unit[$type]['content'] ) ) { |
| 161 | $unit[$type]['content'] = Sanitizer::stripAllTags( $unit[$type]['content'] ); |
| 162 | } |
| 163 | } |
| 164 | } |
| 165 | |
| 166 | return $unit; |
| 167 | } |
| 168 | |
| 169 | public function getRevision(): ?int { |
| 170 | $sectionIdPieces = explode( '_', $this->sectionId ); |
| 171 | |
| 172 | return isset( $sectionIdPieces[0] ) ? (int)$sectionIdPieces[0] : null; |
| 173 | } |
| 174 | |
| 175 | public function getMwSectionNumber(): ?int { |
| 176 | $sectionIdPieces = explode( '_', $this->sectionId ); |
| 177 | |
| 178 | return isset( $sectionIdPieces[1] ) ? (int)$sectionIdPieces[1] : null; |
| 179 | } |
| 180 | |
| 181 | public function getBaseSectionId(): string { |
| 182 | $sectionIdPieces = explode( '_', $this->sectionId ); |
| 183 | |
| 184 | // sectionId should be in the following format: ${revision}_${sectionNumber}_${subSectionId} |
| 185 | if ( count( $sectionIdPieces ) < 3 ) { |
| 186 | throw new UnexpectedValueException( '[CX] Invalid format for section id of the translation unit DTO' ); |
| 187 | } |
| 188 | |
| 189 | return "$sectionIdPieces[0]_$sectionIdPieces[1]"; |
| 190 | } |
| 191 | |
| 192 | } |