Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
62.20% |
79 / 127 |
|
33.33% |
2 / 6 |
CRAP | |
0.00% |
0 / 1 |
| CsvTranslationImporter | |
62.20% |
79 / 127 |
|
33.33% |
2 / 6 |
121.12 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| parseFile | |
90.00% |
54 / 60 |
|
0.00% |
0 / 1 |
17.29 | |||
| importData | |
0.00% |
0 / 38 |
|
0.00% |
0 / 1 |
90 | |||
| getLanguagesFromHeader | |
85.00% |
17 / 20 |
|
0.00% |
0 / 1 |
6.12 | |||
| getMessageHandleIfValid | |
85.71% |
6 / 7 |
|
0.00% |
0 / 1 |
3.03 | |||
| isCsvRowEmpty | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
3 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace MediaWiki\Extension\Translate\MessageGroupProcessing; |
| 5 | |
| 6 | use MediaWiki\CommentStore\CommentStoreComment; |
| 7 | use MediaWiki\Content\ContentHandler; |
| 8 | use MediaWiki\Extension\Translate\MessageLoading\MessageHandle; |
| 9 | use MediaWiki\Extension\Translate\Utilities\Utilities; |
| 10 | use MediaWiki\Page\WikiPageFactory; |
| 11 | use MediaWiki\Permissions\Authority; |
| 12 | use MediaWiki\Revision\SlotRecord; |
| 13 | use MediaWiki\Status\Status; |
| 14 | use MediaWiki\Title\Title; |
| 15 | use SplFileObject; |
| 16 | |
| 17 | /** |
| 18 | * Parse, validate and import translations from a CSV file |
| 19 | * @since 2022.06 |
| 20 | * @license GPL-2.0-or-later |
| 21 | * @author Abijeet Patro |
| 22 | */ |
| 23 | class CsvTranslationImporter { |
| 24 | private WikiPageFactory $wikiPageFactory; |
| 25 | |
| 26 | public function __construct( WikiPageFactory $wikiPageFactory ) { |
| 27 | $this->wikiPageFactory = $wikiPageFactory; |
| 28 | } |
| 29 | |
| 30 | /** Parse and validate the CSV file */ |
| 31 | public function parseFile( string $csvFilePath ): Status { |
| 32 | if ( !file_exists( $csvFilePath ) || !is_file( $csvFilePath ) ) { |
| 33 | return Status::newFatal( |
| 34 | "CSV file path '$csvFilePath' does not exist, is not readable or is not a file" |
| 35 | ); |
| 36 | } |
| 37 | |
| 38 | $indexedLanguageCodes = []; |
| 39 | $currentRowCount = -1; |
| 40 | $importData = []; |
| 41 | $invalidRows = [ |
| 42 | 'emptyTitleRows' => [], |
| 43 | 'invalidTitleRows' => [], |
| 44 | 'groupNotFoundRows' => [] |
| 45 | ]; |
| 46 | |
| 47 | $csvFileContent = new SplFileObject( $csvFilePath, 'r' ); |
| 48 | while ( !$csvFileContent->eof() ) { |
| 49 | // Increment the row count at the beginning since we have a bunch of jump statements |
| 50 | // at various placaes |
| 51 | ++$currentRowCount; |
| 52 | |
| 53 | $csvRow = $csvFileContent->fgetcsv(); |
| 54 | if ( $this->isCsvRowEmpty( $csvRow ) ) { |
| 55 | continue; |
| 56 | } |
| 57 | |
| 58 | if ( $currentRowCount === 0 ) { |
| 59 | // Validate the header |
| 60 | $status = $this->getLanguagesFromHeader( $csvRow ); |
| 61 | if ( !$status->isGood() ) { |
| 62 | return $status; |
| 63 | } |
| 64 | /** @var string[] $indexedLanguageCodes */ |
| 65 | $indexedLanguageCodes = $status->getValue(); |
| 66 | continue; |
| 67 | } |
| 68 | |
| 69 | $rowData = [ 'translations' => [] ]; |
| 70 | $messageTitle = isset( $csvRow[0] ) ? trim( $csvRow[0] ) : null; |
| 71 | if ( !$messageTitle ) { |
| 72 | $invalidRows['emptyTitleRows'][] = $currentRowCount + 1; |
| 73 | continue; |
| 74 | } |
| 75 | |
| 76 | $handle = $this->getMessageHandleIfValid( $messageTitle ); |
| 77 | if ( $handle === null ) { |
| 78 | $invalidRows['invalidTitleRows'][] = $currentRowCount + 1; |
| 79 | continue; |
| 80 | } |
| 81 | |
| 82 | // Ensure that the group is present |
| 83 | $group = $handle->getGroup(); |
| 84 | if ( !$group ) { |
| 85 | $invalidRows['groupNotFoundRows'][] = $currentRowCount + 1; |
| 86 | continue; |
| 87 | } |
| 88 | |
| 89 | $sourceLanguage = $group->getSourceLanguage(); |
| 90 | |
| 91 | $rowData['messageTitle'] = $messageTitle; |
| 92 | foreach ( $indexedLanguageCodes as $languageCode => $index ) { |
| 93 | if ( $sourceLanguage === $languageCode ) { |
| 94 | continue; |
| 95 | } |
| 96 | |
| 97 | $rowData['translations'][$languageCode] = $csvRow[$index] ?? null; |
| 98 | } |
| 99 | $importData[] = $rowData; |
| 100 | } |
| 101 | |
| 102 | $status = new Status(); |
| 103 | if ( $invalidRows['emptyTitleRows'] ) { |
| 104 | $status->fatal( |
| 105 | 'Empty message titles found on row(s): ' . implode( ',', $invalidRows['emptyTitleRows'] ) |
| 106 | ); |
| 107 | } |
| 108 | |
| 109 | if ( $invalidRows['invalidTitleRows'] ) { |
| 110 | $status->fatal( |
| 111 | 'Invalid message title(s) found on row(s): ' . implode( ',', $invalidRows['invalidTitleRows'] ) |
| 112 | ); |
| 113 | } |
| 114 | |
| 115 | if ( $invalidRows['groupNotFoundRows'] ) { |
| 116 | $status->fatal( |
| 117 | 'Group not found for message(s) on row(s) ' . implode( ',', $invalidRows['invalidTitleRows'] ) |
| 118 | ); |
| 119 | } |
| 120 | |
| 121 | if ( !$status->isGood() ) { |
| 122 | return $status; |
| 123 | } |
| 124 | |
| 125 | return Status::newGood( $importData ); |
| 126 | } |
| 127 | |
| 128 | /** Import the data returned from the parseFile method */ |
| 129 | public function importData( |
| 130 | array $messagesWithTranslations, |
| 131 | Authority $authority, |
| 132 | string $comment, |
| 133 | ?callable $progressReporter = null |
| 134 | ): Status { |
| 135 | $commentStoreComment = CommentStoreComment::newUnsavedComment( $comment ); |
| 136 | |
| 137 | // Loop over each translation to import |
| 138 | $importStatus = new Status(); |
| 139 | $failedStatuses = []; |
| 140 | $currentTranslation = 0; |
| 141 | foreach ( $messagesWithTranslations as $messageTranslation ) { |
| 142 | $messageTitleText = $messageTranslation['messageTitle']; |
| 143 | $messageTitle = Title::newFromText( $messageTitleText ); |
| 144 | $messageHandle = new MessageHandle( $messageTitle ); |
| 145 | |
| 146 | $translationImportStatuses = []; |
| 147 | |
| 148 | // Import each translation for the current message |
| 149 | $translations = $messageTranslation['translations']; |
| 150 | foreach ( $translations as $languageCode => $translation ) { |
| 151 | // Skip empty translations |
| 152 | if ( $translation === null || trim( $translation ) === '' ) { |
| 153 | continue; |
| 154 | } |
| 155 | |
| 156 | $translationTitle = $messageHandle->getTitleForLanguage( $languageCode ); |
| 157 | |
| 158 | // Perform the update for the translation page |
| 159 | $updater = $this->wikiPageFactory->newFromTitle( $translationTitle ) |
| 160 | ->newPageUpdater( $authority ); |
| 161 | $content = ContentHandler::makeContent( $translation, $translationTitle ); |
| 162 | $updater->setContent( SlotRecord::MAIN, $content ); |
| 163 | $updater->setFlags( EDIT_FORCE_BOT ); |
| 164 | $updater->saveRevision( $commentStoreComment ); |
| 165 | |
| 166 | $status = $updater->getStatus(); |
| 167 | $translationImportStatuses[] = $status; |
| 168 | if ( !$status->isOK() ) { |
| 169 | $failedStatuses[ $translationTitle->getPrefixedText() ] = $status; |
| 170 | } |
| 171 | } |
| 172 | |
| 173 | ++$currentTranslation; |
| 174 | if ( $progressReporter ) { |
| 175 | call_user_func( |
| 176 | $progressReporter, |
| 177 | $messageTitle, |
| 178 | $translationImportStatuses, |
| 179 | count( $messagesWithTranslations ), |
| 180 | $currentTranslation |
| 181 | ); |
| 182 | } |
| 183 | } |
| 184 | |
| 185 | if ( $failedStatuses ) { |
| 186 | foreach ( $failedStatuses as $failedStatus ) { |
| 187 | $importStatus->merge( $failedStatus ); |
| 188 | } |
| 189 | |
| 190 | $importStatus->setResult( false, $failedStatuses ); |
| 191 | } |
| 192 | |
| 193 | return $importStatus; |
| 194 | } |
| 195 | |
| 196 | private function getLanguagesFromHeader( array $csvHeader ): Status { |
| 197 | if ( count( $csvHeader ) < 2 ) { |
| 198 | return Status::newFatal( |
| 199 | 'CSV has < 2 columns. Assuming that there are no languages to import' |
| 200 | ); |
| 201 | } |
| 202 | |
| 203 | $languageCodesInHeader = array_slice( $csvHeader, 2 ); |
| 204 | if ( $languageCodesInHeader === [] ) { |
| 205 | return Status::newFatal( 'No languages found for import' ); |
| 206 | } |
| 207 | |
| 208 | $invalidLanguageCodes = []; |
| 209 | $indexedLanguageCodes = []; |
| 210 | // First two columns are message title and definition |
| 211 | $originalLanguageIndex = 2; |
| 212 | foreach ( $languageCodesInHeader as $languageCode ) { |
| 213 | if ( !Utilities::isSupportedLanguageCode( strtolower( $languageCode ) ) ) { |
| 214 | $invalidLanguageCodes[] = $languageCode; |
| 215 | } else { |
| 216 | // Language codes maybe in upper case, convert to lower case for further use. |
| 217 | $indexedLanguageCodes[ strtolower( $languageCode ) ] = $originalLanguageIndex; |
| 218 | } |
| 219 | ++$originalLanguageIndex; |
| 220 | } |
| 221 | |
| 222 | if ( $invalidLanguageCodes ) { |
| 223 | return Status::newFatal( |
| 224 | 'Invalid language codes detected in CSV header: ' . implode( ', ', $invalidLanguageCodes ) |
| 225 | ); |
| 226 | } |
| 227 | |
| 228 | return Status::newGood( $indexedLanguageCodes ); |
| 229 | } |
| 230 | |
| 231 | private function getMessageHandleIfValid( string $messageTitle ): ?MessageHandle { |
| 232 | $title = Title::newFromText( $messageTitle ); |
| 233 | if ( $title === null ) { |
| 234 | return null; |
| 235 | } |
| 236 | |
| 237 | $handle = new MessageHandle( $title ); |
| 238 | if ( $handle->isValid() ) { |
| 239 | return $handle; |
| 240 | } |
| 241 | |
| 242 | return null; |
| 243 | } |
| 244 | |
| 245 | private function isCsvRowEmpty( array $csvRow ): bool { |
| 246 | return count( $csvRow ) === 1 && ( $csvRow[0] === null || trim( $csvRow[0] ) === '' ); |
| 247 | } |
| 248 | } |