Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 110 |
|
0.00% |
0 / 8 |
CRAP | |
0.00% |
0 / 1 |
| Converter | |
0.00% |
0 / 110 |
|
0.00% |
0 / 8 |
870 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
12 | |||
| convertAll | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
12 | |||
| convert | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
42 | |||
| isAllowed | |
0.00% |
0 / 13 |
|
0.00% |
0 / 1 |
20 | |||
| doConversion | |
0.00% |
0 / 14 |
|
0.00% |
0 / 1 |
12 | |||
| getPageMovedFrom | |
0.00% |
0 / 16 |
|
0.00% |
0 / 1 |
6 | |||
| movePage | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
12 | |||
| createArchiveCleanupRevision | |
0.00% |
0 / 20 |
|
0.00% |
0 / 1 |
30 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace Flow\Import; |
| 4 | |
| 5 | use Flow\Exception\FlowException; |
| 6 | use MediaWiki\Content\WikitextContent; |
| 7 | use MediaWiki\Exception\MWExceptionHandler; |
| 8 | use MediaWiki\MediaWikiServices; |
| 9 | use MediaWiki\Page\Article; |
| 10 | use MediaWiki\Revision\RevisionRecord; |
| 11 | use MediaWiki\Revision\SlotRecord; |
| 12 | use MediaWiki\Title\Title; |
| 13 | use MediaWiki\User\User; |
| 14 | use Psr\Log\LoggerInterface; |
| 15 | use Traversable; |
| 16 | use Wikimedia\Rdbms\IDBAccessObject; |
| 17 | use Wikimedia\Rdbms\IReadableDatabase; |
| 18 | use Wikimedia\Rdbms\SelectQueryBuilder; |
| 19 | |
| 20 | /** |
| 21 | * Converts provided titles to Flow. This converter is idempotent when |
| 22 | * used with an appropriate SourceStoreInterface, and may be run many times |
| 23 | * without worry for duplicate imports. |
| 24 | * |
| 25 | * Flow does not currently support viewing the history of its page prior |
| 26 | * to being flow enabled. Because of this prior to conversion the current |
| 27 | * wikitext page will be moved to an archive location. |
| 28 | * |
| 29 | * Implementing classes must choose a name for their archive page and |
| 30 | * be able to create an IImportSource when provided a Title. On successful |
| 31 | * import of a page a 'cleanup archive' edit is optionally performed. |
| 32 | * |
| 33 | * Any content changes to the imported content should be provided as part |
| 34 | * of the IImportSource. |
| 35 | */ |
| 36 | class Converter { |
| 37 | /** |
| 38 | * @var IReadableDatabase Primary database of the current wiki. Required |
| 39 | * to lookup past page moves. |
| 40 | */ |
| 41 | protected $db; |
| 42 | |
| 43 | /** |
| 44 | * @var Importer Service capable of turning an IImportSource into |
| 45 | * flow revisions. |
| 46 | */ |
| 47 | protected $importer; |
| 48 | |
| 49 | /** |
| 50 | * @var LoggerInterface |
| 51 | */ |
| 52 | protected $logger; |
| 53 | |
| 54 | /** |
| 55 | * @var User The user for performing maintenance actions like moving |
| 56 | * pages or editing templates onto an archived page. This should be |
| 57 | * a system account and not a normal user. |
| 58 | */ |
| 59 | protected $user; |
| 60 | |
| 61 | /** |
| 62 | * @var IConversionStrategy Interface between this converter and an |
| 63 | * IImportSource implementation. |
| 64 | */ |
| 65 | protected $strategy; |
| 66 | |
| 67 | /** |
| 68 | * @param IReadableDatabase $db Primary wiki database to read from |
| 69 | * @param Importer $importer |
| 70 | * @param LoggerInterface $logger |
| 71 | * @param User $user User for moves and edits related to the conversion process |
| 72 | * @param IConversionStrategy $strategy |
| 73 | * |
| 74 | * @throws ImportException When $user does not have an Id |
| 75 | */ |
| 76 | public function __construct( |
| 77 | IReadableDatabase $db, |
| 78 | Importer $importer, |
| 79 | LoggerInterface $logger, |
| 80 | User $user, |
| 81 | IConversionStrategy $strategy |
| 82 | ) { |
| 83 | if ( !$user->getId() ) { |
| 84 | throw new ImportException( 'User must have id' ); |
| 85 | } |
| 86 | $this->db = $db; |
| 87 | $this->importer = $importer; |
| 88 | $this->logger = $logger; |
| 89 | $this->user = $user; |
| 90 | $this->strategy = $strategy; |
| 91 | |
| 92 | $postprocessor = $strategy->getPostprocessor(); |
| 93 | if ( $postprocessor !== null ) { |
| 94 | // @todo assert we cant cause duplicate postprocessors |
| 95 | $this->importer->addPostprocessor( $postprocessor ); |
| 96 | } |
| 97 | |
| 98 | // Force the importer to use our logger for consistent output. |
| 99 | $this->importer->setLogger( $logger ); |
| 100 | } |
| 101 | |
| 102 | /** |
| 103 | * Converts multiple pages into Flow boards |
| 104 | * |
| 105 | * @param Traversable<Title>|array $titles |
| 106 | * @param bool $dryRun If true, will not make any changes |
| 107 | * @param bool $convertEmpty Convert pages with no threads |
| 108 | */ |
| 109 | public function convertAll( $titles, $dryRun = false, $convertEmpty = false ) { |
| 110 | /** @var Title $title */ |
| 111 | foreach ( $titles as $title ) { |
| 112 | try { |
| 113 | $this->convert( $title, $dryRun, $convertEmpty ); |
| 114 | } catch ( \Exception $e ) { |
| 115 | MWExceptionHandler::logException( $e ); |
| 116 | $this->logger->error( "Exception while importing: {$title}" ); |
| 117 | $this->logger->error( (string)$e ); |
| 118 | } |
| 119 | } |
| 120 | } |
| 121 | |
| 122 | /** |
| 123 | * Converts a page into a Flow board |
| 124 | * |
| 125 | * @param Title $title |
| 126 | * @param bool $dryRun If true, will not make any changes |
| 127 | * @param bool $convertEmpty Convert pages with no threads |
| 128 | * @throws FlowException |
| 129 | */ |
| 130 | public function convert( Title $title, $dryRun = false, $convertEmpty = false ) { |
| 131 | /* |
| 132 | * $title is the title we're currently considering to import. |
| 133 | * It could be a page we need to import, but could also e.g. |
| 134 | * be an archive page of a previous import run (in which case |
| 135 | * $movedFrom will be the Title object of that original page) |
| 136 | */ |
| 137 | $movedFrom = $this->getPageMovedFrom( $title ); |
| 138 | if ( $this->strategy->isConversionFinished( $title, $movedFrom ) ) { |
| 139 | return; |
| 140 | } |
| 141 | |
| 142 | if ( !$this->isAllowed( $title ) ) { |
| 143 | throw new FlowException( "Not allowed to convert: {$title}" ); |
| 144 | } |
| 145 | |
| 146 | if ( !$convertEmpty ) { |
| 147 | $article = new Article( $title ); |
| 148 | $pager = new \LqtDiscussionPager( $article, \TalkpageView::LQT_NEWEST_CHANGES ); |
| 149 | $pager->setLimit( 1 ); |
| 150 | if ( !$pager->getNumRows() ) { |
| 151 | $this->logger->info( "Skipping {$title} as it has no LiquidThreads content" ); |
| 152 | return; |
| 153 | } |
| 154 | } |
| 155 | |
| 156 | if ( !$dryRun ) { |
| 157 | $this->doConversion( $title, $movedFrom ); |
| 158 | } else { |
| 159 | $this->logger->info( "Dry run: Would convert $title" ); |
| 160 | } |
| 161 | } |
| 162 | |
| 163 | /** |
| 164 | * Returns a boolean indicating if we're allowed to import $title. |
| 165 | * |
| 166 | * @param Title $title |
| 167 | * @return bool |
| 168 | */ |
| 169 | protected function isAllowed( Title $title ) { |
| 170 | // Only make changes to wikitext pages |
| 171 | if ( $title->getContentModel() !== CONTENT_MODEL_WIKITEXT ) { |
| 172 | $this->logger->warning( "WARNING: The title '" . $title->getPrefixedDBkey() . |
| 173 | "' is being skipped because it has content model '" . $title->getContentModel() . "''." ); |
| 174 | return false; |
| 175 | } |
| 176 | |
| 177 | if ( !$title->exists() ) { |
| 178 | $this->logger->warning( "WARNING: The title '" . $title->getPrefixedDBkey() . |
| 179 | "' is being skipped because it does not exist." ); |
| 180 | return false; |
| 181 | } |
| 182 | |
| 183 | // At some point we may want to handle these, but for now just |
| 184 | // let them be |
| 185 | if ( $title->isRedirect() ) { |
| 186 | $this->logger->warning( "WARNING: The title '" . $title->getPrefixedDBkey() . |
| 187 | "' is being skipped because it is a redirect." ); |
| 188 | return false; |
| 189 | } |
| 190 | |
| 191 | // Finally, check strategy-specific logic |
| 192 | return $this->strategy->shouldConvert( $title ); |
| 193 | } |
| 194 | |
| 195 | protected function doConversion( Title $title, ?Title $movedFrom = null ) { |
| 196 | if ( $movedFrom ) { |
| 197 | // If the page is moved but has not completed conversion that |
| 198 | // means the previous import failed to complete. Try again. |
| 199 | $archiveTitle = $title; |
| 200 | $title = $movedFrom; |
| 201 | $this->logger->info( "Page previously archived from $title to $archiveTitle" ); |
| 202 | } else { |
| 203 | // The move needs to happen prior to the import because upon starting the |
| 204 | // import the top revision will be a flow-board revision. |
| 205 | $archiveTitle = $this->strategy->decideArchiveTitle( $title ); |
| 206 | $this->logger->info( "Archiving page from $title to $archiveTitle" ); |
| 207 | $this->movePage( $title, $archiveTitle ); |
| 208 | |
| 209 | $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory(); |
| 210 | // Wait for replicas to pick up the page move |
| 211 | $lbFactory->waitForReplication(); |
| 212 | } |
| 213 | |
| 214 | $source = $this->strategy->createImportSource( $archiveTitle ); |
| 215 | if ( $this->importer->import( $source, $title, $this->user, $this->strategy->getSourceStore() ) ) { |
| 216 | $this->createArchiveCleanupRevision( $title, $archiveTitle ); |
| 217 | $this->logger->info( "Completed import to $title from $archiveTitle" ); |
| 218 | } else { |
| 219 | $this->logger->error( "Failed to complete import to $title from $archiveTitle" ); |
| 220 | } |
| 221 | } |
| 222 | |
| 223 | /** |
| 224 | * Looks in the logging table to see if the provided title was last moved |
| 225 | * there by the user provided in the constructor. The provided user should |
| 226 | * be a system user for this task, as this assumes that user has never |
| 227 | * moved these pages outside the conversion process. |
| 228 | * |
| 229 | * This only considers the most recent move and not prior moves. This allows |
| 230 | * for edge cases such as starting an import, canceling it, and manually |
| 231 | * reverting the move by a normal user. |
| 232 | * |
| 233 | * @param Title $title |
| 234 | * @return Title|null |
| 235 | */ |
| 236 | protected function getPageMovedFrom( Title $title ) { |
| 237 | $row = $this->db->newSelectQueryBuilder() |
| 238 | ->select( [ 'log_namespace', 'log_title' ] ) |
| 239 | ->from( 'logging' ) |
| 240 | ->join( 'page', null, 'log_page = page_id' ) |
| 241 | ->where( [ |
| 242 | 'page_namespace' => $title->getNamespace(), |
| 243 | 'page_title' => $title->getDBkey(), |
| 244 | 'log_type' => 'move', |
| 245 | 'log_actor' => $this->user->getActorId() |
| 246 | ] ) |
| 247 | ->caller( __METHOD__ ) |
| 248 | ->orderBy( 'log_timestamp', SelectQueryBuilder::SORT_DESC ) |
| 249 | ->fetchRow(); |
| 250 | |
| 251 | // The page has never been moved or the most recent move was not by our user |
| 252 | if ( !$row ) { |
| 253 | return null; |
| 254 | } |
| 255 | |
| 256 | return Title::makeTitle( $row->log_namespace, $row->log_title ); |
| 257 | } |
| 258 | |
| 259 | /** |
| 260 | * Moves the source page to the destination. Does not leave behind a |
| 261 | * redirect, intending that flow will place a revision there for its new |
| 262 | * board. |
| 263 | * |
| 264 | * @param Title $from |
| 265 | * @param Title $to |
| 266 | * @throws ImportException on failed import |
| 267 | */ |
| 268 | protected function movePage( Title $from, Title $to ) { |
| 269 | $mp = MediaWikiServices::getInstance() |
| 270 | ->getMovePageFactory() |
| 271 | ->newMovePage( $from, $to ); |
| 272 | |
| 273 | $valid = $mp->isValidMove(); |
| 274 | if ( !$valid->isOK() ) { |
| 275 | $this->logger->error( $valid->getMessage()->text() ); |
| 276 | throw new ImportException( "It is not valid to move {$from} to {$to}" ); |
| 277 | } |
| 278 | |
| 279 | // Note that this comment must match the regex in self::getPageMovedFrom |
| 280 | $status = $mp->move( |
| 281 | /* user */ $this->user, |
| 282 | /* reason */ $this->strategy->getMoveComment( $from, $to ), |
| 283 | /* create redirect */ false |
| 284 | ); |
| 285 | |
| 286 | if ( !$status->isGood() ) { |
| 287 | $this->logger->error( $status->getMessage()->text() ); |
| 288 | throw new ImportException( "Failed moving {$from} to {$to}" ); |
| 289 | } |
| 290 | } |
| 291 | |
| 292 | /** |
| 293 | * Creates a new revision of the archived page with strategy-specific changes. |
| 294 | * |
| 295 | * @param Title $title Previous location of the page, before moving |
| 296 | * @param Title $archiveTitle Current location of the page, after moving |
| 297 | * @throws ImportException |
| 298 | */ |
| 299 | protected function createArchiveCleanupRevision( Title $title, Title $archiveTitle ) { |
| 300 | $page = MediaWikiServices::getInstance()->getWikiPageFactory()->newFromTitle( $archiveTitle ); |
| 301 | // doUserEditContent will do this anyway, but we need to now for the revision. |
| 302 | $page->loadPageData( IDBAccessObject::READ_LATEST ); |
| 303 | $revision = $page->getRevisionRecord(); |
| 304 | if ( $revision === null ) { |
| 305 | throw new ImportException( "Expected a revision at {$archiveTitle}" ); |
| 306 | } |
| 307 | |
| 308 | // Do not create revisions based on rev_deleted revisions. |
| 309 | $content = $revision->getContent( SlotRecord::MAIN, RevisionRecord::FOR_PUBLIC ); |
| 310 | if ( !$content instanceof WikitextContent ) { |
| 311 | throw new ImportException( "Expected wikitext content at: {$archiveTitle}" ); |
| 312 | } |
| 313 | |
| 314 | $newContent = $this->strategy->createArchiveCleanupRevisionContent( $content, $title ); |
| 315 | if ( $newContent === null ) { |
| 316 | return; |
| 317 | } |
| 318 | |
| 319 | $status = $page->doUserEditContent( |
| 320 | $newContent, |
| 321 | $this->user, |
| 322 | $this->strategy->getCleanupComment( $title, $archiveTitle ), |
| 323 | EDIT_FORCE_BOT | EDIT_SUPPRESS_RC |
| 324 | ); |
| 325 | |
| 326 | if ( !$status->isGood() ) { |
| 327 | $this->logger->error( $status->getMessage()->text() ); |
| 328 | throw new ImportException( "Failed creating archive cleanup revision at {$archiveTitle}" ); |
| 329 | } |
| 330 | } |
| 331 | } |