includes/export/WikiExporter.php

   1 <?php
   2 /**
   3  * Base class for exporting
   4  *
   5  * Copyright © 2003, 2005, 2006 Brion Vibber <brion@pobox.com>
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  */
  25
  26 /**
  27  * @defgroup Dump Dump
  28  */
  29
  30 use Wikimedia\Rdbms\ResultWrapper;
  31 use Wikimedia\Rdbms\IDatabase;
  32
  33 /**
  34  * @ingroup SpecialPage Dump
  35  */
  36 class WikiExporter {
  37         /** @var bool Return distinct author list (when not returning full history) */
  38         public $list_authors = false;
  39
  40         /** @var bool */
  41         public $dumpUploads = false;
  42
  43         /** @var bool */
  44         public $dumpUploadFileContents = false;
  45
  46         /** @var string */
  47         public $author_list = "";
  48
  49         const FULL = 1;
  50         const CURRENT = 2;
  51         const STABLE = 4; // extension defined
  52         const LOGS = 8;
  53         const RANGE = 16;
  54
  55         const TEXT = 0;
  56         const STUB = 1;
  57
  58         const BATCH_SIZE = 50000;
  59
  60         /** @var int */
  61         public $text;
  62
  63         /** @var DumpOutput */
  64         public $sink;
  65
  66         /** @var XmlDumpWriter */
  67         private $writer;
  68
  69         /**
  70          * Returns the default export schema version, as defined by $wgXmlDumpSchemaVersion.
  71          * @return string
  72          */
  73         public static function schemaVersion() {
  74                 global $wgXmlDumpSchemaVersion;
  75                 return $wgXmlDumpSchemaVersion;
  76         }
  77
  78         /**
  79          * @param IDatabase $db
  80          * @param int|array $history One of WikiExporter::FULL, WikiExporter::CURRENT,
  81          *   WikiExporter::RANGE or WikiExporter::STABLE, or an associative array:
  82          *   - offset: non-inclusive offset at which to start the query
  83          *   - limit: maximum number of rows to return
  84          *   - dir: "asc" or "desc" timestamp order
  85          * @param int $text One of WikiExporter::TEXT or WikiExporter::STUB
  86          */
  87         function __construct( $db, $history = self::CURRENT, $text = self::TEXT ) {
  88                 $this->db = $db;
  89                 $this->history = $history;
  90                 $this->writer = new XmlDumpWriter( $text, self::schemaVersion() );
  91                 $this->sink = new DumpOutput();
  92                 $this->text = $text;
  93         }
  94
  95         /**
  96          * @param string $schemaVersion which schema version the generated XML should comply to.
  97          * One of the values from self::$supportedSchemas, using the XML_DUMP_SCHEMA_VERSION_XX
  98          * constants.
  99          */
 100         public function setSchemaVersion( $schemaVersion ) {
 101                 $this->writer = new XmlDumpWriter( $this->text, $schemaVersion );
 102         }
 103
 104         /**
 105          * Set the DumpOutput or DumpFilter object which will receive
 106          * various row objects and XML output for filtering. Filters
 107          * can be chained or used as callbacks.
 108          *
 109          * @param DumpOutput &$sink
 110          */
 111         public function setOutputSink( &$sink ) {
 112                 $this->sink =& $sink;
 113         }
 114
 115         public function openStream() {
 116                 $output = $this->writer->openStream();
 117                 $this->sink->writeOpenStream( $output );
 118         }
 119
 120         public function closeStream() {
 121                 $output = $this->writer->closeStream();
 122                 $this->sink->writeCloseStream( $output );
 123         }
 124
 125         /**
 126          * Dumps a series of page and revision records for all pages
 127          * in the database, either including complete history or only
 128          * the most recent version.
 129          */
 130         public function allPages() {
 131                 $this->dumpFrom( '' );
 132         }
 133
 134         /**
 135          * Dumps a series of page and revision records for those pages
 136          * in the database falling within the page_id range given.
 137          * @param int $start Inclusive lower limit (this id is included)
 138          * @param int $end Exclusive upper limit (this id is not included)
 139          *   If 0, no upper limit.
 140          * @param bool $orderRevs order revisions within pages in ascending order
 141          */
 142         public function pagesByRange( $start, $end, $orderRevs ) {
 143                 if ( $orderRevs ) {
 144                         $condition = 'rev_page >= ' . intval( $start );
 145                         if ( $end ) {
 146                                 $condition .= ' AND rev_page < ' . intval( $end );
 147                         }
 148                 } else {
 149                         $condition = 'page_id >= ' . intval( $start );
 150                         if ( $end ) {
 151                                 $condition .= ' AND page_id < ' . intval( $end );
 152                         }
 153                 }
 154                 $this->dumpFrom( $condition, $orderRevs );
 155         }
 156
 157         /**
 158          * Dumps a series of page and revision records for those pages
 159          * in the database with revisions falling within the rev_id range given.
 160          * @param int $start Inclusive lower limit (this id is included)
 161          * @param int $end Exclusive upper limit (this id is not included)
 162          *   If 0, no upper limit.
 163          */
 164         public function revsByRange( $start, $end ) {
 165                 $condition = 'rev_id >= ' . intval( $start );
 166                 if ( $end ) {
 167                         $condition .= ' AND rev_id < ' . intval( $end );
 168                 }
 169                 $this->dumpFrom( $condition );
 170         }
 171
 172         /**
 173          * @param Title $title
 174          */
 175         public function pageByTitle( $title ) {
 176                 $this->dumpFrom(
 177                         'page_namespace=' . $title->getNamespace() .
 178                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 179         }
 180
 181         /**
 182          * @param string $name
 183          * @throws MWException
 184          */
 185         public function pageByName( $name ) {
 186                 $title = Title::newFromText( $name );
 187                 if ( is_null( $title ) ) {
 188                         throw new MWException( "Can't export invalid title" );
 189                 } else {
 190                         $this->pageByTitle( $title );
 191                 }
 192         }
 193
 194         /**
 195          * @param array $names
 196          */
 197         public function pagesByName( $names ) {
 198                 foreach ( $names as $name ) {
 199                         $this->pageByName( $name );
 200                 }
 201         }
 202
 203         public function allLogs() {
 204                 $this->dumpFrom( '' );
 205         }
 206
 207         /**
 208          * @param int $start
 209          * @param int $end
 210          */
 211         public function logsByRange( $start, $end ) {
 212                 $condition = 'log_id >= ' . intval( $start );
 213                 if ( $end ) {
 214                         $condition .= ' AND log_id < ' . intval( $end );
 215                 }
 216                 $this->dumpFrom( $condition );
 217         }
 218
 219         /**
 220          * Generates the distinct list of authors of an article
 221          * Not called by default (depends on $this->list_authors)
 222          * Can be set by Special:Export when not exporting whole history
 223          *
 224          * @param array $cond
 225          */
 226         protected function do_list_authors( $cond ) {
 227                 $this->author_list = "<contributors>";
 228                 // rev_deleted
 229
 230                 $revQuery = Revision::getQueryInfo( [ 'page' ] );
 231                 $res = $this->db->select(
 232                         $revQuery['tables'],
 233                         [
 234                                 'rev_user_text' => $revQuery['fields']['rev_user_text'],
 235                                 'rev_user' => $revQuery['fields']['rev_user'],
 236                         ],
 237                         [
 238                                 $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ' = 0',
 239                                 $cond,
 240                         ],
 241                         __METHOD__,
 242                         [ 'DISTINCT' ],
 243                         $revQuery['joins']
 244                 );
 245
 246                 foreach ( $res as $row ) {
 247                         $this->author_list .= "<contributor>" .
 248                                 "<username>" .
 249                                 htmlspecialchars( $row->rev_user_text ) .
 250                                 "</username>" .
 251                                 "<id>" .
 252                                 ( (int)$row->rev_user ) .
 253                                 "</id>" .
 254                                 "</contributor>";
 255                 }
 256                 $this->author_list .= "</contributors>";
 257         }
 258
 259         /**
 260          * @param string $cond
 261          * @param bool $orderRevs
 262          * @throws MWException
 263          * @throws Exception
 264          */
 265         protected function dumpFrom( $cond = '', $orderRevs = false ) {
 266                 if ( $this->history & self::LOGS ) {
 267                         $this->dumpLogs( $cond );
 268                 } else {
 269                         $this->dumpPages( $cond, $orderRevs );
 270                 }
 271         }
 272
 273         /**
 274          * @param string $cond
 275          * @throws Exception
 276          */
 277         protected function dumpLogs( $cond ) {
 278                 $where = [];
 279                 # Hide private logs
 280                 $hideLogs = LogEventsList::getExcludeClause( $this->db );
 281                 if ( $hideLogs ) {
 282                         $where[] = $hideLogs;
 283                 }
 284                 # Add on any caller specified conditions
 285                 if ( $cond ) {
 286                         $where[] = $cond;
 287                 }
 288                 # Get logging table name for logging.* clause
 289                 $logging = $this->db->tableName( 'logging' );
 290
 291                 $result = null; // Assuring $result is not undefined, if exception occurs early
 292
 293                 $commentQuery = CommentStore::getStore()->getJoin( 'log_comment' );
 294                 $actorQuery = ActorMigration::newMigration()->getJoin( 'log_user' );
 295
 296                 $lastLogId = 0;
 297                 while ( true ) {
 298                         $result = $this->db->select(
 299                                 array_merge( [ 'logging' ], $commentQuery['tables'], $actorQuery['tables'], [ 'user' ] ),
 300                                 [ "{$logging}.*", 'user_name' ] + $commentQuery['fields'] + $actorQuery['fields'],
 301                                 array_merge( $where, [ 'log_id > ' . intval( $lastLogId ) ] ),
 302                                 __METHOD__,
 303                                 [
 304                                         'ORDER BY' => 'log_id',
 305                                         'USE INDEX' => [ 'logging' => 'PRIMARY' ],
 306                                         'LIMIT' => self::BATCH_SIZE,
 307                                 ],
 308                                 [
 309                                         'user' => [ 'JOIN', 'user_id = ' . $actorQuery['fields']['log_user'] ]
 310                                 ] + $commentQuery['joins'] + $actorQuery['joins']
 311                         );
 312
 313                         if ( !$result->numRows() ) {
 314                                 break;
 315                         }
 316
 317                         $lastLogId = $this->outputLogStream( $result );
 318                 };
 319         }
 320
 321         /**
 322          * @param string $cond
 323          * @param bool $orderRevs
 324          * @throws MWException
 325          * @throws Exception
 326          */
 327         protected function dumpPages( $cond, $orderRevs ) {
 328                 global $wgMultiContentRevisionSchemaMigrationStage;
 329                 if ( !( $wgMultiContentRevisionSchemaMigrationStage & SCHEMA_COMPAT_WRITE_OLD ) ) {
 330                         // TODO: Make XmlDumpWriter use a RevisionStore! (see T198706 and T174031)
 331                         throw new MWException(
 332                                 'Cannot use WikiExporter with SCHEMA_COMPAT_WRITE_OLD mode disabled!'
 333                                 . ' Support for dumping from the new schema is not implemented yet!'
 334                         );
 335                 }
 336
 337                 $revOpts = [ 'page' ];
 338
 339                 $revQuery = Revision::getQueryInfo( $revOpts );
 340
 341                 // We want page primary rather than revision
 342                 $tables = array_merge( [ 'page' ], array_diff( $revQuery['tables'], [ 'page' ] ) );
 343                 $join = $revQuery['joins'] + [
 344                                 'revision' => $revQuery['joins']['page']
 345                         ];
 346                 unset( $join['page'] );
 347
 348                 $fields = $revQuery['fields'];
 349                 $fields[] = 'page_restrictions';
 350
 351                 if ( $this->text != self::STUB ) {
 352                         $fields['_load_content'] = '1';
 353                 }
 354
 355                 $conds = [];
 356                 if ( $cond !== '' ) {
 357                         $conds[] = $cond;
 358                 }
 359                 $opts = [ 'ORDER BY' => [ 'rev_page ASC', 'rev_id ASC' ] ];
 360                 $opts['USE INDEX'] = [];
 361
 362                 $op = '>';
 363                 if ( is_array( $this->history ) ) {
 364                         # Time offset/limit for all pages/history...
 365                         # Set time order
 366                         if ( $this->history['dir'] == 'asc' ) {
 367                                 $opts['ORDER BY'] = 'rev_timestamp ASC';
 368                         } else {
 369                                 $op = '<';
 370                                 $opts['ORDER BY'] = 'rev_timestamp DESC';
 371                         }
 372                         # Set offset
 373                         if ( !empty( $this->history['offset'] ) ) {
 374                                 $conds[] = "rev_timestamp $op " .
 375                                         $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 376                         }
 377                         # Set query limit
 378                         if ( !empty( $this->history['limit'] ) ) {
 379                                 $maxRowCount = intval( $this->history['limit'] );
 380                         }
 381                 } elseif ( $this->history & self::FULL ) {
 382                         # Full history dumps...
 383                         # query optimization for history stub dumps
 384                         if ( $this->text == self::STUB ) {
 385                                 $tables = $revQuery['tables'];
 386                                 $opts[] = 'STRAIGHT_JOIN';
 387                                 $opts['USE INDEX']['revision'] = 'rev_page_id';
 388                                 unset( $join['revision'] );
 389                                 $join['page'] = [ 'JOIN', 'rev_page=page_id' ];
 390                         }
 391                 } elseif ( $this->history & self::CURRENT ) {
 392                         # Latest revision dumps...
 393                         if ( $this->list_authors && $cond != '' ) { // List authors, if so desired
 394                                 $this->do_list_authors( $cond );
 395                         }
 396                         $join['revision'] = [ 'JOIN', 'page_id=rev_page AND page_latest=rev_id' ];
 397                 } elseif ( $this->history & self::STABLE ) {
 398                         # "Stable" revision dumps...
 399                         # Default JOIN, to be overridden...
 400                         $join['revision'] = [ 'JOIN', 'page_id=rev_page AND page_latest=rev_id' ];
 401                         # One, and only one hook should set this, and return false
 402                         if ( Hooks::run( 'WikiExporter::dumpStableQuery', [ &$tables, &$opts, &$join ] ) ) {
 403                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 404                         }
 405                 } elseif ( $this->history & self::RANGE ) {
 406                         # Dump of revisions within a specified range.  Condition already set in revsByRange().
 407                 } else {
 408                         # Unknown history specification parameter?
 409                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 410                 }
 411
 412                 $result = null; // Assuring $result is not undefined, if exception occurs early
 413                 $done = false;
 414                 $lastRow = null;
 415                 $revPage = 0;
 416                 $revId = 0;
 417                 $rowCount = 0;
 418
 419                 $opts['LIMIT'] = self::BATCH_SIZE;
 420
 421                 Hooks::run( 'ModifyExportQuery',
 422                         [ $this->db, &$tables, &$cond, &$opts, &$join ] );
 423
 424                 while ( !$done ) {
 425                         // If necessary, impose the overall maximum and stop looping after this iteration.
 426                         if ( !empty( $maxRowCount ) && $rowCount + self::BATCH_SIZE > $maxRowCount ) {
 427                                 $opts['LIMIT'] = $maxRowCount - $rowCount;
 428                                 $done = true;
 429                         }
 430
 431                         $queryConds = $conds;
 432                         $queryConds[] = 'rev_page>' . intval( $revPage ) . ' OR (rev_page=' .
 433                                 intval( $revPage ) . ' AND rev_id' . $op . intval( $revId ) . ')';
 434
 435                         # Do the query and process any results, remembering max ids for the next iteration.
 436                         $result = $this->db->select(
 437                                 $tables,
 438                                 $fields,
 439                                 $queryConds,
 440                                 __METHOD__,
 441                                 $opts,
 442                                 $join
 443                         );
 444                         if ( $result->numRows() > 0 ) {
 445                                 $lastRow = $this->outputPageStreamBatch( $result, $lastRow );
 446                                 $rowCount += $result->numRows();
 447                                 $revPage = $lastRow->rev_page;
 448                                 $revId = $lastRow->rev_id;
 449                         } else {
 450                                 $done = true;
 451                         }
 452
 453                         // If we are finished, close off final page element (if any).
 454                         if ( $done && $lastRow ) {
 455                                 $this->finishPageStreamOutput( $lastRow );
 456                         }
 457                 }
 458         }
 459
 460         /**
 461          * Runs through a query result set dumping page and revision records.
 462          * The result set should be sorted/grouped by page to avoid duplicate
 463          * page records in the output.
 464          *
 465          * @param ResultWrapper $results
 466          * @param object $lastRow the last row output from the previous call (or null if none)
 467          * @return object the last row processed
 468          */
 469         protected function outputPageStreamBatch( $results, $lastRow ) {
 470                 foreach ( $results as $row ) {
 471                         if ( $lastRow === null ||
 472                                 $lastRow->page_namespace !== $row->page_namespace ||
 473                                 $lastRow->page_title !== $row->page_title ) {
 474                                 if ( $lastRow !== null ) {
 475                                         $output = '';
 476                                         if ( $this->dumpUploads ) {
 477                                                 $output .= $this->writer->writeUploads( $lastRow, $this->dumpUploadFileContents );
 478                                         }
 479                                         $output .= $this->writer->closePage();
 480                                         $this->sink->writeClosePage( $output );
 481                                 }
 482                                 $output = $this->writer->openPage( $row );
 483                                 $this->sink->writeOpenPage( $row, $output );
 484                         }
 485                         $output = $this->writer->writeRevision( $row );
 486                         $this->sink->writeRevision( $row, $output );
 487                         $lastRow = $row;
 488                 }
 489
 490                 return $lastRow;
 491         }
 492
 493         /**
 494          * Final page stream output, after all batches are complete
 495          *
 496          * @param object $lastRow the last row output from the last batch (or null if none)
 497          */
 498         protected function finishPageStreamOutput( $lastRow ) {
 499                 $output = '';
 500                 if ( $this->dumpUploads ) {
 501                         $output .= $this->writer->writeUploads( $lastRow, $this->dumpUploadFileContents );
 502                 }
 503                 $output .= $this->author_list;
 504                 $output .= $this->writer->closePage();
 505                 $this->sink->writeClosePage( $output );
 506         }
 507
 508         /**
 509          * @param ResultWrapper $resultset
 510          * @return int the log_id value of the last item output, or null if none
 511          */
 512         protected function outputLogStream( $resultset ) {
 513                 foreach ( $resultset as $row ) {
 514                         $output = $this->writer->writeLogItem( $row );
 515                         $this->sink->writeLogItem( $row, $output );
 516                 }
 517                 return isset( $row ) ? $row->log_id : null;
 518         }
 519 }