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 MediaWiki\MediaWikiServices as MediaWikiServicesAlias;
  31 use Wikimedia\Rdbms\IResultWrapper;
  32 use Wikimedia\Rdbms\IDatabase;
  33
  34 /**
  35  * @ingroup SpecialPage Dump
  36  */
  37 class WikiExporter {
  38         /** @var bool Return distinct author list (when not returning full history) */
  39         public $list_authors = false;
  40
  41         /** @var bool */
  42         public $dumpUploads = false;
  43
  44         /** @var bool */
  45         public $dumpUploadFileContents = false;
  46
  47         /** @var string */
  48         public $author_list = "";
  49
  50         const FULL = 1;
  51         const CURRENT = 2;
  52         const STABLE = 4; // extension defined
  53         const LOGS = 8;
  54         const RANGE = 16;
  55
  56         const TEXT = XmlDumpWriter::WRITE_CONTENT;
  57         const STUB = XmlDumpWriter::WRITE_STUB;
  58
  59         const BATCH_SIZE = 50000;
  60
  61         /** @var int */
  62         public $text;
  63
  64         /** @var DumpOutput */
  65         public $sink;
  66
  67         /** @var XmlDumpWriter */
  68         private $writer;
  69
  70         /** @var IDatabase */
  71         protected $db;
  72
  73         /** @var array|int */
  74         protected $history;
  75
  76         /** @var array|null */
  77         protected $limitNamespaces;
  78
  79         /**
  80          * Returns the default export schema version, as defined by $wgXmlDumpSchemaVersion.
  81          * @return string
  82          */
  83         public static function schemaVersion() {
  84                 global $wgXmlDumpSchemaVersion;
  85                 return $wgXmlDumpSchemaVersion;
  86         }
  87
  88         /**
  89          * @param IDatabase $db
  90          * @param int|array $history One of WikiExporter::FULL, WikiExporter::CURRENT,
  91          *   WikiExporter::RANGE or WikiExporter::STABLE, or an associative array:
  92          *   - offset: non-inclusive offset at which to start the query
  93          *   - limit: maximum number of rows to return
  94          *   - dir: "asc" or "desc" timestamp order
  95          * @param int $text One of WikiExporter::TEXT or WikiExporter::STUB
  96          * @param null|array $limitNamespaces Comma-separated list of namespace numbers
  97          *   to limit results
  98          */
  99         function __construct(
 100                 $db,
 101                 $history = self::CURRENT,
 102                 $text = self::TEXT,
 103                 $limitNamespaces = null
 104         ) {
 105                 $this->db = $db;
 106                 $this->history = $history;
 107                 $this->writer = new XmlDumpWriter( $text, self::schemaVersion() );
 108                 $this->sink = new DumpOutput();
 109                 $this->text = $text;
 110                 $this->limitNamespaces = $limitNamespaces;
 111         }
 112
 113         /**
 114          * @param string $schemaVersion which schema version the generated XML should comply to.
 115          * One of the values from self::$supportedSchemas, using the XML_DUMP_SCHEMA_VERSION_XX
 116          * constants.
 117          */
 118         public function setSchemaVersion( $schemaVersion ) {
 119                 $this->writer = new XmlDumpWriter( $this->text, $schemaVersion );
 120         }
 121
 122         /**
 123          * Set the DumpOutput or DumpFilter object which will receive
 124          * various row objects and XML output for filtering. Filters
 125          * can be chained or used as callbacks.
 126          *
 127          * @param DumpOutput &$sink
 128          */
 129         public function setOutputSink( &$sink ) {
 130                 $this->sink =& $sink;
 131         }
 132
 133         public function openStream() {
 134                 $output = $this->writer->openStream();
 135                 $this->sink->writeOpenStream( $output );
 136         }
 137
 138         public function closeStream() {
 139                 $output = $this->writer->closeStream();
 140                 $this->sink->writeCloseStream( $output );
 141         }
 142
 143         /**
 144          * Dumps a series of page and revision records for all pages
 145          * in the database, either including complete history or only
 146          * the most recent version.
 147          */
 148         public function allPages() {
 149                 $this->dumpFrom( '' );
 150         }
 151
 152         /**
 153          * Dumps a series of page and revision records for those pages
 154          * in the database falling within the page_id range given.
 155          * @param int $start Inclusive lower limit (this id is included)
 156          * @param int $end Exclusive upper limit (this id is not included)
 157          *   If 0, no upper limit.
 158          * @param bool $orderRevs order revisions within pages in ascending order
 159          */
 160         public function pagesByRange( $start, $end, $orderRevs ) {
 161                 if ( $orderRevs ) {
 162                         $condition = 'rev_page >= ' . intval( $start );
 163                         if ( $end ) {
 164                                 $condition .= ' AND rev_page < ' . intval( $end );
 165                         }
 166                 } else {
 167                         $condition = 'page_id >= ' . intval( $start );
 168                         if ( $end ) {
 169                                 $condition .= ' AND page_id < ' . intval( $end );
 170                         }
 171                 }
 172                 $this->dumpFrom( $condition, $orderRevs );
 173         }
 174
 175         /**
 176          * Dumps a series of page and revision records for those pages
 177          * in the database with revisions falling within the rev_id range given.
 178          * @param int $start Inclusive lower limit (this id is included)
 179          * @param int $end Exclusive upper limit (this id is not included)
 180          *   If 0, no upper limit.
 181          */
 182         public function revsByRange( $start, $end ) {
 183                 $condition = 'rev_id >= ' . intval( $start );
 184                 if ( $end ) {
 185                         $condition .= ' AND rev_id < ' . intval( $end );
 186                 }
 187                 $this->dumpFrom( $condition );
 188         }
 189
 190         /**
 191          * @param Title $title
 192          */
 193         public function pageByTitle( $title ) {
 194                 $this->dumpFrom(
 195                         'page_namespace=' . $title->getNamespace() .
 196                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 197         }
 198
 199         /**
 200          * @param string $name
 201          * @throws MWException
 202          */
 203         public function pageByName( $name ) {
 204                 $title = Title::newFromText( $name );
 205                 if ( is_null( $title ) ) {
 206                         throw new MWException( "Can't export invalid title" );
 207                 } else {
 208                         $this->pageByTitle( $title );
 209                 }
 210         }
 211
 212         /**
 213          * @param array $names
 214          */
 215         public function pagesByName( $names ) {
 216                 foreach ( $names as $name ) {
 217                         $this->pageByName( $name );
 218                 }
 219         }
 220
 221         public function allLogs() {
 222                 $this->dumpFrom( '' );
 223         }
 224
 225         /**
 226          * @param int $start
 227          * @param int $end
 228          */
 229         public function logsByRange( $start, $end ) {
 230                 $condition = 'log_id >= ' . intval( $start );
 231                 if ( $end ) {
 232                         $condition .= ' AND log_id < ' . intval( $end );
 233                 }
 234                 $this->dumpFrom( $condition );
 235         }
 236
 237         /**
 238          * Generates the distinct list of authors of an article
 239          * Not called by default (depends on $this->list_authors)
 240          * Can be set by Special:Export when not exporting whole history
 241          *
 242          * @param array $cond
 243          */
 244         protected function do_list_authors( $cond ) {
 245                 $this->author_list = "<contributors>";
 246                 // rev_deleted
 247
 248                 $revQuery = Revision::getQueryInfo( [ 'page' ] );
 249                 $res = $this->db->select(
 250                         $revQuery['tables'],
 251                         [
 252                                 'rev_user_text' => $revQuery['fields']['rev_user_text'],
 253                                 'rev_user' => $revQuery['fields']['rev_user'],
 254                         ],
 255                         [
 256                                 $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ' = 0',
 257                                 $cond,
 258                         ],
 259                         __METHOD__,
 260                         [ 'DISTINCT' ],
 261                         $revQuery['joins']
 262                 );
 263
 264                 foreach ( $res as $row ) {
 265                         $this->author_list .= "<contributor>" .
 266                                 "<username>" .
 267                                 htmlspecialchars( $row->rev_user_text ) .
 268                                 "</username>" .
 269                                 "<id>" .
 270                                 ( (int)$row->rev_user ) .
 271                                 "</id>" .
 272                                 "</contributor>";
 273                 }
 274                 $this->author_list .= "</contributors>";
 275         }
 276
 277         /**
 278          * @param string $cond
 279          * @param bool $orderRevs
 280          * @throws MWException
 281          * @throws Exception
 282          */
 283         protected function dumpFrom( $cond = '', $orderRevs = false ) {
 284                 if ( $this->history & self::LOGS ) {
 285                         $this->dumpLogs( $cond );
 286                 } else {
 287                         $this->dumpPages( $cond, $orderRevs );
 288                 }
 289         }
 290
 291         /**
 292          * @param string $cond
 293          * @throws Exception
 294          */
 295         protected function dumpLogs( $cond ) {
 296                 $where = [];
 297                 # Hide private logs
 298                 $hideLogs = LogEventsList::getExcludeClause( $this->db );
 299                 if ( $hideLogs ) {
 300                         $where[] = $hideLogs;
 301                 }
 302                 # Add on any caller specified conditions
 303                 if ( $cond ) {
 304                         $where[] = $cond;
 305                 }
 306                 # Get logging table name for logging.* clause
 307                 $logging = $this->db->tableName( 'logging' );
 308
 309                 $result = null; // Assuring $result is not undefined, if exception occurs early
 310
 311                 $commentQuery = CommentStore::getStore()->getJoin( 'log_comment' );
 312                 $actorQuery = ActorMigration::newMigration()->getJoin( 'log_user' );
 313
 314                 $lastLogId = 0;
 315                 while ( true ) {
 316                         $result = $this->db->select(
 317                                 array_merge( [ 'logging' ], $commentQuery['tables'], $actorQuery['tables'], [ 'user' ] ),
 318                                 [ "{$logging}.*", 'user_name' ] + $commentQuery['fields'] + $actorQuery['fields'],
 319                                 array_merge( $where, [ 'log_id > ' . intval( $lastLogId ) ] ),
 320                                 __METHOD__,
 321                                 [
 322                                         'ORDER BY' => 'log_id',
 323                                         'USE INDEX' => [ 'logging' => 'PRIMARY' ],
 324                                         'LIMIT' => self::BATCH_SIZE,
 325                                 ],
 326                                 [
 327                                         'user' => [ 'JOIN', 'user_id = ' . $actorQuery['fields']['log_user'] ]
 328                                 ] + $commentQuery['joins'] + $actorQuery['joins']
 329                         );
 330
 331                         if ( !$result->numRows() ) {
 332                                 break;
 333                         }
 334
 335                         $lastLogId = $this->outputLogStream( $result );
 336                 }
 337         }
 338
 339         /**
 340          * @param string $cond
 341          * @param bool $orderRevs
 342          * @throws MWException
 343          * @throws Exception
 344          */
 345         protected function dumpPages( $cond, $orderRevs ) {
 346                 global $wgMultiContentRevisionSchemaMigrationStage;
 347                 if ( !( $wgMultiContentRevisionSchemaMigrationStage & SCHEMA_COMPAT_WRITE_OLD ) ) {
 348                         // TODO: Make XmlDumpWriter use a RevisionStore! (see T198706 and T174031)
 349                         throw new MWException(
 350                                 'Cannot use WikiExporter with SCHEMA_COMPAT_WRITE_OLD mode disabled!'
 351                                 . ' Support for dumping from the new schema is not implemented yet!'
 352                         );
 353                 }
 354
 355                 $revQuery = MediaWikiServicesAlias::getInstance()->getRevisionStore()->getQueryInfo(
 356                         [ 'page' ]
 357                 );
 358                 $slotQuery = MediaWikiServicesAlias::getInstance()->getRevisionStore()->getSlotsQueryInfo(
 359                         [ 'content' ]
 360                 );
 361
 362                 // We want page primary rather than revision.
 363                 // We also want to join in the slots and content tables.
 364                 // NOTE: This means we may get multiple rows per revision, and more rows
 365                 // than the batch size! Should be ok, since the max number of slots is
 366                 // fixed and low (dozens at worst).
 367                 $tables = array_merge( [ 'page' ], array_diff( $revQuery['tables'], [ 'page' ] ) );
 368                 $tables = array_merge( $tables, array_diff( $slotQuery['tables'], $tables ) );
 369                 $join = $revQuery['joins'] + [
 370                                 'revision' => $revQuery['joins']['page'],
 371                                 'slots' => [ 'JOIN', [ 'slot_revision_id = rev_id' ] ],
 372                                 'content' => [ 'JOIN', [ 'content_id = slot_content_id' ] ],
 373                         ];
 374                 unset( $join['page'] );
 375
 376                 $fields = array_merge( $revQuery['fields'], $slotQuery['fields'] );
 377                 $fields[] = 'page_restrictions';
 378
 379                 if ( $this->text != self::STUB ) {
 380                         $fields['_load_content'] = '1';
 381                 }
 382
 383                 $conds = [];
 384                 if ( $cond !== '' ) {
 385                         $conds[] = $cond;
 386                 }
 387                 $opts = [ 'ORDER BY' => [ 'rev_page ASC', 'rev_id ASC' ] ];
 388                 $opts['USE INDEX'] = [];
 389
 390                 $op = '>';
 391                 if ( is_array( $this->history ) ) {
 392                         # Time offset/limit for all pages/history...
 393                         # Set time order
 394                         if ( $this->history['dir'] == 'asc' ) {
 395                                 $opts['ORDER BY'] = 'rev_timestamp ASC';
 396                         } else {
 397                                 $op = '<';
 398                                 $opts['ORDER BY'] = 'rev_timestamp DESC';
 399                         }
 400                         # Set offset
 401                         if ( !empty( $this->history['offset'] ) ) {
 402                                 $conds[] = "rev_timestamp $op " .
 403                                         $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 404                         }
 405                         # Set query limit
 406                         if ( !empty( $this->history['limit'] ) ) {
 407                                 $maxRowCount = intval( $this->history['limit'] );
 408                         }
 409                 } elseif ( $this->history & self::FULL ) {
 410                         # Full history dumps...
 411                         # query optimization for history stub dumps
 412                         if ( $this->text == self::STUB ) {
 413                                 $opts[] = 'STRAIGHT_JOIN';
 414                                 $opts['USE INDEX']['revision'] = 'rev_page_id';
 415                                 unset( $join['revision'] );
 416                                 $join['page'] = [ 'JOIN', 'rev_page=page_id' ];
 417                         }
 418                 } elseif ( $this->history & self::CURRENT ) {
 419                         # Latest revision dumps...
 420                         if ( $this->list_authors && $cond != '' ) { // List authors, if so desired
 421                                 $this->do_list_authors( $cond );
 422                         }
 423                         $join['revision'] = [ 'JOIN', 'page_id=rev_page AND page_latest=rev_id' ];
 424                         $opts[ 'ORDER BY' ] = [ 'page_id ASC' ];
 425                 } elseif ( $this->history & self::STABLE ) {
 426                         # "Stable" revision dumps...
 427                         # Default JOIN, to be overridden...
 428                         $join['revision'] = [ 'JOIN', 'page_id=rev_page AND page_latest=rev_id' ];
 429                         # One, and only one hook should set this, and return false
 430                         if ( Hooks::run( 'WikiExporter::dumpStableQuery', [ &$tables, &$opts, &$join ] ) ) {
 431                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 432                         }
 433                 } elseif ( $this->history & self::RANGE ) {
 434                         # Dump of revisions within a specified range.  Condition already set in revsByRange().
 435                 } else {
 436                         # Unknown history specification parameter?
 437                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 438                 }
 439
 440                 $result = null; // Assuring $result is not undefined, if exception occurs early
 441                 $done = false;
 442                 $lastRow = null;
 443                 $revPage = 0;
 444                 $revId = 0;
 445                 $rowCount = 0;
 446
 447                 $opts['LIMIT'] = self::BATCH_SIZE;
 448
 449                 Hooks::run( 'ModifyExportQuery',
 450                         [ $this->db, &$tables, &$cond, &$opts, &$join ] );
 451
 452                 while ( !$done ) {
 453                         // If necessary, impose the overall maximum and stop looping after this iteration.
 454                         if ( !empty( $maxRowCount ) && $rowCount + self::BATCH_SIZE > $maxRowCount ) {
 455                                 $opts['LIMIT'] = $maxRowCount - $rowCount;
 456                                 $done = true;
 457                         }
 458
 459                         $queryConds = $conds;
 460                         $queryConds[] = 'rev_page>' . intval( $revPage ) . ' OR (rev_page=' .
 461                                 intval( $revPage ) . ' AND rev_id' . $op . intval( $revId ) . ')';
 462
 463                         # Do the query and process any results, remembering max ids for the next iteration.
 464                         $result = $this->db->select(
 465                                 $tables,
 466                                 $fields,
 467                                 $queryConds,
 468                                 __METHOD__,
 469                                 $opts,
 470                                 $join
 471                         );
 472                         if ( $result->numRows() > 0 ) {
 473                                 $lastRow = $this->outputPageStreamBatch( $result, $lastRow );
 474                                 $rowCount += $result->numRows();
 475                                 $revPage = $lastRow->rev_page;
 476                                 $revId = $lastRow->rev_id;
 477                         } else {
 478                                 $done = true;
 479                         }
 480
 481                         // If we are finished, close off final page element (if any).
 482                         if ( $done && $lastRow ) {
 483                                 $this->finishPageStreamOutput( $lastRow );
 484                         }
 485                 }
 486         }
 487
 488         /**
 489          * Runs through a query result set dumping page, revision, and slot records.
 490          * The result set should join the page, revision, slots, and content tables,
 491          * and be sorted/grouped by page and revision to avoid duplicate page records in the output.
 492          *
 493          * @param IResultWrapper $results
 494          * @param object $lastRow the last row output from the previous call (or null if none)
 495          * @return object the last row processed
 496          */
 497         protected function outputPageStreamBatch( $results, $lastRow ) {
 498                 $rowCarry = null;
 499                 while ( true ) {
 500                         $slotRows = $this->getSlotRowBatch( $results, $rowCarry );
 501
 502                         if ( !$slotRows ) {
 503                                 break;
 504                         }
 505
 506                         // All revision info is present in all slot rows.
 507                         // Use the first slot row as the revision row.
 508                         $revRow = $slotRows[0];
 509
 510                         if ( $this->limitNamespaces &&
 511                                 !in_array( $revRow->page_namespace, $this->limitNamespaces ) ) {
 512                                 $lastRow = $revRow;
 513                                 continue;
 514                         }
 515
 516                         if ( $lastRow === null ||
 517                                 $lastRow->page_namespace !== $revRow->page_namespace ||
 518                                 $lastRow->page_title !== $revRow->page_title ) {
 519                                 if ( $lastRow !== null ) {
 520                                         $output = '';
 521                                         if ( $this->dumpUploads ) {
 522                                                 $output .= $this->writer->writeUploads( $lastRow, $this->dumpUploadFileContents );
 523                                         }
 524                                         $output .= $this->writer->closePage();
 525                                         $this->sink->writeClosePage( $output );
 526                                 }
 527                                 $output = $this->writer->openPage( $revRow );
 528                                 $this->sink->writeOpenPage( $revRow, $output );
 529                         }
 530                         $output = $this->writer->writeRevision( $revRow, $slotRows );
 531                         $this->sink->writeRevision( $revRow, $output );
 532                         $lastRow = $revRow;
 533                 }
 534
 535                 if ( $rowCarry ) {
 536                         throw new LogicException( 'Error while processing a stream of slot rows' );
 537                 }
 538
 539                 return $lastRow;
 540         }
 541
 542         /**
 543          * Returns all slot rows for a revision.
 544          * Takes and returns a carry row from the last batch;
 545          *
 546          * @param IResultWrapper|array $results
 547          * @param null|object &$carry A row carried over from the last call to getSlotRowBatch()
 548          *
 549          * @return object[]
 550          */
 551         protected function getSlotRowBatch( $results, &$carry = null ) {
 552                 $slotRows = [];
 553                 $prev = null;
 554
 555                 if ( $carry ) {
 556                         $slotRows[] = $carry;
 557                         $prev = $carry;
 558                         $carry = null;
 559                 }
 560
 561                 while ( $row = $results->fetchObject() ) {
 562                         if ( $prev && $prev->rev_id !== $row->rev_id ) {
 563                                 $carry = $row;
 564                                 break;
 565                         }
 566                         $slotRows[] = $row;
 567                         $prev = $row;
 568                 }
 569
 570                 return $slotRows;
 571         }
 572
 573         /**
 574          * Final page stream output, after all batches are complete
 575          *
 576          * @param object $lastRow the last row output from the last batch (or null if none)
 577          */
 578         protected function finishPageStreamOutput( $lastRow ) {
 579                 $output = '';
 580                 if ( $this->dumpUploads ) {
 581                         $output .= $this->writer->writeUploads( $lastRow, $this->dumpUploadFileContents );
 582                 }
 583                 $output .= $this->author_list;
 584                 $output .= $this->writer->closePage();
 585                 $this->sink->writeClosePage( $output );
 586         }
 587
 588         /**
 589          * @param IResultWrapper $resultset
 590          * @return int the log_id value of the last item output, or null if none
 591          */
 592         protected function outputLogStream( $resultset ) {
 593                 foreach ( $resultset as $row ) {
 594                         $output = $this->writer->writeLogItem( $row );
 595                         $this->sink->writeLogItem( $row, $output );
 596                 }
 597                 return isset( $row ) ? $row->log_id : null;
 598         }
 599 }