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         /**
  67          * Returns the export schema version.
  68          * @return string
  69          */
  70         public static function schemaVersion() {
  71                 return "0.10";
  72         }
  73
  74         /**
  75          * @param IDatabase $db
  76          * @param int|array $history One of WikiExporter::FULL, WikiExporter::CURRENT,
  77          *   WikiExporter::RANGE or WikiExporter::STABLE, or an associative array:
  78          *   - offset: non-inclusive offset at which to start the query
  79          *   - limit: maximum number of rows to return
  80          *   - dir: "asc" or "desc" timestamp order
  81          * @param int $text One of WikiExporter::TEXT or WikiExporter::STUB
  82          */
  83         function __construct( $db, $history = self::CURRENT, $text = self::TEXT ) {
  84                 $this->db = $db;
  85                 $this->history = $history;
  86                 $this->writer = new XmlDumpWriter();
  87                 $this->sink = new DumpOutput();
  88                 $this->text = $text;
  89         }
  90
  91         /**
  92          * Set the DumpOutput or DumpFilter object which will receive
  93          * various row objects and XML output for filtering. Filters
  94          * can be chained or used as callbacks.
  95          *
  96          * @param DumpOutput &$sink
  97          */
  98         public function setOutputSink( &$sink ) {
  99                 $this->sink =& $sink;
 100         }
 101
 102         public function openStream() {
 103                 $output = $this->writer->openStream();
 104                 $this->sink->writeOpenStream( $output );
 105         }
 106
 107         public function closeStream() {
 108                 $output = $this->writer->closeStream();
 109                 $this->sink->writeCloseStream( $output );
 110         }
 111
 112         /**
 113          * Dumps a series of page and revision records for all pages
 114          * in the database, either including complete history or only
 115          * the most recent version.
 116          */
 117         public function allPages() {
 118                 $this->dumpFrom( '' );
 119         }
 120
 121         /**
 122          * Dumps a series of page and revision records for those pages
 123          * in the database falling within the page_id range given.
 124          * @param int $start Inclusive lower limit (this id is included)
 125          * @param int $end Exclusive upper limit (this id is not included)
 126          *   If 0, no upper limit.
 127          * @param bool $orderRevs order revisions within pages in ascending order
 128          */
 129         public function pagesByRange( $start, $end, $orderRevs ) {
 130                 if ( $orderRevs ) {
 131                         $condition = 'rev_page >= ' . intval( $start );
 132                         if ( $end ) {
 133                                 $condition .= ' AND rev_page < ' . intval( $end );
 134                         }
 135                 } else {
 136                         $condition = 'page_id >= ' . intval( $start );
 137                         if ( $end ) {
 138                                 $condition .= ' AND page_id < ' . intval( $end );
 139                         }
 140                 }
 141                 $this->dumpFrom( $condition, $orderRevs );
 142         }
 143
 144         /**
 145          * Dumps a series of page and revision records for those pages
 146          * in the database with revisions falling within the rev_id range given.
 147          * @param int $start Inclusive lower limit (this id is included)
 148          * @param int $end Exclusive upper limit (this id is not included)
 149          *   If 0, no upper limit.
 150          */
 151         public function revsByRange( $start, $end ) {
 152                 $condition = 'rev_id >= ' . intval( $start );
 153                 if ( $end ) {
 154                         $condition .= ' AND rev_id < ' . intval( $end );
 155                 }
 156                 $this->dumpFrom( $condition );
 157         }
 158
 159         /**
 160          * @param Title $title
 161          */
 162         public function pageByTitle( $title ) {
 163                 $this->dumpFrom(
 164                         'page_namespace=' . $title->getNamespace() .
 165                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 166         }
 167
 168         /**
 169          * @param string $name
 170          * @throws MWException
 171          */
 172         public function pageByName( $name ) {
 173                 $title = Title::newFromText( $name );
 174                 if ( is_null( $title ) ) {
 175                         throw new MWException( "Can't export invalid title" );
 176                 } else {
 177                         $this->pageByTitle( $title );
 178                 }
 179         }
 180
 181         /**
 182          * @param array $names
 183          */
 184         public function pagesByName( $names ) {
 185                 foreach ( $names as $name ) {
 186                         $this->pageByName( $name );
 187                 }
 188         }
 189
 190         public function allLogs() {
 191                 $this->dumpFrom( '' );
 192         }
 193
 194         /**
 195          * @param int $start
 196          * @param int $end
 197          */
 198         public function logsByRange( $start, $end ) {
 199                 $condition = 'log_id >= ' . intval( $start );
 200                 if ( $end ) {
 201                         $condition .= ' AND log_id < ' . intval( $end );
 202                 }
 203                 $this->dumpFrom( $condition );
 204         }
 205
 206         /**
 207          * Generates the distinct list of authors of an article
 208          * Not called by default (depends on $this->list_authors)
 209          * Can be set by Special:Export when not exporting whole history
 210          *
 211          * @param array $cond
 212          */
 213         protected function do_list_authors( $cond ) {
 214                 $this->author_list = "<contributors>";
 215                 // rev_deleted
 216
 217                 $revQuery = Revision::getQueryInfo( [ 'page' ] );
 218                 $res = $this->db->select(
 219                         $revQuery['tables'],
 220                         [
 221                                 'rev_user_text' => $revQuery['fields']['rev_user_text'],
 222                                 'rev_user' => $revQuery['fields']['rev_user'],
 223                         ],
 224                         [
 225                                 $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ' = 0',
 226                                 $cond,
 227                         ],
 228                         __METHOD__,
 229                         [ 'DISTINCT' ],
 230                         $revQuery['joins']
 231                 );
 232
 233                 foreach ( $res as $row ) {
 234                         $this->author_list .= "<contributor>" .
 235                                 "<username>" .
 236                                 htmlspecialchars( $row->rev_user_text ) .
 237                                 "</username>" .
 238                                 "<id>" .
 239                                 ( (int)$row->rev_user ) .
 240                                 "</id>" .
 241                                 "</contributor>";
 242                 }
 243                 $this->author_list .= "</contributors>";
 244         }
 245
 246         /**
 247          * @param string $cond
 248          * @param bool $orderRevs
 249          * @throws MWException
 250          * @throws Exception
 251          */
 252         protected function dumpFrom( $cond = '', $orderRevs = false ) {
 253                 if ( $this->history & self::LOGS ) {
 254                         $this->dumpLogs( $cond );
 255                 } else {
 256                         $this->dumpPages( $cond, $orderRevs );
 257                 }
 258         }
 259
 260         /**
 261          * @param string $cond
 262          * @throws Exception
 263          */
 264         protected function dumpLogs( $cond ) {
 265                 $where = [];
 266                 # Hide private logs
 267                 $hideLogs = LogEventsList::getExcludeClause( $this->db );
 268                 if ( $hideLogs ) {
 269                         $where[] = $hideLogs;
 270                 }
 271                 # Add on any caller specified conditions
 272                 if ( $cond ) {
 273                         $where[] = $cond;
 274                 }
 275                 # Get logging table name for logging.* clause
 276                 $logging = $this->db->tableName( 'logging' );
 277
 278                 $result = null; // Assuring $result is not undefined, if exception occurs early
 279
 280                 $commentQuery = CommentStore::getStore()->getJoin( 'log_comment' );
 281                 $actorQuery = ActorMigration::newMigration()->getJoin( 'log_user' );
 282
 283                 $lastLogId = 0;
 284                 while ( true ) {
 285                         $result = $this->db->select(
 286                                 array_merge( [ 'logging' ], $commentQuery['tables'], $actorQuery['tables'], [ 'user' ] ),
 287                                 [ "{$logging}.*", 'user_name' ] + $commentQuery['fields'] + $actorQuery['fields'],
 288                                 array_merge( $where, [ 'log_id > ' . intval( $lastLogId ) ] ),
 289                                 __METHOD__,
 290                                 [
 291                                         'ORDER BY' => 'log_id',
 292                                         'USE INDEX' => [ 'logging' => 'PRIMARY' ],
 293                                         'LIMIT' => self::BATCH_SIZE,
 294                                 ],
 295                                 [
 296                                         'user' => [ 'JOIN', 'user_id = ' . $actorQuery['fields']['log_user'] ]
 297                                 ] + $commentQuery['joins'] + $actorQuery['joins']
 298                         );
 299
 300                         if ( !$result->numRows() ) {
 301                                 break;
 302                         }
 303
 304                         $lastLogId = $this->outputLogStream( $result );
 305                 };
 306         }
 307
 308         /**
 309          * @param string $cond
 310          * @param bool $orderRevs
 311          * @throws MWException
 312          * @throws Exception
 313          */
 314         protected function dumpPages( $cond, $orderRevs ) {
 315                 global $wgMultiContentRevisionSchemaMigrationStage;
 316                 if ( !( $wgMultiContentRevisionSchemaMigrationStage & SCHEMA_COMPAT_WRITE_OLD ) ) {
 317                         // TODO: Make XmlDumpWriter use a RevisionStore! (see T198706 and T174031)
 318                         throw new MWException(
 319                                 'Cannot use WikiExporter with SCHEMA_COMPAT_WRITE_OLD mode disabled!'
 320                                 . ' Support for dumping from the new schema is not implemented yet!'
 321                         );
 322                 }
 323
 324                 $revOpts = [ 'page' ];
 325                 if ( $this->text != self::STUB ) {
 326                         // TODO: remove the text and make XmlDumpWriter use a RevisionStore instead! (T198706)
 327                         $revOpts[] = 'text';
 328                 }
 329                 $revQuery = Revision::getQueryInfo( $revOpts );
 330
 331                 // We want page primary rather than revision
 332                 $tables = array_merge( [ 'page' ], array_diff( $revQuery['tables'], [ 'page' ] ) );
 333                 $join = $revQuery['joins'] + [
 334                                 'revision' => $revQuery['joins']['page']
 335                         ];
 336                 unset( $join['page'] );
 337
 338                 // TODO: remove rev_text_id and make XmlDumpWriter use a RevisionStore instead! (T198706)
 339                 $fields = array_merge( $revQuery['fields'], [ 'page_restrictions, rev_text_id' ] );
 340
 341                 $conds = [];
 342                 if ( $cond !== '' ) {
 343                         $conds[] = $cond;
 344                 }
 345                 $opts = [ 'ORDER BY' => [ 'rev_page ASC', 'rev_id ASC' ] ];
 346                 $opts['USE INDEX'] = [];
 347
 348                 $op = '>';
 349                 if ( is_array( $this->history ) ) {
 350                         # Time offset/limit for all pages/history...
 351                         # Set time order
 352                         if ( $this->history['dir'] == 'asc' ) {
 353                                 $opts['ORDER BY'] = 'rev_timestamp ASC';
 354                         } else {
 355                                 $op = '<';
 356                                 $opts['ORDER BY'] = 'rev_timestamp DESC';
 357                         }
 358                         # Set offset
 359                         if ( !empty( $this->history['offset'] ) ) {
 360                                 $conds[] = "rev_timestamp $op " .
 361                                         $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 362                         }
 363                         # Set query limit
 364                         if ( !empty( $this->history['limit'] ) ) {
 365                                 $maxRowCount = intval( $this->history['limit'] );
 366                         }
 367                 } elseif ( $this->history & self::FULL ) {
 368                         # Full history dumps...
 369                         # query optimization for history stub dumps
 370                         if ( $this->text == self::STUB ) {
 371                                 $tables = $revQuery['tables'];
 372                                 $opts[] = 'STRAIGHT_JOIN';
 373                                 $opts['USE INDEX']['revision'] = 'rev_page_id';
 374                                 unset( $join['revision'] );
 375                                 $join['page'] = [ 'INNER JOIN', 'rev_page=page_id' ];
 376                         }
 377                 } elseif ( $this->history & self::CURRENT ) {
 378                         # Latest revision dumps...
 379                         if ( $this->list_authors && $cond != '' ) { // List authors, if so desired
 380                                 $this->do_list_authors( $cond );
 381                         }
 382                         $join['revision'] = [ 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' ];
 383                 } elseif ( $this->history & self::STABLE ) {
 384                         # "Stable" revision dumps...
 385                         # Default JOIN, to be overridden...
 386                         $join['revision'] = [ 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' ];
 387                         # One, and only one hook should set this, and return false
 388                         if ( Hooks::run( 'WikiExporter::dumpStableQuery', [ &$tables, &$opts, &$join ] ) ) {
 389                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 390                         }
 391                 } elseif ( $this->history & self::RANGE ) {
 392                         # Dump of revisions within a specified range.  Condition already set in revsByRange().
 393                 } else {
 394                         # Unknown history specification parameter?
 395                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 396                 }
 397
 398                 $result = null; // Assuring $result is not undefined, if exception occurs early
 399                 $done = false;
 400                 $lastRow = null;
 401                 $revPage = 0;
 402                 $revId = 0;
 403                 $rowCount = 0;
 404
 405                 $opts['LIMIT'] = self::BATCH_SIZE;
 406
 407                 Hooks::run( 'ModifyExportQuery',
 408                         [ $this->db, &$tables, &$cond, &$opts, &$join ] );
 409
 410                 while ( !$done ) {
 411                         // If necessary, impose the overall maximum and stop looping after this iteration.
 412                         if ( !empty( $maxRowCount ) && $rowCount + self::BATCH_SIZE > $maxRowCount ) {
 413                                 $opts['LIMIT'] = $maxRowCount - $rowCount;
 414                                 $done = true;
 415                         }
 416
 417                         $queryConds = $conds;
 418                         $queryConds[] = 'rev_page>' . intval( $revPage ) . ' OR (rev_page=' .
 419                                 intval( $revPage ) . ' AND rev_id' . $op . intval( $revId ) . ')';
 420
 421                         # Do the query and process any results, remembering max ids for the next iteration.
 422                         $result = $this->db->select(
 423                                 $tables,
 424                                 $fields,
 425                                 $queryConds,
 426                                 __METHOD__,
 427                                 $opts,
 428                                 $join
 429                         );
 430                         if ( $result->numRows() > 0 ) {
 431                                 $lastRow = $this->outputPageStreamBatch( $result, $lastRow );
 432                                 $rowCount += $result->numRows();
 433                                 $revPage = $lastRow->rev_page;
 434                                 $revId = $lastRow->rev_id;
 435                         } else {
 436                                 $done = true;
 437                         }
 438
 439                         // If we are finished, close off final page element (if any).
 440                         if ( $done && $lastRow ) {
 441                                 $this->finishPageStreamOutput( $lastRow );
 442                         }
 443                 }
 444         }
 445
 446         /**
 447          * Runs through a query result set dumping page and revision records.
 448          * The result set should be sorted/grouped by page to avoid duplicate
 449          * page records in the output.
 450          *
 451          * @param ResultWrapper $results
 452          * @param object $lastRow the last row output from the previous call (or null if none)
 453          * @return object the last row processed
 454          */
 455         protected function outputPageStreamBatch( $results, $lastRow ) {
 456                 foreach ( $results as $row ) {
 457                         if ( $lastRow === null ||
 458                                 $lastRow->page_namespace != $row->page_namespace ||
 459                                 $lastRow->page_title != $row->page_title ) {
 460                                 if ( $lastRow !== null ) {
 461                                         $output = '';
 462                                         if ( $this->dumpUploads ) {
 463                                                 $output .= $this->writer->writeUploads( $lastRow, $this->dumpUploadFileContents );
 464                                         }
 465                                         $output .= $this->writer->closePage();
 466                                         $this->sink->writeClosePage( $output );
 467                                 }
 468                                 $output = $this->writer->openPage( $row );
 469                                 $this->sink->writeOpenPage( $row, $output );
 470                         }
 471                         $output = $this->writer->writeRevision( $row );
 472                         $this->sink->writeRevision( $row, $output );
 473                         $lastRow = $row;
 474                 }
 475
 476                 return $lastRow;
 477         }
 478
 479         /**
 480          * Final page stream output, after all batches are complete
 481          *
 482          * @param object $lastRow the last row output from the last batch (or null if none)
 483          */
 484         protected function finishPageStreamOutput( $lastRow ) {
 485                 $output = '';
 486                 if ( $this->dumpUploads ) {
 487                         $output .= $this->writer->writeUploads( $lastRow, $this->dumpUploadFileContents );
 488                 }
 489                 $output .= $this->author_list;
 490                 $output .= $this->writer->closePage();
 491                 $this->sink->writeClosePage( $output );
 492         }
 493
 494         /**
 495          * @param ResultWrapper $resultset
 496          * @return int the log_id value of the last item output, or null if none
 497          */
 498         protected function outputLogStream( $resultset ) {
 499                 foreach ( $resultset as $row ) {
 500                         $output = $this->writer->writeLogItem( $row );
 501                         $this->sink->writeLogItem( $row, $output );
 502                 }
 503                 return isset( $row ) ? $row->log_id : null;
 504         }
 505 }