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