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 BUFFER = 0;
  56         const STREAM = 1;
  57
  58         const TEXT = 0;
  59         const STUB = 1;
  60
  61         /** @var int */
  62         public $buffer;
  63
  64         /** @var int */
  65         public $text;
  66
  67         /** @var DumpOutput */
  68         public $sink;
  69
  70         /**
  71          * Returns the export schema version.
  72          * @return string
  73          */
  74         public static function schemaVersion() {
  75                 return "0.10";
  76         }
  77
  78         /**
  79          * If using WikiExporter::STREAM to stream a large amount of data,
  80          * provide a database connection which is not managed by
  81          * LoadBalancer to read from: some history blob types will
  82          * make additional queries to pull source data while the
  83          * main query is still running.
  84          *
  85          * @param IDatabase $db
  86          * @param int|array $history One of WikiExporter::FULL, WikiExporter::CURRENT,
  87          *   WikiExporter::RANGE or WikiExporter::STABLE, or an associative array:
  88          *   - offset: non-inclusive offset at which to start the query
  89          *   - limit: maximum number of rows to return
  90          *   - dir: "asc" or "desc" timestamp order
  91          * @param int $buffer One of WikiExporter::BUFFER or WikiExporter::STREAM
  92          * @param int $text One of WikiExporter::TEXT or WikiExporter::STUB
  93          */
  94         function __construct( $db, $history = self::CURRENT,
  95                         $buffer = self::BUFFER, $text = self::TEXT ) {
  96                 $this->db = $db;
  97                 $this->history = $history;
  98                 $this->buffer = $buffer;
  99                 $this->writer = new XmlDumpWriter();
 100                 $this->sink = new DumpOutput();
 101                 $this->text = $text;
 102         }
 103
 104         /**
 105          * Set the DumpOutput or DumpFilter object which will receive
 106          * various row objects and XML output for filtering. Filters
 107          * can be chained or used as callbacks.
 108          *
 109          * @param DumpOutput &$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                 $res = $this->db->select(
 231                         [ 'page', 'revision' ],
 232                         [ 'DISTINCT rev_user_text', 'rev_user' ],
 233                         [
 234                                 $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ' = 0',
 235                                 $cond,
 236                                 'page_id = rev_id',
 237                         ],
 238                         __METHOD__
 239                 );
 240
 241                 foreach ( $res as $row ) {
 242                         $this->author_list .= "<contributor>" .
 243                                 "<username>" .
 244                                 htmlentities( $row->rev_user_text ) .
 245                                 "</username>" .
 246                                 "<id>" .
 247                                 $row->rev_user .
 248                                 "</id>" .
 249                                 "</contributor>";
 250                 }
 251                 $this->author_list .= "</contributors>";
 252         }
 253
 254         /**
 255          * @param string $cond
 256          * @param bool $orderRevs
 257          * @throws MWException
 258          * @throws Exception
 259          */
 260         protected function dumpFrom( $cond = '', $orderRevs = false ) {
 261                 # For logging dumps...
 262                 if ( $this->history & self::LOGS ) {
 263                         $where = [];
 264                         # Hide private logs
 265                         $hideLogs = LogEventsList::getExcludeClause( $this->db );
 266                         if ( $hideLogs ) {
 267                                 $where[] = $hideLogs;
 268                         }
 269                         # Add on any caller specified conditions
 270                         if ( $cond ) {
 271                                 $where[] = $cond;
 272                         }
 273                         # Get logging table name for logging.* clause
 274                         $logging = $this->db->tableName( 'logging' );
 275
 276                         if ( $this->buffer == self::STREAM ) {
 277                                 $prev = $this->db->bufferResults( false );
 278                         }
 279                         $result = null; // Assuring $result is not undefined, if exception occurs early
 280
 281                         $commentQuery = CommentStore::getStore()->getJoin( 'log_comment' );
 282
 283                         try {
 284                                 $result = $this->db->select( [ 'logging', 'user' ] + $commentQuery['tables'],
 285                                         [ "{$logging}.*", 'user_name' ] + $commentQuery['fields'], // grab the user name
 286                                         $where,
 287                                         __METHOD__,
 288                                         [ 'ORDER BY' => 'log_id', 'USE INDEX' => [ 'logging' => 'PRIMARY' ] ],
 289                                         [ 'user' => [ 'JOIN', 'user_id = log_user' ] ] + $commentQuery['joins']
 290                                 );
 291                                 $this->outputLogStream( $result );
 292                                 if ( $this->buffer == self::STREAM ) {
 293                                         $this->db->bufferResults( $prev );
 294                                 }
 295                         } catch ( Exception $e ) {
 296                                 // Throwing the exception does not reliably free the resultset, and
 297                                 // would also leave the connection in unbuffered mode.
 298
 299                                 // Freeing result
 300                                 try {
 301                                         if ( $result ) {
 302                                                 $result->free();
 303                                         }
 304                                 } catch ( Exception $e2 ) {
 305                                         // Already in panic mode -> ignoring $e2 as $e has
 306                                         // higher priority
 307                                 }
 308
 309                                 // Putting database back in previous buffer mode
 310                                 try {
 311                                         if ( $this->buffer == self::STREAM ) {
 312                                                 $this->db->bufferResults( $prev );
 313                                         }
 314                                 } catch ( Exception $e2 ) {
 315                                         // Already in panic mode -> ignoring $e2 as $e has
 316                                         // higher priority
 317                                 }
 318
 319                                 // Inform caller about problem
 320                                 throw $e;
 321                         }
 322                 # For page dumps...
 323                 } else {
 324                         $tables = [ 'page', 'revision' ];
 325                         $opts = [ 'ORDER BY' => 'page_id ASC' ];
 326                         $opts['USE INDEX'] = [];
 327                         $join = [];
 328                         if ( is_array( $this->history ) ) {
 329                                 # Time offset/limit for all pages/history...
 330                                 $revJoin = 'page_id=rev_page';
 331                                 # Set time order
 332                                 if ( $this->history['dir'] == 'asc' ) {
 333                                         $op = '>';
 334                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 335                                 } else {
 336                                         $op = '<';
 337                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 338                                 }
 339                                 # Set offset
 340                                 if ( !empty( $this->history['offset'] ) ) {
 341                                         $revJoin .= " AND rev_timestamp $op " .
 342                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 343                                 }
 344                                 $join['revision'] = [ 'INNER JOIN', $revJoin ];
 345                                 # Set query limit
 346                                 if ( !empty( $this->history['limit'] ) ) {
 347                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 348                                 }
 349                         } elseif ( $this->history & self::FULL ) {
 350                                 # Full history dumps...
 351                                 # query optimization for history stub dumps
 352                                 if ( $this->text == self::STUB && $orderRevs ) {
 353                                         $tables = [ 'revision', 'page' ];
 354                                         $opts[] = 'STRAIGHT_JOIN';
 355                                         $opts['ORDER BY'] = [ 'rev_page ASC', 'rev_id ASC' ];
 356                                         $opts['USE INDEX']['revision'] = 'rev_page_id';
 357                                         $join['page'] = [ 'INNER JOIN', 'rev_page=page_id' ];
 358                                 } else {
 359                                         $join['revision'] = [ 'INNER JOIN', 'page_id=rev_page' ];
 360                                 }
 361                         } elseif ( $this->history & self::CURRENT ) {
 362                                 # Latest revision dumps...
 363                                 if ( $this->list_authors && $cond != '' ) { // List authors, if so desired
 364                                         $this->do_list_authors( $cond );
 365                                 }
 366                                 $join['revision'] = [ 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' ];
 367                         } elseif ( $this->history & self::STABLE ) {
 368                                 # "Stable" revision dumps...
 369                                 # Default JOIN, to be overridden...
 370                                 $join['revision'] = [ 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' ];
 371                                 # One, and only one hook should set this, and return false
 372                                 if ( Hooks::run( 'WikiExporter::dumpStableQuery', [ &$tables, &$opts, &$join ] ) ) {
 373                                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 374                                 }
 375                         } elseif ( $this->history & self::RANGE ) {
 376                                 # Dump of revisions within a specified range
 377                                 $join['revision'] = [ 'INNER JOIN', 'page_id=rev_page' ];
 378                                 $opts['ORDER BY'] = [ 'rev_page ASC', 'rev_id ASC' ];
 379                         } else {
 380                                 # Unknown history specification parameter?
 381                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 382                         }
 383                         # Query optimization hacks
 384                         if ( $cond == '' ) {
 385                                 $opts[] = 'STRAIGHT_JOIN';
 386                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 387                         }
 388                         # Build text join options
 389                         if ( $this->text != self::STUB ) { // 1-pass
 390                                 $tables[] = 'text';
 391                                 $join['text'] = [ 'INNER JOIN', 'rev_text_id=old_id' ];
 392                         }
 393
 394                         if ( $this->buffer == self::STREAM ) {
 395                                 $prev = $this->db->bufferResults( false );
 396                         }
 397                         $result = null; // Assuring $result is not undefined, if exception occurs early
 398                         try {
 399                                 Hooks::run( 'ModifyExportQuery',
 400                                                 [ $this->db, &$tables, &$cond, &$opts, &$join ] );
 401
 402                                 $commentQuery = CommentStore::getStore()->getJoin( 'rev_comment' );
 403
 404                                 # Do the query!
 405                                 $result = $this->db->select(
 406                                         $tables + $commentQuery['tables'],
 407                                         [ '*' ] + $commentQuery['fields'],
 408                                         $cond,
 409                                         __METHOD__,
 410                                         $opts,
 411                                         $join + $commentQuery['joins']
 412                                 );
 413                                 # Output dump results
 414                                 $this->outputPageStream( $result );
 415
 416                                 if ( $this->buffer == self::STREAM ) {
 417                                         $this->db->bufferResults( $prev );
 418                                 }
 419                         } catch ( Exception $e ) {
 420                                 // Throwing the exception does not reliably free the resultset, and
 421                                 // would also leave the connection in unbuffered mode.
 422
 423                                 // Freeing result
 424                                 try {
 425                                         if ( $result ) {
 426                                                 $result->free();
 427                                         }
 428                                 } catch ( Exception $e2 ) {
 429                                         // Already in panic mode -> ignoring $e2 as $e has
 430                                         // higher priority
 431                                 }
 432
 433                                 // Putting database back in previous buffer mode
 434                                 try {
 435                                         if ( $this->buffer == self::STREAM ) {
 436                                                 $this->db->bufferResults( $prev );
 437                                         }
 438                                 } catch ( Exception $e2 ) {
 439                                         // Already in panic mode -> ignoring $e2 as $e has
 440                                         // higher priority
 441                                 }
 442
 443                                 // Inform caller about problem
 444                                 throw $e;
 445                         }
 446                 }
 447         }
 448
 449         /**
 450          * Runs through a query result set dumping page and revision records.
 451          * The result set should be sorted/grouped by page to avoid duplicate
 452          * page records in the output.
 453          *
 454          * Should be safe for
 455          * streaming (non-buffered) queries, as long as it was made on a
 456          * separate database connection not managed by LoadBalancer; some
 457          * blob storage types will make queries to pull source data.
 458          *
 459          * @param ResultWrapper $resultset
 460          */
 461         protected function outputPageStream( $resultset ) {
 462                 $last = null;
 463                 foreach ( $resultset as $row ) {
 464                         if ( $last === null ||
 465                                 $last->page_namespace != $row->page_namespace ||
 466                                 $last->page_title != $row->page_title ) {
 467                                 if ( $last !== null ) {
 468                                         $output = '';
 469                                         if ( $this->dumpUploads ) {
 470                                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 471                                         }
 472                                         $output .= $this->writer->closePage();
 473                                         $this->sink->writeClosePage( $output );
 474                                 }
 475                                 $output = $this->writer->openPage( $row );
 476                                 $this->sink->writeOpenPage( $row, $output );
 477                                 $last = $row;
 478                         }
 479                         $output = $this->writer->writeRevision( $row );
 480                         $this->sink->writeRevision( $row, $output );
 481                 }
 482                 if ( $last !== null ) {
 483                         $output = '';
 484                         if ( $this->dumpUploads ) {
 485                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 486                         }
 487                         $output .= $this->author_list;
 488                         $output .= $this->writer->closePage();
 489                         $this->sink->writeClosePage( $output );
 490                 }
 491         }
 492
 493         /**
 494          * @param ResultWrapper $resultset
 495          */
 496         protected function outputLogStream( $resultset ) {
 497                 foreach ( $resultset as $row ) {
 498                         $output = $this->writer->writeLogItem( $row );
 499                         $this->sink->writeLogItem( $row, $output );
 500                 }
 501         }
 502 }