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
  32 /**
  33  * @ingroup SpecialPage Dump
  34  */
  35 class WikiExporter {
  36         /** @var bool Return distinct author list (when not returning full history) */
  37         public $list_authors = false;
  38
  39         /** @var bool */
  40         public $dumpUploads = false;
  41
  42         /** @var bool */
  43         public $dumpUploadFileContents = false;
  44
  45         /** @var string */
  46         public $author_list = "";
  47
  48         const FULL = 1;
  49         const CURRENT = 2;
  50         const STABLE = 4; // extension defined
  51         const LOGS = 8;
  52         const RANGE = 16;
  53
  54         const BUFFER = 0;
  55         const STREAM = 1;
  56
  57         const TEXT = 0;
  58         const STUB = 1;
  59
  60         /** @var int */
  61         public $buffer;
  62
  63         /** @var int */
  64         public $text;
  65
  66         /** @var DumpOutput */
  67         public $sink;
  68
  69         /**
  70          * Returns the export schema version.
  71          * @return string
  72          */
  73         public static function schemaVersion() {
  74                 return "0.10";
  75         }
  76
  77         /**
  78          * If using WikiExporter::STREAM to stream a large amount of data,
  79          * provide a database connection which is not managed by
  80          * LoadBalancer to read from: some history blob types will
  81          * make additional queries to pull source data while the
  82          * main query is still running.
  83          *
  84          * @param IDatabase $db
  85          * @param int|array $history One of WikiExporter::FULL, WikiExporter::CURRENT,
  86          *   WikiExporter::RANGE or WikiExporter::STABLE, or an associative array:
  87          *   - offset: non-inclusive offset at which to start the query
  88          *   - limit: maximum number of rows to return
  89          *   - dir: "asc" or "desc" timestamp order
  90          * @param int $buffer One of WikiExporter::BUFFER or WikiExporter::STREAM
  91          * @param int $text One of WikiExporter::TEXT or WikiExporter::STUB
  92          */
  93         function __construct( $db, $history = WikiExporter::CURRENT,
  94                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  95                 $this->db = $db;
  96                 $this->history = $history;
  97                 $this->buffer = $buffer;
  98                 $this->writer = new XmlDumpWriter();
  99                 $this->sink = new DumpOutput();
 100                 $this->text = $text;
 101         }
 102
 103         /**
 104          * Set the DumpOutput or DumpFilter object which will receive
 105          * various row objects and XML output for filtering. Filters
 106          * can be chained or used as callbacks.
 107          *
 108          * @param DumpOutput $sink
 109          */
 110         public function setOutputSink( &$sink ) {
 111                 $this->sink =& $sink;
 112         }
 113
 114         public function openStream() {
 115                 $output = $this->writer->openStream();
 116                 $this->sink->writeOpenStream( $output );
 117         }
 118
 119         public function closeStream() {
 120                 $output = $this->writer->closeStream();
 121                 $this->sink->writeCloseStream( $output );
 122         }
 123
 124         /**
 125          * Dumps a series of page and revision records for all pages
 126          * in the database, either including complete history or only
 127          * the most recent version.
 128          */
 129         public function allPages() {
 130                 $this->dumpFrom( '' );
 131         }
 132
 133         /**
 134          * Dumps a series of page and revision records for those pages
 135          * in the database falling within the page_id range given.
 136          * @param int $start Inclusive lower limit (this id is included)
 137          * @param int $end Exclusive upper limit (this id is not included)
 138          *   If 0, no upper limit.
 139          * @param bool $orderRevs order revisions within pages in ascending order
 140          */
 141         public function pagesByRange( $start, $end, $orderRevs ) {
 142                 if ( $orderRevs ) {
 143                         $condition = 'rev_page >= ' . intval( $start );
 144                         if ( $end ) {
 145                                 $condition .= ' AND rev_page < ' . intval( $end );
 146                         }
 147                 } else {
 148                         $condition = 'page_id >= ' . intval( $start );
 149                         if ( $end ) {
 150                                 $condition .= ' AND page_id < ' . intval( $end );
 151                         }
 152                 }
 153                 $this->dumpFrom( $condition, $orderRevs );
 154         }
 155
 156         /**
 157          * Dumps a series of page and revision records for those pages
 158          * in the database with revisions falling within the rev_id range given.
 159          * @param int $start Inclusive lower limit (this id is included)
 160          * @param int $end Exclusive upper limit (this id is not included)
 161          *   If 0, no upper limit.
 162          */
 163         public function revsByRange( $start, $end ) {
 164                 $condition = 'rev_id >= ' . intval( $start );
 165                 if ( $end ) {
 166                         $condition .= ' AND rev_id < ' . intval( $end );
 167                 }
 168                 $this->dumpFrom( $condition );
 169         }
 170
 171         /**
 172          * @param Title $title
 173          */
 174         public function pageByTitle( $title ) {
 175                 $this->dumpFrom(
 176                         'page_namespace=' . $title->getNamespace() .
 177                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 178         }
 179
 180         /**
 181          * @param string $name
 182          * @throws MWException
 183          */
 184         public function pageByName( $name ) {
 185                 $title = Title::newFromText( $name );
 186                 if ( is_null( $title ) ) {
 187                         throw new MWException( "Can't export invalid title" );
 188                 } else {
 189                         $this->pageByTitle( $title );
 190                 }
 191         }
 192
 193         /**
 194          * @param array $names
 195          */
 196         public function pagesByName( $names ) {
 197                 foreach ( $names as $name ) {
 198                         $this->pageByName( $name );
 199                 }
 200         }
 201
 202         public function allLogs() {
 203                 $this->dumpFrom( '' );
 204         }
 205
 206         /**
 207          * @param int $start
 208          * @param int $end
 209          */
 210         public function logsByRange( $start, $end ) {
 211                 $condition = 'log_id >= ' . intval( $start );
 212                 if ( $end ) {
 213                         $condition .= ' AND log_id < ' . intval( $end );
 214                 }
 215                 $this->dumpFrom( $condition );
 216         }
 217
 218         /**
 219          * Generates the distinct list of authors of an article
 220          * Not called by default (depends on $this->list_authors)
 221          * Can be set by Special:Export when not exporting whole history
 222          *
 223          * @param array $cond
 224          */
 225         protected function do_list_authors( $cond ) {
 226                 $this->author_list = "<contributors>";
 227                 // rev_deleted
 228
 229                 $res = $this->db->select(
 230                         [ 'page', 'revision' ],
 231                         [ 'DISTINCT rev_user_text', 'rev_user' ],
 232                         [
 233                                 $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ' = 0',
 234                                 $cond,
 235                                 'page_id = rev_id',
 236                         ],
 237                         __METHOD__
 238                 );
 239
 240                 foreach ( $res as $row ) {
 241                         $this->author_list .= "<contributor>" .
 242                                 "<username>" .
 243                                 htmlentities( $row->rev_user_text ) .
 244                                 "</username>" .
 245                                 "<id>" .
 246                                 $row->rev_user .
 247                                 "</id>" .
 248                                 "</contributor>";
 249                 }
 250                 $this->author_list .= "</contributors>";
 251         }
 252
 253         /**
 254          * @param string $cond
 255          * @throws MWException
 256          * @throws Exception
 257          */
 258         protected function dumpFrom( $cond = '', $orderRevs = false ) {
 259                 # For logging dumps...
 260                 if ( $this->history & self::LOGS ) {
 261                         $where = [ 'user_id = log_user' ];
 262                         # Hide private logs
 263                         $hideLogs = LogEventsList::getExcludeClause( $this->db );
 264                         if ( $hideLogs ) {
 265                                 $where[] = $hideLogs;
 266                         }
 267                         # Add on any caller specified conditions
 268                         if ( $cond ) {
 269                                 $where[] = $cond;
 270                         }
 271                         # Get logging table name for logging.* clause
 272                         $logging = $this->db->tableName( 'logging' );
 273
 274                         if ( $this->buffer == WikiExporter::STREAM ) {
 275                                 $prev = $this->db->bufferResults( false );
 276                         }
 277                         $result = null; // Assuring $result is not undefined, if exception occurs early
 278                         try {
 279                                 $result = $this->db->select( [ 'logging', 'user' ],
 280                                         [ "{$logging}.*", 'user_name' ], // grab the user name
 281                                         $where,
 282                                         __METHOD__,
 283                                         [ 'ORDER BY' => 'log_id', 'USE INDEX' => [ 'logging' => 'PRIMARY' ] ]
 284                                 );
 285                                 $this->outputLogStream( $result );
 286                                 if ( $this->buffer == WikiExporter::STREAM ) {
 287                                         $this->db->bufferResults( $prev );
 288                                 }
 289                         } catch ( Exception $e ) {
 290                                 // Throwing the exception does not reliably free the resultset, and
 291                                 // would also leave the connection in unbuffered mode.
 292
 293                                 // Freeing result
 294                                 try {
 295                                         if ( $result ) {
 296                                                 $result->free();
 297                                         }
 298                                 } catch ( Exception $e2 ) {
 299                                         // Already in panic mode -> ignoring $e2 as $e has
 300                                         // higher priority
 301                                 }
 302
 303                                 // Putting database back in previous buffer mode
 304                                 try {
 305                                         if ( $this->buffer == WikiExporter::STREAM ) {
 306                                                 $this->db->bufferResults( $prev );
 307                                         }
 308                                 } catch ( Exception $e2 ) {
 309                                         // Already in panic mode -> ignoring $e2 as $e has
 310                                         // higher priority
 311                                 }
 312
 313                                 // Inform caller about problem
 314                                 throw $e;
 315                         }
 316                 # For page dumps...
 317                 } else {
 318                         $tables = [ 'page', 'revision' ];
 319                         $opts = [ 'ORDER BY' => 'page_id ASC' ];
 320                         $opts['USE INDEX'] = [];
 321                         $join = [];
 322                         if ( is_array( $this->history ) ) {
 323                                 # Time offset/limit for all pages/history...
 324                                 $revJoin = 'page_id=rev_page';
 325                                 # Set time order
 326                                 if ( $this->history['dir'] == 'asc' ) {
 327                                         $op = '>';
 328                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 329                                 } else {
 330                                         $op = '<';
 331                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 332                                 }
 333                                 # Set offset
 334                                 if ( !empty( $this->history['offset'] ) ) {
 335                                         $revJoin .= " AND rev_timestamp $op " .
 336                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 337                                 }
 338                                 $join['revision'] = [ 'INNER JOIN', $revJoin ];
 339                                 # Set query limit
 340                                 if ( !empty( $this->history['limit'] ) ) {
 341                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 342                                 }
 343                         } elseif ( $this->history & WikiExporter::FULL ) {
 344                                 # Full history dumps...
 345                                 # query optimization for history stub dumps
 346                                 if ( $this->text == WikiExporter::STUB && $orderRevs ) {
 347                                         $tables = [ 'revision', 'page' ];
 348                                         $opts[] = 'STRAIGHT_JOIN';
 349                                         $opts['ORDER BY'] = [ 'rev_page ASC', 'rev_id ASC' ];
 350                                         $opts['USE INDEX']['revision'] = 'rev_page_id';
 351                                         $join['page'] = [ 'INNER JOIN', 'rev_page=page_id' ];
 352                                 } else {
 353                                         $join['revision'] = [ 'INNER JOIN', 'page_id=rev_page' ];
 354                                 }
 355                         } elseif ( $this->history & WikiExporter::CURRENT ) {
 356                                 # Latest revision dumps...
 357                                 if ( $this->list_authors && $cond != '' ) { // List authors, if so desired
 358                                         $this->do_list_authors( $cond );
 359                                 }
 360                                 $join['revision'] = [ 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' ];
 361                         } elseif ( $this->history & WikiExporter::STABLE ) {
 362                                 # "Stable" revision dumps...
 363                                 # Default JOIN, to be overridden...
 364                                 $join['revision'] = [ 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' ];
 365                                 # One, and only one hook should set this, and return false
 366                                 if ( Hooks::run( 'WikiExporter::dumpStableQuery', [ &$tables, &$opts, &$join ] ) ) {
 367                                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 368                                 }
 369                         } elseif ( $this->history & WikiExporter::RANGE ) {
 370                                 # Dump of revisions within a specified range
 371                                 $join['revision'] = [ 'INNER JOIN', 'page_id=rev_page' ];
 372                                 $opts['ORDER BY'] = [ 'rev_page ASC', 'rev_id ASC' ];
 373                         } else {
 374                                 # Unknown history specification parameter?
 375                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 376                         }
 377                         # Query optimization hacks
 378                         if ( $cond == '' ) {
 379                                 $opts[] = 'STRAIGHT_JOIN';
 380                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 381                         }
 382                         # Build text join options
 383                         if ( $this->text != WikiExporter::STUB ) { // 1-pass
 384                                 $tables[] = 'text';
 385                                 $join['text'] = [ 'INNER JOIN', 'rev_text_id=old_id' ];
 386                         }
 387
 388                         if ( $this->buffer == WikiExporter::STREAM ) {
 389                                 $prev = $this->db->bufferResults( false );
 390                         }
 391                         $result = null; // Assuring $result is not undefined, if exception occurs early
 392                         try {
 393                                 Hooks::run( 'ModifyExportQuery',
 394                                                 [ $this->db, &$tables, &$cond, &$opts, &$join ] );
 395
 396                                 # Do the query!
 397                                 $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 398                                 # Output dump results
 399                                 $this->outputPageStream( $result );
 400
 401                                 if ( $this->buffer == WikiExporter::STREAM ) {
 402                                         $this->db->bufferResults( $prev );
 403                                 }
 404                         } catch ( Exception $e ) {
 405                                 // Throwing the exception does not reliably free the resultset, and
 406                                 // would also leave the connection in unbuffered mode.
 407
 408                                 // Freeing result
 409                                 try {
 410                                         if ( $result ) {
 411                                                 $result->free();
 412                                         }
 413                                 } catch ( Exception $e2 ) {
 414                                         // Already in panic mode -> ignoring $e2 as $e has
 415                                         // higher priority
 416                                 }
 417
 418                                 // Putting database back in previous buffer mode
 419                                 try {
 420                                         if ( $this->buffer == WikiExporter::STREAM ) {
 421                                                 $this->db->bufferResults( $prev );
 422                                         }
 423                                 } catch ( Exception $e2 ) {
 424                                         // Already in panic mode -> ignoring $e2 as $e has
 425                                         // higher priority
 426                                 }
 427
 428                                 // Inform caller about problem
 429                                 throw $e;
 430                         }
 431                 }
 432         }
 433
 434         /**
 435          * Runs through a query result set dumping page and revision records.
 436          * The result set should be sorted/grouped by page to avoid duplicate
 437          * page records in the output.
 438          *
 439          * Should be safe for
 440          * streaming (non-buffered) queries, as long as it was made on a
 441          * separate database connection not managed by LoadBalancer; some
 442          * blob storage types will make queries to pull source data.
 443          *
 444          * @param ResultWrapper $resultset
 445          */
 446         protected function outputPageStream( $resultset ) {
 447                 $last = null;
 448                 foreach ( $resultset as $row ) {
 449                         if ( $last === null ||
 450                                 $last->page_namespace != $row->page_namespace ||
 451                                 $last->page_title != $row->page_title ) {
 452                                 if ( $last !== null ) {
 453                                         $output = '';
 454                                         if ( $this->dumpUploads ) {
 455                                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 456                                         }
 457                                         $output .= $this->writer->closePage();
 458                                         $this->sink->writeClosePage( $output );
 459                                 }
 460                                 $output = $this->writer->openPage( $row );
 461                                 $this->sink->writeOpenPage( $row, $output );
 462                                 $last = $row;
 463                         }
 464                         $output = $this->writer->writeRevision( $row );
 465                         $this->sink->writeRevision( $row, $output );
 466                 }
 467                 if ( $last !== null ) {
 468                         $output = '';
 469                         if ( $this->dumpUploads ) {
 470                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 471                         }
 472                         $output .= $this->author_list;
 473                         $output .= $this->writer->closePage();
 474                         $this->sink->writeClosePage( $output );
 475                 }
 476         }
 477
 478         /**
 479          * @param ResultWrapper $resultset
 480          */
 481         protected function outputLogStream( $resultset ) {
 482                 foreach ( $resultset as $row ) {
 483                         $output = $this->writer->writeLogItem( $row );
 484                         $this->sink->writeLogItem( $row, $output );
 485                 }
 486         }
 487 }