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