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          */
 138         public function pagesByRange( $start, $end ) {
 139                 $condition = 'page_id >= ' . intval( $start );
 140                 if ( $end ) {
 141                         $condition .= ' AND page_id < ' . intval( $end );
 142                 }
 143                 $this->dumpFrom( $condition );
 144         }
 145
 146         /**
 147          * Dumps a series of page and revision records for those pages
 148          * in the database with revisions falling within the rev_id range given.
 149          * @param int $start Inclusive lower limit (this id is included)
 150          * @param int $end Exclusive upper limit (this id is not included)
 151          *   If 0, no upper limit.
 152          */
 153         public function revsByRange( $start, $end ) {
 154                 $condition = 'rev_id >= ' . intval( $start );
 155                 if ( $end ) {
 156                         $condition .= ' AND rev_id < ' . intval( $end );
 157                 }
 158                 $this->dumpFrom( $condition );
 159         }
 160
 161         /**
 162          * @param Title $title
 163          */
 164         public function pageByTitle( $title ) {
 165                 $this->dumpFrom(
 166                         'page_namespace=' . $title->getNamespace() .
 167                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 168         }
 169
 170         /**
 171          * @param string $name
 172          * @throws MWException
 173          */
 174         public function pageByName( $name ) {
 175                 $title = Title::newFromText( $name );
 176                 if ( is_null( $title ) ) {
 177                         throw new MWException( "Can't export invalid title" );
 178                 } else {
 179                         $this->pageByTitle( $title );
 180                 }
 181         }
 182
 183         /**
 184          * @param array $names
 185          */
 186         public function pagesByName( $names ) {
 187                 foreach ( $names as $name ) {
 188                         $this->pageByName( $name );
 189                 }
 190         }
 191
 192         public function allLogs() {
 193                 $this->dumpFrom( '' );
 194         }
 195
 196         /**
 197          * @param int $start
 198          * @param int $end
 199          */
 200         public function logsByRange( $start, $end ) {
 201                 $condition = 'log_id >= ' . intval( $start );
 202                 if ( $end ) {
 203                         $condition .= ' AND log_id < ' . intval( $end );
 204                 }
 205                 $this->dumpFrom( $condition );
 206         }
 207
 208         /**
 209          * Generates the distinct list of authors of an article
 210          * Not called by default (depends on $this->list_authors)
 211          * Can be set by Special:Export when not exporting whole history
 212          *
 213          * @param array $cond
 214          */
 215         protected function do_list_authors( $cond ) {
 216                 $this->author_list = "<contributors>";
 217                 // rev_deleted
 218
 219                 $res = $this->db->select(
 220                         [ 'page', 'revision' ],
 221                         [ 'DISTINCT rev_user_text', 'rev_user' ],
 222                         [
 223                                 $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ' = 0',
 224                                 $cond,
 225                                 'page_id = rev_id',
 226                         ],
 227                         __METHOD__
 228                 );
 229
 230                 foreach ( $res as $row ) {
 231                         $this->author_list .= "<contributor>" .
 232                                 "<username>" .
 233                                 htmlentities( $row->rev_user_text ) .
 234                                 "</username>" .
 235                                 "<id>" .
 236                                 $row->rev_user .
 237                                 "</id>" .
 238                                 "</contributor>";
 239                 }
 240                 $this->author_list .= "</contributors>";
 241         }
 242
 243         /**
 244          * @param string $cond
 245          * @throws MWException
 246          * @throws Exception
 247          */
 248         protected function dumpFrom( $cond = '' ) {
 249                 # For logging dumps...
 250                 if ( $this->history & self::LOGS ) {
 251                         $where = [ 'user_id = log_user' ];
 252                         # Hide private logs
 253                         $hideLogs = LogEventsList::getExcludeClause( $this->db );
 254                         if ( $hideLogs ) {
 255                                 $where[] = $hideLogs;
 256                         }
 257                         # Add on any caller specified conditions
 258                         if ( $cond ) {
 259                                 $where[] = $cond;
 260                         }
 261                         # Get logging table name for logging.* clause
 262                         $logging = $this->db->tableName( 'logging' );
 263
 264                         if ( $this->buffer == WikiExporter::STREAM ) {
 265                                 $prev = $this->db->bufferResults( false );
 266                         }
 267                         $result = null; // Assuring $result is not undefined, if exception occurs early
 268                         try {
 269                                 $result = $this->db->select( [ 'logging', 'user' ],
 270                                         [ "{$logging}.*", 'user_name' ], // grab the user name
 271                                         $where,
 272                                         __METHOD__,
 273                                         [ 'ORDER BY' => 'log_id', 'USE INDEX' => [ 'logging' => 'PRIMARY' ] ]
 274                                 );
 275                                 $this->outputLogStream( $result );
 276                                 if ( $this->buffer == WikiExporter::STREAM ) {
 277                                         $this->db->bufferResults( $prev );
 278                                 }
 279                         } catch ( Exception $e ) {
 280                                 // Throwing the exception does not reliably free the resultset, and
 281                                 // would also leave the connection in unbuffered mode.
 282
 283                                 // Freeing result
 284                                 try {
 285                                         if ( $result ) {
 286                                                 $result->free();
 287                                         }
 288                                 } catch ( Exception $e2 ) {
 289                                         // Already in panic mode -> ignoring $e2 as $e has
 290                                         // higher priority
 291                                 }
 292
 293                                 // Putting database back in previous buffer mode
 294                                 try {
 295                                         if ( $this->buffer == WikiExporter::STREAM ) {
 296                                                 $this->db->bufferResults( $prev );
 297                                         }
 298                                 } catch ( Exception $e2 ) {
 299                                         // Already in panic mode -> ignoring $e2 as $e has
 300                                         // higher priority
 301                                 }
 302
 303                                 // Inform caller about problem
 304                                 throw $e;
 305                         }
 306                 # For page dumps...
 307                 } else {
 308                         $tables = [ 'page', 'revision' ];
 309                         $opts = [ 'ORDER BY' => 'page_id ASC' ];
 310                         $opts['USE INDEX'] = [];
 311                         $join = [];
 312                         if ( is_array( $this->history ) ) {
 313                                 # Time offset/limit for all pages/history...
 314                                 $revJoin = 'page_id=rev_page';
 315                                 # Set time order
 316                                 if ( $this->history['dir'] == 'asc' ) {
 317                                         $op = '>';
 318                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 319                                 } else {
 320                                         $op = '<';
 321                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 322                                 }
 323                                 # Set offset
 324                                 if ( !empty( $this->history['offset'] ) ) {
 325                                         $revJoin .= " AND rev_timestamp $op " .
 326                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 327                                 }
 328                                 $join['revision'] = [ 'INNER JOIN', $revJoin ];
 329                                 # Set query limit
 330                                 if ( !empty( $this->history['limit'] ) ) {
 331                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 332                                 }
 333                         } elseif ( $this->history & WikiExporter::FULL ) {
 334                                 # Full history dumps...
 335                                 $join['revision'] = [ 'INNER JOIN', 'page_id=rev_page' ];
 336                         } elseif ( $this->history & WikiExporter::CURRENT ) {
 337                                 # Latest revision dumps...
 338                                 if ( $this->list_authors && $cond != '' ) { // List authors, if so desired
 339                                         $this->do_list_authors( $cond );
 340                                 }
 341                                 $join['revision'] = [ 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' ];
 342                         } elseif ( $this->history & WikiExporter::STABLE ) {
 343                                 # "Stable" revision dumps...
 344                                 # Default JOIN, to be overridden...
 345                                 $join['revision'] = [ 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' ];
 346                                 # One, and only one hook should set this, and return false
 347                                 if ( Hooks::run( 'WikiExporter::dumpStableQuery', [ &$tables, &$opts, &$join ] ) ) {
 348                                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 349                                 }
 350                         } elseif ( $this->history & WikiExporter::RANGE ) {
 351                                 # Dump of revisions within a specified range
 352                                 $join['revision'] = [ 'INNER JOIN', 'page_id=rev_page' ];
 353                                 $opts['ORDER BY'] = [ 'rev_page ASC', 'rev_id ASC' ];
 354                         } else {
 355                                 # Unknown history specification parameter?
 356                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 357                         }
 358                         # Query optimization hacks
 359                         if ( $cond == '' ) {
 360                                 $opts[] = 'STRAIGHT_JOIN';
 361                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 362                         }
 363                         # Build text join options
 364                         if ( $this->text != WikiExporter::STUB ) { // 1-pass
 365                                 $tables[] = 'text';
 366                                 $join['text'] = [ 'INNER JOIN', 'rev_text_id=old_id' ];
 367                         }
 368
 369                         if ( $this->buffer == WikiExporter::STREAM ) {
 370                                 $prev = $this->db->bufferResults( false );
 371                         }
 372
 373                         $result = null; // Assuring $result is not undefined, if exception occurs early
 374                         try {
 375                                 Hooks::run( 'ModifyExportQuery',
 376                                                 [ $this->db, &$tables, &$cond, &$opts, &$join ] );
 377
 378                                 # Do the query!
 379                                 $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 380                                 # Output dump results
 381                                 $this->outputPageStream( $result );
 382
 383                                 if ( $this->buffer == WikiExporter::STREAM ) {
 384                                         $this->db->bufferResults( $prev );
 385                                 }
 386                         } catch ( Exception $e ) {
 387                                 // Throwing the exception does not reliably free the resultset, and
 388                                 // would also leave the connection in unbuffered mode.
 389
 390                                 // Freeing result
 391                                 try {
 392                                         if ( $result ) {
 393                                                 $result->free();
 394                                         }
 395                                 } catch ( Exception $e2 ) {
 396                                         // Already in panic mode -> ignoring $e2 as $e has
 397                                         // higher priority
 398                                 }
 399
 400                                 // Putting database back in previous buffer mode
 401                                 try {
 402                                         if ( $this->buffer == WikiExporter::STREAM ) {
 403                                                 $this->db->bufferResults( $prev );
 404                                         }
 405                                 } catch ( Exception $e2 ) {
 406                                         // Already in panic mode -> ignoring $e2 as $e has
 407                                         // higher priority
 408                                 }
 409
 410                                 // Inform caller about problem
 411                                 throw $e;
 412                         }
 413                 }
 414         }
 415
 416         /**
 417          * Runs through a query result set dumping page and revision records.
 418          * The result set should be sorted/grouped by page to avoid duplicate
 419          * page records in the output.
 420          *
 421          * Should be safe for
 422          * streaming (non-buffered) queries, as long as it was made on a
 423          * separate database connection not managed by LoadBalancer; some
 424          * blob storage types will make queries to pull source data.
 425          *
 426          * @param ResultWrapper $resultset
 427          */
 428         protected function outputPageStream( $resultset ) {
 429                 $last = null;
 430                 foreach ( $resultset as $row ) {
 431                         if ( $last === null ||
 432                                 $last->page_namespace != $row->page_namespace ||
 433                                 $last->page_title != $row->page_title ) {
 434                                 if ( $last !== null ) {
 435                                         $output = '';
 436                                         if ( $this->dumpUploads ) {
 437                                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 438                                         }
 439                                         $output .= $this->writer->closePage();
 440                                         $this->sink->writeClosePage( $output );
 441                                 }
 442                                 $output = $this->writer->openPage( $row );
 443                                 $this->sink->writeOpenPage( $row, $output );
 444                                 $last = $row;
 445                         }
 446                         $output = $this->writer->writeRevision( $row );
 447                         $this->sink->writeRevision( $row, $output );
 448                 }
 449                 if ( $last !== null ) {
 450                         $output = '';
 451                         if ( $this->dumpUploads ) {
 452                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 453                         }
 454                         $output .= $this->author_list;
 455                         $output .= $this->writer->closePage();
 456                         $this->sink->writeClosePage( $output );
 457                 }
 458         }
 459
 460         /**
 461          * @param ResultWrapper $resultset
 462          */
 463         protected function outputLogStream( $resultset ) {
 464                 foreach ( $resultset as $row ) {
 465                         $output = $this->writer->writeLogItem( $row );
 466                         $this->sink->writeLogItem( $row, $output );
 467                 }
 468         }
 469 }