includes/export/XmlDumpWriter.php

   1 <?php
   2 /**
   3  * XmlDumpWriter
   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 use MediaWiki\MediaWikiServices;
  26 use MediaWiki\Revision\RevisionStore;
  27 use MediaWiki\Storage\SqlBlobStore;
  28
  29 /**
  30  * @ingroup Dump
  31  */
  32 class XmlDumpWriter {
  33         /**
  34          * @var string[] the schema versions supported for output
  35          * @final
  36          */
  37         public static $supportedSchemas = [
  38                 XML_DUMP_SCHEMA_VERSION_10,
  39         ];
  40
  41         /**
  42          * Title of the currently processed page
  43          *
  44          * @var Title|null
  45          */
  46         private $currentTitle = null;
  47
  48         /**
  49          * Opens the XML output stream's root "<mediawiki>" element.
  50          * This does not include an xml directive, so is safe to include
  51          * as a subelement in a larger XML stream. Namespace and XML Schema
  52          * references are included.
  53          *
  54          * Output will be encoded in UTF-8.
  55          *
  56          * @return string
  57          */
  58         function openStream() {
  59                 $ver = WikiExporter::schemaVersion();
  60                 return Xml::element( 'mediawiki', [
  61                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
  62                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
  63                         /*
  64                          * When a new version of the schema is created, it needs staging on mediawiki.org.
  65                          * This requires a change in the operations/mediawiki-config git repo.
  66                          *
  67                          * Create a changeset like https://gerrit.wikimedia.org/r/#/c/149643/ in which
  68                          * you copy in the new xsd file.
  69                          *
  70                          * After it is reviewed, merged and deployed (sync-docroot), the index.html needs purging.
  71                          * echo "https://www.mediawiki.org/xml/index.html" | mwscript purgeList.php --wiki=aawiki
  72                          */
  73                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
  74                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
  75                         'version' => $ver,
  76                         'xml:lang' => MediaWikiServices::getInstance()->getContentLanguage()->getHtmlCode() ],
  77                         null ) .
  78                         "\n" .
  79                         $this->siteInfo();
  80         }
  81
  82         /**
  83          * @return string
  84          */
  85         function siteInfo() {
  86                 $info = [
  87                         $this->sitename(),
  88                         $this->dbname(),
  89                         $this->homelink(),
  90                         $this->generator(),
  91                         $this->caseSetting(),
  92                         $this->namespaces() ];
  93                 return "  <siteinfo>\n    " .
  94                         implode( "\n    ", $info ) .
  95                         "\n  </siteinfo>\n";
  96         }
  97
  98         /**
  99          * @return string
 100          */
 101         function sitename() {
 102                 global $wgSitename;
 103                 return Xml::element( 'sitename', [], $wgSitename );
 104         }
 105
 106         /**
 107          * @return string
 108          */
 109         function dbname() {
 110                 global $wgDBname;
 111                 return Xml::element( 'dbname', [], $wgDBname );
 112         }
 113
 114         /**
 115          * @return string
 116          */
 117         function generator() {
 118                 global $wgVersion;
 119                 return Xml::element( 'generator', [], "MediaWiki $wgVersion" );
 120         }
 121
 122         /**
 123          * @return string
 124          */
 125         function homelink() {
 126                 return Xml::element( 'base', [], Title::newMainPage()->getCanonicalURL() );
 127         }
 128
 129         /**
 130          * @return string
 131          */
 132         function caseSetting() {
 133                 global $wgCapitalLinks;
 134                 // "case-insensitive" option is reserved for future
 135                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 136                 return Xml::element( 'case', [], $sensitivity );
 137         }
 138
 139         /**
 140          * @return string
 141          */
 142         function namespaces() {
 143                 $spaces = "<namespaces>\n";
 144                 $nsInfo = MediaWikiServices::getInstance()->getNamespaceInfo();
 145                 foreach (
 146                         MediaWikiServices::getInstance()->getContentLanguage()->getFormattedNamespaces()
 147                         as $ns => $title
 148                 ) {
 149                         $spaces .= '      ' .
 150                                 Xml::element( 'namespace',
 151                                         [
 152                                                 'key' => $ns,
 153                                                 'case' => $nsInfo->isCapitalized( $ns )
 154                                                         ? 'first-letter' : 'case-sensitive',
 155                                         ], $title ) . "\n";
 156                 }
 157                 $spaces .= "    </namespaces>";
 158                 return $spaces;
 159         }
 160
 161         /**
 162          * Closes the output stream with the closing root element.
 163          * Call when finished dumping things.
 164          *
 165          * @return string
 166          */
 167         function closeStream() {
 168                 return "</mediawiki>\n";
 169         }
 170
 171         /**
 172          * Opens a "<page>" section on the output stream, with data
 173          * from the given database row.
 174          *
 175          * @param object $row
 176          * @return string
 177          */
 178         public function openPage( $row ) {
 179                 $out = "  <page>\n";
 180                 $this->currentTitle = Title::makeTitle( $row->page_namespace, $row->page_title );
 181                 $canonicalTitle = self::canonicalTitle( $this->currentTitle );
 182                 $out .= '    ' . Xml::elementClean( 'title', [], $canonicalTitle ) . "\n";
 183                 $out .= '    ' . Xml::element( 'ns', [], strval( $row->page_namespace ) ) . "\n";
 184                 $out .= '    ' . Xml::element( 'id', [], strval( $row->page_id ) ) . "\n";
 185                 if ( $row->page_is_redirect ) {
 186                         $page = WikiPage::factory( $this->currentTitle );
 187                         $redirect = $page->getRedirectTarget();
 188                         if ( $redirect instanceof Title && $redirect->isValidRedirectTarget() ) {
 189                                 $out .= '    ';
 190                                 $out .= Xml::element( 'redirect', [ 'title' => self::canonicalTitle( $redirect ) ] );
 191                                 $out .= "\n";
 192                         }
 193                 }
 194
 195                 if ( $row->page_restrictions != '' ) {
 196                         $out .= '    ' . Xml::element( 'restrictions', [],
 197                                 strval( $row->page_restrictions ) ) . "\n";
 198                 }
 199
 200                 Hooks::run( 'XmlDumpWriterOpenPage', [ $this, &$out, $row, $this->currentTitle ] );
 201
 202                 return $out;
 203         }
 204
 205         /**
 206          * Closes a "<page>" section on the output stream.
 207          *
 208          * @private
 209          * @return string
 210          */
 211         function closePage() {
 212                 if ( $this->currentTitle !== null ) {
 213                         $linkCache = MediaWikiServices::getInstance()->getLinkCache();
 214                         // In rare cases, link cache has the same key for some pages which
 215                         // might be read as part of the same batch. T220424 and T220316
 216                         $linkCache->clearLink( $this->currentTitle );
 217                 }
 218                 return "  </page>\n";
 219         }
 220
 221         /**
 222          * @return RevisionStore
 223          */
 224         private function getRevisionStore() {
 225                 return MediaWikiServices::getInstance()->getRevisionStore();
 226         }
 227
 228         /**
 229          * @return SqlBlobStore
 230          */
 231         private function getBlobStore() {
 232                 return MediaWikiServices::getInstance()->getBlobStore();
 233         }
 234
 235         /**
 236          * Dumps a "<revision>" section on the output stream, with
 237          * data filled in from the given database row.
 238          *
 239          * @param object $row
 240          * @return string
 241          * @private
 242          */
 243         function writeRevision( $row ) {
 244                 $out = "    <revision>\n";
 245                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 246                 if ( isset( $row->rev_parent_id ) && $row->rev_parent_id ) {
 247                         $out .= "      " . Xml::element( 'parentid', null, strval( $row->rev_parent_id ) ) . "\n";
 248                 }
 249
 250                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 251
 252                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_USER ) ) {
 253                         $out .= "      " . Xml::element( 'contributor', [ 'deleted' => 'deleted' ] ) . "\n";
 254                 } else {
 255                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 256                 }
 257
 258                 if ( isset( $row->rev_minor_edit ) && $row->rev_minor_edit ) {
 259                         $out .= "      <minor/>\n";
 260                 }
 261                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_COMMENT ) ) {
 262                         $out .= "      " . Xml::element( 'comment', [ 'deleted' => 'deleted' ] ) . "\n";
 263                 } else {
 264                         $comment = CommentStore::getStore()->getComment( 'rev_comment', $row )->text;
 265                         if ( $comment != '' ) {
 266                                 $out .= "      " . Xml::elementClean( 'comment', [], strval( $comment ) ) . "\n";
 267                         }
 268                 }
 269
 270                 // TODO: rev_content_model no longer exists with MCR, see T174031
 271                 if ( isset( $row->rev_content_model ) && !is_null( $row->rev_content_model ) ) {
 272                         $content_model = strval( $row->rev_content_model );
 273                 } else {
 274                         // probably using $wgContentHandlerUseDB = false;
 275                         $content_model = ContentHandler::getDefaultModelFor( $this->currentTitle );
 276                 }
 277
 278                 $content_handler = ContentHandler::getForModelID( $content_model );
 279
 280                 // TODO: rev_content_format no longer exists with MCR, see T174031
 281                 if ( isset( $row->rev_content_format ) && !is_null( $row->rev_content_format ) ) {
 282                         $content_format = strval( $row->rev_content_format );
 283                 } else {
 284                         // probably using $wgContentHandlerUseDB = false;
 285                         $content_format = $content_handler->getDefaultFormat();
 286                 }
 287
 288                 $out .= "      " . Xml::element( 'model', null, strval( $content_model ) ) . "\n";
 289                 $out .= "      " . Xml::element( 'format', null, strval( $content_format ) ) . "\n";
 290
 291                 $text = '';
 292                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_TEXT ) ) {
 293                         $out .= "      " . Xml::element( 'text', [ 'deleted' => 'deleted' ] ) . "\n";
 294                 } elseif ( isset( $row->old_text ) ) {
 295                         // Raw text from the database may have invalid chars
 296                         $text = strval( Revision::getRevisionText( $row ) );
 297                         try {
 298                                 $text = $content_handler->exportTransform( $text, $content_format );
 299                         }
 300                         catch ( Exception $ex ) {
 301                                 if ( $ex instanceof MWException || $ex instanceof RuntimeException ) {
 302                                         // leave text as is; that's the way it goes
 303                                         wfLogWarning( 'exportTransform failed on text for revid ' . $row->rev_id . "\n" );
 304                                 } else {
 305                                         throw $ex;
 306                                 }
 307                         }
 308                         $out .= "      " . Xml::elementClean( 'text',
 309                                 [ 'xml:space' => 'preserve', 'bytes' => intval( $row->rev_len ) ],
 310                                 strval( $text ) ) . "\n";
 311                 } elseif ( isset( $row->_load_content ) ) {
 312                         // TODO: make this fully MCR aware, see T174031
 313                         $rev = $this->getRevisionStore()->newRevisionFromRow( $row, 0, $this->currentTitle );
 314                         $slot = $rev->getSlot( 'main' );
 315                         try {
 316                                 $content = $slot->getContent();
 317
 318                                 if ( $content instanceof TextContent ) {
 319                                         // HACK: For text based models, bypass the serialization step.
 320                                         // This allows extensions (like Flow)that use incompatible combinations
 321                                         // of serialization format and content model.
 322                                         $text = $content->getNativeData();
 323                                 } else {
 324                                         $text = $content->serialize( $content_format );
 325                                 }
 326                                 $text = $content_handler->exportTransform( $text, $content_format );
 327                                 $out .= "      " . Xml::elementClean( 'text',
 328                                         [ 'xml:space' => 'preserve', 'bytes' => intval( $slot->getSize() ) ],
 329                                         strval( $text ) ) . "\n";
 330                         }
 331                         catch ( Exception $ex ) {
 332                                 if ( $ex instanceof MWException || $ex instanceof RuntimeException ) {
 333                                         // there's no provsion in the schema for an attribute that will let
 334                                         // the user know this element was unavailable due to error; an empty
 335                                         // tag is the best we can do
 336                                         $out .= "      " . Xml::element( 'text' ) . "\n";
 337                                         wfLogWarning( 'failed to load content for revid ' . $row->rev_id . "\n" );
 338                                 } else {
 339                                         throw $ex;
 340                                 }
 341                         }
 342                 } elseif ( isset( $row->rev_text_id ) ) {
 343                         // Stub output for pre-MCR schema
 344                         // TODO: MCR: rev_text_id only exists in the pre-MCR schema. Remove this when
 345                         // we drop support for the old schema.
 346                         $out .= "      " . Xml::element( 'text',
 347                                 [ 'id' => $row->rev_text_id, 'bytes' => intval( $row->rev_len ) ],
 348                                 "" ) . "\n";
 349                 } else {
 350                         // Backwards-compatible stub output for MCR aware schema
 351                         // TODO: MCR: emit content addresses instead of text ids, see T174031, T199121
 352                         $rev = $this->getRevisionStore()->newRevisionFromRow( $row, 0, $this->currentTitle );
 353                         $slot = $rev->getSlot( 'main' );
 354
 355                         // Note that this is currently the ONLY reason we have a BlobStore here at all.
 356                         // When removing this line, check whether the BlobStore has become unused.
 357                         $textId = $this->getBlobStore()->getTextIdFromAddress( $slot->getAddress() );
 358                         $out .= "      " . Xml::element( 'text',
 359                                         [ 'id' => $textId, 'bytes' => intval( $slot->getSize() ) ],
 360                                         "" ) . "\n";
 361                 }
 362
 363                 if ( isset( $row->rev_sha1 )
 364                         && $row->rev_sha1
 365                         && !( $row->rev_deleted & Revision::DELETED_TEXT )
 366                 ) {
 367                         $out .= "      " . Xml::element( 'sha1', null, strval( $row->rev_sha1 ) ) . "\n";
 368                 } else {
 369                         $out .= "      <sha1/>\n";
 370                 }
 371
 372                 // Avoid PHP 7.1 warning from passing $this by reference
 373                 $writer = $this;
 374                 Hooks::run( 'XmlDumpWriterWriteRevision', [ &$writer, &$out, $row, $text ] );
 375
 376                 $out .= "    </revision>\n";
 377
 378                 return $out;
 379         }
 380
 381         /**
 382          * Dumps a "<logitem>" section on the output stream, with
 383          * data filled in from the given database row.
 384          *
 385          * @param object $row
 386          * @return string
 387          * @private
 388          */
 389         function writeLogItem( $row ) {
 390                 $out = "  <logitem>\n";
 391                 $out .= "    " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 392
 393                 $out .= $this->writeTimestamp( $row->log_timestamp, "    " );
 394
 395                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 396                         $out .= "    " . Xml::element( 'contributor', [ 'deleted' => 'deleted' ] ) . "\n";
 397                 } else {
 398                         $out .= $this->writeContributor( $row->log_user, $row->user_name, "    " );
 399                 }
 400
 401                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 402                         $out .= "    " . Xml::element( 'comment', [ 'deleted' => 'deleted' ] ) . "\n";
 403                 } else {
 404                         $comment = CommentStore::getStore()->getComment( 'log_comment', $row )->text;
 405                         if ( $comment != '' ) {
 406                                 $out .= "    " . Xml::elementClean( 'comment', null, strval( $comment ) ) . "\n";
 407                         }
 408                 }
 409
 410                 $out .= "    " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 411                 $out .= "    " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 412
 413                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 414                         $out .= "    " . Xml::element( 'text', [ 'deleted' => 'deleted' ] ) . "\n";
 415                 } else {
 416                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 417                         $out .= "    " . Xml::elementClean( 'logtitle', null, self::canonicalTitle( $title ) ) . "\n";
 418                         $out .= "    " . Xml::elementClean( 'params',
 419                                 [ 'xml:space' => 'preserve' ],
 420                                 strval( $row->log_params ) ) . "\n";
 421                 }
 422
 423                 $out .= "  </logitem>\n";
 424
 425                 return $out;
 426         }
 427
 428         /**
 429          * @param string $timestamp
 430          * @param string $indent Default to six spaces
 431          * @return string
 432          */
 433         function writeTimestamp( $timestamp, $indent = "      " ) {
 434                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 435                 return $indent . Xml::element( 'timestamp', null, $ts ) . "\n";
 436         }
 437
 438         /**
 439          * @param int $id
 440          * @param string $text
 441          * @param string $indent Default to six spaces
 442          * @return string
 443          */
 444         function writeContributor( $id, $text, $indent = "      " ) {
 445                 $out = $indent . "<contributor>\n";
 446                 if ( $id || !IP::isValid( $text ) ) {
 447                         $out .= $indent . "  " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 448                         $out .= $indent . "  " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 449                 } else {
 450                         $out .= $indent . "  " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 451                 }
 452                 $out .= $indent . "</contributor>\n";
 453                 return $out;
 454         }
 455
 456         /**
 457          * Warning! This data is potentially inconsistent. :(
 458          * @param object $row
 459          * @param bool $dumpContents
 460          * @return string
 461          */
 462         function writeUploads( $row, $dumpContents = false ) {
 463                 if ( $row->page_namespace == NS_FILE ) {
 464                         $img = wfLocalFile( $row->page_title );
 465                         if ( $img && $img->exists() ) {
 466                                 $out = '';
 467                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 468                                         $out .= $this->writeUpload( $ver, $dumpContents );
 469                                 }
 470                                 $out .= $this->writeUpload( $img, $dumpContents );
 471                                 return $out;
 472                         }
 473                 }
 474                 return '';
 475         }
 476
 477         /**
 478          * @param File $file
 479          * @param bool $dumpContents
 480          * @return string
 481          */
 482         function writeUpload( $file, $dumpContents = false ) {
 483                 if ( $file->isOld() ) {
 484                         $archiveName = "      " .
 485                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 486                 } else {
 487                         $archiveName = '';
 488                 }
 489                 if ( $dumpContents ) {
 490                         $be = $file->getRepo()->getBackend();
 491                         # Dump file as base64
 492                         # Uses only XML-safe characters, so does not need escaping
 493                         # @todo Too bad this loads the contents into memory (script might swap)
 494                         $contents = '      <contents encoding="base64">' .
 495                                 chunk_split( base64_encode(
 496                                         $be->getFileContents( [ 'src' => $file->getPath() ] ) ) ) .
 497                                 "      </contents>\n";
 498                 } else {
 499                         $contents = '';
 500                 }
 501                 if ( $file->isDeleted( File::DELETED_COMMENT ) ) {
 502                         $comment = Xml::element( 'comment', [ 'deleted' => 'deleted' ] );
 503                 } else {
 504                         $comment = Xml::elementClean( 'comment', null, strval( $file->getDescription() ) );
 505                 }
 506                 return "    <upload>\n" .
 507                         $this->writeTimestamp( $file->getTimestamp() ) .
 508                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 509                         "      " . $comment . "\n" .
 510                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 511                         $archiveName .
 512                         "      " . Xml::element( 'src', null, $file->getCanonicalUrl() ) . "\n" .
 513                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 514                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 515                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 516                         $contents .
 517                         "    </upload>\n";
 518         }
 519
 520         /**
 521          * Return prefixed text form of title, but using the content language's
 522          * canonical namespace. This skips any special-casing such as gendered
 523          * user namespaces -- which while useful, are not yet listed in the
 524          * XML "<siteinfo>" data so are unsafe in export.
 525          *
 526          * @param Title $title
 527          * @return string
 528          * @since 1.18
 529          */
 530         public static function canonicalTitle( Title $title ) {
 531                 if ( $title->isExternal() ) {
 532                         return $title->getPrefixedText();
 533                 }
 534
 535                 $prefix = MediaWikiServices::getInstance()->getContentLanguage()->
 536                         getFormattedNsText( $title->getNamespace() );
 537
 538                 // @todo Emit some kind of warning to the user if $title->getNamespace() !==
 539                 // NS_MAIN and $prefix === '' (viz. pages in an unregistered namespace)
 540
 541                 if ( $prefix !== '' ) {
 542                         $prefix .= ':';
 543                 }
 544
 545                 return $prefix . $title->getText();
 546         }
 547 }