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                 foreach (
 145                         MediaWikiServices::getInstance()->getContentLanguage()->getFormattedNamespaces()
 146                         as $ns => $title
 147                 ) {
 148                         $spaces .= '      ' .
 149                                 Xml::element( 'namespace',
 150                                         [
 151                                                 'key' => $ns,
 152                                                 'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 153                                         ], $title ) . "\n";
 154                 }
 155                 $spaces .= "    </namespaces>";
 156                 return $spaces;
 157         }
 158
 159         /**
 160          * Closes the output stream with the closing root element.
 161          * Call when finished dumping things.
 162          *
 163          * @return string
 164          */
 165         function closeStream() {
 166                 return "</mediawiki>\n";
 167         }
 168
 169         /**
 170          * Opens a "<page>" section on the output stream, with data
 171          * from the given database row.
 172          *
 173          * @param object $row
 174          * @return string
 175          */
 176         public function openPage( $row ) {
 177                 $out = "  <page>\n";
 178                 $this->currentTitle = Title::makeTitle( $row->page_namespace, $row->page_title );
 179                 $canonicalTitle = self::canonicalTitle( $this->currentTitle );
 180                 $out .= '    ' . Xml::elementClean( 'title', [], $canonicalTitle ) . "\n";
 181                 $out .= '    ' . Xml::element( 'ns', [], strval( $row->page_namespace ) ) . "\n";
 182                 $out .= '    ' . Xml::element( 'id', [], strval( $row->page_id ) ) . "\n";
 183                 if ( $row->page_is_redirect ) {
 184                         $page = WikiPage::factory( $this->currentTitle );
 185                         $redirect = $page->getRedirectTarget();
 186                         if ( $redirect instanceof Title && $redirect->isValidRedirectTarget() ) {
 187                                 $out .= '    ';
 188                                 $out .= Xml::element( 'redirect', [ 'title' => self::canonicalTitle( $redirect ) ] );
 189                                 $out .= "\n";
 190                         }
 191                 }
 192
 193                 if ( $row->page_restrictions != '' ) {
 194                         $out .= '    ' . Xml::element( 'restrictions', [],
 195                                 strval( $row->page_restrictions ) ) . "\n";
 196                 }
 197
 198                 Hooks::run( 'XmlDumpWriterOpenPage', [ $this, &$out, $row, $this->currentTitle ] );
 199
 200                 return $out;
 201         }
 202
 203         /**
 204          * Closes a "<page>" section on the output stream.
 205          *
 206          * @private
 207          * @return string
 208          */
 209         function closePage() {
 210                 if ( $this->currentTitle !== null ) {
 211                         $linkCache = MediaWikiServices::getInstance()->getLinkCache();
 212                         // In rare cases, link cache has the same key for some pages which
 213                         // might be read as part of the same batch. T220424 and T220316
 214                         $linkCache->clearLink( $this->currentTitle );
 215                 }
 216                 return "  </page>\n";
 217         }
 218
 219         /**
 220          * @return RevisionStore
 221          */
 222         private function getRevisionStore() {
 223                 return MediaWikiServices::getInstance()->getRevisionStore();
 224         }
 225
 226         /**
 227          * @return SqlBlobStore
 228          */
 229         private function getBlobStore() {
 230                 return MediaWikiServices::getInstance()->getBlobStore();
 231         }
 232
 233         /**
 234          * Dumps a "<revision>" section on the output stream, with
 235          * data filled in from the given database row.
 236          *
 237          * @param object $row
 238          * @return string
 239          * @private
 240          */
 241         function writeRevision( $row ) {
 242                 $out = "    <revision>\n";
 243                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 244                 if ( isset( $row->rev_parent_id ) && $row->rev_parent_id ) {
 245                         $out .= "      " . Xml::element( 'parentid', null, strval( $row->rev_parent_id ) ) . "\n";
 246                 }
 247
 248                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 249
 250                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_USER ) ) {
 251                         $out .= "      " . Xml::element( 'contributor', [ 'deleted' => 'deleted' ] ) . "\n";
 252                 } else {
 253                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 254                 }
 255
 256                 if ( isset( $row->rev_minor_edit ) && $row->rev_minor_edit ) {
 257                         $out .= "      <minor/>\n";
 258                 }
 259                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_COMMENT ) ) {
 260                         $out .= "      " . Xml::element( 'comment', [ 'deleted' => 'deleted' ] ) . "\n";
 261                 } else {
 262                         $comment = CommentStore::getStore()->getComment( 'rev_comment', $row )->text;
 263                         if ( $comment != '' ) {
 264                                 $out .= "      " . Xml::elementClean( 'comment', [], strval( $comment ) ) . "\n";
 265                         }
 266                 }
 267
 268                 // TODO: rev_content_model no longer exists with MCR, see T174031
 269                 if ( isset( $row->rev_content_model ) && !is_null( $row->rev_content_model ) ) {
 270                         $content_model = strval( $row->rev_content_model );
 271                 } else {
 272                         // probably using $wgContentHandlerUseDB = false;
 273                         $content_model = ContentHandler::getDefaultModelFor( $this->currentTitle );
 274                 }
 275
 276                 $content_handler = ContentHandler::getForModelID( $content_model );
 277
 278                 // TODO: rev_content_format no longer exists with MCR, see T174031
 279                 if ( isset( $row->rev_content_format ) && !is_null( $row->rev_content_format ) ) {
 280                         $content_format = strval( $row->rev_content_format );
 281                 } else {
 282                         // probably using $wgContentHandlerUseDB = false;
 283                         $content_format = $content_handler->getDefaultFormat();
 284                 }
 285
 286                 $out .= "      " . Xml::element( 'model', null, strval( $content_model ) ) . "\n";
 287                 $out .= "      " . Xml::element( 'format', null, strval( $content_format ) ) . "\n";
 288
 289                 $text = '';
 290                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_TEXT ) ) {
 291                         $out .= "      " . Xml::element( 'text', [ 'deleted' => 'deleted' ] ) . "\n";
 292                 } elseif ( isset( $row->old_text ) ) {
 293                         // Raw text from the database may have invalid chars
 294                         $text = strval( Revision::getRevisionText( $row ) );
 295                         try {
 296                                 $text = $content_handler->exportTransform( $text, $content_format );
 297                         }
 298                         catch ( Exception $ex ) {
 299                                 if ( $ex instanceof MWException || $ex instanceof RuntimeException ) {
 300                                         // leave text as is; that's the way it goes
 301                                         wfLogWarning( 'exportTransform failed on text for revid ' . $row->rev_id . "\n" );
 302                                 } else {
 303                                         throw $ex;
 304                                 }
 305                         }
 306                         $out .= "      " . Xml::elementClean( 'text',
 307                                 [ 'xml:space' => 'preserve', 'bytes' => intval( $row->rev_len ) ],
 308                                 strval( $text ) ) . "\n";
 309                 } elseif ( isset( $row->_load_content ) ) {
 310                         // TODO: make this fully MCR aware, see T174031
 311                         $rev = $this->getRevisionStore()->newRevisionFromRow( $row, 0, $this->currentTitle );
 312                         $slot = $rev->getSlot( 'main' );
 313                         try {
 314                                 $content = $slot->getContent();
 315
 316                                 if ( $content instanceof TextContent ) {
 317                                         // HACK: For text based models, bypass the serialization step.
 318                                         // This allows extensions (like Flow)that use incompatible combinations
 319                                         // of serialization format and content model.
 320                                         $text = $content->getNativeData();
 321                                 } else {
 322                                         $text = $content->serialize( $content_format );
 323                                 }
 324                                 $text = $content_handler->exportTransform( $text, $content_format );
 325                                 $out .= "      " . Xml::elementClean( 'text',
 326                                         [ 'xml:space' => 'preserve', 'bytes' => intval( $slot->getSize() ) ],
 327                                         strval( $text ) ) . "\n";
 328                         }
 329                         catch ( Exception $ex ) {
 330                                 if ( $ex instanceof MWException || $ex instanceof RuntimeException ) {
 331                                         // there's no provsion in the schema for an attribute that will let
 332                                         // the user know this element was unavailable due to error; an empty
 333                                         // tag is the best we can do
 334                                         $out .= "      " . Xml::element( 'text' ) . "\n";
 335                                         wfLogWarning( 'failed to load content for revid ' . $row->rev_id . "\n" );
 336                                 } else {
 337                                         throw $ex;
 338                                 }
 339                         }
 340                 } elseif ( isset( $row->rev_text_id ) ) {
 341                         // Stub output for pre-MCR schema
 342                         // TODO: MCR: rev_text_id only exists in the pre-MCR schema. Remove this when
 343                         // we drop support for the old schema.
 344                         $out .= "      " . Xml::element( 'text',
 345                                 [ 'id' => $row->rev_text_id, 'bytes' => intval( $row->rev_len ) ],
 346                                 "" ) . "\n";
 347                 } else {
 348                         // Backwards-compatible stub output for MCR aware schema
 349                         // TODO: MCR: emit content addresses instead of text ids, see T174031, T199121
 350                         $rev = $this->getRevisionStore()->newRevisionFromRow( $row, 0, $this->currentTitle );
 351                         $slot = $rev->getSlot( 'main' );
 352
 353                         // Note that this is currently the ONLY reason we have a BlobStore here at all.
 354                         // When removing this line, check whether the BlobStore has become unused.
 355                         $textId = $this->getBlobStore()->getTextIdFromAddress( $slot->getAddress() );
 356                         $out .= "      " . Xml::element( 'text',
 357                                         [ 'id' => $textId, 'bytes' => intval( $slot->getSize() ) ],
 358                                         "" ) . "\n";
 359                 }
 360
 361                 if ( isset( $row->rev_sha1 )
 362                         && $row->rev_sha1
 363                         && !( $row->rev_deleted & Revision::DELETED_TEXT )
 364                 ) {
 365                         $out .= "      " . Xml::element( 'sha1', null, strval( $row->rev_sha1 ) ) . "\n";
 366                 } else {
 367                         $out .= "      <sha1/>\n";
 368                 }
 369
 370                 // Avoid PHP 7.1 warning from passing $this by reference
 371                 $writer = $this;
 372                 Hooks::run( 'XmlDumpWriterWriteRevision', [ &$writer, &$out, $row, $text ] );
 373
 374                 $out .= "    </revision>\n";
 375
 376                 return $out;
 377         }
 378
 379         /**
 380          * Dumps a "<logitem>" section on the output stream, with
 381          * data filled in from the given database row.
 382          *
 383          * @param object $row
 384          * @return string
 385          * @private
 386          */
 387         function writeLogItem( $row ) {
 388                 $out = "  <logitem>\n";
 389                 $out .= "    " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 390
 391                 $out .= $this->writeTimestamp( $row->log_timestamp, "    " );
 392
 393                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 394                         $out .= "    " . Xml::element( 'contributor', [ 'deleted' => 'deleted' ] ) . "\n";
 395                 } else {
 396                         $out .= $this->writeContributor( $row->log_user, $row->user_name, "    " );
 397                 }
 398
 399                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 400                         $out .= "    " . Xml::element( 'comment', [ 'deleted' => 'deleted' ] ) . "\n";
 401                 } else {
 402                         $comment = CommentStore::getStore()->getComment( 'log_comment', $row )->text;
 403                         if ( $comment != '' ) {
 404                                 $out .= "    " . Xml::elementClean( 'comment', null, strval( $comment ) ) . "\n";
 405                         }
 406                 }
 407
 408                 $out .= "    " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 409                 $out .= "    " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 410
 411                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 412                         $out .= "    " . Xml::element( 'text', [ 'deleted' => 'deleted' ] ) . "\n";
 413                 } else {
 414                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 415                         $out .= "    " . Xml::elementClean( 'logtitle', null, self::canonicalTitle( $title ) ) . "\n";
 416                         $out .= "    " . Xml::elementClean( 'params',
 417                                 [ 'xml:space' => 'preserve' ],
 418                                 strval( $row->log_params ) ) . "\n";
 419                 }
 420
 421                 $out .= "  </logitem>\n";
 422
 423                 return $out;
 424         }
 425
 426         /**
 427          * @param string $timestamp
 428          * @param string $indent Default to six spaces
 429          * @return string
 430          */
 431         function writeTimestamp( $timestamp, $indent = "      " ) {
 432                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 433                 return $indent . Xml::element( 'timestamp', null, $ts ) . "\n";
 434         }
 435
 436         /**
 437          * @param int $id
 438          * @param string $text
 439          * @param string $indent Default to six spaces
 440          * @return string
 441          */
 442         function writeContributor( $id, $text, $indent = "      " ) {
 443                 $out = $indent . "<contributor>\n";
 444                 if ( $id || !IP::isValid( $text ) ) {
 445                         $out .= $indent . "  " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 446                         $out .= $indent . "  " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 447                 } else {
 448                         $out .= $indent . "  " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 449                 }
 450                 $out .= $indent . "</contributor>\n";
 451                 return $out;
 452         }
 453
 454         /**
 455          * Warning! This data is potentially inconsistent. :(
 456          * @param object $row
 457          * @param bool $dumpContents
 458          * @return string
 459          */
 460         function writeUploads( $row, $dumpContents = false ) {
 461                 if ( $row->page_namespace == NS_FILE ) {
 462                         $img = wfLocalFile( $row->page_title );
 463                         if ( $img && $img->exists() ) {
 464                                 $out = '';
 465                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 466                                         $out .= $this->writeUpload( $ver, $dumpContents );
 467                                 }
 468                                 $out .= $this->writeUpload( $img, $dumpContents );
 469                                 return $out;
 470                         }
 471                 }
 472                 return '';
 473         }
 474
 475         /**
 476          * @param File $file
 477          * @param bool $dumpContents
 478          * @return string
 479          */
 480         function writeUpload( $file, $dumpContents = false ) {
 481                 if ( $file->isOld() ) {
 482                         $archiveName = "      " .
 483                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 484                 } else {
 485                         $archiveName = '';
 486                 }
 487                 if ( $dumpContents ) {
 488                         $be = $file->getRepo()->getBackend();
 489                         # Dump file as base64
 490                         # Uses only XML-safe characters, so does not need escaping
 491                         # @todo Too bad this loads the contents into memory (script might swap)
 492                         $contents = '      <contents encoding="base64">' .
 493                                 chunk_split( base64_encode(
 494                                         $be->getFileContents( [ 'src' => $file->getPath() ] ) ) ) .
 495                                 "      </contents>\n";
 496                 } else {
 497                         $contents = '';
 498                 }
 499                 if ( $file->isDeleted( File::DELETED_COMMENT ) ) {
 500                         $comment = Xml::element( 'comment', [ 'deleted' => 'deleted' ] );
 501                 } else {
 502                         $comment = Xml::elementClean( 'comment', null, strval( $file->getDescription() ) );
 503                 }
 504                 return "    <upload>\n" .
 505                         $this->writeTimestamp( $file->getTimestamp() ) .
 506                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 507                         "      " . $comment . "\n" .
 508                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 509                         $archiveName .
 510                         "      " . Xml::element( 'src', null, $file->getCanonicalUrl() ) . "\n" .
 511                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 512                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 513                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 514                         $contents .
 515                         "    </upload>\n";
 516         }
 517
 518         /**
 519          * Return prefixed text form of title, but using the content language's
 520          * canonical namespace. This skips any special-casing such as gendered
 521          * user namespaces -- which while useful, are not yet listed in the
 522          * XML "<siteinfo>" data so are unsafe in export.
 523          *
 524          * @param Title $title
 525          * @return string
 526          * @since 1.18
 527          */
 528         public static function canonicalTitle( Title $title ) {
 529                 if ( $title->isExternal() ) {
 530                         return $title->getPrefixedText();
 531                 }
 532
 533                 $prefix = MediaWikiServices::getInstance()->getContentLanguage()->
 534                         getFormattedNsText( $title->getNamespace() );
 535
 536                 // @todo Emit some kind of warning to the user if $title->getNamespace() !==
 537                 // NS_MAIN and $prefix === '' (viz. pages in an unregistered namespace)
 538
 539                 if ( $prefix !== '' ) {
 540                         $prefix .= ':';
 541                 }
 542
 543                 return $prefix . $title->getText();
 544         }
 545 }