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