| Index: trunk/phase3/maintenance/dumpBackup.php |
| — | — | @@ -40,11 +40,78 @@ |
| 41 | 41 | var $skipFooter = false; // don't output </mediawiki> |
| 42 | 42 | var $startId = 0; |
| 43 | 43 | var $endId = 0; |
| | 44 | + var $sink = null; // Output filters |
| 44 | 45 | |
| 45 | | - function BackupDumper() { |
| | 46 | + function BackupDumper( $args ) { |
| 46 | 47 | $this->stderr = fopen( "php://stderr", "wt" ); |
| | 48 | + $this->sink = $this->processArgs( $args ); |
| 47 | 49 | } |
| 48 | 50 | |
| | 51 | + /** |
| | 52 | + * @param array $args |
| | 53 | + * @return array |
| | 54 | + * @static |
| | 55 | + */ |
| | 56 | + function processArgs( $args ) { |
| | 57 | + $outputTypes = array( |
| | 58 | + 'file' => 'DumpFileOutput', |
| | 59 | + 'gzip' => 'DumpGZipOutput', |
| | 60 | + 'bzip2' => 'DumpBZip2Output', |
| | 61 | + '7zip' => 'Dump7ZipOutput' ); |
| | 62 | + $filterTypes = array( |
| | 63 | + 'latest' => 'DumpLatestFilter', |
| | 64 | + 'notalk' => 'DumpNotalkFilter', |
| | 65 | + 'namespace' => 'DumpNamespaceFilter' ); |
| | 66 | + $sink = null; |
| | 67 | + $sinks = array(); |
| | 68 | + foreach( $args as $arg ) { |
| | 69 | + if( preg_match( '/^--(.+?)(?:=(.+?)(?::(.+?))?)?$/', $arg, $matches ) ) { |
| | 70 | + @list( $full, $opt, $val, $param ) = $matches; |
| | 71 | + switch( $opt ) { |
| | 72 | + case "output": |
| | 73 | + if( !is_null( $sink ) ) { |
| | 74 | + $sinks[] = $sink; |
| | 75 | + } |
| | 76 | + if( !isset( $outputTypes[$val] ) ) { |
| | 77 | + die( "Unrecognized output sink type '$val'\n" ); |
| | 78 | + } |
| | 79 | + $type = $outputTypes[$val]; |
| | 80 | + $sink = new $type( $param ); |
| | 81 | + break; |
| | 82 | + case "filter": |
| | 83 | + if( is_null( $sink ) ) { |
| | 84 | + $this->progress( "Warning: assuming stdout for filter output\n" ); |
| | 85 | + $sink = new DumpOutput(); |
| | 86 | + } |
| | 87 | + if( !isset( $filterTypes[$val] ) ) { |
| | 88 | + die( "Unrecognized filter type '$val'\n" ); |
| | 89 | + } |
| | 90 | + $type = $filterTypes[$val]; |
| | 91 | + $filter = new $type( $sink, $param ); |
| | 92 | + |
| | 93 | + // references are lame in php... |
| | 94 | + unset( $sink ); |
| | 95 | + $sink = $filter; |
| | 96 | + |
| | 97 | + break; |
| | 98 | + default: |
| | 99 | + //die( "Unrecognized dump option'$opt'\n" ); |
| | 100 | + } |
| | 101 | + } |
| | 102 | + } |
| | 103 | + |
| | 104 | + if( is_null( $sink ) ) { |
| | 105 | + $sink = new DumpOutput(); |
| | 106 | + } |
| | 107 | + $sinks[] = $sink; |
| | 108 | + |
| | 109 | + if( count( $sinks ) > 1 ) { |
| | 110 | + return new DumpMultiWriter( $sinks ); |
| | 111 | + } else { |
| | 112 | + return $sink; |
| | 113 | + } |
| | 114 | + } |
| | 115 | + |
| 49 | 116 | function dump( $history ) { |
| 50 | 117 | # This shouldn't happen if on console... ;) |
| 51 | 118 | header( 'Content-type: text/html; charset=UTF-8' ); |
| — | — | @@ -61,9 +128,10 @@ |
| 62 | 129 | |
| 63 | 130 | $db =& $this->backupDb(); |
| 64 | 131 | $exporter = new WikiExporter( $db, $history, MW_EXPORT_STREAM ); |
| 65 | | - $exporter->setPageCallback( array( &$this, 'reportPage' ) ); |
| 66 | | - $exporter->setRevisionCallback( array( &$this, 'revCount' ) ); |
| 67 | 132 | |
| | 133 | + $wrapper = new ExportProgressFilter( $this->sink, $this ); |
| | 134 | + $exporter->setOutputSink( $wrapper ); |
| | 135 | + |
| 68 | 136 | if( !$this->skipHeader ) |
| 69 | 137 | $exporter->openStream(); |
| 70 | 138 | |
| — | — | @@ -100,12 +168,12 @@ |
| 101 | 169 | : $wgDBserver; |
| 102 | 170 | } |
| 103 | 171 | |
| 104 | | - function reportPage( $page ) { |
| | 172 | + function reportPage() { |
| 105 | 173 | $this->pageCount++; |
| 106 | 174 | $this->report(); |
| 107 | 175 | } |
| 108 | 176 | |
| 109 | | - function revCount( $rev ) { |
| | 177 | + function revCount() { |
| 110 | 178 | $this->revCount++; |
| 111 | 179 | } |
| 112 | 180 | |
| — | — | @@ -140,7 +208,25 @@ |
| 141 | 209 | } |
| 142 | 210 | } |
| 143 | 211 | |
| 144 | | -$dumper = new BackupDumper(); |
| | 212 | +class ExportProgressFilter extends DumpFilter { |
| | 213 | + function ExportProgressFilter( &$sink, &$progress ) { |
| | 214 | + parent::DumpFilter( $sink ); |
| | 215 | + $this->progress = $progress; |
| | 216 | + } |
| | 217 | + |
| | 218 | + function writeClosePage( $string ) { |
| | 219 | + parent::writeClosePage( $string ); |
| | 220 | + $this->progress->reportPage(); |
| | 221 | + } |
| | 222 | + |
| | 223 | + function writeRevision( $rev, $string ) { |
| | 224 | + parent::writeRevision( $rev, $string ); |
| | 225 | + $this->progress->revCount(); |
| | 226 | + } |
| | 227 | +} |
| | 228 | + |
| | 229 | +$dumper = new BackupDumper( $argv ); |
| | 230 | + |
| 145 | 231 | if( isset( $options['quiet'] ) ) { |
| 146 | 232 | $dumper->reporting = false; |
| 147 | 233 | } |
| Index: trunk/phase3/includes/Export.php |
| — | — | @@ -0,0 +1,675 @@ |
| | 2 | +<?php |
| | 3 | +# Copyright (C) 2003, 2005 Brion Vibber <brion@pobox.com> |
| | 4 | +# http://www.mediawiki.org/ |
| | 5 | +# |
| | 6 | +# This program is free software; you can redistribute it and/or modify |
| | 7 | +# it under the terms of the GNU General Public License as published by |
| | 8 | +# the Free Software Foundation; either version 2 of the License, or |
| | 9 | +# (at your option) any later version. |
| | 10 | +# |
| | 11 | +# This program is distributed in the hope that it will be useful, |
| | 12 | +# but WITHOUT ANY WARRANTY; without even the implied warranty of |
| | 13 | +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| | 14 | +# GNU General Public License for more details. |
| | 15 | +# |
| | 16 | +# You should have received a copy of the GNU General Public License along |
| | 17 | +# with this program; if not, write to the Free Software Foundation, Inc., |
| | 18 | +# 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. |
| | 19 | +# http://www.gnu.org/copyleft/gpl.html |
| | 20 | +/** |
| | 21 | + * |
| | 22 | + * @package MediaWiki |
| | 23 | + * @subpackage SpecialPage |
| | 24 | + */ |
| | 25 | + |
| | 26 | +/** */ |
| | 27 | +require_once( 'Revision.php' ); |
| | 28 | + |
| | 29 | +define( 'MW_EXPORT_FULL', 0 ); |
| | 30 | +define( 'MW_EXPORT_CURRENT', 1 ); |
| | 31 | + |
| | 32 | +define( 'MW_EXPORT_BUFFER', 0 ); |
| | 33 | +define( 'MW_EXPORT_STREAM', 1 ); |
| | 34 | + |
| | 35 | + |
| | 36 | +/** |
| | 37 | + * @package MediaWiki |
| | 38 | + * @subpackage SpecialPage |
| | 39 | + */ |
| | 40 | +class WikiExporter { |
| | 41 | + /** |
| | 42 | + * If using MW_EXPORT_STREAM to stream a large amount of data, |
| | 43 | + * provide a database connection which is not managed by |
| | 44 | + * LoadBalancer to read from: some history blob types will |
| | 45 | + * make additional queries to pull source data while the |
| | 46 | + * main query is still running. |
| | 47 | + * |
| | 48 | + * @param Database $db |
| | 49 | + * @param int $history one of MW_EXPORT_FULL or MW_EXPORT_CURRENT |
| | 50 | + * @param int $buffer one of MW_EXPORT_BUFFER or MW_EXPORT_STREAM |
| | 51 | + */ |
| | 52 | + function WikiExporter( &$db, $history = MW_EXPORT_CURRENT, |
| | 53 | + $buffer = MW_EXPORT_BUFFER ) { |
| | 54 | + $this->db =& $db; |
| | 55 | + $this->history = $history; |
| | 56 | + $this->buffer = $buffer; |
| | 57 | + $this->writer = new XmlDumpWriter(); |
| | 58 | + $this->sink = new DumpOutput(); |
| | 59 | + } |
| | 60 | + |
| | 61 | + /** |
| | 62 | + * Set the DumpOutput or DumpFilter object which will receive |
| | 63 | + * various row objects and XML output for filtering. Filters |
| | 64 | + * can be chained or used as callbacks. |
| | 65 | + * |
| | 66 | + * @param mixed $callback |
| | 67 | + */ |
| | 68 | + function setOutputSink( &$sink ) { |
| | 69 | + $this->sink =& $sink; |
| | 70 | + } |
| | 71 | + |
| | 72 | + function openStream() { |
| | 73 | + $output = $this->writer->openStream(); |
| | 74 | + $this->sink->writeOpenStream( $output ); |
| | 75 | + } |
| | 76 | + |
| | 77 | + function closeStream() { |
| | 78 | + $output = $this->writer->closeStream(); |
| | 79 | + $this->sink->writeCloseStream( $output ); |
| | 80 | + } |
| | 81 | + |
| | 82 | + /** |
| | 83 | + * Dumps a series of page and revision records for all pages |
| | 84 | + * in the database, either including complete history or only |
| | 85 | + * the most recent version. |
| | 86 | + */ |
| | 87 | + function allPages() { |
| | 88 | + return $this->dumpFrom( '' ); |
| | 89 | + } |
| | 90 | + |
| | 91 | + /** |
| | 92 | + * Dumps a series of page and revision records for those pages |
| | 93 | + * in the database falling within the page_id range given. |
| | 94 | + * @param int $start Inclusive lower limit (this id is included) |
| | 95 | + * @param int $end Exclusive upper limit (this id is not included) |
| | 96 | + * If 0, no upper limit. |
| | 97 | + */ |
| | 98 | + function pagesByRange( $start, $end ) { |
| | 99 | + $condition = 'page_id >= ' . intval( $start ); |
| | 100 | + if( $end ) { |
| | 101 | + $condition .= ' AND page_id < ' . intval( $end ); |
| | 102 | + } |
| | 103 | + return $this->dumpFrom( $condition ); |
| | 104 | + } |
| | 105 | + |
| | 106 | + /** |
| | 107 | + * @param Title $title |
| | 108 | + */ |
| | 109 | + function pageByTitle( $title ) { |
| | 110 | + return $this->dumpFrom( |
| | 111 | + 'page_namespace=' . $title->getNamespace() . |
| | 112 | + ' AND page_title=' . $this->db->addQuotes( $title->getDbKey() ) ); |
| | 113 | + } |
| | 114 | + |
| | 115 | + function pageByName( $name ) { |
| | 116 | + $title = Title::newFromText( $name ); |
| | 117 | + if( is_null( $title ) ) { |
| | 118 | + return new WikiError( "Can't export invalid title" ); |
| | 119 | + } else { |
| | 120 | + return $this->pageByTitle( $title ); |
| | 121 | + } |
| | 122 | + } |
| | 123 | + |
| | 124 | + function pagesByName( $names ) { |
| | 125 | + foreach( $names as $name ) { |
| | 126 | + $this->pageByName( $name ); |
| | 127 | + } |
| | 128 | + } |
| | 129 | + |
| | 130 | + |
| | 131 | + // -------------------- private implementation below -------------------- |
| | 132 | + |
| | 133 | + function dumpFrom( $cond = '' ) { |
| | 134 | + $fname = 'WikiExporter::dumpFrom'; |
| | 135 | + wfProfileIn( $fname ); |
| | 136 | + |
| | 137 | + $page = $this->db->tableName( 'page' ); |
| | 138 | + $revision = $this->db->tableName( 'revision' ); |
| | 139 | + $text = $this->db->tableName( 'text' ); |
| | 140 | + |
| | 141 | + if( $this->history == MW_EXPORT_FULL ) { |
| | 142 | + $join = 'page_id=rev_page'; |
| | 143 | + } elseif( $this->history == MW_EXPORT_CURRENT ) { |
| | 144 | + $join = 'page_id=rev_page AND page_latest=rev_id'; |
| | 145 | + } else { |
| | 146 | + wfProfileOut( $fname ); |
| | 147 | + return new WikiError( "$fname given invalid history dump type." ); |
| | 148 | + } |
| | 149 | + $where = ( $cond == '' ) ? '' : "$cond AND"; |
| | 150 | + |
| | 151 | + if( $this->buffer == MW_EXPORT_STREAM ) { |
| | 152 | + $prev = $this->db->bufferResults( false ); |
| | 153 | + } |
| | 154 | + if( $cond == '' ) { |
| | 155 | + // Optimization hack for full-database dump |
| | 156 | + $pageindex = 'FORCE INDEX (PRIMARY)'; |
| | 157 | + $revindex = 'FORCE INDEX(page_timestamp)'; |
| | 158 | + } else { |
| | 159 | + $pageindex = ''; |
| | 160 | + $revindex = ''; |
| | 161 | + } |
| | 162 | + $result = $this->db->query( |
| | 163 | + "SELECT * FROM |
| | 164 | + $page $pageindex, |
| | 165 | + $revision $revindex, |
| | 166 | + $text |
| | 167 | + WHERE $where $join AND rev_text_id=old_id |
| | 168 | + ORDER BY page_id", $fname ); |
| | 169 | + $wrapper = $this->db->resultObject( $result ); |
| | 170 | + $this->outputStream( $wrapper ); |
| | 171 | + |
| | 172 | + if( $this->buffer == MW_EXPORT_STREAM ) { |
| | 173 | + $this->db->bufferResults( $prev ); |
| | 174 | + } |
| | 175 | + |
| | 176 | + wfProfileOut( $fname ); |
| | 177 | + } |
| | 178 | + |
| | 179 | + /** |
| | 180 | + * Runs through a query result set dumping page and revision records. |
| | 181 | + * The result set should be sorted/grouped by page to avoid duplicate |
| | 182 | + * page records in the output. |
| | 183 | + * |
| | 184 | + * The result set will be freed once complete. Should be safe for |
| | 185 | + * streaming (non-buffered) queries, as long as it was made on a |
| | 186 | + * separate database connection not managed by LoadBalancer; some |
| | 187 | + * blob storage types will make queries to pull source data. |
| | 188 | + * |
| | 189 | + * @param ResultWrapper $resultset |
| | 190 | + * @access private |
| | 191 | + */ |
| | 192 | + function outputStream( $resultset ) { |
| | 193 | + $last = null; |
| | 194 | + while( $row = $resultset->fetchObject() ) { |
| | 195 | + if( is_null( $last ) || |
| | 196 | + $last->page_namespace != $row->page_namespace || |
| | 197 | + $last->page_title != $row->page_title ) { |
| | 198 | + if( isset( $last ) ) { |
| | 199 | + $output = $this->writer->closePage(); |
| | 200 | + $this->sink->writeClosePage( $output ); |
| | 201 | + } |
| | 202 | + $output = $this->writer->openPage( $row ); |
| | 203 | + $this->sink->writeOpenPage( $row, $output ); |
| | 204 | + $last = $row; |
| | 205 | + } |
| | 206 | + $output = $this->writer->writeRevision( $row ); |
| | 207 | + $this->sink->writeRevision( $row, $output ); |
| | 208 | + } |
| | 209 | + if( isset( $last ) ) { |
| | 210 | + $output = $this->writer->closePage(); |
| | 211 | + $this->sink->writeClosePage( $output ); |
| | 212 | + } |
| | 213 | + $resultset->free(); |
| | 214 | + } |
| | 215 | +} |
| | 216 | + |
| | 217 | +class XmlDumpWriter { |
| | 218 | + |
| | 219 | + /** |
| | 220 | + * Returns the export schema version. |
| | 221 | + * @return string |
| | 222 | + */ |
| | 223 | + function schemaVersion() { |
| | 224 | + return "0.3"; |
| | 225 | + } |
| | 226 | + |
| | 227 | + /** |
| | 228 | + * Opens the XML output stream's root <mediawiki> element. |
| | 229 | + * This does not include an xml directive, so is safe to include |
| | 230 | + * as a subelement in a larger XML stream. Namespace and XML Schema |
| | 231 | + * references are included. |
| | 232 | + * |
| | 233 | + * Output will be encoded in UTF-8. |
| | 234 | + * |
| | 235 | + * @return string |
| | 236 | + */ |
| | 237 | + function openStream() { |
| | 238 | + global $wgContLanguageCode; |
| | 239 | + $ver = $this->schemaVersion(); |
| | 240 | + return wfElement( 'mediawiki', array( |
| | 241 | + 'xmlns' => "http://www.mediawiki.org/xml/export-$ver/", |
| | 242 | + 'xmlns:xsi' => "http://www.w3.org/2001/XMLSchema-instance", |
| | 243 | + 'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " . |
| | 244 | + "http://www.mediawiki.org/xml/export-$ver.xsd", |
| | 245 | + 'version' => $ver, |
| | 246 | + 'xml:lang' => $wgContLanguageCode ), |
| | 247 | + null ) . |
| | 248 | + "\n" . |
| | 249 | + $this->siteInfo(); |
| | 250 | + } |
| | 251 | + |
| | 252 | + function siteInfo() { |
| | 253 | + $info = array( |
| | 254 | + $this->sitename(), |
| | 255 | + $this->homelink(), |
| | 256 | + $this->generator(), |
| | 257 | + $this->caseSetting(), |
| | 258 | + $this->namespaces() ); |
| | 259 | + return " <siteinfo>\n " . |
| | 260 | + implode( "\n ", $info ) . |
| | 261 | + "\n </siteinfo>\n"; |
| | 262 | + } |
| | 263 | + |
| | 264 | + function sitename() { |
| | 265 | + global $wgSitename; |
| | 266 | + return wfElement( 'sitename', array(), $wgSitename ); |
| | 267 | + } |
| | 268 | + |
| | 269 | + function generator() { |
| | 270 | + global $wgVersion; |
| | 271 | + return wfElement( 'generator', array(), "MediaWiki $wgVersion" ); |
| | 272 | + } |
| | 273 | + |
| | 274 | + function homelink() { |
| | 275 | + $page = Title::newFromText( wfMsgForContent( 'mainpage' ) ); |
| | 276 | + return wfElement( 'base', array(), $page->getFullUrl() ); |
| | 277 | + } |
| | 278 | + |
| | 279 | + function caseSetting() { |
| | 280 | + global $wgCapitalLinks; |
| | 281 | + // "case-insensitive" option is reserved for future |
| | 282 | + $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive'; |
| | 283 | + return wfElement( 'case', array(), $sensitivity ); |
| | 284 | + } |
| | 285 | + |
| | 286 | + function namespaces() { |
| | 287 | + global $wgContLang; |
| | 288 | + $spaces = " <namespaces>\n"; |
| | 289 | + foreach( $wgContLang->getFormattedNamespaces() as $ns => $title ) { |
| | 290 | + $spaces .= ' ' . wfElement( 'namespace', array( 'key' => $ns ), $title ) . "\n"; |
| | 291 | + } |
| | 292 | + $spaces .= " </namespaces>"; |
| | 293 | + return $spaces; |
| | 294 | + } |
| | 295 | + |
| | 296 | + /** |
| | 297 | + * Closes the output stream with the closing root element. |
| | 298 | + * Call when finished dumping things. |
| | 299 | + */ |
| | 300 | + function closeStream() { |
| | 301 | + return "</mediawiki>\n"; |
| | 302 | + } |
| | 303 | + |
| | 304 | + |
| | 305 | + /** |
| | 306 | + * Opens a <page> section on the output stream, with data |
| | 307 | + * from the given database row. |
| | 308 | + * |
| | 309 | + * @param object $row |
| | 310 | + * @return string |
| | 311 | + * @access private |
| | 312 | + */ |
| | 313 | + function openPage( $row ) { |
| | 314 | + $out = " <page>\n"; |
| | 315 | + $title = Title::makeTitle( $row->page_namespace, $row->page_title ); |
| | 316 | + $out .= ' ' . wfElementClean( 'title', array(), $title->getPrefixedText() ) . "\n"; |
| | 317 | + $out .= ' ' . wfElement( 'id', array(), strval( $row->page_id ) ) . "\n"; |
| | 318 | + if( '' != $row->page_restrictions ) { |
| | 319 | + $out .= ' ' . wfElement( 'restrictions', array(), |
| | 320 | + strval( $row->page_restrictions ) ) . "\n"; |
| | 321 | + } |
| | 322 | + return $out; |
| | 323 | + } |
| | 324 | + |
| | 325 | + /** |
| | 326 | + * Closes a <page> section on the output stream. |
| | 327 | + * |
| | 328 | + * @access private |
| | 329 | + */ |
| | 330 | + function closePage() { |
| | 331 | + return " </page>\n"; |
| | 332 | + } |
| | 333 | + |
| | 334 | + /** |
| | 335 | + * Dumps a <revision> section on the output stream, with |
| | 336 | + * data filled in from the given database row. |
| | 337 | + * |
| | 338 | + * @param object $row |
| | 339 | + * @return string |
| | 340 | + * @access private |
| | 341 | + */ |
| | 342 | + function writeRevision( $row ) { |
| | 343 | + $fname = 'WikiExporter::dumpRev'; |
| | 344 | + wfProfileIn( $fname ); |
| | 345 | + |
| | 346 | + $out = " <revision>\n"; |
| | 347 | + $out .= " " . wfElement( 'id', null, strval( $row->rev_id ) ) . "\n"; |
| | 348 | + |
| | 349 | + $ts = wfTimestamp2ISO8601( strval( $row->rev_timestamp ) ); |
| | 350 | + $out .= " " . wfElement( 'timestamp', null, $ts ) . "\n"; |
| | 351 | + |
| | 352 | + $out .= " <contributor>\n"; |
| | 353 | + if( $row->rev_user ) { |
| | 354 | + $out .= " " . wfElementClean( 'username', null, strval( $row->rev_user_text ) ) . "\n"; |
| | 355 | + $out .= " " . wfElement( 'id', null, strval( $row->rev_user ) ) . "\n"; |
| | 356 | + } else { |
| | 357 | + $out .= " " . wfElementClean( 'ip', null, strval( $row->rev_user_text ) ) . "\n"; |
| | 358 | + } |
| | 359 | + $out .= " </contributor>\n"; |
| | 360 | + |
| | 361 | + if( $row->rev_minor_edit ) { |
| | 362 | + $out .= " <minor/>\n"; |
| | 363 | + } |
| | 364 | + if( $row->rev_comment != '' ) { |
| | 365 | + $out .= " " . wfElementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n"; |
| | 366 | + } |
| | 367 | + |
| | 368 | + $text = strval( Revision::getRevisionText( $row ) ); |
| | 369 | + $out .= " " . wfElementClean( 'text', |
| | 370 | + array( 'xml:space' => 'preserve' ), |
| | 371 | + strval( $text ) ) . "\n"; |
| | 372 | + |
| | 373 | + $out .= " </revision>\n"; |
| | 374 | + |
| | 375 | + wfProfileOut( $fname ); |
| | 376 | + return $out; |
| | 377 | + } |
| | 378 | + |
| | 379 | +} |
| | 380 | + |
| | 381 | + |
| | 382 | +/** |
| | 383 | + * Base class for output stream; prints to stdout or buffer or whereever. |
| | 384 | + */ |
| | 385 | +class DumpOutput { |
| | 386 | + function writeOpenStream( $string ) { |
| | 387 | + $this->write( $string ); |
| | 388 | + } |
| | 389 | + |
| | 390 | + function writeCloseStream( $string ) { |
| | 391 | + $this->write( $string ); |
| | 392 | + } |
| | 393 | + |
| | 394 | + function writeOpenPage( $page, $string ) { |
| | 395 | + $this->write( $string ); |
| | 396 | + } |
| | 397 | + |
| | 398 | + function writeClosePage( $string ) { |
| | 399 | + $this->write( $string ); |
| | 400 | + } |
| | 401 | + |
| | 402 | + function writeRevision( $rev, $string ) { |
| | 403 | + $this->write( $string ); |
| | 404 | + } |
| | 405 | + |
| | 406 | + /** |
| | 407 | + * Override to write to a different stream type. |
| | 408 | + * @return bool |
| | 409 | + */ |
| | 410 | + function write( $string ) { |
| | 411 | + print $string; |
| | 412 | + } |
| | 413 | +} |
| | 414 | + |
| | 415 | +/** |
| | 416 | + * Stream outputter to send data to a file. |
| | 417 | + */ |
| | 418 | +class DumpFileOutput extends DumpOutput { |
| | 419 | + var $handle; |
| | 420 | + |
| | 421 | + function DumpFileOutput( $file ) { |
| | 422 | + $this->handle = fopen( $file, "wt" ); |
| | 423 | + } |
| | 424 | + |
| | 425 | + function write( $string ) { |
| | 426 | + fputs( $this->handle, $string ); |
| | 427 | + } |
| | 428 | +} |
| | 429 | + |
| | 430 | +/** |
| | 431 | + * Stream outputter to send data to a file via some filter program. |
| | 432 | + * Even if compression is available in a library, using a separate |
| | 433 | + * program can allow us to make use of a multi-processor system. |
| | 434 | + */ |
| | 435 | +class DumpPipeOutput extends DumpFileOutput { |
| | 436 | + function DumpPipeOutput( $command, $file = null ) { |
| | 437 | + if( !is_null( $file ) ) { |
| | 438 | + $command .= " > " . wfEscapeShellArg( $file ); |
| | 439 | + } |
| | 440 | + $this->handle = popen( $command, "w" ); |
| | 441 | + } |
| | 442 | +} |
| | 443 | + |
| | 444 | +/** |
| | 445 | + * Sends dump output via the gzip compressor. |
| | 446 | + */ |
| | 447 | +class DumpGZipOutput extends DumpPipeOutput { |
| | 448 | + function DumpGZipOutput( $file ) { |
| | 449 | + parent::DumpPipeOutput( "gzip", $file ); |
| | 450 | + } |
| | 451 | +} |
| | 452 | + |
| | 453 | +/** |
| | 454 | + * Sends dump output via the bgzip2 compressor. |
| | 455 | + */ |
| | 456 | +class DumpBZip2Output extends DumpPipeOutput { |
| | 457 | + function DumpBZip2Output( $file ) { |
| | 458 | + parent::DumpPipeOutput( "bzip2", $file ); |
| | 459 | + } |
| | 460 | +} |
| | 461 | + |
| | 462 | +/** |
| | 463 | + * Sends dump output via the p7zip compressor. |
| | 464 | + */ |
| | 465 | +class Dump7ZipOutput extends DumpPipeOutput { |
| | 466 | + function Dump7ZipOutput( $file ) { |
| | 467 | + $command = "7za a -si " . wfEscapeShellArg( $file ); |
| | 468 | + parent::DumpPipeOutput( $command ); |
| | 469 | + } |
| | 470 | +} |
| | 471 | + |
| | 472 | + |
| | 473 | + |
| | 474 | +/** |
| | 475 | + * Dump output filter class. |
| | 476 | + * This just does output filtering and streaming; XML formatting is done |
| | 477 | + * higher up, so be careful in what you do. |
| | 478 | + */ |
| | 479 | +class DumpFilter { |
| | 480 | + function DumpFilter( &$sink ) { |
| | 481 | + $this->sink =& $sink; |
| | 482 | + } |
| | 483 | + |
| | 484 | + function writeOpenStream( $string ) { |
| | 485 | + $this->sink->writeOpenStream( $string ); |
| | 486 | + } |
| | 487 | + |
| | 488 | + function writeCloseStream( $string ) { |
| | 489 | + $this->sink->writeCloseStream( $string ); |
| | 490 | + } |
| | 491 | + |
| | 492 | + function writeOpenPage( $page, $string ) { |
| | 493 | + $this->sendingThisPage = $this->pass( $page, $string ); |
| | 494 | + if( $this->sendingThisPage ) { |
| | 495 | + $this->sink->writeOpenPage( $page, $string ); |
| | 496 | + } |
| | 497 | + } |
| | 498 | + |
| | 499 | + function writeClosePage( $string ) { |
| | 500 | + if( $this->sendingThisPage ) { |
| | 501 | + $this->sink->writeClosePage( $string ); |
| | 502 | + $this->sendingThisPage = false; |
| | 503 | + } |
| | 504 | + } |
| | 505 | + |
| | 506 | + function writeRevision( $rev, $string ) { |
| | 507 | + if( $this->sendingThisPage ) { |
| | 508 | + $this->sink->writeRevision( $rev, $string ); |
| | 509 | + } |
| | 510 | + } |
| | 511 | + |
| | 512 | + /** |
| | 513 | + * Override for page-based filter types. |
| | 514 | + * @return bool |
| | 515 | + */ |
| | 516 | + function pass( $page, $string ) { |
| | 517 | + return true; |
| | 518 | + } |
| | 519 | +} |
| | 520 | + |
| | 521 | +/** |
| | 522 | + * Simple dump output filter to exclude all talk pages. |
| | 523 | + */ |
| | 524 | +class DumpNotalkFilter extends DumpFilter { |
| | 525 | + function pass( $page ) { |
| | 526 | + return Namespace::isTalk( $page->page_namespace ); |
| | 527 | + } |
| | 528 | +} |
| | 529 | + |
| | 530 | +/** |
| | 531 | + * Dump output filter to include or exclude pages in a given set of namespaces. |
| | 532 | + */ |
| | 533 | +class DumpNamespaceFilter extends DumpFilter { |
| | 534 | + var $invert = false; |
| | 535 | + var $match = array(); |
| | 536 | + |
| | 537 | + function DumpNamespaceFilter( &$sink, $param ) { |
| | 538 | + parent::DumpFilter( $sink ); |
| | 539 | + |
| | 540 | + $constants = array( |
| | 541 | + "NS_MAIN" => NS_MAIN, |
| | 542 | + "NS_TALK" => NS_TALK, |
| | 543 | + "NS_USER" => NS_USER, |
| | 544 | + "NS_USER_TALK" => NS_USER_TALK, |
| | 545 | + "NS_PROJECT" => NS_PROJECT, |
| | 546 | + "NS_PROJECT_TALK" => NS_PROJECT_TALK, |
| | 547 | + "NS_IMAGE" => NS_IMAGE, |
| | 548 | + "NS_IMAGE_TALK" => NS_IMAGE_TALK, |
| | 549 | + "NS_MEDIAWIKI" => NS_MEDIAWIKI, |
| | 550 | + "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK, |
| | 551 | + "NS_TEMPLATE" => NS_TEMPLATE, |
| | 552 | + "NS_TEMPLATE_TALK" => NS_TEMPLATE_TALK, |
| | 553 | + "NS_HELP" => NS_HELP, |
| | 554 | + "NS_HELP_TALK" => NS_HELP_TALK, |
| | 555 | + "NS_CATEGORY" => NS_CATEGORY, |
| | 556 | + "NS_CATEGORY_TALK" => NS_CATEGORY_TALK ); |
| | 557 | + |
| | 558 | + if( $param{0} == '!' ) { |
| | 559 | + $this->invert = true; |
| | 560 | + $param = substr( $param, 1 ); |
| | 561 | + } |
| | 562 | + |
| | 563 | + foreach( explode( ',', $param ) as $key ) { |
| | 564 | + $key = trim( $key ); |
| | 565 | + if( isset( $contants[$key] ) ) { |
| | 566 | + $ns = $constants[$key]; |
| | 567 | + $this->namespaces[$ns] = true; |
| | 568 | + } elseif( is_numeric( $key ) ) { |
| | 569 | + $ns = intval( $key ); |
| | 570 | + $this->namespaces[$ns] = true; |
| | 571 | + } |
| | 572 | + } |
| | 573 | + } |
| | 574 | + |
| | 575 | + function pass( $page ) { |
| | 576 | + $match = isset( $this->namespaces[$page->page_namespace] ); |
| | 577 | + return $this->invert xor $match; |
| | 578 | + } |
| | 579 | +} |
| | 580 | + |
| | 581 | + |
| | 582 | +/** |
| | 583 | + * Dump output filter to include only the last revision in each page sequence. |
| | 584 | + */ |
| | 585 | +class DumpLatestFilter extends DumpFilter { |
| | 586 | + var $page, $pageString, $rev, $revString; |
| | 587 | + |
| | 588 | + function writeOpenPage( $page, $string ) { |
| | 589 | + $this->page = $page; |
| | 590 | + $this->pageString = $string; |
| | 591 | + } |
| | 592 | + |
| | 593 | + function writeClosePage( $string ) { |
| | 594 | + if( $this->rev ) { |
| | 595 | + $this->sink->writeOpenPage( $this->page, $this->pageString ); |
| | 596 | + $this->sink->writeRevision( $this->rev, $this->revString ); |
| | 597 | + $this->sink->writeClosePage( $string ); |
| | 598 | + } |
| | 599 | + $this->rev = null; |
| | 600 | + $this->revString = null; |
| | 601 | + $this->page = null; |
| | 602 | + $this->pageString = null; |
| | 603 | + } |
| | 604 | + |
| | 605 | + function writeRevision( $rev, $string ) { |
| | 606 | + if( $rev->rev_id == $this->page->page_latest ) { |
| | 607 | + $this->rev = $rev; |
| | 608 | + $this->revString = $string; |
| | 609 | + } |
| | 610 | + } |
| | 611 | +} |
| | 612 | + |
| | 613 | +/** |
| | 614 | + * Base class for output stream; prints to stdout or buffer or whereever. |
| | 615 | + */ |
| | 616 | +class DumpMultiWriter { |
| | 617 | + function DumpMultiWriter( $sinks ) { |
| | 618 | + $this->sinks = $sinks; |
| | 619 | + $this->count = count( $sinks ); |
| | 620 | + } |
| | 621 | + |
| | 622 | + function writeOpenStream( $string ) { |
| | 623 | + for( $i = 0; $i < $this->count; $i++ ) { |
| | 624 | + $this->sinks[$i]->writeOpenStream( $string ); |
| | 625 | + } |
| | 626 | + } |
| | 627 | + |
| | 628 | + function writeCloseStream( $string ) { |
| | 629 | + for( $i = 0; $i < $this->count; $i++ ) { |
| | 630 | + $this->sinks[$i]->writeCloseStream( $string ); |
| | 631 | + } |
| | 632 | + } |
| | 633 | + |
| | 634 | + function writeOpenPage( $page, $string ) { |
| | 635 | + for( $i = 0; $i < $this->count; $i++ ) { |
| | 636 | + $this->sinks[$i]->writeOpenPage( $page, $string ); |
| | 637 | + } |
| | 638 | + } |
| | 639 | + |
| | 640 | + function writeClosePage( $string ) { |
| | 641 | + for( $i = 0; $i < $this->count; $i++ ) { |
| | 642 | + $this->sinks[$i]->writeClosePage( $string ); |
| | 643 | + } |
| | 644 | + } |
| | 645 | + |
| | 646 | + function writeRevision( $rev, $string ) { |
| | 647 | + for( $i = 0; $i < $this->count; $i++ ) { |
| | 648 | + $this->sinks[$i]->writeRevision( $rev, $string ); |
| | 649 | + } |
| | 650 | + } |
| | 651 | +} |
| | 652 | + |
| | 653 | + |
| | 654 | + |
| | 655 | +function wfTimestamp2ISO8601( $ts ) { |
| | 656 | + #2003-08-05T18:30:02Z |
| | 657 | + return preg_replace( '/^(....)(..)(..)(..)(..)(..)$/', '$1-$2-$3T$4:$5:$6Z', wfTimestamp( TS_MW, $ts ) ); |
| | 658 | +} |
| | 659 | + |
| | 660 | +function xmlsafe( $string ) { |
| | 661 | + $fname = 'xmlsafe'; |
| | 662 | + wfProfileIn( $fname ); |
| | 663 | + |
| | 664 | + /** |
| | 665 | + * The page may contain old data which has not been properly normalized. |
| | 666 | + * Invalid UTF-8 sequences or forbidden control characters will make our |
| | 667 | + * XML output invalid, so be sure to strip them out. |
| | 668 | + */ |
| | 669 | + $string = UtfNormal::cleanUp( $string ); |
| | 670 | + |
| | 671 | + $string = htmlspecialchars( $string ); |
| | 672 | + wfProfileOut( $fname ); |
| | 673 | + return $string; |
| | 674 | +} |
| | 675 | + |
| | 676 | +?> |
| Property changes on: trunk/phase3/includes/Export.php |
| ___________________________________________________________________ |
| Added: svn:eol-style |
| 1 | 677 | + native |
| Added: svn:keywords |
| 2 | 678 | + Author Date Id Revision |
| Index: trunk/phase3/includes/SpecialExport.php |
| — | — | @@ -24,6 +24,7 @@ |
| 25 | 25 | |
| 26 | 26 | /** */ |
| 27 | 27 | require_once( 'Revision.php' ); |
| | 28 | +require_once( 'Export.php' ); |
| 28 | 29 | |
| 29 | 30 | /** |
| 30 | 31 | * |
| — | — | @@ -67,378 +68,4 @@ |
| 68 | 69 | " ); |
| 69 | 70 | } |
| 70 | 71 | |
| 71 | | -define( 'MW_EXPORT_FULL', 0 ); |
| 72 | | -define( 'MW_EXPORT_CURRENT', 1 ); |
| 73 | | - |
| 74 | | -define( 'MW_EXPORT_BUFFER', 0 ); |
| 75 | | -define( 'MW_EXPORT_STREAM', 1 ); |
| 76 | | - |
| 77 | | -/** |
| 78 | | - * @package MediaWiki |
| 79 | | - * @subpackage SpecialPage |
| 80 | | - */ |
| 81 | | -class WikiExporter { |
| 82 | | - var $pageCallback = null; |
| 83 | | - var $revCallback = null; |
| 84 | | - |
| 85 | | - /** |
| 86 | | - * If using MW_EXPORT_STREAM to stream a large amount of data, |
| 87 | | - * provide a database connection which is not managed by |
| 88 | | - * LoadBalancer to read from: some history blob types will |
| 89 | | - * make additional queries to pull source data while the |
| 90 | | - * main query is still running. |
| 91 | | - * |
| 92 | | - * @param Database $db |
| 93 | | - * @param int $history one of MW_EXPORT_FULL or MW_EXPORT_CURRENT |
| 94 | | - * @param int $buffer one of MW_EXPORT_BUFFER or MW_EXPORT_STREAM |
| 95 | | - */ |
| 96 | | - function WikiExporter( &$db, $history = MW_EXPORT_CURRENT, |
| 97 | | - $buffer = MW_EXPORT_BUFFER ) { |
| 98 | | - $this->db =& $db; |
| 99 | | - $this->history = $history; |
| 100 | | - $this->buffer = $buffer; |
| 101 | | - } |
| 102 | | - |
| 103 | | - /** |
| 104 | | - * Set a callback to be called after each page in the output |
| 105 | | - * stream is closed. The callback will be passed a database row |
| 106 | | - * object with the last revision output. |
| 107 | | - * |
| 108 | | - * A set callback can be removed by passing null here. |
| 109 | | - * |
| 110 | | - * @param mixed $callback |
| 111 | | - */ |
| 112 | | - function setPageCallback( $callback ) { |
| 113 | | - $this->pageCallback = $callback; |
| 114 | | - } |
| 115 | | - |
| 116 | | - /** |
| 117 | | - * Set a callback to be called after each revision in the output |
| 118 | | - * stream is closed. The callback will be passed a database row |
| 119 | | - * object with the revision data. |
| 120 | | - * |
| 121 | | - * A set callback can be removed by passing null here. |
| 122 | | - * |
| 123 | | - * @param mixed $callback |
| 124 | | - */ |
| 125 | | - function setRevisionCallback( $callback ) { |
| 126 | | - $this->revCallback = $callback; |
| 127 | | - } |
| 128 | | - |
| 129 | | - /** |
| 130 | | - * Returns the export schema version. |
| 131 | | - * @return string |
| 132 | | - */ |
| 133 | | - function schemaVersion() { |
| 134 | | - return "0.3"; |
| 135 | | - } |
| 136 | | - |
| 137 | | - /** |
| 138 | | - * Opens the XML output stream's root <mediawiki> element. |
| 139 | | - * This does not include an xml directive, so is safe to include |
| 140 | | - * as a subelement in a larger XML stream. Namespace and XML Schema |
| 141 | | - * references are included. |
| 142 | | - * |
| 143 | | - * To capture the stream to a string, use PHP's output buffering |
| 144 | | - * functions. Output will be encoded in UTF-8. |
| 145 | | - */ |
| 146 | | - function openStream() { |
| 147 | | - global $wgContLanguageCode; |
| 148 | | - $ver = $this->schemaVersion(); |
| 149 | | - print wfElement( 'mediawiki', array( |
| 150 | | - 'xmlns' => "http://www.mediawiki.org/xml/export-$ver/", |
| 151 | | - 'xmlns:xsi' => "http://www.w3.org/2001/XMLSchema-instance", |
| 152 | | - 'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " . |
| 153 | | - "http://www.mediawiki.org/xml/export-$ver.xsd", |
| 154 | | - 'version' => $ver, |
| 155 | | - 'xml:lang' => $wgContLanguageCode ), |
| 156 | | - null ) . "\n"; |
| 157 | | - $this->siteInfo(); |
| 158 | | - } |
| 159 | | - |
| 160 | | - function siteInfo() { |
| 161 | | - $info = array( |
| 162 | | - $this->sitename(), |
| 163 | | - $this->homelink(), |
| 164 | | - $this->generator(), |
| 165 | | - $this->caseSetting(), |
| 166 | | - $this->namespaces() ); |
| 167 | | - print "<siteinfo>\n"; |
| 168 | | - foreach( $info as $item ) { |
| 169 | | - print " $item\n"; |
| 170 | | - } |
| 171 | | - print "</siteinfo>\n"; |
| 172 | | - } |
| 173 | | - |
| 174 | | - function sitename() { |
| 175 | | - global $wgSitename; |
| 176 | | - return wfElement( 'sitename', array(), $wgSitename ); |
| 177 | | - } |
| 178 | | - |
| 179 | | - function generator() { |
| 180 | | - global $wgVersion; |
| 181 | | - return wfElement( 'generator', array(), "MediaWiki $wgVersion" ); |
| 182 | | - } |
| 183 | | - |
| 184 | | - function homelink() { |
| 185 | | - $page = Title::newFromText( wfMsgForContent( 'mainpage' ) ); |
| 186 | | - return wfElement( 'base', array(), $page->getFullUrl() ); |
| 187 | | - } |
| 188 | | - |
| 189 | | - function caseSetting() { |
| 190 | | - global $wgCapitalLinks; |
| 191 | | - // "case-insensitive" option is reserved for future |
| 192 | | - $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive'; |
| 193 | | - return wfElement( 'case', array(), $sensitivity ); |
| 194 | | - } |
| 195 | | - |
| 196 | | - function namespaces() { |
| 197 | | - global $wgContLang; |
| 198 | | - $spaces = "<namespaces>\n"; |
| 199 | | - foreach( $wgContLang->getFormattedNamespaces() as $ns => $title ) { |
| 200 | | - $spaces .= ' ' . wfElement( 'namespace', array( 'key' => $ns ), $title ) . "\n"; |
| 201 | | - } |
| 202 | | - $spaces .= " </namespaces>"; |
| 203 | | - return $spaces; |
| 204 | | - } |
| 205 | | - |
| 206 | | - /** |
| 207 | | - * Closes the output stream with the closing root element. |
| 208 | | - * Call when finished dumping things. |
| 209 | | - */ |
| 210 | | - function closeStream() { |
| 211 | | - print "</mediawiki>\n"; |
| 212 | | - } |
| 213 | | - |
| 214 | | - /** |
| 215 | | - * Dumps a series of page and revision records for all pages |
| 216 | | - * in the database, either including complete history or only |
| 217 | | - * the most recent version. |
| 218 | | - */ |
| 219 | | - function allPages() { |
| 220 | | - return $this->dumpFrom( '' ); |
| 221 | | - } |
| 222 | | - |
| 223 | | - /** |
| 224 | | - * Dumps a series of page and revision records for those pages |
| 225 | | - * in the database falling within the page_id range given. |
| 226 | | - * @param int $start Inclusive lower limit (this id is included) |
| 227 | | - * @param int $end Exclusive upper limit (this id is not included) |
| 228 | | - * If 0, no upper limit. |
| 229 | | - */ |
| 230 | | - function pagesByRange( $start, $end ) { |
| 231 | | - $condition = 'page_id >= ' . intval( $start ); |
| 232 | | - if( $end ) { |
| 233 | | - $condition .= ' AND page_id < ' . intval( $end ); |
| 234 | | - } |
| 235 | | - return $this->dumpFrom( $condition ); |
| 236 | | - } |
| 237 | | - |
| 238 | | - /** |
| 239 | | - * @param Title $title |
| 240 | | - */ |
| 241 | | - function pageByTitle( $title ) { |
| 242 | | - return $this->dumpFrom( |
| 243 | | - 'page_namespace=' . $title->getNamespace() . |
| 244 | | - ' AND page_title=' . $this->db->addQuotes( $title->getDbKey() ) ); |
| 245 | | - } |
| 246 | | - |
| 247 | | - function pageByName( $name ) { |
| 248 | | - $title = Title::newFromText( $name ); |
| 249 | | - if( is_null( $title ) ) { |
| 250 | | - return new WikiError( "Can't export invalid title" ); |
| 251 | | - } else { |
| 252 | | - return $this->pageByTitle( $title ); |
| 253 | | - } |
| 254 | | - } |
| 255 | | - |
| 256 | | - function pagesByName( $names ) { |
| 257 | | - foreach( $names as $name ) { |
| 258 | | - $this->pageByName( $name ); |
| 259 | | - } |
| 260 | | - } |
| 261 | | - |
| 262 | | - |
| 263 | | - // -------------------- private implementation below -------------------- |
| 264 | | - |
| 265 | | - function dumpFrom( $cond = '' ) { |
| 266 | | - $fname = 'WikiExporter::dumpFrom'; |
| 267 | | - wfProfileIn( $fname ); |
| 268 | | - |
| 269 | | - $page = $this->db->tableName( 'page' ); |
| 270 | | - $revision = $this->db->tableName( 'revision' ); |
| 271 | | - $text = $this->db->tableName( 'text' ); |
| 272 | | - |
| 273 | | - if( $this->history == MW_EXPORT_FULL ) { |
| 274 | | - $join = 'page_id=rev_page'; |
| 275 | | - } elseif( $this->history == MW_EXPORT_CURRENT ) { |
| 276 | | - $join = 'page_id=rev_page AND page_latest=rev_id'; |
| 277 | | - } else { |
| 278 | | - wfProfileOut( $fname ); |
| 279 | | - return new WikiError( "$fname given invalid history dump type." ); |
| 280 | | - } |
| 281 | | - $where = ( $cond == '' ) ? '' : "$cond AND"; |
| 282 | | - |
| 283 | | - if( $this->buffer == MW_EXPORT_STREAM ) { |
| 284 | | - $prev = $this->db->bufferResults( false ); |
| 285 | | - } |
| 286 | | - if( $cond == '' ) { |
| 287 | | - // Optimization hack for full-database dump |
| 288 | | - $pageindex = 'FORCE INDEX (PRIMARY)'; |
| 289 | | - $revindex = 'FORCE INDEX(page_timestamp)'; |
| 290 | | - } else { |
| 291 | | - $pageindex = ''; |
| 292 | | - $revindex = ''; |
| 293 | | - } |
| 294 | | - $result = $this->db->query( |
| 295 | | - "SELECT * FROM |
| 296 | | - $page $pageindex, |
| 297 | | - $revision $revindex, |
| 298 | | - $text |
| 299 | | - WHERE $where $join AND rev_text_id=old_id |
| 300 | | - ORDER BY page_id", $fname ); |
| 301 | | - $wrapper = $this->db->resultObject( $result ); |
| 302 | | - $this->outputStream( $wrapper ); |
| 303 | | - |
| 304 | | - if( $this->buffer == MW_EXPORT_STREAM ) { |
| 305 | | - $this->db->bufferResults( $prev ); |
| 306 | | - } |
| 307 | | - |
| 308 | | - wfProfileOut( $fname ); |
| 309 | | - } |
| 310 | | - |
| 311 | | - /** |
| 312 | | - * Runs through a query result set dumping page and revision records. |
| 313 | | - * The result set should be sorted/grouped by page to avoid duplicate |
| 314 | | - * page records in the output. |
| 315 | | - * |
| 316 | | - * The result set will be freed once complete. Should be safe for |
| 317 | | - * streaming (non-buffered) queries, as long as it was made on a |
| 318 | | - * separate database connection not managed by LoadBalancer; some |
| 319 | | - * blob storage types will make queries to pull source data. |
| 320 | | - * |
| 321 | | - * @param ResultWrapper $resultset |
| 322 | | - * @access private |
| 323 | | - */ |
| 324 | | - function outputStream( $resultset ) { |
| 325 | | - $last = null; |
| 326 | | - while( $row = $resultset->fetchObject() ) { |
| 327 | | - if( is_null( $last ) || |
| 328 | | - $last->page_namespace != $row->page_namespace || |
| 329 | | - $last->page_title != $row->page_title ) { |
| 330 | | - if( isset( $last ) ) { |
| 331 | | - $this->closePage( $last ); |
| 332 | | - } |
| 333 | | - $this->openPage( $row ); |
| 334 | | - $last = $row; |
| 335 | | - } |
| 336 | | - $this->dumpRev( $row ); |
| 337 | | - } |
| 338 | | - if( isset( $last ) ) { |
| 339 | | - $this->closePage( $last ); |
| 340 | | - } |
| 341 | | - $resultset->free(); |
| 342 | | - } |
| 343 | | - |
| 344 | | - /** |
| 345 | | - * Opens a <page> section on the output stream, with data |
| 346 | | - * from the given database row. |
| 347 | | - * |
| 348 | | - * @param object $row |
| 349 | | - * @access private |
| 350 | | - */ |
| 351 | | - function openPage( $row ) { |
| 352 | | - print "<page>\n"; |
| 353 | | - $title = Title::makeTitle( $row->page_namespace, $row->page_title ); |
| 354 | | - print ' ' . wfElementClean( 'title', array(), $title->getPrefixedText() ) . "\n"; |
| 355 | | - print ' ' . wfElement( 'id', array(), strval( $row->page_id ) ) . "\n"; |
| 356 | | - if( '' != $row->page_restrictions ) { |
| 357 | | - print ' ' . wfElement( 'restrictions', array(), |
| 358 | | - strval( $row->page_restrictions ) ) . "\n"; |
| 359 | | - } |
| 360 | | - } |
| 361 | | - |
| 362 | | - /** |
| 363 | | - * Closes a <page> section on the output stream. |
| 364 | | - * If a per-page callback has been set, it will be called |
| 365 | | - * and passed the last database row used for this page. |
| 366 | | - * |
| 367 | | - * @param object $row |
| 368 | | - * @access private |
| 369 | | - */ |
| 370 | | - function closePage( $row ) { |
| 371 | | - print "</page>\n"; |
| 372 | | - if( isset( $this->pageCallback ) ) { |
| 373 | | - call_user_func( $this->pageCallback, $row ); |
| 374 | | - } |
| 375 | | - } |
| 376 | | - |
| 377 | | - /** |
| 378 | | - * Dumps a <revision> section on the output stream, with |
| 379 | | - * data filled in from the given database row. |
| 380 | | - * |
| 381 | | - * @param object $row |
| 382 | | - * @access private |
| 383 | | - */ |
| 384 | | - function dumpRev( $row ) { |
| 385 | | - $fname = 'WikiExporter::dumpRev'; |
| 386 | | - wfProfileIn( $fname ); |
| 387 | | - |
| 388 | | - print " <revision>\n"; |
| 389 | | - print " " . wfElement( 'id', null, strval( $row->rev_id ) ) . "\n"; |
| 390 | | - |
| 391 | | - $ts = wfTimestamp2ISO8601( strval( $row->rev_timestamp ) ); |
| 392 | | - print " " . wfElement( 'timestamp', null, $ts ) . "\n"; |
| 393 | | - |
| 394 | | - print " <contributor>\n"; |
| 395 | | - if( $row->rev_user ) { |
| 396 | | - print " " . wfElementClean( 'username', null, strval( $row->rev_user_text ) ) . "\n"; |
| 397 | | - print " " . wfElement( 'id', null, strval( $row->rev_user ) ) . "\n"; |
| 398 | | - } else { |
| 399 | | - print " " . wfElementClean( 'ip', null, strval( $row->rev_user_text ) ) . "\n"; |
| 400 | | - } |
| 401 | | - print " </contributor>\n"; |
| 402 | | - |
| 403 | | - if( $row->rev_minor_edit ) { |
| 404 | | - print " <minor/>\n"; |
| 405 | | - } |
| 406 | | - if( $row->rev_comment != '' ) { |
| 407 | | - print " " . wfElementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n"; |
| 408 | | - } |
| 409 | | - |
| 410 | | - $text = strval( Revision::getRevisionText( $row ) ); |
| 411 | | - print " " . wfElementClean( 'text', array( 'xml:space' => 'preserve' ), $text ) . "\n"; |
| 412 | | - |
| 413 | | - print " </revision>\n"; |
| 414 | | - |
| 415 | | - wfProfileOut( $fname ); |
| 416 | | - |
| 417 | | - if( isset( $this->revCallback ) ) { |
| 418 | | - call_user_func( $this->revCallback, $row ); |
| 419 | | - } |
| 420 | | - } |
| 421 | | - |
| 422 | | -} |
| 423 | | - |
| 424 | | -function wfTimestamp2ISO8601( $ts ) { |
| 425 | | - #2003-08-05T18:30:02Z |
| 426 | | - return preg_replace( '/^(....)(..)(..)(..)(..)(..)$/', '$1-$2-$3T$4:$5:$6Z', wfTimestamp( TS_MW, $ts ) ); |
| 427 | | -} |
| 428 | | - |
| 429 | | -function xmlsafe( $string ) { |
| 430 | | - $fname = 'xmlsafe'; |
| 431 | | - wfProfileIn( $fname ); |
| 432 | | - |
| 433 | | - /** |
| 434 | | - * The page may contain old data which has not been properly normalized. |
| 435 | | - * Invalid UTF-8 sequences or forbidden control characters will make our |
| 436 | | - * XML output invalid, so be sure to strip them out. |
| 437 | | - */ |
| 438 | | - $string = UtfNormal::cleanUp( $string ); |
| 439 | | - |
| 440 | | - $string = htmlspecialchars( $string ); |
| 441 | | - wfProfileOut( $fname ); |
| 442 | | - return $string; |
| 443 | | -} |
| 444 | | - |
| 445 | 72 | ?> |
| Index: trunk/phase3/RELEASE-NOTES |
| — | — | @@ -124,6 +124,8 @@ |
| 125 | 125 | * (bug 3503) Update LanguageSq.php from sq.wikipedia.org messages |
| 126 | 126 | * Added EditFilter hook, and output callback on EditPage::showEditForm() |
| 127 | 127 | for a place to add in captcha-type extensions in the edit flow |
| | 128 | +* Added filter options, compression piping, and multiple output streams for |
| | 129 | + dumpBackup.php |
| 128 | 130 | |
| 129 | 131 | |
| 130 | 132 | === Caveats === |