* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
* http://www.gnu.org/copyleft/gpl.html
*
- * @package MediaWiki
- * @subpackage SpecialPage
+ * @file
+ * @ingroup Maintenance
*/
$originalDir = getcwd();
-require_once( 'commandLine.inc' );
-require_once( 'SpecialExport.php' );
-require_once( 'maintenance/backup.inc' );
+require_once( dirname(__FILE__) . '/commandLine.inc' );
+require_once( 'backup.inc' );
/**
- * Stream wrapper around 7za filter program.
- * Required since we can't pass an open file resource to XMLReader->open()
- * which is used for the text prefetch.
+ * @ingroup Maintenance
*/
-class SevenZipStream {
- var $stream;
-
- private function stripPath( $path ) {
- $prefix = 'mediawiki.compress.7z://';
- return substr( $path, strlen( $prefix ) );
- }
-
- function stream_open( $path, $mode, $options, &$opened_path ) {
- if( $mode{0} == 'r' ) {
- $options = 'e -bd -so';
- } elseif( $mode{0} == 'w' ) {
- $options = 'a -bd -si';
- } else {
- return false;
- }
- $arg = wfEscapeShellArg( $this->stripPath( $path ) );
- $command = "7za $options $arg";
- if( !wfIsWindows() ) {
- // Suppress the stupid messages on stderr
- $command .= ' 2>/dev/null';
- }
- $this->stream = popen( $command, $mode );
- return ($this->stream !== false);
- }
-
- function url_stat( $path, $flags ) {
- return stat( $this->stripPath( $path ) );
- }
-
- // This is all so lame; there should be a default class we can extend
-
- function stream_close() {
- return fclose( $this->stream );
- }
-
- function stream_flush() {
- return fflush( $this->stream );
- }
-
- function stream_read( $count ) {
- return fread( $this->stream, $count );
- }
-
- function stream_write( $data ) {
- return fwrite( $this->stream, $data );
- }
-
- function stream_tell() {
- return ftell( $this->stream );
- }
-
- function stream_eof() {
- return feof( $this->stream );
- }
-
- function stream_seek( $offset, $whence ) {
- return fseek( $this->stream, $offset, $whence );
- }
-}
-stream_wrapper_register( 'mediawiki.compress.7z', 'SevenZipStream' );
-
-
class TextPassDumper extends BackupDumper {
var $prefetch = null;
var $input = "php://stdin";
- var $history = MW_EXPORT_FULL;
+ var $history = WikiExporter::FULL;
var $fetchCount = 0;
var $prefetchCount = 0;
var $failures = 0;
var $maxFailures = 200;
var $failureTimeout = 5; // Seconds to sleep after db failure
+
+ var $php = "php";
+ var $spawn = false;
+ var $spawnProc = false;
+ var $spawnWrite = false;
+ var $spawnRead = false;
+ var $spawnErr = false;
function dump() {
# This shouldn't happen if on console... ;)
# Notice messages will foul up your XML output even if they're
# relatively harmless.
-// ini_set( 'display_errors', false );
+ if( ini_get( 'display_errors' ) )
+ ini_set( 'display_errors', 'stderr' );
$this->initProgress( $this->history );
- $this->db =& $this->backupDb();
+ $this->db = $this->backupDb();
$this->egress = new ExportProgressFilter( $this->sink, $this );
if( WikiError::isError( $result ) ) {
wfDie( $result->getMessage() );
}
+
+ if( $this->spawnProc ) {
+ $this->closeSpawn();
+ }
$this->report( true );
}
switch( $opt ) {
case 'prefetch':
- require_once 'maintenance/backupPrefetch.inc';
+ global $IP;
+ require_once "$IP/maintenance/backupPrefetch.inc";
$this->prefetch = new BaseDump( $url );
break;
case 'stub':
$this->input = $url;
break;
case 'current':
- $this->history = MW_EXPORT_CURRENT;
+ $this->history = WikiExporter::CURRENT;
break;
case 'full':
- $this->history = MW_EXPORT_FULL;
+ $this->history = WikiExporter::FULL;
+ break;
+ case 'spawn':
+ $this->spawn = true;
+ if( $val ) {
+ $this->php = $val;
+ }
break;
}
}
$etats = '-';
$fetchrate = '-';
}
- global $wgDBname;
$this->progress( sprintf( "%s: %s %d pages (%0.3f/sec), %d revs (%0.3f/sec), %0.1f%% prefetched, ETA %s [max %d]",
- $now, $wgDBname, $this->pageCount, $rate, $this->revCount, $revrate, $fetchrate, $etats, $this->maxCount ) );
+ $now, wfWikiID(), $this->pageCount, $rate, $this->revCount, $revrate, $fetchrate, $etats, $this->maxCount ) );
}
}
return $text;
}
}
+ return $this->doGetText( $id );
+ }
+
+ private function doGetText( $id ) {
+ if( $this->spawn ) {
+ return $this->getTextSpawned( $id );
+ } else {
+ return $this->getTextDbSafe( $id );
+ }
+ }
+
+ /**
+ * Fetch a text revision from the database, retrying in case of failure.
+ * This may survive some transitory errors by reconnecting, but
+ * may not survive a long-term server outage.
+ */
+ private function getTextDbSafe( $id ) {
while( true ) {
try {
- return $this->doGetText( $id );
+ $text = $this->getTextDb( $id );
+ $ex = new MWException("Graceful storage failure");
} catch (DBQueryError $ex) {
+ $text = false;
+ }
+ if( $text === false ) {
$this->failures++;
if( $this->failures > $this->maxFailures ) {
throw $ex;
} else {
$this->progress( "Database failure $this->failures " .
- "of allowed $this->maxFailures! " .
+ "of allowed $this->maxFailures for revision $id! " .
"Pausing $this->failureTimeout seconds..." );
sleep( $this->failureTimeout );
}
+ } else {
+ return $text;
}
}
}
/**
* May throw a database error if, say, the server dies during query.
*/
- private function doGetText( $id ) {
+ private function getTextDb( $id ) {
$id = intval( $id );
$row = $this->db->selectRow( 'text',
array( 'old_text', 'old_flags' ),
array( 'old_id' => $id ),
'TextPassDumper::getText' );
$text = Revision::getRevisionText( $row );
+ if( $text === false ) {
+ return false;
+ }
+ $stripped = str_replace( "\r", "", $text );
+ $normalized = UtfNormal::cleanUp( $stripped );
+ return $normalized;
+ }
+
+ private function getTextSpawned( $id ) {
+ wfSuppressWarnings();
+ if( !$this->spawnProc ) {
+ // First time?
+ $this->openSpawn();
+ }
+ while( true ) {
+
+ $text = $this->getTextSpawnedOnce( $id );
+ if( !is_string( $text ) ) {
+ $this->progress("Database subprocess failed. Respawning...");
+
+ $this->closeSpawn();
+ sleep( $this->failureTimeout );
+ $this->openSpawn();
+
+ continue;
+ }
+ wfRestoreWarnings();
+ return $text;
+ }
+ }
+
+ function openSpawn() {
+ global $IP, $wgDBname;
+
+ $cmd = implode( " ",
+ array_map( 'wfEscapeShellArg',
+ array(
+ $this->php,
+ "$IP/maintenance/fetchText.php",
+ $wgDBname ) ) );
+ $spec = array(
+ 0 => array( "pipe", "r" ),
+ 1 => array( "pipe", "w" ),
+ 2 => array( "file", "/dev/null", "a" ) );
+ $pipes = array();
+
+ $this->progress( "Spawning database subprocess: $cmd" );
+ $this->spawnProc = proc_open( $cmd, $spec, $pipes );
+ if( !$this->spawnProc ) {
+ // shit
+ $this->progress( "Subprocess spawn failed." );
+ return false;
+ }
+ list(
+ $this->spawnWrite, // -> stdin
+ $this->spawnRead, // <- stdout
+ ) = $pipes;
+
+ return true;
+ }
+
+ private function closeSpawn() {
+ wfSuppressWarnings();
+ if( $this->spawnRead )
+ fclose( $this->spawnRead );
+ $this->spawnRead = false;
+ if( $this->spawnWrite )
+ fclose( $this->spawnWrite );
+ $this->spawnWrite = false;
+ if( $this->spawnErr )
+ fclose( $this->spawnErr );
+ $this->spawnErr = false;
+ if( $this->spawnProc )
+ pclose( $this->spawnProc );
+ $this->spawnProc = false;
+ wfRestoreWarnings();
+ }
+
+ private function getTextSpawnedOnce( $id ) {
+ $ok = fwrite( $this->spawnWrite, "$id\n" );
+ //$this->progress( ">> $id" );
+ if( !$ok ) return false;
+
+ $ok = fflush( $this->spawnWrite );
+ //$this->progress( ">> [flush]" );
+ if( !$ok ) return false;
+
+ $len = fgets( $this->spawnRead );
+ //$this->progress( "<< " . trim( $len ) );
+ if( $len === false ) return false;
+
+ $nbytes = intval( $len );
+ $text = "";
+
+ // Subprocess may not send everything at once, we have to loop.
+ while( $nbytes > strlen( $text ) ) {
+ $buffer = fread( $this->spawnRead, $nbytes - strlen( $text ) );
+ if( $buffer === false ) break;
+ $text .= $buffer;
+ }
+
+ $gotbytes = strlen( $text );
+ if( $gotbytes != $nbytes ) {
+ $this->progress( "Expected $nbytes bytes from database subprocess, got $gotbytes ");
+ return false;
+ }
+
+ // Do normalization in the dump thread...
$stripped = str_replace( "\r", "", $text );
$normalized = UtfNormal::cleanUp( $stripped );
return $normalized;
function clearOpenElement( $style ) {
if( $this->openElement ) {
- $this->buffer .= wfElement( $this->openElement[0], $this->openElement[1], $style );
+ $this->buffer .= Xml::element( $this->openElement[0], $this->openElement[1], $style );
$this->openElement = false;
}
}
if( true ) {
$dumper->dump();
} else {
- $dumper->progress( <<<END
+ $dumper->progress( <<<ENDS
This script postprocesses XML dumps from dumpBackup.php to add
page text which was stubbed out (using --stub).
(Default: 100)
--server=h Force reading from MySQL server h
--current Base ETA on number of pages in database instead of all revisions
-END
+ --spawn Spawn a subprocess for loading text records
+ENDS
);
}
-?>
+