* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
* http://www.gnu.org/copyleft/gpl.html
*
- * @addtogroup SpecialPage
+ * @file
+ * @ingroup Maintenance
*/
$originalDir = getcwd();
-require_once( 'commandLine.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 $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... ;)
header( 'Content-type: text/html; charset=UTF-8' );
# 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 );
$input = fopen( $this->input, "rt" );
$result = $this->readDump( $input );
- if( WikiError::isError( $result ) ) {
+ 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':
case 'full':
$this->history = WikiExporter::FULL;
break;
+ case 'spawn':
+ $this->spawn = true;
+ if ( $val ) {
+ $this->php = $val;
+ }
+ break;
}
}
* Overridden to include prefetch ratio if enabled.
*/
function showReport() {
- if( !$this->prefetch ) {
+ if ( !$this->prefetch ) {
return parent::showReport();
}
- if( $this->reporting ) {
+ if ( $this->reporting ) {
$delta = wfTime() - $this->startTime;
$now = wfTimestamp( TS_DB );
- if( $delta ) {
+ if ( $delta ) {
$rate = $this->pageCount / $delta;
$revrate = $this->revCount / $delta;
$portion = $this->revCount / $this->maxCount;
$bufferSize = 512 * 1024;
do {
$chunk = fread( $input, $bufferSize );
- if( !xml_parse( $parser, $chunk, feof( $input ) ) ) {
+ if ( !xml_parse( $parser, $chunk, feof( $input ) ) ) {
wfDebug( "TextDumpPass::readDump encountered XML parsing error\n" );
return new WikiXmlError( $parser, 'XML import parse failure', $chunk, $offset );
}
$offset += strlen( $chunk );
- } while( $chunk !== false && !feof( $input ) );
+ } while ( $chunk !== false && !feof( $input ) );
xml_parser_free( $parser );
return true;
function getText( $id ) {
$this->fetchCount++;
- if( isset( $this->prefetch ) ) {
+ if ( isset( $this->prefetch ) ) {
$text = $this->prefetch->prefetch( $this->thisPage, $this->thisRev );
- if( $text === null ) {
+ if ( $text === null ) {
// Entry missing from prefetch dump
- } elseif( $text === "" ) {
+ } elseif ( $text === "" ) {
// Blank entries may indicate that the prior dump was broken.
// To be safe, reload it.
} else {
return $text;
}
}
- while( true ) {
+ 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 );
- } catch (DBQueryError $ex) {
+ $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 ) {
+ 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 ) {
+ global $wgContLang;
$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 = $wgContLang->normalize( $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 ) {
+ global $wgContLang;
+
+ $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 );
+ $normalized = $wgContLang->normalize( $stripped );
return $normalized;
}
$this->clearOpenElement( null );
$this->lastName = $name;
- if( $name == 'revision' ) {
+ if ( $name == 'revision' ) {
$this->state = $name;
$this->egress->writeOpenPage( null, $this->buffer );
$this->buffer = "";
- } elseif( $name == 'page' ) {
+ } elseif ( $name == 'page' ) {
$this->state = $name;
- if( $this->atStart ) {
+ if ( $this->atStart ) {
$this->egress->writeOpenStream( $this->buffer );
$this->buffer = "";
$this->atStart = false;
}
}
- if( $name == "text" && isset( $attribs['id'] ) ) {
+ if ( $name == "text" && isset( $attribs['id'] ) ) {
$text = $this->getText( $attribs['id'] );
$this->openElement = array( $name, array( 'xml:space' => 'preserve' ) );
- if( strlen( $text ) > 0 ) {
+ if ( strlen( $text ) > 0 ) {
$this->characterData( $parser, $text );
}
} else {
}
function endElement( $parser, $name ) {
- if( $this->openElement ) {
+ if ( $this->openElement ) {
$this->clearOpenElement( "" );
} else {
$this->buffer .= "</$name>";
}
- if( $name == 'revision' ) {
+ if ( $name == 'revision' ) {
$this->egress->writeRevision( null, $this->buffer );
$this->buffer = "";
$this->thisRev = "";
- } elseif( $name == 'page' ) {
+ } elseif ( $name == 'page' ) {
$this->egress->writeClosePage( $this->buffer );
$this->buffer = "";
$this->thisPage = "";
- } elseif( $name == 'mediawiki' ) {
+ } elseif ( $name == 'mediawiki' ) {
$this->egress->writeCloseStream( $this->buffer );
$this->buffer = "";
}
function characterData( $parser, $data ) {
$this->clearOpenElement( null );
- if( $this->lastName == "id" ) {
- if( $this->state == "revision" ) {
+ if ( $this->lastName == "id" ) {
+ if ( $this->state == "revision" ) {
$this->thisRev .= $data;
- } elseif( $this->state == "page" ) {
+ } elseif ( $this->state == "page" ) {
$this->thisPage .= $data;
}
}
}
function clearOpenElement( $style ) {
- if( $this->openElement ) {
- $this->buffer .= wfElement( $this->openElement[0], $this->openElement[1], $style );
+ if ( $this->openElement ) {
+ $this->buffer .= Xml::element( $this->openElement[0], $this->openElement[1], $style );
$this->openElement = false;
}
}
$dumper = new TextPassDumper( $argv );
-if( true ) {
+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
);
}