$originalDir = getcwd();
-require_once( dirname(__FILE__) . '/commandLine.inc' );
+require_once( dirname( __FILE__ ) . '/commandLine.inc' );
require_once( 'backup.inc' );
-require_once( '7zip.inc' );
/**
* @ingroup Maintenance
var $history = WikiExporter::FULL;
var $fetchCount = 0;
var $prefetchCount = 0;
-
+
var $failures = 0;
- var $maxFailures = 200;
+ var $maxFailures = 5;
+ var $failedTextRetrievals = 0;
+ var $maxConsecutiveFailedTextRetrievals = 200;
var $failureTimeout = 5; // Seconds to sleep after db failure
-
+
var $php = "php";
var $spawn = false;
var $spawnProc = false;
# Notice messages will foul up your XML output even if they're
# relatively harmless.
- if( ini_get( 'display_errors' ) )
+ 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 ) {
+
+ if ( $this->spawnProc ) {
$this->closeSpawn();
}
}
function processOption( $opt, $val, $param ) {
+ global $IP;
$url = $this->processFileOpt( $val, $param );
-
+
switch( $opt ) {
case 'prefetch':
- global $IP;
require_once "$IP/maintenance/backupPrefetch.inc";
$this->prefetch = new BaseDump( $url );
break;
break;
case 'spawn':
$this->spawn = true;
- if( $val ) {
+ if ( $val ) {
$this->php = $val;
}
break;
}
}
-
+
function processFileOpt( $val, $param ) {
switch( $val ) {
case "file":
* Overridden to include prefetch ratio if enabled.
*/
function showReport() {
- if( !$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 {
- $this->prefetchCount++;
- return $text;
+ $dbr = wfGetDB( DB_SLAVE );
+ $revID = intval($this->thisRev);
+ $revLength = $dbr->selectField( 'revision', 'rev_len', array('rev_id' => $revID ) );
+ // if length of rev text in file doesn't match length in db, we reload
+ // this avoids carrying forward broken data from previous xml dumps
+ if( strlen($text) == $revLength ) {
+ $this->prefetchCount++;
+ return $text;
+ }
}
}
return $this->doGetText( $id );
}
-
+
private function doGetText( $id ) {
- if( $this->spawn ) {
- return $this->getTextSpawned( $id );
- } else {
- return $this->getTextDbSafe( $id );
+
+ $id = intval( $id );
+ $this->failures = 0;
+ $ex = new MWException( "Graceful storage failure" );
+ while (true) {
+ if ( $this->spawn ) {
+ if ($this->failures) {
+ // we don't know why it failed, could be the child process
+ // borked, could be db entry busted, could be db server out to lunch,
+ // so cover all bases
+ $this->closeSpawn();
+ $this->openSpawn();
+ }
+ $text = $this->getTextSpawned( $id );
+ } else {
+ $text = $this->getTextDbSafe( $id );
+ }
+ if ( $text === false ) {
+ $this->failures++;
+ if ( $this->failures > $this->maxFailures) {
+ $this->progress( "Failed to retrieve revision text for text id ".
+ "$id after $this->maxFailures tries, giving up" );
+ // were there so many bad retrievals in a row we want to bail?
+ // at some point we have to declare the dump irretrievably broken
+ $this->failedTextRetrievals++;
+ if ($this->failedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals) {
+ throw $ex;
+ }
+ else {
+ // would be nice to return something better to the caller someday,
+ // log what we know about the failure and about the revision
+ return("");
+ }
+ } else {
+ $this->progress( "Error $this->failures " .
+ "of allowed $this->maxFailures retrieving revision text for text id $id! " .
+ "Pausing $this->failureTimeout seconds before retry..." );
+ sleep( $this->failureTimeout );
+ }
+ } else {
+ $this->failedTextRetrievals= 0;
+ return( $text );
+ }
}
+
}
-
+
/**
* 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 ) {
+ while ( true ) {
try {
$text = $this->getTextDb( $id );
- $ex = new MWException("Graceful storage failure");
- } catch (DBQueryError $ex) {
+ $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 for revision $id! " .
- "Pausing $this->failureTimeout seconds..." );
- sleep( $this->failureTimeout );
- }
- } else {
- return $text;
- }
+ return $text;
}
}
-
+
/**
* May throw a database error if, say, the server dies during query.
*/
private function getTextDb( $id ) {
- $id = intval( $id );
+ global $wgContLang;
$row = $this->db->selectRow( 'text',
array( 'old_text', 'old_flags' ),
array( 'old_id' => $id ),
'TextPassDumper::getText' );
$text = Revision::getRevisionText( $row );
- if( $text === false ) {
+ if ( $text === false ) {
return false;
}
$stripped = str_replace( "\r", "", $text );
- $normalized = UtfNormal::cleanUp( $stripped );
+ $normalized = $wgContLang->normalize( $stripped );
return $normalized;
}
-
+
private function getTextSpawned( $id ) {
wfSuppressWarnings();
- if( !$this->spawnProc ) {
+ 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;
- }
+ $text = $this->getTextSpawnedOnce( $id );
+ wfRestoreWarnings();
+ return $text;
}
-
+
function openSpawn() {
global $IP, $wgDBname;
-
+
$cmd = implode( " ",
array_map( 'wfEscapeShellArg',
array(
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 ) {
+ if ( !$this->spawnProc ) {
// shit
$this->progress( "Subprocess spawn failed." );
return false;
$this->spawnWrite, // -> stdin
$this->spawnRead, // <- stdout
) = $pipes;
-
+
return true;
}
-
+
private function closeSpawn() {
wfSuppressWarnings();
- if( $this->spawnRead )
+ if ( $this->spawnRead )
fclose( $this->spawnRead );
$this->spawnRead = false;
- if( $this->spawnWrite )
+ if ( $this->spawnWrite )
fclose( $this->spawnWrite );
$this->spawnWrite = false;
- if( $this->spawnErr )
+ if ( $this->spawnErr )
fclose( $this->spawnErr );
$this->spawnErr = false;
- if( $this->spawnProc )
+ 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;
-
+ // $this->progress( ">> $id" );
+ if ( !$ok ) return false;
+
$ok = fflush( $this->spawnWrite );
- //$this->progress( ">> [flush]" );
- if( !$ok ) return false;
-
+ // $this->progress( ">> [flush]" );
+ if ( !$ok ) return false;
+
+ // check that the text id they are sending is the one we asked for
+ // this avoids out of sync revision text errors we have encountered in the past
+ $newId = fgets( $this->spawnRead );
+ if ( $newId === false ) {
+ return false;
+ }
+ if ( $id != intval( $newId ) ) {
+ return false;
+ }
+
$len = fgets( $this->spawnRead );
- //$this->progress( "<< " . trim( $len ) );
- if( $len === false ) return false;
-
+ // $this->progress( "<< " . trim( $len ) );
+ if ( $len === false ) return false;
+
$nbytes = intval( $len );
+ // actual error, not zero-length text
+ if ($nbytes < 0 ) return false;
+
$text = "";
-
+
// Subprocess may not send everything at once, we have to loop.
- while( $nbytes > strlen( $text ) ) {
+ while ( $nbytes > strlen( $text ) ) {
$buffer = fread( $this->spawnRead, $nbytes - strlen( $text ) );
- if( $buffer === false ) break;
+ if ( $buffer === false ) break;
$text .= $buffer;
}
-
+
$gotbytes = strlen( $text );
- if( $gotbytes != $nbytes ) {
- $this->progress( "Expected $nbytes bytes from database subprocess, got $gotbytes ");
+ 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 ) {
+ 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( <<<ENDS
Options:
--stub=<type>:<file> To load a compressed stub dump instead of stdin
--prefetch=<type>:<file> Use a prior dump file as a text source, to save
- pressure on the database.
- (Requires PHP 5.0+ and the XMLReader PECL extension)
- --quiet Don't dump status reports to stderr.
+ pressure on the database.
+ (Requires PHP 5.0+ and the XMLReader PECL extension)
+ --quiet Don't dump status reports to stderr.
--report=n Report position and speed after every n pages processed.
- (Default: 100)
+ (Default: 100)
--server=h Force reading from MySQL server h
- --current Base ETA on number of pages in database instead of all revisions
- --spawn Spawn a subprocess for loading text records
+ --current Base ETA on number of pages in database instead of all revisions
+ --spawn Spawn a subprocess for loading text records
ENDS
);
}