http://www.mediawiki.org/wiki/Special:Code/MediaWiki/96556
Revision: 96556
Author: reedy
Date: 2011-09-08 12:28:26 +0000 (Thu, 08 Sep 2011)
Log Message:
-----------
MFT r95260, r95272, r95288, r95290, r95443, r95601, r95604, r95634, r95720,
r95810
Modified Paths:
--------------
branches/REL1_18/phase3/includes/Export.php
branches/REL1_18/phase3/maintenance/backup.inc
branches/REL1_18/phase3/maintenance/dumpTextPass.php
Modified: branches/REL1_18/phase3/includes/Export.php
===================================================================
--- branches/REL1_18/phase3/includes/Export.php 2011-09-08 12:23:06 UTC (rev
96555)
+++ branches/REL1_18/phase3/includes/Export.php 2011-09-08 12:28:26 UTC (rev
96556)
@@ -354,6 +354,9 @@
* @ingroup Dump
*/
class XmlDumpWriter {
+ var $firstPageWritten = 0;
+ var $lastPageWritten = 0;
+ var $pageInProgress = 0;
/**
* Returns the export schema version.
@@ -458,6 +461,7 @@
$title = Title::makeTitle( $row->page_namespace,
$row->page_title );
$out .= ' ' . Xml::elementClean( 'title', array(),
$title->getPrefixedText() ) . "\n";
$out .= ' ' . Xml::element( 'id', array(), strval(
$row->page_id ) ) . "\n";
+ $this->pageInProgress = $row->page_id;
if ( $row->page_is_redirect ) {
$out .= ' ' . Xml::element( 'redirect', array() ) .
"\n";
}
@@ -478,6 +482,10 @@
*/
function closePage() {
return " </page>\n";
+ if (! $this->firstPageWritten) {
+ $this->firstPageWritten = $this->pageInProgress;
+ }
+ $this->lastPageWritten = $this->pageInProgress;
}
/**
@@ -691,6 +699,22 @@
function write( $string ) {
print $string;
}
+
+ function closeRenameAndReopen( $newname ) {
+ return;
+ }
+
+ function closeAndRename( $newname ) {
+ return;
+ }
+
+ function rename( $newname ) {
+ return;
+ }
+
+ function getFilename() {
+ return NULL;
+ }
}
/**
@@ -699,14 +723,71 @@
*/
class DumpFileOutput extends DumpOutput {
var $handle;
+ var $filename;
function __construct( $file ) {
$this->handle = fopen( $file, "wt" );
+ $this->filename = $file;
}
function write( $string ) {
fputs( $this->handle, $string );
}
+
+ /**
+ * Close the old file, move it to a specified name,
+ * and reopen new file with the old name. Use this
+ * for writing out a file in multiple pieces
+ * at specified checkpoints (e.g. every n hours).
+ */
+ function closeRenameAndReopen( $newname ) {
+ if ( is_array($newname) ) {
+ if (count($newname) > 1) {
+ WfDie("Export closeRenameAndReopen: passed
multiple argumnts for rename of single file\n");
+ }
+ else {
+ $newname = $newname[0];
+ }
+ }
+ if ( $newname ) {
+ fclose( $this->handle );
+ rename( $this->filename, $newname );
+ $this->handle = fopen( $this->filename, "wt" );
+ }
+ }
+
+ function closeAndRename( $newname ) {
+ if ( is_array($newname) ) {
+ if (count($newname) > 1) {
+ throw new MWException("Export
closeRenameAndReopen: passed multiple argumnts for rename of single file\n");
+ }
+ else {
+ $newname = $newname[0];
+ }
+ }
+ if ( $newname ) {
+ fclose( $this->handle );
+ rename( $this->filename, $newname );
+ }
+ }
+
+ function rename( $newname ) {
+ if ( is_array($newname) ) {
+ if (count($newname) > 1) {
+ WfDie("Export closeRenameAndReopen: passed
multiple argumnts for rename of single file\n");
+ }
+ else {
+ $newname = $newname[0];
+ }
+ }
+ if ( $newname ) {
+ rename( $this->filename, $newname );
+ }
+ }
+
+ function getFilename() {
+ return $this->filename;
+ }
}
/**
@@ -716,12 +797,80 @@
* @ingroup Dump
*/
class DumpPipeOutput extends DumpFileOutput {
+ var $command;
+
function __construct( $command, $file = null ) {
if ( !is_null( $file ) ) {
$command .= " > " . wfEscapeShellArg( $file );
}
- $this->handle = popen( $command, "w" );
+
+ $this->startCommand($command);
+ $this->command = $command;
+ $this->filename = $file;
}
+
+ function startCommand($command) {
+ $spec = array(
+ 0 => array( "pipe", "r" ),
+ );
+ $pipes = array();
+ $this->procOpenResource = proc_open( $command, $spec, $pipes );
+ $this->handle = $pipes[0];
+ }
+
+ /**
+ * Close the old file, move it to a specified name,
+ * and reopen new file with the old name.
+ */
+ function closeRenameAndReopen( $newname ) {
+ if ( is_array($newname) ) {
+ if (count($newname) > 1) {
+ WfDie("Export closeRenameAndReopen: passed
multiple argumnts for rename of single file\n");
+ }
+ else {
+ $newname = $newname[0];
+ }
+ }
+ if ( $newname ) {
+ fclose( $this->handle );
+ proc_close($this->procOpenResource);
+ rename( $this->filename, $newname );
+ $command = $this->command;
+ $command .= " > " . wfEscapeShellArg( $this->filename
);
+ $this->startCommand($command);
+ }
+ }
+
+ function closeAndRename( $newname ) {
+ if ( is_array($newname) ) {
+ if (count($newname) > 1) {
+ throw new MWException("Export
closeRenameAndReopen: passed multiple argumnts for rename of single file\n");
+ }
+ else {
+ $newname = $newname[0];
+ }
+ }
+ if ( $newname ) {
+# pclose( $this->handle );
+ fclose( $this->handle );
+ proc_close($this->procOpenResource);
+ rename( $this->filename, $newname );
+ }
+ }
+
+ function rename( $newname ) {
+ if ( is_array($newname) ) {
+ if (count($newname) > 1) {
+ WfDie("Export closeRenameAndReopen: passed
multiple argumnts for rename of single file\n");
+ }
+ else {
+ $newname = $newname[0];
+ }
+ }
+ if ( $newname ) {
+ rename( $this->filename, $newname );
+ }
+ }
}
/**
@@ -749,13 +898,65 @@
* @ingroup Dump
*/
class Dump7ZipOutput extends DumpPipeOutput {
+ var $filename;
+
function __construct( $file ) {
$command = "7za a -bd -si " . wfEscapeShellArg( $file );
// Suppress annoying useless crap from p7zip
// Unfortunately this could suppress real error messages too
$command .= ' >' . wfGetNull() . ' 2>&1';
parent::__construct( $command );
+ $this->filename = $file;
}
+
+ function closeRenameAndReopen( $newname ) {
+ if ( is_array($newname) ) {
+ if (count($newname) > 1) {
+ WfDie("Export closeRenameAndReopen: passed
multiple argumnts for rename of single file\n");
+ }
+ else {
+ $newname = $newname[0];
+ }
+ }
+ if ( $newname ) {
+ fclose( $this->handle );
+ proc_close($this->procOpenResource);
+ rename( $this->filename, $newname );
+ $command = "7za a -bd -si " . wfEscapeShellArg( $file );
+ $command .= ' >' . wfGetNull() . ' 2>&1';
+ $this->startCommand($command);
+ }
+ }
+
+ function closeAndRename( $newname ) {
+ if ( is_array($newname) ) {
+ if (count($newname) > 1) {
+ throw new MWException("Export
closeRenameAndReopen: passed multiple argumnts for rename of single file\n");
+ }
+ else {
+ $newname = $newname[0];
+ }
+ }
+ if ( $newname ) {
+ fclose( $this->handle );
+ proc_close($this->procOpenResource);
+ rename( $this->filename, $newname );
+ }
+ }
+
+ function rename( $newname ) {
+ if ( is_array($newname) ) {
+ if (count($newname) > 1) {
+ WfDie("Export closeRenameAndReopen: passed
multiple argumnts for rename of single file\n");
+ }
+ else {
+ $newname = $newname[0];
+ }
+ }
+ if ( $newname ) {
+ rename( $this->filename, $newname );
+ }
+ }
}
@@ -803,6 +1004,22 @@
$this->sink->writeRevision( $rev, $string );
}
+ function closeRenameAndReopen( $newname ) {
+ $this->sink->closeRenameAndReopen( $newname );
+ }
+
+ function closeAndRename( $newname ) {
+ $this->sink->closeAndRename( $newname );
+ }
+
+ function rename( $newname ) {
+ $this->sink->rename( $newname );
+ }
+
+ function getFilename() {
+ return $this->sink->getFilename();
+ }
+
/**
* Override for page-based filter types.
* @return bool
@@ -950,6 +1167,32 @@
$this->sinks[$i]->writeRevision( $rev, $string );
}
}
+
+ function closeRenameAndReopen( $newnames ) {
+ for( $i = 0; $i < $this->count; $i++ ) {
+ $this->sinks[$i]->closeRenameAndReopen( $newnames[$i] );
+ }
+ }
+
+ function closeAndRename( $newname ) {
+ for( $i = 0; $i < $this->count; $i++ ) {
+ $this->sinks[$i]->closeAndRename( $newnames[$i] );
+ }
+ }
+ function rename( $newnames ) {
+ for( $i = 0; $i < $this->count; $i++ ) {
+ $this->sinks[$i]->rename( $newnames[$i] );
+ }
+ }
+
+ function getFilename() {
+ $filenames = array();
+ for( $i = 0; $i < $this->count; $i++ ) {
+ $filenames[] = $this->sinks[$i]->getFilename();
+ }
+ return $filenames;
+ }
+
}
function xmlsafe( $string ) {
Modified: branches/REL1_18/phase3/maintenance/backup.inc
===================================================================
--- branches/REL1_18/phase3/maintenance/backup.inc 2011-09-08 12:23:06 UTC
(rev 96555)
+++ branches/REL1_18/phase3/maintenance/backup.inc 2011-09-08 12:28:26 UTC
(rev 96556)
@@ -51,6 +51,10 @@
var $stubText = false; // include rev_text_id instead of text; for
2-pass dump
var $dumpUploads = false;
var $dumpUploadFileContents = false;
+ var $lastTime = 0;
+ var $pageCountLast = 0;
+ var $revCountLast = 0;
+ var $ID = 0;
function BackupDumper( $args ) {
$this->stderr = fopen( "php://stderr", "wt" );
@@ -233,6 +237,8 @@
$dbr = wfGetDB( DB_SLAVE );
$this->maxCount = $dbr->selectField( $table, "MAX($field)", '',
__METHOD__ );
$this->startTime = wfTime();
+ $this->lastTime = $this->startTime;
+ $this->ID = getmypid();
}
/**
@@ -281,21 +287,35 @@
function showReport() {
if ( $this->reporting ) {
- $delta = wfTime() - $this->startTime;
$now = wfTimestamp( TS_DB );
- if ( $delta ) {
- $rate = $this->pageCount / $delta;
- $revrate = $this->revCount / $delta;
+ $nowts = wfTime();
+ $deltaAll = wfTime() - $this->startTime;
+ $deltaPart = wfTime() - $this->lastTime;
+ $this->pageCountPart = $this->pageCount -
$this->pageCountLast;
+ $this->revCountPart = $this->revCount -
$this->revCountLast;
+
+ if ( $deltaAll ) {
$portion = $this->revCount / $this->maxCount;
- $eta = $this->startTime + $delta / $portion;
+ $eta = $this->startTime + $deltaAll / $portion;
$etats = wfTimestamp( TS_DB, intval( $eta ) );
+ $pageRate = $this->pageCount / $deltaAll;
+ $revRate = $this->revCount / $deltaAll;
} else {
- $rate = '-';
- $revrate = '-';
+ $pageRate = '-';
+ $revRate = '-';
$etats = '-';
}
- $this->progress( sprintf( "%s: %s %d pages (%0.3f/sec),
%d revs (%0.3f/sec), ETA %s [max %d]",
- $now, wfWikiID(), $this->pageCount, $rate,
$this->revCount, $revrate, $etats, $this->maxCount ) );
+ if ( $deltaPart ) {
+ $pageRatePart = $this->pageCountPart /
$deltaPart;
+ $revRatePart = $this->revCountPart / $deltaPart;
+ } else {
+ $pageRatePart = '-';
+ $revRatePart = '-';
+ }
+ $this->progress( sprintf( "%s: %s (ID %d) %d pages
(%0.1f|%0.1f/sec all|curr), %d revs (%0.1f|%0.1f/sec all|curr), ETA %s [max
%d]",
+ $now, wfWikiID(), $this->ID,
$this->pageCount, $pageRate, $pageRatePart, $this->revCount, $revRate,
$revRatePart, $etats, $this->maxCount ) );
+ $this->lastTime = $nowts;
+ $this->revCountLast = $this->revCount;
}
}
Modified: branches/REL1_18/phase3/maintenance/dumpTextPass.php
===================================================================
--- branches/REL1_18/phase3/maintenance/dumpTextPass.php 2011-09-08
12:23:06 UTC (rev 96555)
+++ branches/REL1_18/phase3/maintenance/dumpTextPass.php 2011-09-08
12:28:26 UTC (rev 96556)
@@ -2,7 +2,7 @@
/**
* Script that postprocesses XML dumps from dumpBackup.php to add page text
*
- * Copyright © 2005 Brion Vibber <[email protected]>, 2010 Alexandre Emsenhuber
+ * Copyright \xA9 2005 Brion Vibber <[email protected]>, 2010 Alexandre
Emsenhuber
* http://www.mediawiki.org/
*
* This program is free software; you can redistribute it and/or modify
@@ -35,11 +35,9 @@
class TextPassDumper extends BackupDumper {
var $prefetch = null;
var $input = "php://stdin";
+ var $history = WikiExporter::FULL;
var $fetchCount = 0;
var $prefetchCount = 0;
- var $lastTime = 0;
- var $pageCountLast = 0;
- var $revCountLast = 0;
var $prefetchCountLast = 0;
var $fetchCountLast = 0;
@@ -56,12 +54,21 @@
var $spawnRead = false;
var $spawnErr = false;
- var $ID = 0;
+ var $xmlwriterobj = false;
+ # when we spend more than maxTimeAllowed seconds on this run, we
continue
+ # processing until we write out the next complete page, then save
output file(s),
+ # rename it/them and open new one(s)
+ var $maxTimeAllowed = 0; // 0 = no limit
+ var $timeExceeded = false;
+ var $firstPageWritten = false;
+ var $lastPageWritten = false;
+ var $checkpointJustWritten = false;
+ var $checkpointFiles = array();
+
function initProgress( $history ) {
parent::initProgress();
- $this->ID = getmypid();
- $this->lastTime = $this->startTime;
+ $this->timeOfCheckpoint = $this->startTime;
}
function dump( $history, $text = WikiExporter::TEXT ) {
@@ -73,12 +80,25 @@
if ( ini_get( 'display_errors' ) )
ini_set( 'display_errors', 'stderr' );
- $this->initProgress( $history );
+ $this->initProgress( $this->history );
$this->db = $this->backupDb();
- $this->readDump();
+ $this->egress = new ExportProgressFilter( $this->sink, $this );
+ # it would be nice to do it in the constructor, oh well. need
egress set
+ $this->finalOptionCheck();
+
+ # we only want this so we know how to close a stream :-P
+ $this->xmlwriterobj = new XmlDumpWriter();
+
+ $input = fopen( $this->input, "rt" );
+ $result = $this->readDump( $input );
+
+ if ( WikiError::isError( $result ) ) {
+ throw new MWException( $result->getMessage() );
+ }
+
if ( $this->spawnProc ) {
$this->closeSpawn();
}
@@ -98,6 +118,18 @@
case 'stub':
$this->input = $url;
break;
+ case 'maxtime':
+ $this->maxTimeAllowed = intval($val)*60;
+ break;
+ case 'checkpointfile':
+ $this->checkpointFiles[] = $val;
+ break;
+ case 'current':
+ $this->history = WikiExporter::CURRENT;
+ break;
+ case 'full':
+ $this->history = WikiExporter::FULL;
+ break;
case 'spawn':
$this->spawn = true;
if ( $val ) {
@@ -142,6 +174,7 @@
if ( $this->reporting ) {
$now = wfTimestamp( TS_DB );
+ $nowts = wfTime();
$deltaAll = wfTime() - $this->startTime;
$deltaPart = wfTime() - $this->lastTime;
$this->pageCountPart = $this->pageCount -
$this->pageCountLast;
@@ -180,86 +213,98 @@
$pageRatePart = '-';
$revRatePart = '-';
}
- $this->progress( sprintf( "%s: %s (ID %d) %d pages
(%0.1f|%0.1f/sec all|curr), %d revs (%0.1f|%0.1f/sec all|curr), %0.1f%%|%0.1f%%
prefetched (all|curr), ETA %s [max %d]",-
+ $this->progress( sprintf( "%s: %s (ID %d) %d pages
(%0.1f|%0.1f/sec all|curr), %d revs (%0.1f|%0.1f/sec all|curr), %0.1f%%|%0.1f%%
prefetched (all|curr), ETA %s [max %d]",
$now, wfWikiID(), $this->ID,
$this->pageCount, $pageRate, $pageRatePart, $this->revCount, $revRate,
$revRatePart, $fetchRate, $fetchRatePart, $etats, $this->maxCount ) );
- $this->lastTime = $now;
- $this->partCountLast = $this->partCount;
+ $this->lastTime = $nowts;
$this->revCountLast = $this->revCount;
$this->prefetchCountLast = $this->prefetchCount;
$this->fetchCountLast = $this->fetchCount;
}
}
- function readDump() {
- $state = '';
- $lastName = '';
- $this->thisPage = 0;
- $this->thisRev = 0;
+ function setTimeExceeded() {
+ $this->timeExceeded = True;
+ }
- $reader = new XMLReader();
- $reader->open( $this->input );
- $writer = new XMLWriter();
- $writer->openMemory();
+ function checkIfTimeExceeded() {
+ if ( $this->maxTimeAllowed && ( $this->lastTime -
$this->timeOfCheckpoint > $this->maxTimeAllowed ) ) {
+ return True;
+ }
+ return False;
+ }
+ function finalOptionCheck() {
+ if (($this->checkpointFiles && ! $this->maxTimeAllowed) ||
+ ($this->maxTimeAllowed && !$this->checkpointFiles)) {
+ throw new MWException("Options checkpointfile and
maxtime must be specified together.\n");
+ }
+ foreach ($this->checkpointFiles as $checkpointFile) {
+ $count = substr_count ($checkpointFile,"%s");
+ if (substr_count ($checkpointFile,"%s") != 2) {
+ throw new MWException("Option checkpointfile
must contain two '%s' for substitution of first and last pageids, count is
$count instead, file is $checkpointFile.\n");
+ }
+ }
- while ( $reader->read() ) {
- $tag = $reader->name;
- $type = $reader->nodeType;
+ if ($this->checkpointFiles) {
+ $filenameList = $this->egress->getFilename();
+ if (! is_array($filenameList)) {
+ $filenameList = array( $filenameList );
+ }
+ if (count($filenameList) !=
count($this->checkpointFiles)) {
+ throw new MWException("One checkpointfile must
be specified for each output option, if maxtime is used.\n");
+ }
+ }
+ }
- if ( $type == XmlReader::END_ELEMENT ) {
- $writer->endElement();
+ function readDump( $input ) {
+ $this->buffer = "";
+ $this->openElement = false;
+ $this->atStart = true;
+ $this->state = "";
+ $this->lastName = "";
+ $this->thisPage = 0;
+ $this->thisRev = 0;
- if ( $tag == 'revision' ) {
- $this->revCount();
- $this->thisRev = '';
- } elseif ( $tag == 'page' ) {
- $this->reportPage();
- $this->thisPage = '';
- }
- } elseif ( $type == XmlReader::ELEMENT ) {
- $attribs = array();
- if ( $reader->hasAttributes ) {
- for ( $i = 0;
$reader->moveToAttributeNo( $i ); $i++ ) {
- $attribs[$reader->name] =
$reader->value;
- }
- }
+ $parser = xml_parser_create( "UTF-8" );
+ xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false
);
- if ( $reader->isEmptyElement && $tag == 'text'
&& isset( $attribs['id'] ) ) {
- $writer->startElement( 'text' );
- $writer->writeAttribute( 'xml:space',
'preserve' );
- $text = $this->getText( $attribs['id']
);
- if ( strlen( $text ) ) {
- $writer->text( $text );
- }
- $writer->endElement();
- } else {
- $writer->startElement( $tag );
- foreach( $attribs as $name => $val ) {
- $writer->writeAttribute( $name,
$val );
- }
- if ( $reader->isEmptyElement ) {
- $writer->endElement();
- }
- }
+ xml_set_element_handler( $parser, array( &$this, 'startElement'
), array( &$this, 'endElement' ) );
+ xml_set_character_data_handler( $parser, array( &$this,
'characterData' ) );
- $lastName = $tag;
- if ( $tag == 'revision' ) {
- $state = 'revision';
- } elseif ( $tag == 'page' ) {
- $state = 'page';
+ $offset = 0; // for context extraction on error reporting
+ $bufferSize = 512 * 1024;
+ do {
+ if ($this->checkIfTimeExceeded()) {
+ $this->setTimeExceeded();
+ }
+ $chunk = fread( $input, $bufferSize );
+ 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( $input ) );
+ if ($this->maxTimeAllowed) {
+ $filenameList = $this->egress->getFilename();
+ # we wrote some stuff after last checkpoint that needs
renamed */
+ if (! is_array($filenameList)) {
+ $filenameList = array( $filenameList );
+ }
+ if (file_exists($filenameList[0])) {
+ $newFilenames = array();
+ $firstPageID =
str_pad($this->firstPageWritten,9,"0",STR_PAD_LEFT);
+ $lastPageID =
str_pad($this->lastPageWritten,9,"0",STR_PAD_LEFT);
+ for ($i =0; $i < count($filenameList); $i++) {
+ $checkpointNameFilledIn =
sprintf($this->checkpointFiles[$i], $firstPageID, $lastPageID);
+ $fileinfo = pathinfo($filenameList[$i]);
+ $newFilenames[] = $fileinfo{'dirname'}
. '/' . $checkpointNameFilledIn;
}
- } elseif ( $type == XMLReader::SIGNIFICANT_WHITESPACE
|| $type == XMLReader::TEXT ) {
- if ( $lastName == 'id' ) {
- if ( $state == 'revision' ) {
- $this->thisRev .=
$reader->value;
- } elseif ( $state == 'page' ) {
- $this->thisPage .=
$reader->value;
- }
- }
- $writer->text( $reader->value );
+ $this->egress->closeAndRename( $newFilenames );
}
- $this->sink->write( $writer->outputMemory() );
}
+ xml_parser_free( $parser );
+
+ return true;
}
function getText( $id ) {
@@ -282,6 +327,7 @@
}
private function doGetText( $id ) {
+
$id = intval( $id );
$this->failures = 0;
$ex = new MWException( "Graceful storage failure" );
@@ -469,13 +515,133 @@
$normalized = $wgContLang->normalize( $stripped );
return $normalized;
}
+
+ function startElement( $parser, $name, $attribs ) {
+ $this->checkpointJustWritten = false;
+
+ $this->clearOpenElement( null );
+ $this->lastName = $name;
+
+ if ( $name == 'revision' ) {
+ $this->state = $name;
+ $this->egress->writeOpenPage( null, $this->buffer );
+ $this->buffer = "";
+ } elseif ( $name == 'page' ) {
+ $this->state = $name;
+ if ( $this->atStart ) {
+ $this->egress->writeOpenStream( $this->buffer );
+ $this->buffer = "";
+ $this->atStart = false;
+ }
+ }
+
+ if ( $name == "text" && isset( $attribs['id'] ) ) {
+ $text = $this->getText( $attribs['id'] );
+ $this->openElement = array( $name, array( 'xml:space'
=> 'preserve' ) );
+ if ( strlen( $text ) > 0 ) {
+ $this->characterData( $parser, $text );
+ }
+ } else {
+ $this->openElement = array( $name, $attribs );
+ }
+ }
+
+ function endElement( $parser, $name ) {
+ $this->checkpointJustWritten = false;
+
+ if ( $this->openElement ) {
+ $this->clearOpenElement( "" );
+ } else {
+ $this->buffer .= "</$name>";
+ }
+
+ if ( $name == 'revision' ) {
+ $this->egress->writeRevision( null, $this->buffer );
+ $this->buffer = "";
+ $this->thisRev = "";
+ } elseif ( $name == 'page' ) {
+ if (! $this->firstPageWritten) {
+ $this->firstPageWritten = trim($this->thisPage);
+ }
+ $this->lastPageWritten = trim($this->thisPage);
+ if ($this->timeExceeded) {
+ $this->egress->writeClosePage( $this->buffer );
+ # nasty hack, we can't just write the chardata
after the
+ # page tag, it will include leading blanks from
the next line
+ $this->egress->sink->write("\n");
+
+ $this->buffer =
$this->xmlwriterobj->closeStream();
+ $this->egress->writeCloseStream( $this->buffer
);
+
+ $this->buffer = "";
+ $this->thisPage = "";
+ /* this could be more than one file if we had
more than one output arg */
+ $checkpointFilenames = array();
+ $filenameList = $this->egress->getFilename();
+
+ if (! is_array($filenameList)) {
+ $filenameList = array( $filenameList );
+ }
+ $newFilenames = array();
+ $firstPageID =
str_pad($this->firstPageWritten,9,"0",STR_PAD_LEFT);
+ $lastPageID =
str_pad($this->lastPageWritten,9,"0",STR_PAD_LEFT);
+ for ($i =0; $i < count($filenameList); $i++) {
+ $checkpointNameFilledIn =
sprintf($this->checkpointFiles[$i], $firstPageID, $lastPageID);
+ $fileinfo = pathinfo($filenameList[$i]);
+ $newFilenames[] = $fileinfo{'dirname'}
. '/' . $checkpointNameFilledIn;
+ }
+ $this->egress->closeRenameAndReopen(
$newFilenames );
+ $this->buffer =
$this->xmlwriterobj->openStream();
+ $this->timeExceeded = false;
+ $this->timeOfCheckpoint = $this->lastTime;
+ $this->firstPageWritten = false;
+ $this->checkpointJustWritten = true;
+ }
+ else {
+ $this->egress->writeClosePage( $this->buffer );
+ $this->buffer = "";
+ $this->thisPage = "";
+ }
+
+ } 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" ) {
+ $this->thisRev .= $data;
+ } elseif ( $this->state == "page" ) {
+ $this->thisPage .= $data;
+ }
+ }
+ # have to skip the newline left over from closepagetag line of
+ # end of checkpoint files. nasty hack!!
+ if ($this->checkpointJustWritten) {
+ if ($data[0] == "\n") {
+ $data = substr($data,1);
+ }
+ $this->checkpointJustWritten = false;
+ }
+ $this->buffer .= htmlspecialchars( $data );
+ }
+
+ function clearOpenElement( $style ) {
+ if ( $this->openElement ) {
+ $this->buffer .= Xml::element( $this->openElement[0],
$this->openElement[1], $style );
+ $this->openElement = false;
+ }
+ }
}
$dumper = new TextPassDumper( $argv );
if ( !isset( $options['help'] ) ) {
- $dumper->dump( WikiExporter::FULL );
+ $dumper->dump( true );
} else {
$dumper->progress( <<<ENDS
This script postprocesses XML dumps from dumpBackup.php to add
@@ -489,17 +655,20 @@
--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 the XMLReader extension)
+ --maxtime=<minutes> Write out checkpoint file after this many minutes
(writing
+ out complete page, closing xml file properly, and opening new
one
+ with header). This option requires the checkpointfile option.
+ --checkpointfile=<filenamepattern> Use this string for checkpoint filenames,
+ substituting first pageid written for the first %s
(required) and the
+ last pageid written for the second %s if it exists.
--quiet Don't dump status reports to stderr.
--report=n Report position and speed after every n pages processed.
(Default: 100)
--server=h Force reading from MySQL server h
- --output=<type>:<file> Write to a file instead of stdout
- <type>s: file, gzip, bzip2, 7zip
--current Base ETA on number of pages in database instead of all
revisions
--spawn Spawn a subprocess for loading text records
--help Display this help message
ENDS
);
}
-
-
_______________________________________________
MediaWiki-CVS mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/mediawiki-cvs
