var $history = WikiExporter::FULL;
var $fetchCount = 0;
var $prefetchCount = 0;
+ var $prefetchCountLast = 0;
+ var $fetchCountLast = 0;
var $failures = 0;
var $maxFailures = 5;
var $spawnRead = false;
var $spawnErr = false;
+ 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();
+
+ /**
+ * @var DatabaseBase
+ */
+ protected $db;
+
+ function initProgress( $history ) {
+ parent::initProgress();
+ $this->timeOfCheckpoint = $this->startTime;
+ }
+
function dump( $history, $text = WikiExporter::TEXT ) {
- # This shouldn't happen if on console... ;)
+ // 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.
+ // Notice messages will foul up your XML output even if they're
+ // relatively harmless.
if ( ini_get( 'display_errors' ) )
ini_set( 'display_errors', 'stderr' );
$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 ) ) {
- wfDie( $result->getMessage() );
+ throw new MWException( $result->getMessage() );
}
if ( $this->spawnProc ) {
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;
}
function processFileOpt( $val, $param ) {
- switch( $val ) {
- case "file":
- return $param;
- case "gzip":
- return "compress.zlib://$param";
- case "bzip2":
- return "compress.bzip2://$param";
- case "7zip":
- return "mediawiki.compress.7z://$param";
- default:
- return $val;
+ $fileURIs = explode(';',$param);
+ foreach ( $fileURIs as $URI ) {
+ switch( $val ) {
+ case "file":
+ $newURI = $URI;
+ break;
+ case "gzip":
+ $newURI = "compress.zlib://$URI";
+ break;
+ case "bzip2":
+ $newURI = "compress.bzip2://$URI";
+ break;
+ case "7zip":
+ $newURI = "mediawiki.compress.7z://$URI";
+ break;
+ default:
+ $newURI = $URI;
+ }
+ $newFileURIs[] = $newURI;
}
+ $val = implode( ';', $newFileURIs );
+ return $val;
}
/**
*/
function showReport() {
if ( !$this->prefetch ) {
- return parent::showReport();
+ parent::showReport();
+ return;
}
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 ) );
- $fetchrate = 100.0 * $this->prefetchCount / $this->fetchCount;
+ if ( $this->fetchCount ) {
+ $fetchRate = 100.0 * $this->prefetchCount / $this->fetchCount;
+ } else {
+ $fetchRate = '-';
+ }
+ $pageRate = $this->pageCount / $deltaAll;
+ $revRate = $this->revCount / $deltaAll;
} else {
- $rate = '-';
- $revrate = '-';
+ $pageRate = '-';
+ $revRate = '-';
$etats = '-';
- $fetchrate = '-';
+ $fetchRate = '-';
+ }
+ if ( $deltaPart ) {
+ if ( $this->fetchCountLast ) {
+ $fetchRatePart = 100.0 * $this->prefetchCountLast / $this->fetchCountLast;
+ } else {
+ $fetchRatePart = '-';
+ }
+ $pageRatePart = $this->pageCountPart / $deltaPart;
+ $revRatePart = $this->revCountPart / $deltaPart;
+
+ } else {
+ $fetchRatePart = '-';
+ $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]",
+ $now, wfWikiID(), $this->ID, $this->pageCount, $pageRate, $pageRatePart, $this->revCount, $revRate, $revRatePart, $fetchRate, $fetchRatePart, $etats, $this->maxCount ) );
+ $this->lastTime = $nowts;
+ $this->revCountLast = $this->revCount;
+ $this->prefetchCountLast = $this->prefetchCount;
+ $this->fetchCountLast = $this->fetchCount;
+ }
+ }
+
+ function setTimeExceeded() {
+ $this->timeExceeded = True;
+ }
+
+ 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 ( $count != 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");
+ }
+ }
+
+ if ( $this->checkpointFiles ) {
+ $filenameList = (array)$this->egress->getFilenames();
+ if ( count( $filenameList ) != count( $this->checkpointFiles ) ) {
+ throw new MWException("One checkpointfile must be specified for each output option, if maxtime is used.\n");
}
- $this->progress( sprintf( "%s: %s %d pages (%0.3f/sec), %d revs (%0.3f/sec), %0.1f%% prefetched, ETA %s [max %d]",
- $now, wfWikiID(), $this->pageCount, $rate, $this->revCount, $revrate, $fetchrate, $etats, $this->maxCount ) );
}
}
$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" );
}
$offset += strlen( $chunk );
} while ( $chunk !== false && !feof( $input ) );
+ if ($this->maxTimeAllowed) {
+ $filenameList = (array)$this->egress->getFilenames();
+ // we wrote some stuff after last checkpoint that needs renamed
+ if (file_exists($filenameList[0])) {
+ $newFilenames = array();
+ # we might have just written the header and footer and had no
+ # pages or revisions written... perhaps they were all deleted
+ # there's no pageID 0 so we use that. the caller is responsible
+ # for deciding what to do with a file containing only the
+ # siteinfo information and the mw tags.
+ if (! $this->firstPageWritten) {
+ $firstPageID = str_pad(0,9,"0",STR_PAD_LEFT);
+ $lastPageID = str_pad(0,9,"0",STR_PAD_LEFT);
+ }
+ else {
+ $firstPageID = str_pad($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->closeAndRename( $newFilenames );
+ }
+ }
xml_parser_free( $parser );
return true;
$this->fetchCount++;
if ( isset( $this->prefetch ) ) {
$text = $this->prefetch->prefetch( $this->thisPage, $this->thisRev );
- if ( $text === null ) {
- // Entry missing from prefetch dump
- } elseif ( $text === "" ) {
- // Blank entries may indicate that the prior dump was broken.
- // To be safe, reload it.
- } else {
+ if ( $text !== null ) { // Entry missing from prefetch dump
$dbr = wfGetDB( DB_SLAVE );
- $revID = intval($this->thisRev);
- $revLength = $dbr->selectField( 'revision', 'rev_len', array('rev_id' => $revID ) );
+ $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 ) {
+ if( strlen( $text ) == $revLength ) {
$this->prefetchCount++;
return $text;
}
}
private function doGetText( $id ) {
-
$id = intval( $id );
$this->failures = 0;
$ex = new MWException( "Graceful storage failure" );
$this->closeSpawn();
$this->openSpawn();
}
- $text = $this->getTextSpawned( $id );
+ $text = $this->getTextSpawned( $id );
} else {
- $text = $this->getTextDbSafe( $id );
+ $text = $this->getTextDbSafe( $id );
}
if ( $text === false ) {
$this->failures++;
$this->failedTextRetrievals++;
if ($this->failedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals) {
throw $ex;
- }
- else {
+ } else {
// would be nice to return something better to the caller someday,
// log what we know about the failure and about the revision
- return("");
+ return "";
}
} else {
$this->progress( "Error $this->failures " .
}
} else {
$this->failedTextRetrievals= 0;
- return( $text );
+ return $text;
}
}
-
+ return '';
}
/**
* 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.
+ *
+ * FIXME: WTF? Why is it using a loop and then returning unconditionally?
*/
private function getTextDbSafe( $id ) {
while ( true ) {
try {
$text = $this->getTextDb( $id );
- $ex = new MWException( "Graceful storage failure" );
} catch ( DBQueryError $ex ) {
$text = false;
}
/**
* May throw a database error if, say, the server dies during query.
+ * @param $id
+ * @return bool|string
*/
private function getTextDb( $id ) {
global $wgContLang;
}
function openSpawn() {
- global $IP, $wgDBname;
-
- $cmd = implode( " ",
- array_map( 'wfEscapeShellArg',
- array(
- $this->php,
- "$IP/maintenance/fetchText.php",
- $wgDBname ) ) );
+ global $IP;
+
+ if ( file_exists( "$IP/../multiversion/MWScript.php" ) ) {
+ $cmd = implode( " ",
+ array_map( 'wfEscapeShellArg',
+ array(
+ $this->php,
+ "$IP/../multiversion/MWScript.php",
+ "fetchText.php",
+ '--wiki', wfWikiID() ) ) );
+ }
+ else {
+ $cmd = implode( " ",
+ array_map( 'wfEscapeShellArg',
+ array(
+ $this->php,
+ "$IP/maintenance/fetchText.php",
+ '--wiki', wfWikiID() ) ) );
+ }
$spec = array(
0 => array( "pipe", "r" ),
1 => array( "pipe", "w" ),
}
function startElement( $parser, $name, $attribs ) {
+ $this->checkpointJustWritten = false;
+
$this->clearOpenElement( null );
$this->lastName = $name;
}
function endElement( $parser, $name ) {
+ $this->checkpointJustWritten = false;
+
if ( $this->openElement ) {
$this->clearOpenElement( "" );
} else {
$this->buffer = "";
$this->thisRev = "";
} elseif ( $name == 'page' ) {
- $this->egress->writeClosePage( $this->buffer );
- $this->buffer = "";
- $this->thisPage = "";
+ 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
+
+ $filenameList = (array)$this->egress->getFilenames();
+ $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 = "";
$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 );
}
--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)
dépôts / lhc / web / wiklou.git / blobdiff