* @ingroup Dump Maintenance
*/
-/**
- * @ingroup Dump Maintenance
- */
-class DumpDBZip2Output extends DumpPipeOutput {
- function __construct( $file ) {
- parent::__construct( "dbzip2", $file );
- }
-}
+require_once __DIR__ . '/Maintenance.php';
+require_once __DIR__ . '/../includes/export/DumpFilter.php';
/**
* @ingroup Dump Maintenance
*/
-class BackupDumper {
- var $reportingInterval = 100;
- var $reporting = true;
- var $pageCount = 0;
- var $revCount = 0;
- var $server = null; // use default
- var $pages = null; // all pages
- var $skipHeader = false; // don't output <mediawiki> and <siteinfo>
- var $skipFooter = false; // don't output </mediawiki>
- var $startId = 0;
- var $endId = 0;
- var $revStartId = 0;
- var $revEndId = 0;
- var $sink = null; // Output filters
- 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;
-
- var $outputTypes = array(), $filterTypes = array();
+class BackupDumper extends Maintenance {
+ public $reporting = true;
+ public $pages = null; // all pages
+ public $skipHeader = false; // don't output <mediawiki> and <siteinfo>
+ public $skipFooter = false; // don't output </mediawiki>
+ public $startId = 0;
+ public $endId = 0;
+ public $revStartId = 0;
+ public $revEndId = 0;
+ public $dumpUploads = false;
+ public $dumpUploadFileContents = false;
+
+ protected $reportingInterval = 100;
+ protected $pageCount = 0;
+ protected $revCount = 0;
+ protected $server = null; // use default
+ protected $sink = null; // Output filters
+ protected $lastTime = 0;
+ protected $pageCountLast = 0;
+ protected $revCountLast = 0;
+
+ protected $outputTypes = array();
+ protected $filterTypes = array();
+
+ protected $ID = 0;
/**
* The dependency-injected database to use.
*
* @var DatabaseBase|null
*
- * @see self::setDb
+ * @see self::setDB
*/
protected $forcedDb = null;
- /**
- * @var LoadBalancer
- */
+ /** @var LoadBalancer */
protected $lb;
- function __construct( $args ) {
+ // @todo Unused?
+ private $stubText = false; // include rev_text_id instead of text; for 2-pass dump
+
+ /**
+ * @param array $args For backward compatibility
+ */
+ function __construct( $args = null ) {
+ parent::__construct();
$this->stderr = fopen( "php://stderr", "wt" );
// Built-in output and filter plugins
$this->registerFilter( 'notalk', 'DumpNotalkFilter' );
$this->registerFilter( 'namespace', 'DumpNamespaceFilter' );
- $this->sink = $this->processArgs( $args );
+ // These three can be specified multiple times
+ $this->addOption( 'plugin', 'Load a dump plugin class. Specify as <class>[:<file>].',
+ false, true, false, true );
+ $this->addOption( 'output', 'Begin a filtered output stream; Specify as <type>:<file>. ' .
+ '<type>s: file, gzip, bzip2, 7zip, dbzip2', false, true, false, true );
+ $this->addOption( 'filter', 'Add a filter on an output branch. Specify as ' .
+ '<type>[:<options>]. <types>s: latest, notalk, namespace', false, true, false, true );
+ $this->addOption( 'report', 'Report position and speed after every n pages processed. ' .
+ 'Default: 100.', false, true );
+ $this->addOption( 'server', 'Force reading from MySQL server', false, true );
+ $this->addOption( '7ziplevel', '7zip compression level for all 7zip outputs. Used for ' .
+ '-mx option to 7za command.', false, true );
+
+ if ( $args ) {
+ // Args should be loaded and processed so that dump() can be called directly
+ // instead of execute()
+ $this->loadWithArgv( $args );
+ $this->processOptions();
+ }
}
/**
- * @param $name String
- * @param $class String: name of output filter plugin class
+ * @param string $name
+ * @param string $class Name of output filter plugin class
*/
function registerOutput( $name, $class ) {
$this->outputTypes[$name] = $class;
}
/**
- * @param $name String
- * @param $class String: name of filter plugin class
+ * @param string $name
+ * @param string $class Name of filter plugin class
*/
function registerFilter( $name, $class ) {
$this->filterTypes[$name] = $class;
/**
* Load a plugin and register it
*
- * @param $class String: name of plugin class; must have a static 'register'
- * method that takes a BackupDumper as a parameter.
- * @param $file String: full or relative path to the PHP file to load, or empty
+ * @param string $class Name of plugin class; must have a static 'register'
+ * method that takes a BackupDumper as a parameter.
+ * @param string $file Full or relative path to the PHP file to load, or empty
*/
function loadPlugin( $class, $file ) {
if ( $file != '' ) {
call_user_func_array( $register, array( &$this ) );
}
+ function execute() {
+ throw new MWException( 'execute() must be overridden in subclasses' );
+ }
+
/**
- * @param $args Array
- * @return Array
+ * Processes arguments and sets $this->$sink accordingly
*/
- function processArgs( $args ) {
+ function processOptions() {
$sink = null;
$sinks = array();
- foreach ( $args as $arg ) {
- $matches = array();
- if ( preg_match( '/^--(.+?)(?:=(.+?)(?::(.+?))?)?$/', $arg, $matches ) ) {
- @list( /* $full */ , $opt, $val, $param ) = $matches;
- switch ( $opt ) {
- case "plugin":
- $this->loadPlugin( $val, $param );
+
+ $options = $this->orderedOptions;
+ foreach ( $options as $arg ) {
+ $opt = $arg[0];
+ $param = $arg[1];
+
+ switch ( $opt ) {
+ case 'plugin':
+ $val = explode( ':', $param );
+
+ if ( count( $val ) === 1 ) {
+ $this->loadPlugin( $val[0] );
+ } elseif ( count( $val ) === 2 ) {
+ $this->loadPlugin( $val[0], $val[1] );
+ } else {
+ $this->fatalError( 'Invalid plugin parameter' );
+ return;
+ }
+
break;
- case "output":
+ case 'output':
+ $split = explode( ':', $param, 2 );
+ if ( count( $split ) !== 2 ) {
+ $this->fatalError( 'Invalid output parameter' );
+ }
+ list( $type, $file ) = $split;
if ( !is_null( $sink ) ) {
$sinks[] = $sink;
}
- if ( !isset( $this->outputTypes[$val] ) ) {
- $this->fatalError( "Unrecognized output sink type '$val'" );
+ if ( !isset( $this->outputTypes[$type] ) ) {
+ $this->fatalError( "Unrecognized output sink type '$type'" );
}
- $type = $this->outputTypes[$val];
- $sink = new $type( $param );
+ $class = $this->outputTypes[$type];
+ if ( $type === "7zip" ) {
+ $sink = new $class( $file, intval( $this->getOption( '7ziplevel' ) ) );
+ } else {
+ $sink = new $class( $file );
+ }
+
break;
- case "filter":
+ case 'filter':
if ( is_null( $sink ) ) {
$sink = new DumpOutput();
}
- if ( !isset( $this->filterTypes[$val] ) ) {
- $this->fatalError( "Unrecognized filter type '$val'" );
+
+ $split = explode( ':', $param );
+ $key = $split[0];
+
+ if ( !isset( $this->filterTypes[$key] ) ) {
+ $this->fatalError( "Unrecognized filter type '$key'" );
+ }
+
+ $type = $this->filterTypes[$key];
+
+ if ( count( $split ) === 1 ) {
+ $filter = new $type( $sink );
+ } elseif ( count( $split ) === 2 ) {
+ $filter = new $type( $sink, $split[1] );
+ } else {
+ $this->fatalError( 'Invalid filter parameter' );
}
- $type = $this->filterTypes[$val];
- $filter = new $type( $sink, $param );
// references are lame in php...
unset( $sink );
$sink = $filter;
break;
- case "report":
- $this->reportingInterval = intval( $val );
- break;
- case "server":
- $this->server = $val;
- break;
- case "force-normal":
- if ( !function_exists( 'utf8_normalize' ) ) {
- $this->fatalError( "UTF-8 normalization extension not loaded. " .
- "Install or remove --force-normal parameter to use slower code." );
- }
- break;
- default:
- $this->processOption( $opt, $val, $param );
- }
}
}
+ if ( $this->hasOption( 'report' ) ) {
+ $this->reportingInterval = intval( $this->getOption( 'report' ) );
+ }
+
+ if ( $this->hasOption( 'server' ) ) {
+ $this->server = $this->getOption( 'server' );
+ }
+
if ( is_null( $sink ) ) {
$sink = new DumpOutput();
}
$sinks[] = $sink;
if ( count( $sinks ) > 1 ) {
- return new DumpMultiWriter( $sinks );
+ $this->sink = new DumpMultiWriter( $sinks );
} else {
- return $sink;
+ $this->sink = $sink;
}
}
- function processOption( $opt, $val, $param ) {
- // extension point for subclasses to add options
- }
-
function dump( $history, $text = WikiExporter::TEXT ) {
# Notice messages will foul up your XML output even if they're
# relatively harmless.
} else {
$exporter->allLogs();
}
- # Page dumps: all or by page ID range
} elseif ( is_null( $this->pages ) ) {
+ # Page dumps: all or by page ID range
if ( $this->startId || $this->endId ) {
$exporter->pagesByRange( $this->startId, $this->endId );
} elseif ( $this->revStartId || $this->revEndId ) {
} else {
$exporter->allPages();
}
- # Dump of specific pages
} else {
+ # Dump of specific pages
$exporter->pagesByName( $this->pages );
}
* Initialise starting time and maximum revision count.
* We'll make ETA calculations based an progress, assuming relatively
* constant per-revision rate.
- * @param $history Integer: WikiExporter::CURRENT or WikiExporter::FULL
+ * @param int $history WikiExporter::CURRENT or WikiExporter::FULL
*/
function initProgress( $history = WikiExporter::FULL ) {
$table = ( $history == WikiExporter::CURRENT ) ? 'page' : 'revision';
* Force the dump to use the provided database connection for database
* operations, wherever possible.
*
- * @param $db DatabaseBase|null: (Optional) the database connection to
- * use. If null, resort to use the globally provided ways to
- * get database connections.
+ * @param DatabaseBase|null $db (Optional) the database connection to use. If null, resort to
+ * use the globally provided ways to get database connections.
*/
- function setDb( DatabaseBase $db = null ) {
+ function setDB( IDatabase $db = null ) {
+ parent::setDB( $db );
$this->forcedDb = $db;
}
function backupServer() {
global $wgDBserver;
+
return $this->server
? $this->server
: $wgDBserver;
$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->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;
}
}
function progress( $string ) {
- fwrite( $this->stderr, $string . "\n" );
+ if ( $this->reporting ) {
+ fwrite( $this->stderr, $string . "\n" );
+ }
}
function fatalError( $msg ) {
- $this->progress( "$msg\n" );
- die( 1 );
+ $this->error( "$msg\n", 1 );
}
}