checkstorage.php

来自「php 开发的内容管理系统」· PHP 代码 · 共 469 行 · 第 1/2 页

<?php

/**
 * Fsck for MediaWiki
 */

define( 'CONCAT_HEADER', 'O:27:"concatenatedgziphistoryblob"' );

if ( !defined( 'MEDIAWIKI' ) ) {
	require_once( dirname(__FILE__) . '/../commandLine.inc' );
	require_once( 'ExternalStore.php' );
	require_once( 'ExternalStoreDB.php' );
	require_once( 'SpecialImport.php' );

	$cs = new CheckStorage;
	$fix = isset( $options['fix'] );
	if ( isset( $args[0] ) ) {
		$xml = $args[0];
	} else {
		$xml = false;
	}
	$cs->check( $fix, $xml );
}


//----------------------------------------------------------------------------------

class CheckStorage
{
	var $oldIdMap, $errors;
	var $dbStore = null;

	var $errorDescriptions = array(
		'restore text' => 'Damaged text, need to be restored from a backup',
		'restore revision' => 'Damaged revision row, need to be restored from a backup',
		'unfixable' => 'Unexpected errors with no automated fixing method',
		'fixed' => 'Errors already fixed',
		'fixable' => 'Errors which would already be fixed if --fix was specified',
	);		

	function check( $fix = false, $xml = '' ) {
		$fname = 'checkStorage';
		$dbr =& wfGetDB( DB_SLAVE );
		if ( $fix ) {
			$dbw =& wfGetDB( DB_MASTER );
			print "Checking, will fix errors if possible...\n";
		} else {
			print "Checking...\n";
		}
		$maxRevId = $dbr->selectField( 'revision', 'MAX(rev_id)', false, $fname );
		$chunkSize = 1000;
		$flagStats = array();
		$objectStats = array();
		$knownFlags = array( 'external', 'gzip', 'object', 'utf-8' );
		$this->errors = array(
			'restore text' => array(),
			'restore revision' => array(),
			'unfixable' => array(),
			'fixed' => array(),
			'fixable' => array(),
		);

		for ( $chunkStart = 1 ; $chunkStart < $maxRevId; $chunkStart += $chunkSize ) {
			$chunkEnd = $chunkStart + $chunkSize - 1;
			//print "$chunkStart of $maxRevId\n";

			// Fetch revision rows
			$this->oldIdMap = array();
			$dbr->ping();		
			$res = $dbr->select( 'revision', array( 'rev_id', 'rev_text_id' ), 
				array( "rev_id BETWEEN $chunkStart AND $chunkEnd" ), $fname );
			while ( $row = $dbr->fetchObject( $res ) ) {
				$this->oldIdMap[$row->rev_id] = $row->rev_text_id;
			}
			$dbr->freeResult( $res );

			if ( !count( $this->oldIdMap ) ) {
				continue;
			}

			// Fetch old_flags
			$missingTextRows = array_flip( $this->oldIdMap );
			$externalRevs = array();
			$objectRevs = array();
			$res = $dbr->select( 'text', array( 'old_id', 'old_flags' ), 
				'old_id IN (' . implode( ',', $this->oldIdMap ) . ')', $fname );
			while ( $row = $dbr->fetchObject( $res ) ) {
				$flags = $row->old_flags;
				$id = $row->old_id;

				// Create flagStats row if it doesn't exist
				$flagStats = $flagStats + array( $flags => 0 );
				// Increment counter
				$flagStats[$flags]++;

				// Not missing
				unset( $missingTextRows[$row->old_id] );

				// Check for external or object
				if ( $flags == '' ) {
					$flagArray = array();
				} else {
					$flagArray = explode( ',', $flags );
				}
				if ( in_array( 'external', $flagArray ) ) {
					$externalRevs[] = $id;
				} elseif ( in_array( 'object', $flagArray ) ) {
					$objectRevs[] = $id;
				}

				// Check for unrecognised flags
				if ( $flags == '0' ) {
					// This is a known bug from 2004
					// It's safe to just erase the old_flags field
					if ( $fix ) {
						$this->error( 'fixed', "Warning: old_flags set to 0", $id );
						$dbw->ping();
						$dbw->update( 'text', array( 'old_flags' => '' ), 
							array( 'old_id' => $id ), $fname );
						echo "Fixed\n";
					} else {
						$this->error( 'fixable', "Warning: old_flags set to 0", $id );
					}
				} elseif ( count( array_diff( $flagArray, $knownFlags ) ) ) {
					$this->error( 'unfixable', "Error: invalid flags field \"$flags\"", $id );
				}
			}
			$dbr->freeResult( $res );

			// Output errors for any missing text rows
			foreach ( $missingTextRows as $oldId => $revId ) {
				$this->error( 'restore revision', "Error: missing text row", $oldId );
			}

			// Verify external revisions
			$externalConcatBlobs = array();
			$externalNormalBlobs = array();
			if ( count( $externalRevs ) ) {
				$res = $dbr->select( 'text', array( 'old_id', 'old_flags', 'old_text' ), 
					array( 'old_id IN (' . implode( ',', $externalRevs ) . ')' ), $fname );
				while ( $row = $dbr->fetchObject( $res ) ) {
					$urlParts = explode( '://', $row->old_text, 2 );
					if ( count( $urlParts ) !== 2 || $urlParts[1] == '' ) {
						$this->error( 'restore text', "Error: invalid URL \"{$row->old_text}\"", $row->old_id );
						continue;
					}
					list( $proto, $path ) = $urlParts;
					if ( $proto != 'DB' ) {
						$this->error( 'restore text', "Error: invalid external protocol \"$proto\"", $row->old_id );
						continue;
					}
					$path = explode( '/', $row->old_text );
					$cluster = $path[2];
					$id = $path[3];
					if ( isset( $path[4] ) ) {
						$externalConcatBlobs[$cluster][$id][] = $row->old_id;
					} else {
						$externalNormalBlobs[$cluster][$id][] = $row->old_id;
					}
				}
				$dbr->freeResult( $res );
			}

			// Check external concat blobs for the right header
			$this->checkExternalConcatBlobs( $externalConcatBlobs );
			
			// Check external normal blobs for existence
			if ( count( $externalNormalBlobs ) ) {
				if ( is_null( $this->dbStore ) ) {
					$this->dbStore = new ExternalStoreDB;
				}
				foreach ( $externalConcatBlobs as $cluster => $xBlobIds ) {
					$blobIds = array_keys( $xBlobIds );
					$extDb =& $this->dbStore->getSlave( $cluster );
					$blobsTable = $this->dbStore->getTable( $extDb );
					$res = $extDb->select( $blobsTable, 
						array( 'blob_id' ), 
						array( 'blob_id IN( ' . implode( ',', $blobIds ) . ')' ), $fname );
					while ( $row = $extDb->fetchObject( $res ) ) {
						unset( $xBlobIds[$row->blob_id] );
					}
					$extDb->freeResult( $res );
					// Print errors for missing blobs rows
					foreach ( $xBlobIds as $blobId => $oldId ) {
						$this->error( 'restore text', "Error: missing target $blobId for one-part ES URL", $oldId );
					}
				}
			}

			// Check local objects
			$dbr->ping();
			$concatBlobs = array();
			$curIds = array();
			if ( count( $objectRevs ) ) {
				$headerLength = 300;
				$res = $dbr->select( 'text', array( 'old_id', 'old_flags', "LEFT(old_text, $headerLength) AS header" ), 
					array( 'old_id IN (' . implode( ',', $objectRevs ) . ')' ), $fname );
				while ( $row = $dbr->fetchObject( $res ) ) {
					$oldId = $row->old_id;
					if ( !preg_match( '/^O:(\d+):"(\w+)"/', $row->header, $matches ) ) {
						$this->error( 'restore text', "Error: invalid object header", $oldId );
						continue;
					}

					$className = strtolower( $matches[2] );
					if ( strlen( $className ) != $matches[1] ) {
						$this->error( 'restore text', "Error: invalid object header, wrong class name length", $oldId );
						continue;
					}

					$objectStats = $objectStats + array( $className => 0 );
					$objectStats[$className]++;

					switch ( $className ) {
						case 'concatenatedgziphistoryblob':
							// Good
							break;
						case 'historyblobstub':
						case 'historyblobcurstub':
							if ( strlen( $row->header ) == $headerLength ) {
								$this->error( 'unfixable', "Error: overlong stub header", $oldId );
								continue;
							}
							$stubObj = unserialize( $row->header );
							if ( !is_object( $stubObj ) ) {
								$this->error( 'restore text', "Error: unable to unserialize stub object", $oldId );
								continue;
							}
							if ( $className == 'historyblobstub' ) {
								$concatBlobs[$stubObj->mOldId][] = $oldId;
							} else {
								$curIds[$stubObj->mCurId][] = $oldId;
							}
							break;
						default: