--- /dev/null
+<?php\r
+\r
+/**\r
+ * Fsck for MediaWiki\r
+ */\r
+\r
+define( 'CONCAT_HEADER', 'O:27:"concatenatedgziphistoryblob"' );\r
+\r
+if ( !defined( 'MEDIAWIKI' ) ) {\r
+ require_once( dirname(__FILE__) . '/../commandLine.inc' );\r
+ require_once('ExternalStore.php');\r
+ require_once( 'ExternalStoreDB.php' );\r
+\r
+ checkStorage();\r
+}\r
+\r
+\r
+//----------------------------------------------------------------------------------\r
+\r
+function checkError( $msg, $ids ) {\r
+ global $oldIdMap;\r
+ if ( is_array( $ids ) && count( $ids ) == 1 ) {\r
+ $ids = reset( $ids );\r
+ }\r
+ if ( is_array( $ids ) ) {\r
+ $revIds = array();\r
+ foreach ( $ids as $id ) {\r
+ $revIds = array_merge( $revIds, array_keys( $oldIdMap, $id ) );\r
+ }\r
+ print "$msg in text rows " . implode( ', ', $ids ) . \r
+ ", revisions " . implode( ', ', $revIds ) . "\n";\r
+ } else {\r
+ $id = $ids;\r
+ $revIds = array_keys( $oldIdMap, $id );\r
+ if ( count( $revIds ) == 1 ) {\r
+ print "$msg in old_id $id, rev_id {$revIds[0]}\n";\r
+ } else {\r
+ print "$msg in old_id $id, revisions " . implode( ', ', $revIds ) . "\n";\r
+ }\r
+ }\r
+}\r
+\r
+function checkStorage() {\r
+ global $oldIdMap;\r
+ \r
+ $fname = 'checkStorage';\r
+ $dbr =& wfGetDB( DB_SLAVE );\r
+ $maxRevId = $dbr->selectField( 'revision', 'MAX(rev_id)', false, $fname );\r
+ $chunkSize = 1000;\r
+ $flagStats = array();\r
+ $objectStats = array();\r
+ $knownFlags = array( 'external', 'gzip', 'object', 'utf-8' );\r
+ $dbStore = null;\r
+\r
+ for ( $chunkStart = 1 ; $chunkStart < $maxRevId; $chunkStart += $chunkSize ) {\r
+ $chunkEnd = $chunkStart + $chunkSize - 1;\r
+ //print "$chunkStart to $chunkEnd of $maxRevId\n";\r
+\r
+ // Fetch revision rows\r
+ $oldIdMap = array();\r
+ $res = $dbr->select( 'revision', array( 'rev_id', 'rev_text_id' ), \r
+ array( "rev_id BETWEEN $chunkStart AND $chunkEnd" ), $fname );\r
+ while ( $row = $dbr->fetchObject( $res ) ) {\r
+ $oldIdMap[$row->rev_id] = $row->rev_text_id;\r
+ }\r
+ $dbr->freeResult( $res );\r
+\r
+ // Fetch old_flags\r
+ $missingTextRows = array_flip( $oldIdMap );\r
+ $externalRevs = array();\r
+ $objectRevs = array();\r
+ $flagsFields = array();\r
+ $res = $dbr->select( 'text', array( 'old_id', 'old_flags' ), \r
+ 'old_id IN (' . implode( ',', $oldIdMap ) . ')', $fname );\r
+ while ( $row = $dbr->fetchObject( $res ) ) {\r
+ $flags = $row->old_flags;\r
+ $id = $row->old_id;\r
+\r
+ // Create flagStats row if it doesn't exist\r
+ $flagStats = $flagStats + array( $flags => 0 );\r
+ // Increment counter\r
+ $flagStats[$flags]++;\r
+\r
+ // Not missing\r
+ unset( $missingTextRows[$row->old_id] );\r
+\r
+ // Check for external or object\r
+ if ( $flags == '' ) {\r
+ $flagArray = array();\r
+ } else {\r
+ $flagArray = explode( ',', $flags );\r
+ }\r
+ if ( in_array( 'external', $flagArray ) ) {\r
+ $flagsFields[$id] = $flags; // is this needed?\r
+ $externalRevs[] = $id;\r
+ } elseif ( in_array( 'object', $flagArray ) ) {\r
+ $flagsFields[$id] = $flags; // is this needed?\r
+ $objectRevs[] = $id;\r
+ }\r
+\r
+ // Check for unrecognised flags\r
+ if ( count( array_diff( $flagArray, $knownFlags ) ) ) {\r
+ print_r( array_diff( $flagArray, $knownFlags ) );\r
+ checkError( "Warning: invalid flags field \"$flags\"", $id );\r
+ }\r
+ }\r
+ $dbr->freeResult( $res );\r
+\r
+ // Output errors for any missing text rows\r
+ foreach ( $missingTextRows as $oldId => $revId ) {\r
+ print "Error: missing text row $oldId for revision $revId\n";\r
+ }\r
+\r
+ // Verify external revisions\r
+ $externalConcatBlobs = array();\r
+ $externalNormalBlobs = array();\r
+ if ( count( $externalRevs ) ) {\r
+ $res = $dbr->select( 'text', array( 'old_id', 'old_flags', 'old_text' ), \r
+ array( 'old_id IN (' . implode( ',', $externalRevs ) . ')' ), $fname );\r
+ while ( $row = $dbr->fetchObject( $res ) ) {\r
+ $urlParts = explode( '://', $row->old_text, 2 );\r
+ if ( count( $urlParts ) !== 2 || $urlParts[1] == '' ) {\r
+ checkError( "Error: invalid URL \"{$row->old_text}\"", $row->old_id );\r
+ continue;\r
+ }\r
+ list( $proto, $path ) = $urlParts;\r
+ if ( $proto != 'DB' ) {\r
+ checkError( "Error: invalid external protocol \"$proto\"", $row->old_id );\r
+ continue;\r
+ }\r
+ $path = explode( '/', $row->old_text );\r
+ $cluster = $path[2];\r
+ $id = $path[3];\r
+ if ( isset( $path[4] ) ) {\r
+ $externalConcatBlobs[$cluster][$id][] = $row->old_id;\r
+ } else {\r
+ $externalNormalBlobs[$cluster][$id][] = $row->old_id;\r
+ }\r
+ }\r
+ $dbr->freeResult( $res );\r
+ }\r
+\r
+ // Check external concat blobs for the right header\r
+ checkExternalConcatBlobs( $externalConcatBlobs );\r
+ \r
+\r
+ // Check external normal blobs for existence\r
+ if ( count( $externalNormalBlobs ) ) {\r
+ if ( is_null( $dbStore ) ) {\r
+ $dbStore = new ExternalStoreDB;\r
+ }\r
+ foreach ( $externalConcatBlobs as $cluster => $xBlobIds ) {\r
+ $blobIds = array_keys( $xBlobIds );\r
+ $extDb =& $dbStore->getSlave( $cluster );\r
+ $blobsTable = $dbStore->getTable( $extDb );\r
+ $res = $extDb->select( $blobsTable, \r
+ array( 'blob_id' ), \r
+ array( 'blob_id IN( ' . implode( ',', $blobIds ) . ')' ), $fname );\r
+ while ( $row = $extDb->fetchObject( $res ) ) {\r
+ unset( $xBlobIds[$row->blob_id] );\r
+ }\r
+ $extDb->freeResult( $res );\r
+ // Print errors for missing blobs rows\r
+ foreach ( $xBlobIds as $blobId => $oldId ) {\r
+ checkError( "Error: missing target $blobId for one-part ES URL", $oldId );\r
+ }\r
+ }\r
+ }\r
+\r
+ // Check local objects\r
+ $dbr->ping();\r
+ $concatBlobs = array();\r
+ $curIds = array();\r
+ if ( count( $objectRevs ) ) {\r
+ $headerLength = 300;\r
+ $res = $dbr->select( 'text', array( 'old_id', 'old_flags', "LEFT(old_text, $headerLength) AS header" ), \r
+ array( 'old_id IN (' . implode( ',', $objectRevs ) . ')' ), $fname );\r
+ while ( $row = $dbr->fetchObject( $res ) ) {\r
+ $oldId = $row->old_id;\r
+ if ( !preg_match( '/^O:(\d+):"(\w+)"/', $row->header, $matches ) ) {\r
+ checkError( "Error: invalid object header", $oldId );\r
+ continue;\r
+ }\r
+\r
+ $className = $matches[2];\r
+ if ( strlen( $className ) != $matches[1] ) {\r
+ checkError( "Error: invalid object header, wrong class name length", $oldId );\r
+ continue;\r
+ }\r
+\r
+ $objectStats = $objectStats + array( $className => 0 );\r
+ $objectStats[$className]++;\r
+\r
+ switch ( $className ) {\r
+ case 'concatenatedgziphistoryblob':\r
+ // Good\r
+ break;\r
+ case 'historyblobstub':\r
+ case 'historyblobcurstub':\r
+ if ( strlen( $row->header ) == $headerLength ) {\r
+ checkError( "Error: overlong stub header", $oldId );\r
+ continue;\r
+ }\r
+ $stubObj = unserialize( $row->header );\r
+ if ( !is_object( $stubObj ) ) {\r
+ checkError( "Error: unable to unserialize stub object", $oldId );\r
+ continue;\r
+ }\r
+ if ( $className == 'historyblobstub' ) {\r
+ $concatBlobs[$stubObj->mOldId][] = $oldId;\r
+ } else {\r
+ $curIds[$stubObj->mCurId][] = $oldId;\r
+ }\r
+ break;\r
+ default:\r
+ checkError( "Error: unrecognised object class \"$className\"", $oldId );\r
+ }\r
+ }\r
+ $dbr->freeResult( $res );\r
+ }\r
+\r
+ // Check local concat blob validity\r
+ $externalConcatBlobs = array();\r
+ if ( count( $concatBlobs ) ) {\r
+ $headerLength = 300;\r
+ $res = $dbr->select( 'text', array( 'old_id', 'old_flags', "LEFT(old_text, $headerLength) AS header" ), \r
+ array( 'old_id IN (' . implode( ',', array_keys( $concatBlobs ) ) . ')' ), $fname );\r
+ while ( $row = $dbr->fetchObject( $res ) ) {\r
+ $flags = explode( ',', $row->old_flags );\r
+ if ( in_array( 'external', $flags ) ) {\r
+ // Concat blob is in external storage?\r
+ if ( in_array( 'object', $flags ) ) {\r
+ $urlParts = explode( '/', $row->header );\r
+ if ( $urlParts[0] != 'DB:' ) {\r
+ checkError( "Error: unrecognised external storage type \"{$urlParts[0]}", $row->old_id );\r
+ } else {\r
+ $cluster = $urlParts[2];\r
+ $id = $urlParts[3];\r
+ if ( !isset( $externalConcatBlobs[$cluster][$id] ) ) {\r
+ $externalConcatBlobs[$cluster][$id] = array();\r
+ }\r
+ $externalConcatBlobs[$cluster][$id] = array_merge( \r
+ $externalConcatBlobs[$cluster][$id], $concatBlobs[$row->old_id]\r
+ );\r
+ }\r
+ } else {\r
+ checkError( "Error: invalid flags \"{$row->old_flags}\" on concat bulk row {$row->old_id}",\r
+ $concatBlobs[$row->old_id] );\r
+ }\r
+ } elseif ( substr( $row->header, 0, strlen( CONCAT_HEADER ) ) != CONCAT_HEADER ) {\r
+ checkError( "Error: Incorrect object header for concat bulk row {$row->old_id}", \r
+ $concatBlobs[$row->old_id] );\r
+ } # else good\r
+\r
+ unset( $concatBlobs[$row->old_id] );\r
+ }\r
+ $dbr->freeResult( $res );\r
+ }\r
+\r
+ // Check targets of unresolved stubs\r
+ checkExternalConcatBlobs( $externalConcatBlobs );\r
+ $dbr->ping();\r
+\r
+ // next chunk\r
+ }\r
+\r
+ print "\n\nFlag statistics:\n";\r
+ $total = array_sum( $flagStats );\r
+ foreach ( $flagStats as $flag => $count ) {\r
+ printf( "%-30s %10d %5.2f%%\n", $flag, $count, $count / $total * 100 );\r
+ }\r
+ print "\nObject statistics:\n";\r
+ $total = array_sum( $objectStats );\r
+ foreach ( $objectStats as $className => $count ) {\r
+ printf( "%-30s %10d %5.2f%%\n", $className, $count, $count / $total * 100 );\r
+ }\r
+}\r
+\r
+\r
+function checkExternalConcatBlobs( $externalConcatBlobs ) {\r
+ static $dbStore = null;\r
+ $fname = 'checkExternalConcatBlobs';\r
+ if ( !count( $externalConcatBlobs ) ) {\r
+ return;\r
+ }\r
+\r
+ if ( is_null( $dbStore ) ) {\r
+ $dbStore = new ExternalStoreDB;\r
+ }\r
+ \r
+ foreach ( $externalConcatBlobs as $cluster => $oldIds ) {\r
+ $blobIds = array_keys( $oldIds );\r
+ $extDb =& $dbStore->getSlave( $cluster );\r
+ $blobsTable = $dbStore->getTable( $extDb );\r
+ $headerLength = strlen( CONCAT_HEADER );\r
+ $res = $extDb->select( $blobsTable, \r
+ array( 'blob_id', "LEFT(blob_text, $headerLength) AS header" ), \r
+ array( 'blob_id IN( ' . implode( ',', $blobIds ) . ')' ), $fname );\r
+ while ( $row = $extDb->fetchObject( $res ) ) {\r
+ unset( $oldIds[$row->blob_id] );\r
+ if ( $row->header != CONCAT_HEADER ) {\r
+ checkError( "Error: invalid header on target of two-part ES URL", \r
+ $oldIds[$row->blob_id] );\r
+ }\r
+ }\r
+ $extDb->freeResult( $res );\r
+\r
+ // Print errors for missing blobs rows\r
+ foreach ( $oldIds as $blobId => $oldIds ) {\r
+ checkError( "Error: missing target $blobId for two-part ES URL", $oldIds );\r
+ }\r
+ }\r
+}\r
+\r
+?>\r