+global\CheckStorage::check()

Complexity

Conditions	48
Paths	3672

Size

Total Lines	298
Code Lines	203

Duplication

Lines	0
Ratio	0 %

Importance

Changes

0

<?php
	function check( $fix = false, $xml = '' ) {
		$dbr = wfGetDB( DB_REPLICA );
		if ( $fix ) {
			print "Checking, will fix errors if possible...\n";
		} else {
			print "Checking...\n";
		}
		$maxRevId = $dbr->selectField( 'revision', 'MAX(rev_id)', false, __METHOD__ );
		$chunkSize = 1000;
		$flagStats = [];
		$objectStats = [];
		$knownFlags = [ 'external', 'gzip', 'object', 'utf-8' ];
		$this->errors = [
			'restore text' => [],
			'restore revision' => [],
			'unfixable' => [],
			'fixed' => [],
			'fixable' => [],
		];

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

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

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

			// Fetch old_flags
			$missingTextRows = array_flip( $this->oldIdMap );
			$externalRevs = [];
			$objectRevs = [];
			$res = $dbr->select( 'text', [ 'old_id', 'old_flags' ],
				'old_id IN (' . implode( ',', $this->oldIdMap ) . ')', __METHOD__ );
			foreach ( $res as $row ) {
				/**
				 * @var $flags int
				 */
				$flags = $row->old_flags;
				$id = $row->old_id;

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

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

				// Check for external or object
				if ( $flags == '' ) {
					$flagArray = [];
				} 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 = wfGetDB( DB_MASTER );
						$dbw->ping();
						$dbw->update( 'text', [ 'old_flags' => '' ],
							[ 'old_id' => $id ], __METHOD__ );
						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 = [];
			$externalNormalBlobs = [];
			if ( count( $externalRevs ) ) {
				$res = $dbr->select( 'text', [ 'old_id', 'old_flags', 'old_text' ],
					[ 'old_id IN (' . implode( ',', $externalRevs ) . ')' ], __METHOD__ );
				foreach ( $res as $row ) {
					$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, ) = $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,
						[ 'blob_id' ],
						[ 'blob_id IN( ' . implode( ',', $blobIds ) . ')' ], __METHOD__ );
					foreach ( $res as $row ) {
						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 = [];
			$curIds = [];
			if ( count( $objectRevs ) ) {
				$headerLength = 300;
				$res = $dbr->select(
					'text',
					[ 'old_id', 'old_flags', "LEFT(old_text, $headerLength) AS header" ],
					[ 'old_id IN (' . implode( ',', $objectRevs ) . ')' ],
					__METHOD__
				);
				foreach ( $res as $row ) {
					$oldId = $row->old_id;
					$matches = [];
					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 + [ $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:
							$this->error( 'unfixable', "Error: unrecognised object class \"$className\"", $oldId );
					}
				}
				$dbr->freeResult( $res );
			}

			// Check local concat blob validity
			$externalConcatBlobs = [];
			if ( count( $concatBlobs ) ) {
				$headerLength = 300;
				$res = $dbr->select(
					'text',
					[ 'old_id', 'old_flags', "LEFT(old_text, $headerLength) AS header" ],
					[ 'old_id IN (' . implode( ',', array_keys( $concatBlobs ) ) . ')' ],
					__METHOD__
				);
				foreach ( $res as $row ) {
					$flags = explode( ',', $row->old_flags );
					if ( in_array( 'external', $flags ) ) {
						// Concat blob is in external storage?
						if ( in_array( 'object', $flags ) ) {
							$urlParts = explode( '/', $row->header );
							if ( $urlParts[0] != 'DB:' ) {
								$this->error(
									'unfixable',
									"Error: unrecognised external storage type \"{$urlParts[0]}",
									$row->old_id
								);
							} else {
								$cluster = $urlParts[2];
								$id = $urlParts[3];
								if ( !isset( $externalConcatBlobs[$cluster][$id] ) ) {
									$externalConcatBlobs[$cluster][$id] = [];
								}
								$externalConcatBlobs[$cluster][$id] = array_merge(
									$externalConcatBlobs[$cluster][$id], $concatBlobs[$row->old_id]
								);
							}
						} else {
							$this->error(
								'unfixable',
								"Error: invalid flags \"{$row->old_flags}\" on concat bulk row {$row->old_id}",
								$concatBlobs[$row->old_id] );
						}
					} elseif ( strcasecmp(
						substr( $row->header, 0, strlen( self::CONCAT_HEADER ) ),
						self::CONCAT_HEADER
					) ) {
						$this->error(
							'restore text',
							"Error: Incorrect object header for concat bulk row {$row->old_id}",
							$concatBlobs[$row->old_id]
						);
					} # else good

					unset( $concatBlobs[$row->old_id] );
				}
				$dbr->freeResult( $res );
			}

			// Check targets of unresolved stubs
			$this->checkExternalConcatBlobs( $externalConcatBlobs );
			// next chunk
		}

		print "\n\nErrors:\n";
		foreach ( $this->errors as $name => $errors ) {
			if ( count( $errors ) ) {
				$description = $this->errorDescriptions[$name];
				echo "$description: " . implode( ',', array_keys( $errors ) ) . "\n";
			}
		}

		if ( count( $this->errors['restore text'] ) && $fix ) {
			if ( (string)$xml !== '' ) {
				$this->restoreText( array_keys( $this->errors['restore text'] ), $xml );
			} else {
				echo "Can't fix text, no XML backup specified\n";
			}
		}

		print "\nFlag statistics:\n";
		$total = array_sum( $flagStats );
		foreach ( $flagStats as $flag => $count ) {
			printf( "%-30s %10d %5.2f%%\n", $flag, $count, $count / $total * 100 );
		}
		print "\nLocal object statistics:\n";
		$total = array_sum( $objectStats );
		foreach ( $objectStats as $className => $count ) {
			printf( "%-30s %10d %5.2f%%\n", $className, $count, $count / $total * 100 );
		}
	}