In addition to Gzip compression support when exporting SQL dumps add the ability to use Bzip2 compression.

Other changes include:

+ Enable the use of export compression (Gzip and Bzip2) for all export formats.
+ Move the compression options in the export dialog to the 'Advanced' export settings view.
+ Simplify the setting of common exporter properties (e.g. the connection, use of compression).
+ Fix a potential memory leak in the dot exporter.
+ Update the data importer to recognise Bzip2 compressed files.
+ Fix several display issues on export dialog.
+ Restore the default .csv file extension of CSV exports.
+ Correctly update the default export filename when selecting a output compression type.

The addition of Bzip2 compression support implements issue #688.

1 files changed, 177 insertions, 62 deletions

diff --git a/Source/SPFileHandle.m b/Source/SPFileHandle.m
index bceae2aa..ac812ac7 100644
--- a/Source/SPFileHandle.m
+++ b/Source/SPFileHandle.m
@@ -25,6 +25,7 @@
 
 #import "SPFileHandle.h"
 #import "zlib.1.2.4.h"
+#import "bzlib.h"
 #import "pthread.h"
 
 // Define the maximum size of the background write buffer before the writing thread
@@ -32,9 +33,10 @@
 #define SPFH_MAX_WRITE_BUFFER_SIZE 1048576
 
 @interface SPFileHandle (PrivateAPI)
-- (void) _writeBufferToData;
-@end
 
+- (void)_writeBufferToData;
+
+@end
 
 @implementation SPFileHandle
 
@@ -42,13 +44,16 @@
 #pragma mark Setup and teardown
 
 /**
- * Initialises and returns a SPFileHandle with a specified file (FILE or gzFile).
+ * Initialises and returns a SPFileHandle with a specified file (FILE, gzFile or BZFILE).
  * "mode" indicates the file interaction mode - currently only read-only
  * or write-only are supported.
- * On reading, theFile should always be a gzFile; on writing, theFile is a FILE
- * when compression is disabled, or a gzFile when enbled.
+ *
+ * On reading, theFile can either be one of FILE, gzFile or BZFILE depending on the attempt to
+ * determine whether or not the file is in a compressed format (gzip or bzip2). On writing, 
+ * theFile is a FILE when compression is disabled, a gzFile when gzip compression is enabled
+ * or a BZFILE when bzip2 compression is enabled.
  */
-- (id) initWithFile:(void *)theFile fromPath:(const char *)path mode:(int)mode
+- (id)initWithFile:(void *)theFile fromPath:(const char *)path mode:(int)mode
 {
 	if (self = [super init]) {
 		dataWritten = NO;
@@ -61,26 +66,77 @@
 
 		// Check and set the mode
 		fileMode = mode;
+		
 		if (fileMode != O_RDONLY && fileMode != O_WRONLY) {
 			[NSException raise:NSInvalidArgumentException format:@"SPFileHandle only supports read-only and write-only file modes"];
 		}
 
 		// Instantiate the buffer
 		pthread_mutex_init(&bufferLock, NULL);
+		
 		buffer = [[NSMutableData alloc] init];
 		bufferDataLength = 0;
 		bufferPosition = 0;
 		endOfFile = NO;
+		
+		useCompression = NO;
+		compressionFormat = SPNoCompression;
 
 		// If in read mode, set up the buffer
 		if (fileMode == O_RDONLY) {
-			gzbuffer(wrappedFile, 131072);
-			useGzip = !gzdirect(wrappedFile);
+			
+			int i, c;
+			char *bzbuf = malloc(4);
+			const char *charFileMode = (fileMode == O_WRONLY) ? "wb" : "rb";
+			
+			BZFILE *bzfile;
+			gzFile *gzfile = gzopen(path, charFileMode);
+						
+			// Set gzip buffer
+			gzbuffer(gzfile, 131072);
+			
+			// Get the first 3 bytes from the file
+			for (i = 0; (c = getc(wrappedFile)) != EOF && i < 4; bzbuf[i++] = c);
+			
+			// Test to see if the file is gzip compressed
+			BOOL isGzip = (!gzdirect(gzfile));
+			
+			// Test to see if the first 2 bytes extracted from the file match the Bzip2 signature/magic number
+			// (BZ). The 3rd byte should be either 'h' (Huffman encoding) or 0 (Bzip1 - deprecated) to 
+			// indicate the Bzip version. Finally, the 4th byte should be a number between 1 and 9 that indicates
+			// the block size used.
+			BOOL isBzip2 = ((bzbuf[0] == 'B')   && (bzbuf[1] == 'Z')  && 
+							((bzbuf[2] == 'h')  || (bzbuf[2] == '0')) &&
+							((bzbuf[3] >= 0x31) && (bzbuf[3] <= 0x39)));
+			
+			free(bzbuf);
+			
+			if (isBzip2) bzfile = BZ2_bzopen(path, charFileMode);
+						
+			useCompression = (isGzip || isBzip2);
+									
+			if (useCompression) {
+				if (isGzip) {
+					compressionFormat = SPGzipCompression;
+					wrappedFile = gzfile;
+				}
+				else if (isBzip2) {
+					compressionFormat = SPBzip2Compression;
+					wrappedFile = bzfile;
+					gzclose(gzfile);
+				}
+				
+				fclose(theFile);
+			}
+			else {
+				gzclose(gzfile);
+			}
+			
 			processingThread = nil;
-
+		} 
 		// In write mode, set up a thread to handle writing in the background
-		} else if (fileMode == O_WRONLY) {
-			useGzip = NO;
+		else if (fileMode == O_WRONLY) {
+			useCompression = NO;
 			processingThread = [[NSThread alloc] initWithTarget:self selector:@selector(_writeBufferToData) object:nil];
 			[processingThread start];
 		}
@@ -89,7 +145,10 @@
 	return self;
 }
 
-- (void) dealloc
+/**
+ * Dealloc.
+ */
+- (void)dealloc
 {
 	[self closeFile];
 	if (processingThread) [processingThread release];
@@ -106,7 +165,7 @@
  * Retrieve and return a SPFileHandle for reading a file at the supplied
  * path.  Returns nil if the file could not be found or opened.
  */
-+ (id) fileHandleForReadingAtPath:(NSString *)path
++ (id)fileHandleForReadingAtPath:(NSString *)path
 {
 	return [self fileHandleForPath:path mode:O_RDONLY];
 }
@@ -115,57 +174,63 @@
  * Retrieve and return a SPFileHandle for writing a file at the supplied
  * path.  Returns nil if the file could not be found or opened.
  */
-+ (id) fileHandleForWritingAtPath:(NSString *)path
++ (id)fileHandleForWritingAtPath:(NSString *)path
 {
 	return [self fileHandleForPath:path mode:O_WRONLY];
 }
 
 /**
  * Retrieve and return a SPFileHandle for a file at the specified path,
- * using the supplied file status flag.  Returns nil if the file could
+ * using the supplied file status flag. Returns nil if the file could
  * not be found or opened.
  */
-+ (id) fileHandleForPath:(NSString *)path mode:(int)mode
++ (id)fileHandleForPath:(NSString *)path mode:(int)mode
 {
-
 	// Retrieves the path in a filesystem-appropriate format and encoding
 	const char *pathRepresentation = [path fileSystemRepresentation];
 	if (!pathRepresentation) return nil;
 
 	// Open the file to get a file descriptor, returning on failure
-	FILE *theFile;
-	const char *theMode;
-	if (mode == O_WRONLY) {
-		theMode = "wb";
-		theFile = fopen(pathRepresentation, theMode);
-	} else {
-		theMode = "rb";
-		theFile = gzopen(pathRepresentation, theMode);
-	}
-	if (theFile == NULL) return nil;
+	const char *theMode = (mode == O_WRONLY) ? "wb" : "rb";
+	
+	FILE *file = fopen(pathRepresentation, theMode);
+	
+	if (file == NULL) return nil;
 
 	// Return an autoreleased file handle
-	return [[[self alloc] initWithFile:theFile fromPath:pathRepresentation mode:mode] autorelease];
+	return [[[self alloc] initWithFile:file fromPath:pathRepresentation mode:mode] autorelease];
 }
 
-
 #pragma mark -
 #pragma mark Data reading
 
 /**
  * Reads data up to a specified number of uncompressed bytes from the file.
  */
-- (NSMutableData *) readDataOfLength:(NSUInteger)length
-{
+- (NSMutableData *)readDataOfLength:(NSUInteger)length
+{	
+	long theDataLength;
 	void *theData = malloc(length);
-	long theDataLength = gzread(wrappedFile, theData, length);
+			
+	if (useCompression) {
+		if (compressionFormat == SPGzipCompression) {
+			theDataLength = gzread(wrappedFile, theData, length);
+		}
+		else if (compressionFormat == SPBzip2Compression) {
+			theDataLength = BZ2_bzread(wrappedFile, theData, length);
+		}		
+	}
+	else {
+		theDataLength = fread(theData, 1, length, wrappedFile);
+	}
+		
 	return [NSMutableData dataWithBytesNoCopy:theData length:theDataLength freeWhenDone:YES];
 }
 
 /**
  * Returns all the data to the end of the file.
  */
-- (NSMutableData *) readDataToEndOfFile
+- (NSMutableData *)readDataToEndOfFile
 {
 	return [self readDataOfLength:NSUIntegerMax];
 }
@@ -175,10 +240,16 @@
  * This includes any compression headers within the data, and can be used
  * for progress bars when processing files.
  */
-- (NSUInteger) realDataReadLength
+- (NSUInteger)realDataReadLength
 {
-	if (fileMode == O_WRONLY) return 0;
-	return gzoffset(wrappedFile);
+	if ((fileMode == O_WRONLY) || (compressionFormat == SPBzip2Compression)) return 0;
+	
+	if (useCompression && (compressionFormat == SPGzipCompression)) {
+		return gzoffset(wrappedFile);
+	}
+	else {
+		return ftell(wrappedFile);
+	}
 }
 
 #pragma mark -
@@ -189,31 +260,49 @@
  * to NO on a fresh object. If this is called after data has been
  * written, an exception is thrown.
  */
-- (void) setShouldWriteWithGzipCompression:(BOOL)shouldUseGzip
+- (void)setShouldWriteWithCompressionFormat:(SPFileCompressionFormat)useCompressionFormat
 {
-	if (shouldUseGzip == useGzip) return;
+	if (compressionFormat == useCompressionFormat) return;
 
-	if (dataWritten) [NSException raise:NSInternalInconsistencyException format:@"Cannot change compression settings when data has already been written"];
-
-	if (shouldUseGzip) {
+	// Regardless of the supplied argument, close the current file according to how it was previously opened
+	if (useCompression) {
+		if (compressionFormat == SPGzipCompression) {
+			gzclose(wrappedFile);
+		}
+		else if (compressionFormat == SPBzip2Compression) {
+			BZ2_bzclose(wrappedFile);
+		}
+	}
+	else {
 		fclose(wrappedFile);
-		wrappedFile = gzopen(wrappedFilePath, "wb");
-		gzbuffer(wrappedFile, 131072);
-	} else {
-		gzclose(wrappedFile);
+	}
+	
+	if (dataWritten) [NSException raise:NSInternalInconsistencyException format:@"Cannot change compression settings when data has already been written."];
+
+	useCompression = ((useCompressionFormat == SPGzipCompression) || (useCompressionFormat == SPBzip2Compression));
+	
+	compressionFormat = useCompressionFormat;
+	
+	if (useCompression) {		
+		if (compressionFormat == SPGzipCompression) {
+			wrappedFile = gzopen(wrappedFilePath, "wb");
+			gzbuffer(wrappedFile, 131072);
+		}
+		else if (compressionFormat == SPBzip2Compression) {
+			wrappedFile = BZ2_bzopen(wrappedFilePath, "wb");
+		}
+	} 
+	else {
 		wrappedFile = fopen(wrappedFilePath, "wb");
 	}
-	useGzip = shouldUseGzip;
 }
 
-
 /**
  * Write the supplied data to the file.  The data may not be written to the
  * disk at once (see synchronizeFile).
  */
-- (void) writeData:(NSData *)data
+- (void)writeData:(NSData *)data
 {
-
 	// Throw an exception if the file is closed
 	if (fileIsClosed) [NSException raise:NSInternalInconsistencyException format:@"Cannot write to a file handle after it has been closed"];
 
@@ -237,7 +326,7 @@
 /**
  * Blocks until all data has been written to disk.
  */
-- (void) synchronizeFile
+- (void)synchronizeFile
 {
 	pthread_mutex_lock(&bufferLock);
 	while (!allDataWritten) {
@@ -252,35 +341,51 @@
  * Ensure all data is written out, close any file handles, and prevent any
  * more data from being written to the file.
  */
-- (void) closeFile
+- (void)closeFile
 {
 	if (!fileIsClosed) {
 		[self synchronizeFile];
-		if (useGzip || fileMode == O_RDONLY) {
-			gzclose(wrappedFile);
-		} else {
+		
+		if (useCompression) {
+			if (compressionFormat == SPGzipCompression) {
+				gzclose(wrappedFile);
+			}
+			else if (compressionFormat == SPBzip2Compression) {
+				BZ2_bzclose(wrappedFile);
+			}
+		} 
+		else {
 			fclose(wrappedFile);
 		}
+		
 		if (processingThread) {
 			if ([processingThread isExecuting]) {
 				[processingThread cancel];
 				while ([processingThread isExecuting]) usleep(100);
 			}
 		}
+		
 		fileIsClosed = YES;
 	}
 }
 
-
 #pragma mark -
 #pragma mark File information
 
 /**
- * Returns whether gzip compression is enabled on the file.
+ * Returns whether compression is enabled on the file.
  */
-- (BOOL) isCompressed
+- (BOOL)isCompressed
 {
-	return useGzip;
+	return useCompression;
+}
+
+/**
+ * Returns the compression format being used. Currently gzip or bzip2 only.
+ */
+- (SPFileCompressionFormat)compressionFormat
+{
+	return compressionFormat;
 }
 
 @end
@@ -292,7 +397,7 @@
  * up in a buffer and write to disk in chunks as the buffer fills.  This allows
  * background compression of the data when using Gzip compression.
  */
-- (void) _writeBufferToData
+- (void)_writeBufferToData
 {
 	NSAutoreleasePool *writePool = [[NSAutoreleasePool alloc] init];
 
@@ -315,9 +420,19 @@
 
 		// Write out the data
 		long bufferLengthWrittenOut;
-		if (useGzip) {
-			bufferLengthWrittenOut = gzwrite(wrappedFile, [dataToBeWritten bytes], [dataToBeWritten length]);
-		} else {
+				
+		if (useCompression) {
+			switch (compressionFormat) 
+			{
+				case SPGzipCompression:
+					bufferLengthWrittenOut = gzwrite(wrappedFile, [dataToBeWritten bytes], [dataToBeWritten length]);
+					break;
+				case SPBzip2Compression:
+					bufferLengthWrittenOut = BZ2_bzwrite(wrappedFile, [dataToBeWritten bytes], [dataToBeWritten length]);
+					break;
+			}
+		} 
+		else {
 			bufferLengthWrittenOut = fwrite([dataToBeWritten bytes], 1, [dataToBeWritten length], wrappedFile);
 		}