Add fine-grained filename encoding

* Add entry encoding property based on the EFS flag of the general purpose bit flags. * Add entry fileNameWithEncoding: methods to force a particular entry-level encoding for the file name. * Remove archive-level encoding for file name. * Default encoding is now CP-437, as per the Zip file format specification. This is a breaking change since previous default encoding was UTF-8. * Update unit tests. * Update podspec to 8.1.
pixelglow · Nov 7, 2015 · fe86f85 · fe86f85
1 parent 3e345d6
commit fe86f85
Show file tree

Hide file tree

Showing 14 changed files with 76 additions and 45 deletions.
diff --git a/ZipZap/ZZArchive.mm b/ZipZap/ZZArchive.mm
@@ -35,7 +35,6 @@ - (BOOL)loadCanMiss:(BOOL)canMiss error:(out NSError**)error;
 @implementation ZZArchive
 {
 	id<ZZChannel> _channel;
-	NSStringEncoding _encoding;
 }
 
 + (instancetype)archiveWithURL:(NSURL*)URL
@@ -80,9 +79,6 @@ - (instancetype)initWithChannel:(id<ZZChannel>)channel
 	{
 		_channel = channel;
 
-		NSNumber* encoding = options[ZZOpenOptionsEncodingKey];
-		_encoding = encoding ? encoding.unsignedIntegerValue : NSUTF8StringEncoding;
-
 		NSNumber* createIfMissing = options[ZZOpenOptionsCreateIfMissingKey];
 		if (![self loadCanMiss:createIfMissing.boolValue error:error])
 			return nil;
@@ -164,8 +160,7 @@ - (BOOL)loadCanMiss:(BOOL)canMiss error:(out NSError**)error
 																	  + nextCentralFileHeader->relativeOffsetOfLocalHeader);
 
 		[entries addObject:[[ZZOldArchiveEntry alloc] initWithCentralFileHeader:nextCentralFileHeader
-																localFileHeader:nextLocalFileHeader
-																	   encoding:_encoding]];
+																localFileHeader:nextLocalFileHeader]];
 
 		nextCentralFileHeader = nextCentralFileHeader->nextCentralFileHeader();
 	}

diff --git a/ZipZap/ZZArchiveEntry.h b/ZipZap/ZZArchiveEntry.h
@@ -83,7 +83,7 @@ NS_ASSUME_NONNULL_BEGIN
 @property (readonly, nonatomic) mode_t fileMode;
 
 /**
- * The file name of the entry.
+ * The file name of the entry. Encoding is either CP-437 or UTF-8, depending on the EFS flag of the general purpose bit flags.
  */
 @property (readonly, nonatomic) NSString* fileName;
 
@@ -92,6 +92,11 @@ NS_ASSUME_NONNULL_BEGIN
  */
 @property (readonly, nonatomic) NSData* rawFileName;
 
+/**
+ * The encoding of filename and comment fields, according to the EFS flag of the general purpose bit flags.
+ */
+@property (readonly, nonatomic) NSStringEncoding encoding;
+
 /**
  * Creates a new file entry from a streaming callback.
  *
@@ -169,6 +174,13 @@ NS_ASSUME_NONNULL_BEGIN
  */
 - (BOOL)check:(out NSError**)error;
 
+/**
+ * The file name of the entry with the given encoding.
+ *
+ * @param encoding The encoding for the filename.
+ */
+- (NSString*)fileNameWithEncoding:(NSStringEncoding)encoding;
+
 /**
  * Creates a stream to represent the entry file.
  *

diff --git a/ZipZap/ZZArchiveEntry.m b/ZipZap/ZZArchiveEntry.m
@@ -117,14 +117,19 @@ - (mode_t)fileMode
 
 - (NSString*)fileName
 {
-	return nil;
+	return [self fileNameWithEncoding:self.encoding];
 }
 
 - (NSData*)rawFileName
 {
 	return nil;
 }
 
+- (NSStringEncoding)encoding
+{
+	return (NSStringEncoding)0;
+}
+
 - (NSInputStream*)newStreamWithError:(NSError**)error
 {
 	return [self newStreamWithPassword:nil error:error];
@@ -140,6 +145,11 @@ - (BOOL)check:(NSError**)error
 	return YES;
 }
 
+- (NSString*)fileNameWithEncoding:(NSStringEncoding)encoding
+{
+	return nil;
+}
+
 - (NSData*)newDataWithError:(NSError**)error
 {
 	return [self newDataWithPassword:nil error:error];

diff --git a/ZipZap/ZZConstants.h b/ZipZap/ZZConstants.h
@@ -73,11 +73,6 @@ typedef NS_ENUM(uint8_t, ZZAESEncryptionStrength)
 	ZZAESEncryptionStrength256 = 0x03
 };
 
-/**
- * An NSNumber object that contains the string encoding to use for entry file names and comments. Default is NSUTF8StringEncoding.
- */
-extern NSString* const ZZOpenOptionsEncodingKey;
-
 /**
  * An NSNumber object that determines whether to create the archive file if it is missing. Creation occurs during -[ZZArchive updateEntries:error:]. Default is @NO.
  */

diff --git a/ZipZap/ZZConstants.m b/ZipZap/ZZConstants.m
@@ -8,6 +8,4 @@
 
 #import "ZZConstants.h"
 
-NSString* const ZZOpenOptionsEncodingKey = @"ZZOpenOptionsEncodingKey";
-
 NSString* const ZZOpenOptionsCreateIfMissingKey = @"ZZOpenOptionsCreateIfMissingKey";
diff --git a/ZipZap/ZZHeaders.h b/ZipZap/ZZHeaders.h
@@ -42,7 +42,7 @@ enum class ZZGeneralPurposeBitFlag: uint16_t
 	superFastCompression = (1 << 1) | (1 << 2),
 	sizeInDataDescriptor = 1 << 3,
 	encryptionStrong = 1 << 6,
-	fileNameUTF8Encoded = 1 << 11
+	languageEncoding = 1 << 11
 };
 
 inline ZZGeneralPurposeBitFlag operator|(ZZGeneralPurposeBitFlag lhs, ZZGeneralPurposeBitFlag rhs)

diff --git a/ZipZap/ZZNewArchiveEntry.h b/ZipZap/ZZNewArchiveEntry.h
@@ -23,7 +23,8 @@ NS_ASSUME_NONNULL_BEGIN
 @property (readonly, nonatomic) BOOL compressed;
 @property (readonly, nonatomic) NSDate* lastModified;
 @property (readonly, nonatomic) mode_t fileMode;
-@property (readonly, nonatomic) NSString* fileName;
+@property (readonly, nonatomic) NSData* rawFileName;
+@property (readonly, nonatomic) NSStringEncoding encoding;
 
 - (instancetype)init NS_UNAVAILABLE;
 
@@ -35,6 +36,8 @@ NS_ASSUME_NONNULL_BEGIN
 					 streamBlock:(nullable BOOL(^)(NSOutputStream* stream, NSError** error))streamBlock
 			   dataConsumerBlock:(nullable BOOL(^)(CGDataConsumerRef dataConsumer, NSError** error))dataConsumerBlock NS_DESIGNATED_INITIALIZER;
 
+- (NSString*)fileNameWithEncoding:(NSStringEncoding)encoding;
+
 @end
 
 NS_ASSUME_NONNULL_END
diff --git a/ZipZap/ZZNewArchiveEntry.mm b/ZipZap/ZZNewArchiveEntry.mm
@@ -56,9 +56,14 @@ - (mode_t)fileMode
 	return _fileMode;
 }
 
-- (NSString*)fileName
+- (NSData*)rawFileName
 {
-	return _fileName;
+	return [_fileName dataUsingEncoding:NSUTF8StringEncoding];
+}
+
+- (NSStringEncoding)encoding
+{
+	return NSUTF8StringEncoding;
 }
 
 - (id<ZZArchiveEntryWriter>)newWriterCanSkipLocalFile:(BOOL)canSkipLocalFile
@@ -72,5 +77,10 @@ - (NSString*)fileName
 									   dataConsumerBlock:_dataConsumerBlock];
 }
 
+- (NSString*)fileNameWithEncoding:(NSStringEncoding)encoding
+{
+	return _fileName;
+}
+
 @end
 
diff --git a/ZipZap/ZZNewArchiveEntryWriter.mm b/ZipZap/ZZNewArchiveEntryWriter.mm
@@ -102,7 +102,7 @@ - (instancetype)initWithFileName:(NSString*)fileName
 				compressionFlag = ZZGeneralPurposeBitFlag::maximumCompression;
 				break;
 		}
-		centralFileHeader->generalPurposeBitFlag = localFileHeader->generalPurposeBitFlag = compressionFlag | ZZGeneralPurposeBitFlag::sizeInDataDescriptor | ZZGeneralPurposeBitFlag::fileNameUTF8Encoded;
+		centralFileHeader->generalPurposeBitFlag = localFileHeader->generalPurposeBitFlag = compressionFlag | ZZGeneralPurposeBitFlag::sizeInDataDescriptor | ZZGeneralPurposeBitFlag::languageEncoding;
 
 		centralFileHeader->compressionMethod = localFileHeader->compressionMethod = compressionLevel ? ZZCompressionMethod::deflated : ZZCompressionMethod::stored;
 

diff --git a/ZipZap/ZZOldArchiveEntry.h b/ZipZap/ZZOldArchiveEntry.h
@@ -29,14 +29,15 @@ NS_ASSUME_NONNULL_BEGIN
 @property (readonly, nonatomic) NSUInteger compressedSize;
 @property (readonly, nonatomic) NSUInteger uncompressedSize;
 @property (readonly, nonatomic) mode_t fileMode;
-@property (readonly, nonatomic) NSString* fileName;
 @property (readonly, nonatomic) NSData* rawFileName;
+@property (readonly, nonatomic) NSStringEncoding encoding;
 
 - (instancetype)init NS_UNAVAILABLE;
 
 - (instancetype)initWithCentralFileHeader:(struct ZZCentralFileHeader*)centralFileHeader
-						  localFileHeader:(struct ZZLocalFileHeader*)localFileHeader
-								 encoding:(NSStringEncoding)encoding NS_DESIGNATED_INITIALIZER;
+						  localFileHeader:(struct ZZLocalFileHeader*)localFileHeader NS_DESIGNATED_INITIALIZER;
+
+- (NSString*)fileNameWithEncoding:(NSStringEncoding)encoding;
 
 @end
 

diff --git a/ZipZap/ZZOldArchiveEntry.mm b/ZipZap/ZZOldArchiveEntry.mm
@@ -24,7 +24,6 @@
 @interface ZZOldArchiveEntry ()
 
 - (NSData*)fileData;
-- (NSString*)stringWithBytes:(uint8_t*)bytes length:(NSUInteger)length;
 
 - (BOOL)checkEncryptionAndCompression:(out NSError**)error;
 - (NSInputStream*)streamForData:(NSData*)data withPassword:(NSString*)password error:(out NSError**)error;
@@ -35,19 +34,16 @@ @implementation ZZOldArchiveEntry
 {
 	ZZCentralFileHeader* _centralFileHeader;
 	ZZLocalFileHeader* _localFileHeader;
-	NSStringEncoding _encoding;
 	ZZEncryptionMode _encryptionMode;
 }
 
 - (instancetype)initWithCentralFileHeader:(struct ZZCentralFileHeader*)centralFileHeader
 						  localFileHeader:(struct ZZLocalFileHeader*)localFileHeader
-								 encoding:(NSStringEncoding)encoding
 {
 	if ((self = [super init]))
 	{
 		_centralFileHeader = centralFileHeader;
 		_localFileHeader = localFileHeader;
-		_encoding = encoding;
 
 		if ((_centralFileHeader->generalPurposeBitFlag & ZZGeneralPurposeBitFlag::encrypted) != ZZGeneralPurposeBitFlag::none)
 		{
@@ -91,14 +87,6 @@ - (NSData*)fileData
 	return [NSData dataWithBytesNoCopy:dataStart length:dataLength freeWhenDone:NO];
 }
 
-- (NSString*)stringWithBytes:(uint8_t*)bytes length:(NSUInteger)length
-{
-	// if EFS bit is set, use UTF-8; otherwise use fallback encoding
-	return [[NSString alloc] initWithBytes:bytes
-									length:length
-								  encoding:(_centralFileHeader->generalPurposeBitFlag & ZZGeneralPurposeBitFlag::fileNameUTF8Encoded) == ZZGeneralPurposeBitFlag::none ? _encoding :NSUTF8StringEncoding];
-}
-
 - (ZZCompressionMethod)compressionMethod
 {
     if (_encryptionMode == ZZEncryptionModeWinZipAES)
@@ -169,15 +157,17 @@ - (mode_t)fileMode
 	}
 }
 
-- (NSString*)fileName
+- (NSData*)rawFileName
 {
-	return [self stringWithBytes:_centralFileHeader->fileName()
+	return [NSData dataWithBytes:_centralFileHeader->fileName()
 						  length:_centralFileHeader->fileNameLength];
 }
 
-- (NSData*)rawFileName
+- (NSStringEncoding)encoding
 {
-	return [NSData dataWithBytes:_centralFileHeader->fileName() length:_centralFileHeader->fileNameLength];
+	return (_centralFileHeader->generalPurposeBitFlag & ZZGeneralPurposeBitFlag::languageEncoding) == ZZGeneralPurposeBitFlag::none ?
+		CFStringConvertEncodingToNSStringEncoding(kCFStringEncodingDOSLatinUS) : // CP-437
+		NSUTF8StringEncoding; // UTF-8
 }
 
 - (BOOL)check:(out NSError**)error
@@ -258,6 +248,13 @@ - (BOOL)check:(out NSError**)error
 	return YES;
 }
 
+- (NSString*)fileNameWithEncoding:(NSStringEncoding)encoding
+{
+	return [[NSString alloc] initWithBytes:_centralFileHeader->fileName()
+									length:_centralFileHeader->fileNameLength
+								  encoding:encoding];
+}
+
 - (BOOL)checkEncryptionAndCompression:(out NSError**)error
 {
 	switch (_encryptionMode)

diff --git a/ZipZapTests/ZZUnzipTests.m b/ZipZapTests/ZZUnzipTests.m
@@ -138,11 +138,16 @@ - (void)testZipEntryConsistentWithOriginalFile
 		ZZArchiveEntry* nextEntry = _zipFile.entries[index];
 		NSString* nextEntryFilePath = _entryFilePaths[index];
 
-		XCTAssertEqualObjects(nextEntry.fileName,
+		XCTAssertEqualObjects([nextEntry fileNameWithEncoding:NSUTF8StringEncoding],
 							 nextEntryFilePath,
 							 @"zipFile.entries[%lu].fileName must match the original file name.",
 							 (unsigned long)index);
-
+
+		XCTAssertEqualObjects([nextEntry rawFileName],
+							  [nextEntryFilePath dataUsingEncoding:NSUTF8StringEncoding],
+							  @"zipFile.entries[%lu].rawFileName must match the original raw file name.",
+							  (unsigned long)index);
+
 		NSData* fileData = [self dataAtFilePath:nextEntryFilePath];
 		XCTAssertEqual(nextEntry.crc32,
 					   crc32(0, (const Bytef*)fileData.bytes, (uInt)fileData.length),

diff --git a/ZipZapTests/ZZZipTests.m b/ZipZapTests/ZZZipTests.m
@@ -57,7 +57,8 @@ - (NSArray*)recordsForZipEntries:(NSArray*)zipEntries
 		 @"fileMode": @(zipEntry.fileMode),
 		 @"compressed": @(zipEntry.compressed),
 		 @"lastModified": zipEntry.lastModified,
-		 @"fileName": zipEntry.fileName
+		 @"fileName": [zipEntry fileNameWithEncoding:NSUTF8StringEncoding],
+		 @"rawFileName": zipEntry.rawFileName
 		 }];
 	return records;
 }
@@ -124,7 +125,11 @@ - (void)checkZipEntryRecords:(NSArray*)newEntryRecords
 							 nextNewEntry[@"fileName"],
 							 @"Zip entry #%lu file name must match new entry file name.",
 							 (unsigned long)index);
-
+
+		XCTAssertEqualObjects([nextZipInfo[8] dataUsingEncoding:NSUTF8StringEncoding],
+							  nextNewEntry[@"rawFileName"],
+							  @"Zip entry #%lu raw file name must match new entry raw file name.",
+							  (unsigned long)index);
 
 		if (checkerBlock)
 		{

diff --git a/zipzap.podspec.json b/zipzap.podspec.json
@@ -1,14 +1,14 @@
 {
   "name": "zipzap",
-  "version": "8.0.6",
+  "version": "8.1",
   "authors": {
     "Pixelglow Software": "[email protected]"
   },
   "license": "BSD",
   "homepage": "https://github.com/pixelglow/ZipZap",
   "source": {
     "git": "https://github.com/pixelglow/ZipZap.git",
-    "tag": "8.0.6"
+    "tag": "8.1"
   },
   "summary": "ZipZap is a zip file I/O library for Mac OS X and iOS.",
   "description": "The zip file is an ideal container for compound Objective-C documents. Zip files are widely used and well understood. You can randomly access their parts. The format compresses decently and has extensive operating system and tool support. So we want to make this format an even easier choice for you. Thus, the library features:\n\n- Easy-to-use interface: The public API offers just three classes! Yet you can look through zip files using familiar NSArray collections and properties. And you can zip, unzip and rezip zip files through familiar NSData, NSStream and Image I/O classes.\n- Efficient implementation: We've optimized zip file reading and writing to reduce virtual memory pressure and disk file thrashing. Depending on how your compound document is organized, updating a single entry can be faster than writing the same data to a separate file.\n- File format compatibility: Since ZipZap closely follows the zip file format specification, it is works with most Mac, Linux and Windows zip tools.\n",
Original file line number	Diff line number	Diff line change
Expand Up		@@ -8,6 +8,4 @@

		#import "ZZConstants.h"

		NSString* const ZZOpenOptionsEncodingKey = @"ZZOpenOptionsEncodingKey";

		NSString* const ZZOpenOptionsCreateIfMissingKey = @"ZZOpenOptionsCreateIfMissingKey";