Fix for Issue ZipArchive#16

Fixes issues with files being truncated when their headers have values
that represent smaller uncompressed sizes then the actual data. This
generally fixes issues where the whatever program zipped the file
fucked up the headers. The fix makes ssziparchive behave like other
unzip utilities that ignore the header and decompress the actual data
until it's finished.

I've tested a bunch and all of the tests included pass, if any other
issues arrive I'd be glad to take additional time to look into it.

Includes a file, IncorrectHeaders.zip, that is a zip of a folder called
IncorrectHeaders which includes a single file, Readme.txt. I've
intentionally changed the header for that file to read 50 bytes intend
of the actual 59 for testing purposes.

Author: Julius Parishy

Tests/IncorrectHeaders.zip

@@ -18,6 +18,7 @@
 		B215FB6B143AD576003AC546 /* zip.c in Sources */ = {isa = PBXBuildFile; fileRef = B215FB55143AD460003AC546 /* zip.c */; };
 		B215FB6D143AD6FF003AC546 /* TestArchive.zip in Resources */ = {isa = PBXBuildFile; fileRef = B215FB6C143AD6FF003AC546 /* TestArchive.zip */; };
 		B23FCC7F1558F1B70026375C /* TestPasswordArchive.zip in Resources */ = {isa = PBXBuildFile; fileRef = B23FCC7E1558F1B70026375C /* TestPasswordArchive.zip */; };
+		C5AE4E64155A12760045F3ED /* IncorrectHeaders.zip in Resources */ = {isa = PBXBuildFile; fileRef = C5AE4E63155A12760045F3ED /* IncorrectHeaders.zip */; };
 /* End PBXBuildFile section */
 
 /* Begin PBXFileReference section */
@@ -40,6 +41,7 @@
 		B215FB64143AD527003AC546 /* libz.dylib */ = {isa = PBXFileReference; lastKnownFileType = "compiled.mach-o.dylib"; name = libz.dylib; path = usr/lib/libz.dylib; sourceTree = SDKROOT; };
 		B215FB6C143AD6FF003AC546 /* TestArchive.zip */ = {isa = PBXFileReference; lastKnownFileType = archive.zip; path = TestArchive.zip; sourceTree = "<group>"; };
 		B23FCC7E1558F1B70026375C /* TestPasswordArchive.zip */ = {isa = PBXFileReference; lastKnownFileType = archive.zip; path = TestPasswordArchive.zip; sourceTree = "<group>"; };
+		C5AE4E63155A12760045F3ED /* IncorrectHeaders.zip */ = {isa = PBXFileReference; lastKnownFileType = archive.zip; path = IncorrectHeaders.zip; sourceTree = "<group>"; };
 /* End PBXFileReference section */
 
 /* Begin PBXFrameworksBuildPhase section */
@@ -116,6 +118,7 @@
 			children = (
 				B215FB61143AD514003AC546 /* SSZipArchiveTests.m */,
 				B215FB5F143AD514003AC546 /* SSZipArchiveTests-Info.plist */,
+				C5AE4E63155A12760045F3ED /* IncorrectHeaders.zip */,
 				B215FB6C143AD6FF003AC546 /* TestArchive.zip */,
 				B23FCC7E1558F1B70026375C /* TestPasswordArchive.zip */,
 			);
@@ -176,6 +179,7 @@
 			files = (
 				B215FB6D143AD6FF003AC546 /* TestArchive.zip in Resources */,
 				B23FCC7F1558F1B70026375C /* TestPasswordArchive.zip in Resources */,
+				C5AE4E64155A12760045F3ED /* IncorrectHeaders.zip in Resources */,
 			);
 			runOnlyForDeploymentPostprocessing = 0;
 		};

Tests/SSZipArchive.xcodeproj/project.pbxproj

@@ -8,11 +8,14 @@
 
 #import "SSZipArchive.h"
 #import <SenTestingKit/SenTestingKit.h>
+#import <CommonCrypto/CommonDigest.h>
 
 @interface SSZipArchiveTests : SenTestCase <SSZipArchiveDelegate>
 
 - (NSString *)_cachesPath:(NSString *)directory;
 
+- (NSString *)_calculateMD5Digest:(NSData *)data;
+
 @end
 
 @implementation SSZipArchiveTests
@@ -66,6 +69,21 @@ - (void)testUnzippingWithPassword {
 	STAssertTrue([fileManager fileExistsAtPath:testPath], @"LICENSE unzipped");
 }
 
+- (void)testUnzippingTruncatedFileFix {
+    NSString* zipPath = [[NSBundle bundleForClass:[self class]] pathForResource:@"IncorrectHeaders" ofType:@"zip"];
+    NSString* outputPath = [self _cachesPath:@"IncorrectHeaders"];
+    
+    [SSZipArchive unzipFileAtPath:zipPath toDestination:outputPath delegate:self];
+    
+    NSString* intendedReadmeTxtMD5 = @"31ac96301302eb388070c827447290b5";
+    
+    NSString* filePath = [outputPath stringByAppendingPathComponent:@"IncorrectHeaders/Readme.txt"];
+    NSData* data = [NSData dataWithContentsOfFile:filePath];
+    
+    NSString* actualReadmeTxtMD5 = [self _calculateMD5Digest:data];
+    STAssertTrue([actualReadmeTxtMD5 isEqualToString:intendedReadmeTxtMD5], @"Readme.txt MD5 digest should match original.");
+}
+
 
 // Commented out to avoid checking in several gig file into the repository. Simply add a file named
 // `LargeArchive.zip` to the project and uncomment out these lines to test.
@@ -117,4 +135,17 @@ - (NSString *)_cachesPath:(NSString *)directory {
 	return path;
 }
 
+- (NSString *)_calculateMD5Digest:(NSData *)data
+{
+    unsigned char buffer[CC_MD5_DIGEST_LENGTH];
+    CC_MD5([data bytes], [data length], buffer);
+    
+    NSMutableString* digest = [NSMutableString string];
+    
+    for(int i = 0; i < CC_MD5_DIGEST_LENGTH; ++i)
+        [digest appendFormat:@"%02x", buffer[i]];
+    
+    return digest;
+}
+
 @end

Tests/SSZipArchiveTests.m

@@ -1718,10 +1718,24 @@ extern int ZEXPORT unzReadCurrentFile  (unzFile file, voidp buf, unsigned len)
 
     pfile_in_zip_read_info->stream.avail_out = (uInt)len;
 
+    // NOTE:
+    // This bit of code seems to try to set the amount of space in the output buffer based on the
+    // value stored in the headers stored in the .zip file. However, if those values are incorrect
+    // it may result in a loss of data when uncompresssing that file. The compressed data is still
+    // legit and will deflate without knowing the uncompressed code so this tidbit is unnecessary and
+    // may cause issues for some .zip files.
+    //
+    // It's removed in here to fix those issues.
+    //
+    // See: https://github.com/samsoffes/ssziparchive/issues/16
+    //
+    
+    /*
     if ((len>pfile_in_zip_read_info->rest_read_uncompressed) &&
         (!(pfile_in_zip_read_info->raw)))
         pfile_in_zip_read_info->stream.avail_out =
             (uInt)pfile_in_zip_read_info->rest_read_uncompressed;
+     */
 
     if ((len>pfile_in_zip_read_info->rest_read_compressed+
            pfile_in_zip_read_info->stream.avail_in) &&