Get/set the NSString text encoding of a file using xattr.

XAttrEncoding.m

//
//  NSString+FileTextEncodingAttribute.h
//  Adapted from http://github.com/scrod/nv/blob/master/NSString_NV.m
//

#import <Foundation/Foundation.h>
#include <sys/xattr.h>

@interface NSString (FileTextEncodingAttribute)

// See: http://stackoverflow.com/questions/3534729/nsutf8stringencoding-causes-file-to-render-improperly-in-textedit-pages-but-not
// Fixes as suggested by https://gist.github.com/mwaterfall/543667#comment-1286524

+ (BOOL)setTextEncodingAttribute:(NSStringEncoding)encoding atPath:(NSString *)filePath;
+ (NSStringEncoding)textEncodingAttributeOfPath:(NSString *)filePath;

@end

@implementation NSString (FileTextEncodingAttribute)

+ (BOOL)setTextEncodingAttribute:(NSStringEncoding)encoding atPath:(NSString *)filePath {
	
	// Get C string
	if (!filePath) return NO;
	const char* path = [filePath fileSystemRepresentation];
	
	// Get correct encoding
	CFStringEncoding cfStringEncoding = CFStringConvertNSStringEncodingToEncoding(encoding);
	if (cfStringEncoding == kCFStringEncodingInvalidId) {
		NSLog(@"%s: encoding %lu is invalid!", _cmd, encoding);
		return NO;
	}
	
	// Get text encoding string value
	NSString *textEncStr = [(NSString *)CFStringConvertEncodingToIANACharSetName(cfStringEncoding) 
							stringByAppendingFormat:@";%@", [[NSNumber numberWithInt:cfStringEncoding] stringValue]];
	const char *textEncUTF8Str = [textEncStr UTF8String];
	
	// Set attribute
	if (setxattr(path, "com.apple.TextEncoding", textEncUTF8Str, strlen(textEncUTF8Str), 0, 0) < 0) {
		NSLog(@"couldn't set text encoding attribute of %s to '%s': %d", path, textEncUTF8Str, errno);
		return NO;
	}
	return YES;
	
}

+ (NSStringEncoding)textEncodingAttributeOfPath:(NSString *)filePath {
	
	// Get C string
	if (!filePath) return 0;
	const char* path = [filePath fileSystemRepresentation];
	
	// We could query the size of the attribute, but that would require a second system call
	// and the value for this key shouldn't need to be anywhere near this large, anyway.
	// It could be, but it probably won't. If it is, then we won't get the encoding. Too bad.
	char xattrValueBytes[128] = { 0 };
	if (getxattr(path, "com.apple.TextEncoding", xattrValueBytes, sizeof(xattrValueBytes), 0, 0) < 0) {
		if (ENOATTR != errno) NSLog(@"couldn't get text encoding attribute of %s: %d", path, errno);
		return 0;
	}
	NSString *encodingStr = [NSString stringWithUTF8String:xattrValueBytes];
	if (!encodingStr) {
		NSLog(@"couldn't make attribute data from %s into a string", path);
		return 0;
	}
	NSArray *segs = [encodingStr componentsSeparatedByString:@";"];
	
	if ([segs count] >= 2 && [(NSString*)[segs objectAtIndex:1] length] > 1) {
		return CFStringConvertEncodingToNSStringEncoding([[segs objectAtIndex:1] intValue]);
	} else if ([(NSString*)[segs objectAtIndex:0] length] > 1) {
		CFStringEncoding theCFEncoding = CFStringConvertIANACharSetNameToEncoding((CFStringRef)[segs objectAtIndex:0]);
		if (theCFEncoding == kCFStringEncodingInvalidId) {
			NSLog(@"couldn't convert IANA charset");
			return 0;
		}
		return CFStringConvertEncodingToNSStringEncoding(theCFEncoding);
	} else {
		return 0; // error
	}

}

@end