CODEWITHSUNDEEP
iosobjective-cutf-8printfnslog
Nicolas Miari
Nicolas Miari

Reputation: 16246

NSLog() vs printf() when printing C string (UTF-8)

I have noticed that if I try to print the byte array containing the representation of a string in UTF-8, using the format specifier "%s", printf() gets it right but NSLog() gets it garbled (i.e., each byte printed as-is, so for example "¥" gets printed as the 2 characters: "¬•"). This is curious, because I always thought that NSLog() is just printf(), plus:

  1. The first parameter (the 'format') is an Objective-C string, not a C string (hence the "@").
  2. The timestamp and app name prepended.
  3. The newline automatically added at the end.
  4. The ability to print Objective-C objects (using the format "%@").

My code:

NSString* string; 

// (...fill string with unicode string...)

const char* stringBytes = [string cStringUsingEncoding:NSUTF8Encoding];

NSUInteger stringByteLength = [string lengthOfBytesUsingEncoding:NSUTF8Encoding];
stringByteLength += 1; // add room for '\0' terminator

char* buffer = calloc(sizeof(char), stringByteLength);

memcpy(buffer, stringBytes, stringByteLength);

NSLog(@"Buffer after copy: %s", buffer);
// (renders ascii, no matter what)

printf("Buffer after copy: %s\n", buffer);
// (renders correctly, e.g. japanese text)

Somehow, it looks as if printf() is "smarter" than NSLog(). Does anyone know the underlying cause, and if this feature is documented anywhere? (Couldn't find)

Upvotes: 3

Views: 5666

Answers (1)

Martin R
Martin R

Reputation: 539685

NSLog() and stringWithFormat: seem to expect the string for %s in the "system encoding" (for example "Mac Roman" on my computer):

NSString *string = @"¥";
NSStringEncoding enc = CFStringConvertEncodingToNSStringEncoding(CFStringGetSystemEncoding());
const char* stringBytes = [string cStringUsingEncoding:enc];
NSString *log = [NSString stringWithFormat:@"%s", stringBytes];
NSLog(@"%@", log);

// Output: ¥

Of course this will fail if some characters are not representable in the system encoding. I could not find an official documentation for this behavior, but one can see that using %s in stringWithFormat: or NSLog() does not reliably work with arbitrary UTF-8 strings.

If you want to check the contents of a char buffer containing an UTF-8 string, then this would work with arbitrary characters (using the boxed expression syntax to create an NSString from a UTF-8 string):

NSLog(@"%@", @(utf8Buffer));

Upvotes: 3

Related Questions