Detect if a user has typed an emoji character in UITextView

Over the years these emoji-detecting solutions keep breaking as Apple adds new emojis w/ new methods (like skin-toned emojis built by pre-cursing a character with an additional character), etc.

I finally broke down and just wrote the following method which works for all current emojis and should work for all future emojis.

The solution creates a UILabel with the character and a black background. CG then takes a snapshot of the label and I scan all pixels in the snapshot for any non solid-black pixels. The reason I add the black background is to avoid issues of false-coloring due to Subpixel Rendering

The solution runs VERY fast on my device, I can check hundreds of characters a second, but it should be noted that this is a CoreGraphics solution and should not be used heavily like you could with a regular text method. Graphics processing is data heavy so checking thousands of characters at once could result in noticeable lag.

-(BOOL)isEmoji:(NSString *)character {

    UILabel *characterRender = [[UILabel alloc] initWithFrame:CGRectMake(0, 0, 1, 1)];
    characterRender.text = character;
    characterRender.backgroundColor = [UIColor blackColor];//needed to remove subpixel rendering colors
    [characterRender sizeToFit];

    CGRect rect = [characterRender bounds];
    UIGraphicsBeginImageContextWithOptions(rect.size,YES,0.0f);
    CGContextRef contextSnap = UIGraphicsGetCurrentContext();
    [characterRender.layer renderInContext:contextSnap];
    UIImage *capturedImage = UIGraphicsGetImageFromCurrentImageContext();
    UIGraphicsEndImageContext();

    CGImageRef imageRef = [capturedImage CGImage];
    NSUInteger width = CGImageGetWidth(imageRef);
    NSUInteger height = CGImageGetHeight(imageRef);
    CGColorSpaceRef colorSpace = CGColorSpaceCreateDeviceRGB();
    unsigned char *rawData = (unsigned char*) calloc(height * width * 4, sizeof(unsigned char));
    NSUInteger bytesPerPixel = 4;
    NSUInteger bytesPerRow = bytesPerPixel * width;
    NSUInteger bitsPerComponent = 8;
    CGContextRef context = CGBitmapContextCreate(rawData, width, height,
                                                 bitsPerComponent, bytesPerRow, colorSpace,
                                                 kCGImageAlphaPremultipliedLast | kCGBitmapByteOrder32Big);
    CGColorSpaceRelease(colorSpace);

    CGContextDrawImage(context, CGRectMake(0, 0, width, height), imageRef);
    CGContextRelease(context);

    BOOL colorPixelFound = NO;

    int x = 0;
    int y = 0;
    while (y < height && !colorPixelFound) {
        while (x < width && !colorPixelFound) {

            NSUInteger byteIndex = (bytesPerRow * y) + x * bytesPerPixel;

            CGFloat red = (CGFloat)rawData[byteIndex];
            CGFloat green = (CGFloat)rawData[byteIndex+1];
            CGFloat blue = (CGFloat)rawData[byteIndex+2];

            CGFloat h, s, b, a;
            UIColor *c = [UIColor colorWithRed:red green:green blue:blue alpha:1.0f];
            [c getHue:&h saturation:&s brightness:&b alpha:&a];

            b /= 255.0f;

            if (b > 0) {
                colorPixelFound = YES;
            }

            x++;
        }
        x=0;
        y++;
    }

    return colorPixelFound;

}

First let's address your "55357 method" – and why it works for many emoji characters.

In Cocoa, an NSString is a collection of unichars, and unichar is just a typealias for unsigned short which is the same as UInt16. Since the maximum value of UInt16 is 0xffff, this rules out quite a few emoji from being able to fit into one unichar, as only two out of the six main Unicode blocks used for emoji fall under this range:

• Miscellaneous Symbols (U+2600–U+26FF)
• Dingbats (U+2700–U+27BF)

These blocks contain 113 emoji, and an additional 66 emoji that can be represented as a single unichar can be found spread around various other blocks. However, these 179 characters only represent a fraction of the 1126 emoji base characters, the rest of which must be represented by more than one unichar.

Let's analyse your code:

unichar unicodevalue = [text characterAtIndex:0];

What's happening is that you're simply taking the first unichar of the string, and while this works for the previously mentioned 179 characters, it breaks apart when you encounter a UTF-32 character, since NSString converts everything into UTF-16 encoding. The conversion works by substituting the UTF-32 value with surrogate pairs, which means that the NSString now contains two unichars.

And now we're getting to why the number 55357, or 0xd83d, appears for many emoji: when you only look at the first UTF-16 value of a UTF-32 character you get the high surrogate, each of which have a span of 1024 low surrogates. The range for the high surrogate 0xd83d is U+1F400–U+1F7FF, which starts in the middle of the largest emoji block, Miscellaneous Symbols and Pictographs (U+1F300–U+1F5FF), and continues all the way up to Geometric Shapes Extended (U+1F780–U+1F7FF) – containing a total of 563 emoji, and 333 non-emoji characters within this range.

So, an impressive 50% of emoji base characters have the the high surrogate 0xd83d, but these deduction methods still leave 384 emoji characters unhandled, along with giving false positives for at least as many.

---

So, how can you detect whether a character is an emoji or not?

I recently answered a somewhat related question with a Swift implementation, and if you want to, you can look at how emoji are detected in this framework, which I created for the purpose of replacing standard emoji with custom images.

Anyhow, what you can do is extract the UTF-32 code point from the characters, which we'll do according to the specification:

- (BOOL)textView:(UITextView *)textView shouldChangeTextInRange:(NSRange)range replacementText:(NSString *)text {

    // Get the UTF-16 representation of the text.
    unsigned long length = text.length;
    unichar buffer[length];
    [text getCharacters:buffer];

    // Initialize array to hold our UTF-32 values.
    NSMutableArray *array = [[NSMutableArray alloc] init];

    // Temporary stores for the UTF-32 and UTF-16 values.
    UTF32Char utf32 = 0;
    UTF16Char h16 = 0, l16 = 0;

    for (int i = 0; i < length; i++) {
        unichar surrogate = buffer[i];

        // High surrogate.
        if (0xd800 <= surrogate && surrogate <= 0xd83f) {
            h16 = surrogate;
            continue;
        }
        // Low surrogate.
        else if (0xdc00 <= surrogate && surrogate <= 0xdfff) {
            l16 = surrogate;

            // Convert surrogate pair to UTF-32 encoding.
            utf32 = ((h16 - 0xd800) << 10) + (l16 - 0xdc00) + 0x10000;
        }
        // Normal UTF-16.
        else {
            utf32 = surrogate;
        }

        // Add UTF-32 value to array.
        [array addObject:[NSNumber numberWithUnsignedInteger:utf32]];
    }

    NSLog(@"%@ contains values:", text);

    for (int i = 0; i < array.count; i++) {
        UTF32Char character = (UTF32Char)[[array objectAtIndex:i] unsignedIntegerValue];
        NSLog(@"\t- U+%x", character);
    }

    return YES;
}

Typing "😎" into the UITextView writes this to console:

😎 contains values:
    - U+1f60e

With that logic, just compare the value of character to your data source of emoji code points, and you'll know exactly if the character is an emoji or not.

---

P.S.

There are a few "invisible" characters, namely Variation Selectors and zero-width joiners, that also should be handled, so I recommend studying those to learn how they behave.