Truncate delimited NSString without removing delimiters

Question

I have some data in an NSString, separated by colons:

@"John:Doe:1970:Male:Dodge:Durango"

I need to limit the total length of this string to 100 characters. But I also need to ensure the correct number of colons are present.

What would be a reasonable to way to truncate the string but also add the extra colons so I can parse it into the correct number of fields on the other side?

For example, if my limit was 18, you would end up with something like this:

@"John:Doe:1970:Ma::"

Here's an updated version of my own latest pass at this. Uses @blinkenlights algorithm:

+ (NSUInteger)occurrencesOfSubstring:(NSString *)substring inString:(NSString *)string {
    // http://stackoverflow.com/a/5310084/878969
    return [string length] - [[string stringByReplacingOccurrencesOfString:substring withString:@""] length] / [substring length];
}

+ (NSString *)truncateString:(NSString *)string toLength:(NSUInteger)length butKeepDelmiter:(NSString *)delimiter {
    if (string.length <= length)
        return string;
    NSAssert(delimiter.length == 1, @"Expected delimiter to be a string containing a single character");
    int numDelimitersInOriginal = [[self class] occurrencesOfSubstring:delimiter inString:string];

    NSMutableString *truncatedString = [[string substringToIndex:length] mutableCopy];
    int numDelimitersInTruncated = [[self class] occurrencesOfSubstring:delimiter inString:truncatedString];
    int numDelimitersToAdd = numDelimitersInOriginal - numDelimitersInTruncated;
    int index = length - 1;
    while (numDelimitersToAdd > 0) { // edge case not handled here
        NSRange nextRange = NSMakeRange(index, 1);
        index -= 1;
        NSString *rangeSubstring = [truncatedString substringWithRange:nextRange];
        if ([rangeSubstring isEqualToString:delimiter])
            continue;
        [truncatedString replaceCharactersInRange:nextRange withString:delimiter];
        numDelimitersToAdd -= 1;
    }
    return truncatedString;
}

Note that I don't think this solution handles the edge case from CRD where the number of delimiters is less than the limit.

The reason I need the correct number of colons is the code on the server will split on colon and expect to get 5 strings back.

You can assume the components of the colon separated string do not themselves contain colons.

Answer 1

Your current algorithm will not produce the correct result when one or more of the characters among the last colonsToAdd is a colon.

You can use this approach instead:

• Cut the string at 100 characters, and store the characters in an NSMutableString
• Count the number of colons, and subtract that number from the number that you need
• Starting at the back of the string, replace non-colon characters with colons until you have the right number of colons.

Answer 2

I tend towards @dasblinkenlight, it's just an algorithm after all, but here's some code. Few modern shorthands - used an old compiler. ARC assumed. Won't claim it's efficient, or beautiful, but it does work and handles edge cases (repeated colons, too many fields for limit):

- (NSString *)abbreviate:(NSString *)input limit:(NSUInteger)limit
{
    NSMutableArray *fields = [[input componentsSeparatedByString:@":"] mutableCopy];
    NSUInteger colonCount = fields.count - 1;

    if (colonCount >= limit)
        return [@"" stringByPaddingToLength:limit withString:@":" startingAtIndex:0];

    NSUInteger nonColonsRemaining = limit - colonCount;
    for (NSUInteger ix = 0; ix <= colonCount; ix++)
    {
        if (nonColonsRemaining > 0)
        {
            NSString *fieldValue = [fields objectAtIndex:ix];
            NSUInteger fieldLength = fieldValue.length;
            if (fieldLength <= nonColonsRemaining)
                nonColonsRemaining -= fieldLength;
            else
            {
                [fields replaceObjectAtIndex:ix withObject:[fieldValue substringToIndex:nonColonsRemaining]];
                nonColonsRemaining = 0;
            }
        }
        else
            [fields replaceObjectAtIndex:ix withObject:@""];
    }

    return [fields componentsJoinedByString:@":"];  
}