Volume Number: 18 (2002)
Issue Number: 01
Column Tag: Mac OS X

Smart Quotes

by ©2001 Andrew C. Stone. All Rights Reserved.

Adding automated curly quotes to Cocoa's Text system

Cocoa's Text System architecture is open, open-ended, and contains hooks for easy behavior modification. Because Cocoa Text is a unicode based system, it is by default international and lets your applications work correctly in a variety of scripts and languages. This article will teach you how to inspect the user's typed stream of keys, and convert straight quotes to open and closed curly quotes as the user types:

"Love is ‘THE' answer," she said.

Remembering all the modifier keys to create a curly quote is troublesome for end users, so we'll add the functionality to replace the boring straight quotes with fancy curly ones as they type. We'll also include a mechanism to "toggle through" the various options for those odd border cases, such as words beginning with a quote: 'Tis.

Cocoa's Text system relies on Unicode which allows each character in most standard font to be mapped to a unique short integer, a "unichar" defined in NSString.h:

typedef unsigned short unichar;

For example, each glyph in the Symbol font and Dingbats font are represented by unicode characters. The HTML standard also allows use of unicode characters, and here are the Unicode and HTML codes (which correspond to the unicode value) for curly quotes:

Character	Unicode	HTML	Name	How to type in manually
"	8220	“	double open quote	Option-[
'	8216	‘	single open quote	Option-]
"	8221	”	double close	Option-Shift-[
'	8217	’	single close	Option-Shift-]

Now that we know what will replace the straight quotes, we need a strategy for determining if a quote should be open or closed! It turns out that a very simple heuristic will work:

if character typed is a Straight Quote then

If it's the first character or is preceded by whitespace, turn it into Open Curly Quote

       else turn it into Close Curly Quote

So, when a newline, tab or space precedes a quote, it's probably an open quote. Note that this fails with ‘Tis! So, let's refine our heuristic to include a mechanism to "move through" the options straight, open and closed if the user has the quote in question solely selected:

if character typed is ‘ or " then

if selection contains a  single quote of any style,  replace with the next style in a loop (straight -> 
open -> closed -> straight and so on)

    else

If it's the first character or is preceded by whitespace, turn into Open Quote (single or double)

       else turn it into Close Quote

We have one more feature to add - the user should be able to DIRECTLY type in a straight quote when fancy quotes are enabled. We'll let them do this by changing any typed fancy quote to its straight counterpart (see How to type in manually in table above).

One aspect of Cocoa that reigns supreme is the ease in which functionality can added, as well as the many different approaches that can be taken to accomplish the same task.

Where you put code depends on your architectural design. While you can always override a class, it's often easier to simply look at the "Delegate Methods" defined in various classes. These methods are optional: if an object, such as a text view, has a delegate, and that delegate implements one of these methods, then that method will be called at the appropriate time by the text view. If you overrode "keyDown" in NSTextView in a subclass, your text wouldn't swap the quotes on pasted text, since pasting does not call keyDown! So, instead we'll use this NSTextView delegate method that gets called anytime text is inserted either via keys or paste:

- (BOOL)textView:(NSTextView *)textView shouldChangeTextInRange:(NSRange)affectedCharRange 
replacementString:(NSString *)replacementString;

    // Delegate only.  If characters are changing, replacementString is what will replace the 
    affectedCharRange.  If attributes only are changing, replacementString will be nil.

This delegate method will work for us since we are only replacing single characters with another single character. If I had designed this method, instead of returning a boolean, I would have had this method return the actual new string to be inserted. Working with the method the way it is, we'll simply brute force stick in the character of our choice and return NO.

// Code is always more readable if you can turn the magic numbers into English!
#define SINGLE_QUOTE      ‘\''
#define SINGLE_OPEN_QUOTE   8216
#define SINGLE_CLOSE_QUOTE   8217
#define DOUBLE_QUOTE      ‘"‘
#define DOUBLE_OPEN_QUOTE   8220
#define DOUBLE_CLOSE_QUOTE   8221

// Smart Quote support - in our NSTextView's delegate class, we implement this method
// Be sure to connect the TextView's delegate outlet to your custom delegate class in Interface Builder
// or use  [myTextView setDelegate:self]; in your delegate class so this gets called:

- (BOOL)textView:(NSTextView *)textView shouldChangeTextInRange:(NSRange)affectedCharRange 
replacementString:(NSString *)replacementString;

    // Delegate only.  If characters are changing, replacementString is what will replace the 
    affectedCharRange.  If attributes only are changing, replacementString will be nil.
{

   // if we're just changing text attributes then we don't enter our processing loop
   // Also, provide a defaults mechanism to turn this fancy quoting off:
   
    if (replacementString && [[NSUserDefaults standardUserDefaults]boolForKey:SmartQuotes]) {
    
       // This is what is in our text object before anything is added:
        NSString *text = [[textView textStorage] string];
   
      // We want to know if we are at the very first character:
        unsigned int textLength = [text length];
   
      // how much are  we actually adding at this time:
        unsigned int i, length = [replacementString length];
        unichar c;
   
      // Should we change the string, we'll use this mutable string to hold the new values:
      // If it's non-nil when we get done, that means we've got work to do!
        NSMutableString *s = nil;
            
       // First, our toggle through mechanism:
        // special case: one char typed OVER a smart quote -> toggle them around 3 way
        // plain -> open -> closed -> plain
        // First deal with the case where they want another type of quote or a plain quote!
        if (length == 1 &&  affectedCharRange.length == 1) {
   
            if (((c = [text characterAtIndex:affectedCharRange.location])==DOUBLE_OPEN_QUOTE) || 
            (c == SINGLE_OPEN_QUOTE)) {
                // they had an open quote -> make it a closed one
                s = [NSString stringWithFormat:@"%C", c == DOUBLE_OPEN_QUOTE ? 
                DOUBLE_CLOSE_QUOTE : SINGLE_CLOSE_QUOTE];
      
            } else if (c == DOUBLE_CLOSE_QUOTE || c == SINGLE_CLOSE_QUOTE) {
                // they had a closed quote -> make it plain
                s = [NSString stringWithFormat:@"%C", c == DOUBLE_CLOSE_QUOTE  ? 
                DOUBLE_QUOTE : SINGLE_QUOTE];
      
            } else if (c == SINGLE_QUOTE || c == DOUBLE_QUOTE) {
             // they had a straight quote -> make it open
                 s = [NSString stringWithFormat:@"%C", c == SINGLE_QUOTE  ?  
                 SINGLE_OPEN_QUOTE : DOUBLE_OPEN_QUOTE];
           }
        } else {
   
        // otherwise go through replacement string one by one - paste can put in many characters at 
        one time!
            NSCharacterSet *startSet = [NSCharacterSet whitespaceAndNewlineCharacterSet];
            for (i = 0; i < length; i++) {
       
                unichar theChar = [replacementString characterAtIndex:i];
                unichar previousChar;
                
            // Find out the character which preceeds this one - context is everything!
                if(i == 0) {
                    if (affectedCharRange.location == 0 || textLength==0) previousChar = 0; 
                    // first char
                    else previousChar = [text characterAtIndex:affectedCharRange.location - 1];
                } else previousChar = [replacementString characterAtIndex:i-1];
                
            // When we encounter a straight quote, we decide whether it should be open or closed:
                if ((theChar == SINGLE_QUOTE) || (theChar == DOUBLE_QUOTE)){
      
               // lazily allocate the mutable string if we find something interesting
                    if (!s) s = [NSMutableString stringWithString:replacementString]; 
                    
                    if (previousChar == 0 || [startSet characterIsMember:previousChar] || 
                    (previousChar == DOUBLE_OPEN_QUOTE && theChar == SINGLE_QUOTE) || 
                    (previousChar == SINGLE_OPEN_QUOTE && theChar == DOUBLE_QUOTE))
                        c =  (theChar  == SINGLE_QUOTE  ?  SINGLE_OPEN_QUOTE : DOUBLE_OPEN_QUOTE);
                    else c =  (theChar  == SINGLE_QUOTE  ?  SINGLE_CLOSE_QUOTE : DOUBLE_CLOSE_QUOTE);
                    
                    [s replaceCharactersInRange:NSMakeRange(i,1) 
                    withString:[NSString stringWithFormat:@"%C", c]];
          
                } else if ((i==0) && (length == 1)) {
      
                    // we don't want to do this unless they are typing - paste may contain curly's 
                    already!
                    // reverse the meaning - they want to type in a plain one from the keyboard 
                    directly:
          
                    if ((theChar == SINGLE_CLOSE_QUOTE) || (theChar == SINGLE_OPEN_QUOTE)) {
                            s = [NSMutableString stringWithFormat:@"%C", SINGLE_QUOTE];
                    } else if ((theChar == DOUBLE_CLOSE_QUOTE) || (theChar == DOUBLE_OPEN_QUOTE)) {
                            s = [NSMutableString stringWithFormat:@"%C", DOUBLE_QUOTE];
                    }
                }
                        
            }
        }
        if (s) {
                // We'll be responsible for inserting the changed text ourselves
            // ideally, this method would return the string desired, but it doesn't
            // so we'll just pop the changes in directly ourselves:
                [[textView textStorage] replaceCharactersInRange:affectedCharRange withString:s];
                return NO;               
        }
    }
    // Otherwise, let the text system insert the text as typed...
    return YES;
}

The Cocoa Text system is easy to use and very extensible in a straightforward manner. An improvement you might consider for this smart quoting technique is to allow for internationalization. For example, in Germany, the open quote is along the text baseline, and other languages might use ‹› or other characters for quoting. To accomplish this, you would use NSLocalizedStringFromTable() and create an entry in the .strings file for each of the quotes in the languages that you support - refer to my article on Localization in MacTech a few months back for more information.

Andrew Stone, CEO of Stone Design, www.stone.com, has been coding in Cocoa as an independent software developer for over 13 years.