It has always bothered me that Cocoa doesn't have Regular Expressions built in. Using PCRE 6.7, I've created a small Cocoa Framework that does Regular Expressions. I know there's the AGRegex that's up on sourceforge, but I decided to take my crack at making something that might be a little different. It's not complete yet, but I figured I'd write about it since I just created a subversion repository for it.

Here are the basics:

• There's a PCRegex object
• All matching is done via calls on strings (not on the Regex object)

I chose to call methods on the strings instead of the regex objects because this is a little more like all other object oriented regular expression implementations and because I think it makes more sense that way.

Creating a regular expression is easy as:
PCRegex *regex = [PCRegex regexWithPattern:@"pattern"];

Matching regular expressions against strings is easy, too. With the previous regular expression defined:

#import <ObjCRegex/ObjCRegex.h>
...
[@"This is the string that I want to match" match:regex];
[@"This string will match because it contains the word pattern" match:regex];

With connivence methods everything is extremely simple:

#import <ObjCRegex/ObjCRegex.h>
...
[@"This is the string that I want to match" matchPattern:@"pattern"];
[@"This is the string that I want to match" matchPattern:@"str.*"];

Of course if you want something more powerful, you're wondering what I did to deal with backreferences. Again, it's pretty simple. Here's an example:

#import <ObjCRegex/ObjCRegex.h>
...
NSArray *backrefArray;
PCRegex *regex = [PCRegex regexWithPattern:@"([^ ]*) ([^ ]*)"];
[@"first second third" match:regex backreferences:backrefArray];
NSString *first_match = [[backrefArray objectAtIndex:1] string];
NSString *second_match = [[backrefArray objectAtIndex:2] string];

Like many other implementations, the 0^th item in the backreference array is the fully matched string, so in the previous example it would be @"first second".

There are some things that are incomplete and/or you should know before you get excited about using it:

• There's no support for named backreference (yet)
• UTF-8 isn't supported (thought PCRE dose support it, so if someone wants to add support, that'd be great)
• I've done little testing with it, though it seems to work just fine
• GPL license as always

So if you're still reading, you may be interested in checking out the code which can be done via:
svn co svn://fadingred.org/objcregex/trunk objcregex

Enjoy!

2 Comments Tags: Cocoa, Code, Objective-C, Regex