Question

Now my app targets >iOS4 I am trying to remove dependencies on non Apple frameworks where possible.

I have this code currently

    NSString *destination =@"HHH-DDDD>dddd,ffff";
    NSString *searchString = destination;
    NSString *regexString  = @"[^a-zA-Z\?]";
    NSArray  *splitArray   = NULL;
    splitArray = [searchString componentsSeparatedByRegex:regexString];

This produces an array with string elements between all of the characters. i.e. “HHH”,”DDDD”,”dddd”,”ffff”.

There doesn’t seem to be an equivalent in the NSRegularExpression at all or am I reading the documentation wrong?

Answer 1

seems you’re right but you could use a category:

#import "NSString+RegEx_Array.h"

@implementation NSString (RegEx_Array)

- (NSArray*)componentsSeparatedByRegex2:(NSString *)pattern
{
    NSUInteger pos = 0;
    NSRange area = NSMakeRange(0, [self length]);

    NSRegularExpression *regEx = [NSRegularExpression 
                                  regularExpressionWithPattern:pattern 
                                  options:0 error:nil];

    NSArray *matchResults = [regEx matchesInString:self options:0 range:area];

    NSMutableArray *returnArray = [NSMutableArray arrayWithCapacity:matchResults.count];

    for (NSTextCheckingResult *result in matchResults) {
        NSRange substrRange = NSMakeRange(pos, [result range].location-pos);
        [returnArray addObject:[self substringWithRange:substrRange]];
        pos = [result range].location + [result range].length;
    }

    if (pos < area.length) {
        [returnArray addObject:[self substringFromIndex:pos]];
    }

    return returnArray;
}

@end

but I have to admit, it isn’t that performant like RegexKitLite. Heres the output for a nonsense file:

2012-02-08 00:16:32.251 shittyagain[96057:c03] String-Length: 36355200
2012-02-08 00:16:36.420 shittyagain[96057:c03] Time taken RegexKitLite: 4.167
2012-02-08 00:16:42.989 shittyagain[96057:c03] Time taken NSRegularExp: 6.568

and the testing code:

NSString *regx = @"[^a-zA-Z\?]";
NSString *file = [[NSBundle mainBundle]pathForResource:@"reallylong" ofType:nil];
NSString *searchString = [NSString stringWithContentsOfFile:file encoding:NSUTF8StringEncoding error:nil];

NSLog(@"String-Length: %ld", searchString.length);

NSDate *preDate = [NSDate date];
[searchString componentsSeparatedByRegex:regx];
NSLog(@"Time taken RegexKitLite: %1.3f", [preDate timeIntervalSinceNow]*-1);

NSDate *preDate2 = [NSDate date];
[searchString componentsSeparatedByRegex2:regx];
NSLog(@"Time taken NSRegularExp: %1.3f", [preDate2 timeIntervalSinceNow]*-1);

I’m not sure if the difference really matters because the file has a size of ~36mb but just in case someone needs it really fast he has to use the RegexKitLite. I would like if someone could post some real performance test and/or for for default NSRegularExpression methods. I want to know which one is faster =)