regexkit-discussion

[Regexkit-discussion] Regex string results in an infinite loop

[Aaron W. <pap...@gm...>]

Hello,

The following code block results in an infinite loop. The subject  
string that is commented out, seems to work ok. I'have tried the same  
expression on other engines without this problem. I should note that  
this expression comes from the textmate c code bundle. Am I missing  
anything here?

{
		// capture preprocessor macro
		NSString *regexString = @"(?ms)((?x)^\\s*\\#\\s*(define)\\s+((?<id> 
[a-zA-Z_][a-zA-Z0-9_]*))(?:(\\()(\\s*\\g{id}\\s*((,)\\s*\\g{id}\\s*)* 
(?:\\.\\.\\.)?)(\\)))?.*?(?=(?://|/\\*))|$)";
	NSString *subjectString = @"/";
	// NSString *subjectString = @"#define HELLO(a,b) hello(a,b) // say  
hello";
	NSUInteger matchCount = 0;
	
	// make sure regexString doesn't cause an infinate loop
	RKEnumerator *matchEnumerator = [subjectString  
matchEnumeratorWithRegex:regexString];	
	while([matchEnumerator nextRanges] != NULL) {
		matchCount++;
		if(matchCount++ > 10) {
			assert(0);
	}
}



thanks,
kube

Re: [Regexkit-discussion] Regex string results in an infinite loop

[John E. <joh...@gm...>]

On Sat, Mar 14, 2009 at 6:51 PM, Aaron Warner <pap...@gm...> wrote:
> Hello,
> The following code block results in an infinite loop. The subject string
> that is commented out, seems to work ok. I'have tried the same expression on
> other engines without this problem. I should note that this expression comes
> from the textmate c code bundle. Am I missing anything here?
> {
> // capture preprocessor macro
> NSString *regexString
> = @"(?ms)((?x)^\\s*\\#\\s*(define)\\s+((?<id>[a-zA-Z_][a-zA-Z0-9_]*))(?:(\\()(\\s*\\g{id}\\s*((,)\\s*\\g{id}\\s*)*(?:\\.\\.\\.)?)(\\)))?.*?(?=(?://|/\\*))|$)";
> NSString *subjectString = @"/";
> // NSString *subjectString = @"#define HELLO(a,b) hello(a,b) // say hello";
> NSUInteger matchCount = 0;
>
> // make sure regexString doesn't cause an infinate loop
> RKEnumerator *matchEnumerator = [subjectString
> matchEnumeratorWithRegex:regexString];
> while([matchEnumerator nextRanges] != NULL) {
> matchCount++;
> if(matchCount++ > 10) {
> assert(0);
> }
> }

This is a bug in RegexKit.framework <= 0.6.  It's a rather simple fix
if you're willing to edit the sources.  The problem, and a fix, is
covered in the following forum message:
http://sourceforge.net/forum/forum.php?thread_id=2660738&forum_id=731002

There is also a bug opened on this issue as well:
http://sourceforge.net/tracker/index.php?func=detail&aid=1958025&group_id=204582&atid=990188

I really should get around to patching up this and a few other
miscelanous bugs in RegexKit.framework, I just haven't gotten around
to it.  Most of my effort has been on RegexKitLite (which uses the
system supplied ICU regex matcher, not PCRE) because it's fairly
popular with iPhone developers (much smaller size since the system
supplies the regex matcher in a shared library, and no apple imposed
'no external frameworks' limitations).