faster splitter

[Chris via Digitalmars-d]

On Friday, 27 May 2016 at 14:06:09 UTC, Chris wrote:
>
> This outperforms both `manual_find` and the improved `std find`
>
> string findStringS_Manual(string haystack, string needle)
> {
> 	
> 	if (needle.length > haystack.length)
> 		return haystack[$..$];
> 	outer:
> 	for (auto i = 0; i < haystack.length; i++)
> 	{
> 		if (haystack[i] != needle[0])
> 			continue;
> 		for (size_t j = i+1, k = 1; k < needle.length; ++j, ++k)
> 			if (haystack[j] != needle[k])
> 			continue outer;
> 		return haystack[i..$];
> 	}
> 	return haystack[$..$];
> }
>

Little bug fix:

for (size_t j = (i+1 < haystack.length) ? i+1 : i, k = 1; k < 
needle.length; ++j, ++k)

Else it will be out or range when you look for "za" and the 
string ends with "z". There is a slight performance penalty, but 
it's still faster than the rest. Maybe there are cleverer ways. 
But every little check and decision comes with a performance 
penalty.

The improved `std find` comes very close to the manual function 
above now, sometimes it's even faster or at least as fast.

std find    	took     12573666
manual find 	took      7418454
my std find 	took      6903854 <===
findStringS 	took      7166720
findStringS_  	took      6579529 <===

std find    	took     11849892
manual find 	took      7407091
my std find 	took      6573102 <===
findStringS 	took      7296610
findStringS_  	took      6573214 <===

std find    	took     10744361
manual find 	took      7576133
my std find 	took      6332014 <===
findStringS 	took      7224946
findStringS_  	took      6555976 <===