Re: how to count the number of repeats in a string (really!)

The regex engine is quite adept at backtracking to find all inputs that satisfy a set of conditions. The shortest solution should be a regex one.

my $str = 'aabcdabcabcecdecd';

local our %counts;
$str =~ /
   (.{2,})   # or (.+)
   (?(?{ !$counts{$1} })(?=.*\1))
   (?{ ++$counts{$1} })
   (?!)
/x;

use Data::Dumper;
print Dumper \%counts;
[download]

Update:
(?(?{ !$counts{$1} })(?=.*\1))
might be more efficient as
(?> (?(?{ !$counts{$1} })(?=.*\1)) )

Update: The above doesn't work. $str='ababa' fails. My simpler version doesn't suffer from this bug.

Comment on Re: how to count the number of repeats in a string (really!) Select or Download Code

Replies are listed 'Best First'.
Re^2: how to count the number of repeats in a string (really!) by ikegami (Patriarch) on Nov 14, 2007 at 19:28 UTC
I guess I didn't explain why mine works and why the OP's doesn't work with ".+". Compare `while ($str =~ /($re)/g) { print("$1\n"); }` [download] `use re 'eval'; $str =~ / ($re) (?{ print("$1\n"); }) (?!) /x;` [download] Looks similar? But for the same input, they produce different results. Input: `my $str = 'aabcdabcabce'; my $re = qr/a[^a]*/;` [download] Output from `/.../g`: `a abcd abc abce` [download] Output from `/...(?{ save results })(?!)/`: `a abcd abc ab a abc ab a abce abc ab a` [download] `/...(?{ save results })(?!)/` is key in finding all possible matches. It forces the regex to try everything to obtain a match.	[reply] [d/l] [select]
Re^3: how to count the number of repeats in a string (really!) by blazar (Canon) on Nov 14, 2007 at 21:49 UTC
I guess I didn't explain why mine works and why the OP's doesn't work with ".+". I personally believe that blokhead already did. It was just a trivial mistake/overview on my part.	[reply]


"be consistent"
	PerlMonks