计算Perl或Ruby中重叠的正则表达式匹配项

Question

This is a follow-up to that question . I've learned that finding overlapping regex matches in Python is not straight-forward, so decided to do an additional inquiry to see how Perl and Ruby stand up to this task.

I'd like to count the number of all possible matches of a regex against a certain string. And by "all" I mean that the result should take into account both overlapping and non-unique matches. Here are some examples:

• a.*k should be matched twice in "akka"
• "bbboob" tested against b.*o.*b should yield 6

As a reference, here's a Perl one-liner suggested by tchrist - it outputs the correct matches and their count:

() = "bbboobb" =~ /(b.*o.*b)(?{push @all, $1})(*FAIL)/g; printf "got %d matches: %s\n", scalar(@all), "@all";

The only problem with this is that it eats up too much resources for test cases where the resulting match count is in the order of millions or more. But I understand it is due to the fact that all the matches are first groupped and only counted afterwards. I'm looking for a resource-efficient solution that only returns the count .

Answer 1

It looks like tchrist has done all the hard work. If storing the matches and counting them afterwards is eating too much resource, then you could just change the regex-embedded code to just count the matches:

my $count = 0;

"bbboobb" =~ /(b.*o.*b)(?{$count++})(*FAIL)/g;

print "got $count matches\n";