Find overlapping Regexp matches

Question

I want to find all matches within a given string including overlapping matches. How could I achieve it?

# Example
"a-b-c-d".???(/\w-\w/)  # => ["a-b", "b-c", "c-d"] expected

# Solution without overlapped results
"a-b-c-d".scan(/\w-\w/) # => ["a-b", "c-d"], but "b-c" is missing

Mil

Answer 1

Use capturing inside a positive lookahead:

"a-b-c-d".scan(/(?=(\w-\w))/).flatten
 # => ["a-b", "b-c", "c-d"]

See Ruby demo

Photochronograph

Works perfectly, also with complex patterns. "abaca".scan(/(?=(\w)(?:(?!\1)(\w))\1)/) #=> [["a", "b"], ["a", "c"]]. – Mil

...or positive lookbehind: "a-b-c-d".scan(/(?<=(\w-\w))/).flatten. – Rabbin

Answer 2

I suggest a non-regex solution:

"a-b-c-d".delete('-').each_char.each_cons(2).map { |s| s.join('-') }
  #=> ["a-b", "b-c", "c-d"]

or

"a-b-c-d".each_char.each_cons(3).select.with_index { |_,i| i.even? }.map(&:join)
  #=> ["a-b", "b-c", "c-d"]

or

enum = "a-b-c-d".each_char
a = []
loop do
  a << "%s%s%s" % [enum.next, enum.next, enum.peek]
end
a #=> ["a-b", "b-c", "c-d"]

Rabbin

Your last technique is very very neat. – Haploid

Nice answer, but I was explicitly asking for a regexp. The pattern above was just the simplest one I found to explain my problem. The pattern I actually use, is (\w)(?:(?!\1)(\w))\1. – Mil