ruby regex scan versus =~

Question

The Ruby (1.9.3) documentation seems to imply that scan is equivalent to =~ except that

1. scan returns multiple matches, while =~ returns only the first occurrence, and
2. scan returns the match data, while =~ returns the index.

However, in the following example, the two methods seem to return different results for the same string and expression. Why is that?

1.9.3p0 :002 > str = "Perl and Python - the two languages"
 => "Perl and Python - the two languages" 
1.9.3p0 :008 > exp = /P(erl|ython)/
 => /P(erl|ython)/ 
1.9.3p0 :009 > str =~ exp
 => 0 
1.9.3p0 :010 > str.scan exp
 => [["erl"], ["ython"]] 

If the index of first match is 0, shouldn't scan return "Perl" and "Python" instead of "erl" and "python"?

Thanks

Answer 1

When given a regular expression without capturing groups, scan will return an array of strings, where each string represents a match of the regular expression. If you use scan(/P(?:erl|ython)/) (which is the same as your regex except without capturing groups), you'll get ["Perl", "Python"], which is what you expect.

However when given a regex with capturing groups, scan will return an array of arrays, where each sub-array contains the captures of a given match. So if you have for example the regex (\w*):(\w*), you'll get an array of arrays where each sub-array contains two strings: the part before the colon and the part after the colon. And in your example each sub-array contains one string: the part matched by (erl|ython).