Most Irreal readers have probably already seen Mickey's post on Evaluating Lisp Forms in Regular Expressions, but he mentions one trick that is too good to miss so I'm going to repeat it.
Not too long ago I was trying to find a way to swap several occurrences of two words in a buffer (I think it was for the VimGolf lamb had a little Mary challenge). I did find a way but I wasn't very happy with it. In his post, Mickey shows a way that is, in retrospect, obvious but which didn't occur to me. Suppose you have
Jill and Jack went up the hill to fetch a pail of water. Jill fell down and broke his crown and Jack came tumbling after.
and you want to switch Jill and Jack so that you have the traditional rhyme. Here's Mickey's trick: do a
query-replace-regexp) with a FROM string of
and a TO string of
\,(if \1 "Jack" "Jill")
Notice that we capture only one subgroup and use it to determine which replacement word to use. If
nil then we didn't match
Jill so we must have matched
Jack and we want to use
Jill as the replacement. On the other hand, if
\1 is not
nil then we matched
Jill and want to replace it with
Jack. Very nice.
Mickey's got some other nice tricks you can do with the form-based TO string in
replace-regexp so be sure to read his post if you haven't already.