On Sun, May 25, 2014 at 4:08 AM, John Cowan <xxxxxx@mercury.ccil.org> wrote:

Maybe `regex-replace` should accept an optional integer argument
indicating which occurrence in the main string to replace.

I've implemented this in the reference implementation:

(test "-abc \t\n d ef "

(regexp-replace '(+ space) " abc \t\n d ef " "-" 0))

(test "-abc \t\n d ef "

(regexp-replace '(+ space) " abc \t\n d ef " "-" 0 #f 0))

(test " abc-d ef "

(regexp-replace '(+ space) " abc \t\n d ef " "-" 0 #f 1))

(test " abc \t\n d-ef "

(regexp-replace '(+ space) " abc \t\n d ef " "-" 0 #f 2))

(test " abc \t\n d ef-"

(regexp-replace '(+ space) " abc \t\n d ef " "-" 0 #f 3))

(test " abc \t\n d ef "

(regexp-replace '(+ space) " abc \t\n d ef " "-" 0 #f 4))

(test " abc d ef " (regexp-replace-all '(+ space) " abc \t\n d ef " " "))

There are three optional positional parameters which is

a little on the unwieldy side:

(regexp-replace re str subst [start end count])

but for ease of use I'm allowing #f for the end (0 is short

enough for the start).

Alex