How can I match strings using "match with" and regex in OCaml?

Question

My OCaml .ml code looks like this:

open Str

let idregex = Str.regexp ['a'-'z' 'A'-'Z']+ ['a'-'z' 'A'-'Z' '0'-'9' '_']*;

let evalT (x,y) = (match x with 
    Str.regexp "Id(" (idregex as var) ")" -> (x,y)

Why does the above code not work? How can I get it to work?

EDIT:

I don't need to do a lot of parsing. So, I want it to remain in a OCaml .ml file and not a OCamllex file

Answer 1

Does the following work and do what you want?

open Str

let evalT (x, y) =
  let idexp = "[a-zA-Z]+ [a-zA-Z0-9_]*" in
  let full_idexp = "^Id(" ^ idexp ^ ")$" in
  let id_match x = string_match (regexp full_idexp) x 0 in
  if id_match x then Some (x, y) else None;;

I'm an OCaml noob, but your code doesn't look like OCaml at all. I see that your question is very old, but it still comes up high enough on some searches, so I figured it might help someone else if not the OP.

Answer 2

The match keyword works with OCaml patterns. A regex isn't an OCaml pattern, it's a different kind of pattern, so you don't use match for them.

In the same Str module with the regexp function are the matching functions.

If you have a lot of regular expression matching to do, you can use ocamllex, which reads a file of definitions similar to your (unfortunately invalid) definition of idregex, and generates OCaml code to do the matching.

Here's a session showing how to do a simple match of your pattern using the Str module.

$ ocaml
        OCaml version 4.01.0

# #load "str.cma";;
# let idregex = Str.regexp "[a-zA-Z]+[a-zA-Z0-9_]*";;
val idregex : Str.regexp = <abstr>
# Str.string_match idregex "a_32" 0;;
- : bool = true
# Str.string_match idregex "32" 0;;
- : bool = false

As a side comment, your code doesn't really look anything like OCaml. It looks something like a mix of OCaml and ocamllex. There actually is a system a little bit like this called micmatch. It seems you're planning to use the stock OCaml language (which I applaud), but it might be interesting to look at micmatch at some point.