nsplitp

gistfile1.txt

(* String.nsplitp - split by a predicate on bytes *)

let nsplitp str pred =
  if str = "" then []
  else
    (* str is non empty *)
    let rec aux acc ofs idx =
      (* ofs is current position to check, idx is beginning of last token *)
      if ofs >= 0 then (
        if pred str.[ofs] then (* sep found *)
          let token = sub str (ofs + 1) (idx - ofs) in
          aux (token::acc) (ofs - 1) (ofs - 1)
        else (* sep NOT found at position ofs*)
          aux acc (ofs - 1) idx
      )
      else (* we've run out of string to search, we're done *)
        sub str 0 (idx+1) :: acc
    in
    aux [] (length str - 1 ) (length str - 1)

(*$= nsplitp & ~printer:(IO.to_string (List.print String.print))
  ["a"; "b"; "c"] (nsplitp "a b c" Char.is_whitespace)
*)

(* Text.nsplitp - split by a predicate on unicode characters *)


  let nsplitp str pred =
    if is_empty str then []
    else
    (* str is non empty *)
      let rec aux acc ofs =
        match try Some (rindexp_from str ofs pred)
          with Invalid_rope -> None
        with
          | Some idx -> (* found the delimiter *)
          (* trailing delimiter or multiple seperator characters in a row *)
          if idx = ofs || idx = ofs - 1 then aux acc idx else
            let token = sub str (idx + 1) (ofs - idx) in
            aux (token::acc) (idx - 1)
          | None -> (* pred sep NOT found *)
            sub str 0 (ofs + 1) :: acc
      in
      aux [] (length str - 1)