add various functions on CCUtf8_string

src/core/CCUtf8_string.ml

@@ -29,6 +29,8 @@ module Dec = struct
     { s=s; i=idx; len=String.length s; }
 end
 
+let n_bytes = length
+
 exception Malformed of string * int
 (** Malformed string at given offset *)
 
@@ -134,6 +136,8 @@ let fold ?idx f acc s =
   in
   aux acc
 
+let n_chars = fold (fun x _ -> x+1) 0
+
 let to_list ?(idx=0) s : uchar list =
   fold ~idx (fun acc x -> x :: acc) [] s |> List.rev
 
@@ -185,6 +189,27 @@ let of_list l : t =
   List.iter (code_to_string buf) l;
   Buffer.contents buf
 
+let map f s : t =
+  let buf = Buffer.create (n_bytes s) in
+  iter (fun c -> code_to_string buf (f c)) s;
+  Buffer.contents buf
+
+let filter_map f s : t =
+  let buf = Buffer.create (n_bytes s) in
+  iter
+    (fun c -> match f c with
+       | None -> ()
+       | Some c -> code_to_string buf c)
+    s;
+  Buffer.contents buf
+
+let flat_map f s : t =
+  let buf = Buffer.create (n_bytes s) in
+  iter (fun c -> iter (code_to_string buf) (f c)) s;
+  Buffer.contents buf
+
+let append = Pervasives.(^)
+
 let unsafe_of_string s = s
 
 let is_valid (s:string) : bool =

src/core/CCUtf8_string.mli

@@ -53,6 +53,22 @@ val fold : ?idx:int -> ('a -> uchar -> 'a) -> 'a -> t -> 'a
 
 val iter : ?idx:int -> (uchar -> unit) -> t -> unit
 
+val n_chars : t -> int
+(** Number of characters *)
+
+val n_bytes : t -> int
+(** Number of bytes *)
+
+val map : (uchar -> uchar) -> t -> t
+
+val filter_map : (uchar -> uchar option) -> t -> t
+
+val flat_map : (uchar -> t) -> t -> t
+
+val append : t -> t -> t
+
+val concat : t -> t list -> t
+
 val of_seq : uchar sequence -> t
 
 val of_gen : uchar gen -> t