[r2003-11-24 14:25:16 by szach] - added concatenation function

- added Pcre module which wraps some Pcre functions so that they can
  work on Utf8.t values

Original author: szach
Date: 2003-11-24 14:25:16+00:00

misc/encodings.ml

@@ -183,4 +183,18 @@ struct
   let serialize_sub t s i j =
     Serialize.Put.substring t s i (j - i)
 
+  let concat s1 s2 = s1 ^ s2
+
+  module Pcre =
+    struct
+      let pcre_replace ~rex ?templ s =
+        match templ with
+        | None -> mk (Pcre.replace ~rex (get_str s))
+        | Some templ ->
+            mk (Pcre.replace ~rex ~templ:(get_str templ) (get_str s))
+      let pcre_extract ~rex s = Array.map mk (Pcre.extract ~rex (get_str s))
+      let pcre_regexp s = Pcre.regexp ~flags:[`UTF8] s
+      let pcre_split ~rex s = List.map mk (Pcre.split ~rex (get_str s))
+    end
+
 end

misc/encodings.mli

@@ -23,9 +23,24 @@ sig
   val advance: t -> uindex -> uindex
   val next: t -> uindex -> uchar * uindex 
 
+  val concat: t -> t -> t
+
   val store: Buffer.t -> uchar -> unit
   val copy: Buffer.t -> t -> uindex -> uindex -> unit
   val get_substr: t -> uindex -> uindex -> string
 
   val serialize_sub: Serialize.Put.t -> t -> uindex -> uindex -> unit
+
+  module Pcre:
+    (* Wrappers for some Pcre functions working on Utf8.t values.
+     * Given Pcre.regexp regular expressions should be compiled with `UTF8 flag
+     * or with pcre_regexp below *)
+  sig
+    val pcre_regexp: string -> Pcre.regexp  (* compile using `UTF8 flag *)
+
+    val pcre_replace: rex:Pcre.regexp -> ?templ:t -> t -> t
+    val pcre_extract: rex:Pcre.regexp -> t -> t array
+    val pcre_split: rex:Pcre.regexp -> t -> t list
+  end
+
 end