[r2002-11-06 10:10:54 by cvscast] Empty log message

Original author: cvscast
Date: 2002-11-06 10:11:15+00:00

Makefile

@@ -5,7 +5,7 @@ CLEAN_DIRS = $(DIRS) tools tests
 
 # Objects to build
 
-MISC = misc/pool.cmo
+MISC = misc/pool.cmo misc/encodings.cmo
 
 PARSER = parser/lexer.cmo parser/location.cmo parser/ast.cmo parser/parser.cmo
 

depend

+misc/encodings.cmo: misc/encodings.cmi 
+misc/encodings.cmx: misc/encodings.cmi 
 misc/pool.cmo: misc/pool.cmi 
 misc/pool.cmx: misc/pool.cmi 
 parser/ast.cmo: parser/location.cmi types/patterns.cmi types/types.cmi 

misc/encodings.ml

+type uchar = int
+
+module type T =
+sig
+  val get: string -> int -> uchar
+  val next: string -> int -> int
+
+  val put: string -> int -> uchar -> int
+  val bytes: uchar -> int
+end
+
+
+module Iso88591 =
+struct
+  let get s i = Char.code s.[i]
+  let next s i = succ i
+  let put s i c = s.[i] <- Char.chr i; succ i
+  let bytes c = 1
+end
+
+module Utf8 =
+struct
+(* TODO: handle 5,6 bytes chars; report malformed UTF-8 *)
+  let get s i =
+    match s.[i] with
+      | '\000'..'\127' as c -> 
+	  Char.code c
+      | '\192'..'\223' as c ->
+	  ((Char.code c - 192) lsl 6) lor
+	  ((Char.code s.[i+1] - 128))
+      | '\224'..'\239' as c ->
+	  ((Char.code c - 192) lsl 12) lor
+	  ((Char.code s.[i+1] - 128) lsl 6) lor
+	  ((Char.code s.[i+2] - 128))
+      | '\240'..'\248' as c ->
+	  ((Char.code c - 192) lsl 18) lor
+	  ((Char.code s.[i+1] - 128) lsl 12) lor
+	  ((Char.code s.[i+2] - 128) lsl 6) lor
+	  ((Char.code s.[i+3] - 128))
+      | _ -> failwith "Malformed UTF-8 bufffer"
+
+  let width = Array.create 256 1
+  let () = 
+    for i = 192 to 223 do width.(i) <- 2 done;
+    for i = 224 to 249 do width.(i) <- 3 done;
+    for i = 240 to 248 do width.(i) <- 4 done
+
+  let next s i =
+    Array.unsafe_get width (Char.code s.[i])
+
+  let put s i c =
+    failwith "Encodings.Utf8.put: not yet implemented"
+
+  let bytes c = 
+    failwith "Encodings.Utf8.bytes: not yet implemented"
+end

misc/encodings.mli

+type uchar = int
+
+module type T =
+sig
+  val get: string -> int -> uchar
+  val next: string -> int -> int
+
+  val put: string -> int -> uchar -> int
+  val bytes: uchar -> int
+end
+
+module Iso88591 : T
+module Utf8 : T