From 6b649aba925b6f7462da07599fe67ebb12a3460e Mon Sep 17 00:00:00 2001
From: Samuel Mimram <samuel.mimram@ens-lyon.org>
Date: Wed, 28 Jul 2004 21:54:47 +0000
Subject: Imported Upstream version 8.0pl1

---
 ide/utf8_convert.mll | 51 +++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 51 insertions(+)
 create mode 100644 ide/utf8_convert.mll

(limited to 'ide/utf8_convert.mll')

diff --git a/ide/utf8_convert.mll b/ide/utf8_convert.mll
new file mode 100644
index 00000000..4c88adc5
--- /dev/null
+++ b/ide/utf8_convert.mll
@@ -0,0 +1,51 @@
+(************************************************************************)
+(*  v      *   The Coq Proof Assistant  /  The Coq Development Team     *)
+(* <O___,, * CNRS-Ecole Polytechnique-INRIA Futurs-Universite Paris Sud *)
+(*   \VV/  **************************************************************)
+(*    //   *      This file is distributed under the terms of the       *)
+(*         *       GNU Lesser General Public License Version 2.1        *)
+(************************************************************************)
+
+(* $Id: utf8_convert.mll,v 1.1.2.1 2004/07/16 19:30:21 herbelin Exp $ *)
+
+{
+  open Lexing 
+  let b = Buffer.create 127
+
+}
+
+(* Replace all occurences of \x{iiii} and \x{iiiiiiii} by UTF-8 valid chars *)
+
+let digit = ['0'-'9''A'-'Z''a'-'z']
+let short = digit digit digit digit
+let long = short short
+
+rule entry = parse
+  | "\\x{" (short | long ) '}'
+      { let s = lexeme lexbuf in
+	let n = String.length s in
+	let code = 
+	  try Glib.Utf8.from_unichar 
+	    (int_of_string ("0x"^(String.sub s 3 (n - 4)))) 
+	  with _ -> s
+	in
+	let c = if Glib.Utf8.validate code then code else s in
+	Buffer.add_string b c;
+	entry lexbuf
+      }
+  | _ 
+      { let s = lexeme lexbuf in
+	Buffer.add_string b s;
+	entry lexbuf}
+  | eof
+      {
+	let s = Buffer.contents b in Buffer.reset b ; s
+      }
+
+
+{
+  let f s =
+   let lb = from_string s in
+   Buffer.reset b;
+   entry lb
+}
-- 
cgit v1.2.3