notation fixed to be NON associative by default

[helm.git] / helm / software / components / content_pres / cicNotationLexer.ml

diff --git a/helm/software/components/content_pres/cicNotationLexer.ml b/helm/software/components/content_pres/cicNotationLexer.ml
index f157af40696876a7f38b65946cb6c72acc588590..00696cc00d2cec8b36e4a3a4a1c8d45000331a52 100644 (file)
--- a/helm/software/components/content_pres/cicNotationLexer.ml
+++ b/helm/software/components/content_pres/cicNotationLexer.ml
@@ -30,7 +30,7 @@ open Printf
 exception Error of int * int * string
 
 let regexp number = xml_digit+
-let regexp utf8_blank = " " | "\n" | "\t" | [160] (* this is a nbsp *)
+let regexp utf8_blank = " " | "\r\n" | "\n" | "\t" | [160] (* this is a nbsp *)
 
   (* ZACK: breaks unicode's binder followed by an ascii letter without blank *)
 (* let regexp ident_letter = xml_letter *)
@@ -50,6 +50,10 @@ let is_ligature_char =
       true
     with Not_found -> false))
 
+let regexp we_proved = "we" utf8_blank+ "proved"
+let regexp we_have = "we" utf8_blank+ "have"
+let regexp let_rec = "let" utf8_blank+ "rec" 
+let regexp let_corec = "let" utf8_blank+  "corec"
 let regexp ident_decoration = '\'' | '?' | '`'
 let regexp ident_cont = ident_letter | xml_digit | '_'
 let regexp ident = ident_letter ident_cont* ident_decoration*
@@ -109,8 +113,8 @@ let level2_meta_keywords =
 let level2_ast_keywords = Hashtbl.create 23
 let _ =
   List.iter (fun k -> Hashtbl.add level2_ast_keywords k ())
-  [ "CProp"; "Prop"; "Type"; "Set"; "let"; "rec"; "corec"; "match";
-    "with"; "in"; "and"; "to"; "as"; "on"; "return" ]
+  [ "CProp"; "Prop"; "Type"; "Set"; "let"; "match";
+  "with"; "in"; "and"; "to"; "as"; "on"; "return"; "done" ]
 
 let add_level2_ast_keyword k = Hashtbl.add level2_ast_keywords k ()
 let remove_level2_ast_keyword k = Hashtbl.remove level2_ast_keywords k
@@ -144,14 +148,8 @@ let error_at_end lexbuf msg =
   raise (Error (begin_cnum, end_cnum, msg))
 
 let return_with_loc token begin_cnum end_cnum =
-  (* TODO handle line/column numbers *)
-  let flocation_begin =
-    { Lexing.pos_fname = "";
-      Lexing.pos_lnum = -1; Lexing.pos_bol = -1;
-      Lexing.pos_cnum = begin_cnum }
-  in
-  let flocation_end = { flocation_begin with Lexing.pos_cnum = end_cnum } in
-  (token, (flocation_begin, flocation_end))
+  let flocation = HExtlib.floc_of_loc (begin_cnum,end_cnum) in
+   token, flocation
 
 let return lexbuf token =
   let begin_cnum, end_cnum = Ulexing.loc lexbuf in
@@ -169,17 +167,17 @@ let mk_lexer token =
 (*     let lexbuf = Ulexing.from_utf8_stream stream in *)
 (** XXX Obj.magic rationale.
  * The problem.
- *  camlp4 constraints the tok_func field of Token.glexer to have type:
+ *  camlp5 constraints the tok_func field of Token.glexer to have type:
  *    Stream.t char -> (Stream.t 'te * flocation_function)
  *  In order to use ulex we have (in theory) to instantiate a new lexbuf each
  *  time a char Stream.t is passed, destroying the previous lexbuf which may
  *  have consumed a character from the old stream which is lost forever :-(
  * The "solution".
- *  Instead of passing to camlp4 a char Stream.t we pass a lexbuf, casting it to
+ *  Instead of passing to camlp5 a char Stream.t we pass a lexbuf, casting it to
  *  char Stream.t with Obj.magic where needed.
  *)
     let lexbuf = Obj.magic stream in
-    Token.make_stream_and_flocation
+    Token.make_stream_and_location
       (fun () ->
         try
           token lexbuf
@@ -280,6 +278,10 @@ let rec ligatures_token k =
 
 and level2_ast_token =
   lexer
+  | let_rec -> return lexbuf ("LETREC","")
+  | let_corec -> return lexbuf ("LETCOREC","")
+  | we_proved -> return lexbuf ("WEPROVED","")
+  | we_have -> return lexbuf ("WEHAVE","")
   | utf8_blank+ -> ligatures_token level2_ast_token lexbuf
   | meta ->
      let s = Ulexing.utf8_lexeme lexbuf in