snapshot, notably:

[helm.git] / helm / ocaml / cic_notation / cicNotationLexer.ml

diff --git a/helm/ocaml/cic_notation/cicNotationLexer.ml b/helm/ocaml/cic_notation/cicNotationLexer.ml
index 92204cfb74221f9ae195c4ee2cd8b29f408a3f9c..b5cd026d6946cc45c2f5e5b811e5be727e847951 100644 (file)
--- a/helm/ocaml/cic_notation/cicNotationLexer.ml
+++ b/helm/ocaml/cic_notation/cicNotationLexer.ml
@@ -29,19 +29,24 @@ exception Error of int * int * string
 
 let regexp number = xml_digit+
 
+  (* ZACK: breaks unicode's binder followed by an ascii letter without blank *)
+(* let regexp ident_letter = xml_letter *)
+
+let regexp ident_letter = [ 'a' - 'z' 'A' - 'Z' ]
+
 let regexp ident_decoration = '\'' | '!' | '?' | '`'
-let regexp ident_cont = xml_letter | xml_digit | '_'
-let regexp ident = xml_letter ident_cont* ident_decoration*
+let regexp ident_cont = ident_letter | xml_digit | '_'
+let regexp ident = ident_letter ident_cont* ident_decoration*
 
 let regexp tex_token = '\\' ident
 
 let regexp delim_begin = "\\["
 let regexp delim_end = "\\]"
 
-let regexp keyword = '"' ident '"'
 let regexp qkeyword = "'" ident "'"
 
 let regexp implicit = '?'
+let regexp placeholder = '%'
 let regexp meta = implicit number
 
 let regexp csymbol = '\'' ident
@@ -55,6 +60,11 @@ let regexp meta_ident = "$" ident
 let regexp meta_anonymous = "$_"
 let regexp qstring = '"' [^ '"']* '"'
 
+let regexp begincomment = "(**" xml_blank
+let regexp endcomment = "*)"
+let regexp comment_char = [^'*'] | '*'[^')']
+let regexp note = "(*" ([^'*'] | "**") comment_char* "*)"
+
 let level1_layouts = 
   [ "sub"; "sup";
     "below"; "above";
@@ -150,8 +160,6 @@ let expand_macro lexbuf =
     ("SYMBOL", Utf8Macro.expand macro)
   with Utf8Macro.Macro_not_found _ -> "SYMBOL", Ulexing.utf8_lexeme lexbuf
 
-let keyword lexbuf = "KEYWORD", remove_quotes (Ulexing.utf8_lexeme lexbuf)
-
 let remove_quotes s = String.sub s 1 (String.length s - 2)
 let remove_left_quote s = String.sub s 1 (String.length s - 1)
 
@@ -196,7 +204,8 @@ let rec level2_meta_token = lexer
 let rec level2_ast_token = lexer
   | xml_blank+ -> level2_ast_token lexbuf
   | meta -> return lexbuf ("META", Ulexing.utf8_lexeme lexbuf)
-  | implicit -> return lexbuf ("IMPLICIT", Ulexing.utf8_lexeme lexbuf)
+  | implicit -> return lexbuf ("IMPLICIT", "")
+  | placeholder -> return lexbuf ("PLACEHOLDER", "")
   | ident ->
       let lexeme = Ulexing.utf8_lexeme lexbuf in
       if Hashtbl.mem level2_ast_keywords lexeme then
@@ -204,7 +213,6 @@ let rec level2_ast_token = lexer
       else
         return lexbuf ("IDENT", lexeme)
   | number -> return lexbuf ("NUMBER", Ulexing.utf8_lexeme lexbuf)
-  | keyword -> return lexbuf (keyword lexbuf)
   | tex_token -> return lexbuf (expand_macro lexbuf)
   | uri -> return lexbuf ("URI", Ulexing.utf8_lexeme lexbuf)
   | qstring ->
@@ -218,8 +226,15 @@ let rec level2_ast_token = lexer
       return lexbuf ("UNPARSED_META",
         remove_left_quote (Ulexing.utf8_lexeme lexbuf))
   | meta_anonymous -> return lexbuf ("UNPARSED_META", "anonymous")
-  | _ -> return lexbuf ("SYMBOL", Ulexing.utf8_lexeme lexbuf)
+  | note -> 
+      let comment =
+        Ulexing.utf8_sub_lexeme lexbuf 2 (Ulexing.lexeme_length lexbuf - 4)
+      in
+      return lexbuf ("NOTE", comment)
+  | begincomment -> return lexbuf ("BEGINCOMMENT","")
+  | endcomment -> return lexbuf ("ENDCOMMENT","")
   | eof -> return lexbuf ("EOI", "")
+  | _ -> return lexbuf ("SYMBOL", Ulexing.utf8_lexeme lexbuf)
 
 let rec level1_pattern_token = lexer
   | xml_blank+ -> level1_pattern_token lexbuf