first moogle template checkin

[helm.git] / helm / ocaml / cic_disambiguation / cicTextualLexer2.ml

diff --git a/helm/ocaml/cic_disambiguation/cicTextualLexer2.ml b/helm/ocaml/cic_disambiguation/cicTextualLexer2.ml
index 21f0205203aea6e410b54d283a8ed60f1a92bf1f..22c911eaf5b543e16bb7b1a35d21b3a758eee43c 100644 (file)
--- a/helm/ocaml/cic_disambiguation/cicTextualLexer2.ml
+++ b/helm/ocaml/cic_disambiguation/cicTextualLexer2.ml
@@ -35,18 +35,23 @@ let regexp blanks = blank+
 let regexp num = digit+
 let regexp tex_token = '\\' alpha+
 let regexp symbol = [^ 'a' - 'z' 'A' - 'Z' '0' - '9' ' ' '\t' '\n' ]
-let regexp ident_cont = alpha | num | '_'
+let regexp ident_cont = alpha | num | '_' | '\''
 let regexp ident_cont' = ident_cont | tex_token
 let regexp ident = (alpha ident_cont*) | ('_' ident_cont+)
 let regexp ident' = ((alpha | tex_token) ident_cont'*) | ('_' ident_cont'+)
 let regexp paren = [ '(' '[' '{' ')' ']' '}' ]
+let regexp implicit = '?'
 let regexp meta = '?' num
+let regexp qstring = '"' [^ '"']* '"'
+let regexp uri =
+  (*      schema      *) (*     path     *) (*  ext   *) (*    xpointer     *)
+  ("cic:/" | "theory:/") ident ('/' ident)* ('.' ident)+ ('#' num ('/' num)*)?
 (* let regexp catchall = .* *)
 
 let keywords = Hashtbl.create 17
 let _ =
   List.iter (fun keyword -> Hashtbl.add keywords keyword ("", keyword))
-    [ "Prop"; "Type"; "Set"; "let"; "rec"; "using"; "match"; "with" ]
+    [ "Prop"; "Type"; "Set"; "let"; "Let"; "rec"; "using"; "match"; "with" ]
 
 let error lexbuf msg =
   raise (Error (Ulexing.lexeme_start lexbuf, Ulexing.lexeme_end lexbuf, msg))
@@ -86,6 +91,7 @@ and token = lexer
 *)
 let rec token = lexer
   | blanks -> token lexbuf
+  | uri -> return lexbuf ("URI", Ulexing.utf8_lexeme lexbuf)
   | ident ->
       let lexeme = Ulexing.utf8_lexeme lexbuf in
       (try
@@ -94,14 +100,19 @@ let rec token = lexer
   | num -> return lexbuf ("NUM", Ulexing.utf8_lexeme lexbuf)
   | paren -> return lexbuf ("PAREN", Ulexing.utf8_lexeme lexbuf)
   | meta -> return lexbuf ("META", Ulexing.utf8_lexeme lexbuf)
+  | implicit -> return lexbuf ("IMPLICIT", Ulexing.utf8_lexeme lexbuf)
+  | qstring ->
+      let lexeme = Ulexing.utf8_lexeme lexbuf in
+      let s = String.sub lexeme 1 (String.length lexeme - 2) in
+      return lexbuf ("QSTRING", s)
   | symbol -> return lexbuf ("SYMBOL", Ulexing.utf8_lexeme lexbuf)
   | tex_token ->
       let macro =
         Ulexing.utf8_sub_lexeme lexbuf 1 (Ulexing.lexeme_length lexbuf - 1)
       in
       (try
-        return lexbuf ("SYMBOL", Macro.expand macro)
-      with Macro.Macro_not_found _ ->
+        return lexbuf ("SYMBOL", CicTextualParser2Macro.expand macro)
+      with CicTextualParser2Macro.Macro_not_found _ ->
         return lexbuf ("SYMBOL", Ulexing.utf8_lexeme lexbuf))
   | eof -> return lexbuf ("EOI", "")
   | _ -> error lexbuf "Invalid character"
@@ -116,7 +127,7 @@ let tok_func stream =
       | Ulexing.Error -> error_at_end lexbuf "Unexpected character"
       | Ulexing.InvalidCodepoint i -> error_at_end lexbuf "Invalid code point")
 
-let lex =
+let cic_lexer =
   { 
     Token.tok_func = tok_func;
     Token.tok_using = (fun _ -> ());
@@ -125,3 +136,4 @@ let lex =
     Token.tok_text = Token.lexer_text;
     Token.tok_comm = None;
   }
+