renamed Http_client to Http_user_agent to avoid clashes with Gerd's

[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 5363d2180e72426f31bd329eeaf3f406da343db0..8a6faa19339fe2979b75b5407ebeca108f551ed6 100644 (file)
--- a/helm/ocaml/cic_disambiguation/cicTextualLexer2.ml
+++ b/helm/ocaml/cic_disambiguation/cicTextualLexer2.ml
@@ -30,31 +30,46 @@ exception Not_an_extended_ident
 let regexp alpha = [ 'a' - 'z' 'A' - 'Z' ]
 let regexp digit = [ '0' - '9' ]
 let regexp blank = [ ' ' '\t' '\n' ]
+let regexp paren = [ '(' '[' '{' ')' ']' '}' ]
+let regexp implicit = '?'
+let regexp symbol_char =
+  [^ 'a' - 'z' 'A' - 'Z' '0' - '9' ' ' '\t' '\n' '\\' '(' '[' '{' ')' ']' '}' ]
 
 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 symbol = symbol_char+
+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 lparen = [ '(' '[' '{' ]
-let regexp rparen = [ ')' ']' '}' ]
-let regexp meta = '?' num
+let regexp meta = implicit 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";
+      "in"; "and" ]
 
 let error lexbuf msg =
   raise (Error (Ulexing.lexeme_start lexbuf, Ulexing.lexeme_end lexbuf, msg))
 let error_at_end lexbuf msg =
   raise (Error (Ulexing.lexeme_end lexbuf, Ulexing.lexeme_end lexbuf, msg))
 
-let return lexbuf token = (token, Ulexing.loc lexbuf)
+let return lexbuf token =
+  let flocation_begin =
+    { Lexing.pos_fname = ""; Lexing.pos_lnum = -1; Lexing.pos_bol = -1;
+      Lexing.pos_cnum = Ulexing.lexeme_start lexbuf }
+  in
+  let flocation_end =
+    { flocation_begin with Lexing.pos_cnum = Ulexing.lexeme_end lexbuf }
+  in
+  (token, (flocation_begin, flocation_end))
 
 (*
 let parse_ext_ident ident =
@@ -87,30 +102,35 @@ 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
         return lexbuf (Hashtbl.find keywords lexeme)
       with Not_found -> return lexbuf ("IDENT", lexeme))
   | num -> return lexbuf ("NUM", Ulexing.utf8_lexeme lexbuf)
-  | lparen -> return lexbuf ("LPAREN", Ulexing.utf8_lexeme lexbuf)
-  | rparen -> return lexbuf ("RPAREN", 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", Utf8Macro.expand macro)
+      with Utf8Macro.Macro_not_found _ ->
         return lexbuf ("SYMBOL", Ulexing.utf8_lexeme lexbuf))
   | eof -> return lexbuf ("EOI", "")
   | _ -> error lexbuf "Invalid character"
 
 let tok_func stream =
   let lexbuf = Ulexing.from_utf8_stream stream in
-  Token.make_stream_and_location
+  Token.make_stream_and_flocation
     (fun () ->
       try
        token lexbuf
@@ -118,7 +138,7 @@ 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 _ -> ());
@@ -127,3 +147,4 @@ let lex =
     Token.tok_text = Token.lexer_text;
     Token.tok_comm = None;
   }
+