[r2004-06-25 12:15:47 by afrisch] Clean lexer global state in case of error

Original author: afrisch
Date: 2004-06-25 12:15:48+00:00

depend

@@ -220,18 +220,42 @@ driver/librarian.cmx: types/builtin.cmx compile/compile.cmx \
     misc/encodings.cmx runtime/eval.cmx types/ident.cmx compile/lambda.cmx \
     parser/location.cmx parser/parser.cmx misc/serialize.cmx typing/typer.cmx \
     types/types.cmx runtime/value.cmx driver/librarian.cmi 
+cdo2cmo/ml_types.cmo: types/types.cmi 
+cdo2cmo/ml_types.cmx: types/types.cmx 
+cdo2cmo/ml_cduce.cmo: types/atoms.cmi types/builtin_defs.cmi types/chars.cmi \
+    compile/compile.cmi misc/encodings.cmi types/ident.cmo \
+    types/intervals.cmi compile/lambda.cmo driver/librarian.cmi \
+    cdo2cmo/ml_types.cmo misc/ns.cmi types/sequence.cmi typing/typer.cmi \
+    types/types.cmi 
+cdo2cmo/ml_cduce.cmx: types/atoms.cmx types/builtin_defs.cmx types/chars.cmx \
+    compile/compile.cmx misc/encodings.cmx types/ident.cmx \
+    types/intervals.cmx compile/lambda.cmx driver/librarian.cmx \
+    cdo2cmo/ml_types.cmx misc/ns.cmx types/sequence.cmx typing/typer.cmx \
+    types/types.cmx 
+cdo2cmo/ml_ocaml.cmo: driver/librarian.cmi cdo2cmo/ml_types.cmo 
+cdo2cmo/ml_ocaml.cmx: driver/librarian.cmx cdo2cmo/ml_types.cmx 
+cdo2cmo/ml_checker.cmo: cdo2cmo/ml_cduce.cmo cdo2cmo/ml_ocaml.cmo \
+    cdo2cmo/ml_types.cmo 
+cdo2cmo/ml_checker.cmx: cdo2cmo/ml_cduce.cmx cdo2cmo/ml_ocaml.cmx \
+    cdo2cmo/ml_types.cmx 
+cdo2cmo/ml_generator.cmo: cdo2cmo/ml_cduce.cmo cdo2cmo/ml_ocaml.cmo \
+    cdo2cmo/ml_types.cmo 
+cdo2cmo/ml_generator.cmx: cdo2cmo/ml_cduce.cmx cdo2cmo/ml_ocaml.cmx \
+    cdo2cmo/ml_types.cmx 
 driver/cduce.cmo: parser/ast.cmo types/builtin.cmi compile/compile.cmi \
     misc/encodings.cmi runtime/eval.cmi runtime/explain.cmi types/ident.cmo \
-    driver/librarian.cmi parser/location.cmi misc/ns.cmi parser/parser.cmi \
-    types/patterns.cmi types/sample.cmi schema/schema_common.cmi \
-    misc/state.cmi typing/typer.cmi types/types.cmi parser/ulexer.cmi \
-    runtime/value.cmi driver/cduce.cmi 
+    driver/librarian.cmi parser/location.cmi cdo2cmo/ml_cduce.cmo \
+    cdo2cmo/ml_checker.cmo cdo2cmo/ml_generator.cmo cdo2cmo/ml_ocaml.cmo \
+    misc/ns.cmi parser/parser.cmi types/patterns.cmi types/sample.cmi \
+    schema/schema_common.cmi misc/state.cmi typing/typer.cmi types/types.cmi \
+    parser/ulexer.cmi runtime/value.cmi driver/cduce.cmi 
 driver/cduce.cmx: parser/ast.cmx types/builtin.cmx compile/compile.cmx \
     misc/encodings.cmx runtime/eval.cmx runtime/explain.cmx types/ident.cmx \
-    driver/librarian.cmx parser/location.cmx misc/ns.cmx parser/parser.cmx \
-    types/patterns.cmx types/sample.cmx schema/schema_common.cmx \
-    misc/state.cmx typing/typer.cmx types/types.cmx parser/ulexer.cmx \
-    runtime/value.cmx driver/cduce.cmi 
+    driver/librarian.cmx parser/location.cmx cdo2cmo/ml_cduce.cmx \
+    cdo2cmo/ml_checker.cmx cdo2cmo/ml_generator.cmx cdo2cmo/ml_ocaml.cmx \
+    misc/ns.cmx parser/parser.cmx types/patterns.cmx types/sample.cmx \
+    schema/schema_common.cmx misc/state.cmx typing/typer.cmx types/types.cmx \
+    parser/ulexer.cmx runtime/value.cmx driver/cduce.cmi 
 query/query.cmo: parser/ast.cmo types/atoms.cmi types/builtin_defs.cmi \
     types/chars.cmi types/ident.cmo types/intervals.cmi parser/location.cmi \
     parser/parser.cmi types/types.cmi query/query.cmi 

parser/ulexer.ml

@@ -144,6 +144,12 @@ and string start double = lexer
 let lexbuf = ref None
 let last_tok = ref ("","")
 
+let raise_clean e =
+  clear_buff ();
+  in_comment := false;
+  (* reinit encoding ? *)
+  raise e
+
 let tok_func cs =
   let lb = L.from_var_enc_stream enc cs in
   (lexer ("#!" [^ '\n']* "\n")? -> ()) lb;
@@ -153,11 +159,13 @@ let tok_func cs =
       try token lb
       with
 	| Ulexing.Error -> 
-	    raise (Error (Ulexing.lexeme_end lb, Ulexing.lexeme_end lb,
+	    raise_clean (Error (Ulexing.lexeme_end lb, Ulexing.lexeme_end lb,
 			  "Unexpected character"))
 	| Ulexing.InvalidCodepoint i ->
-	    raise (Error (Ulexing.lexeme_end lb, Ulexing.lexeme_end lb,
-			  "Code point invalid for the current encoding")) in
+	    raise_clean (Error (Ulexing.lexeme_end lb, Ulexing.lexeme_end lb,
+			  "Code point invalid for the current encoding"))
+	| e -> raise_clean e
+    in
     last_tok := fst tok;
     tok
   in