Index: Lib/test/test_compiler.py
===================================================================
--- Lib/test/test_compiler.py	(revision 43130)
+++ Lib/test/test_compiler.py	(working copy)
@@ -74,6 +74,32 @@
         self.assertEquals(flatten([1, [2]]), [1, 2])
         self.assertEquals(flatten((1, (2,))), [1, 2])
 
+    def _testErrEnc(self, src, text, offset):
+        try:
+            compile(src, "", "exec")
+        except SyntaxError, e:
+            self.assertEquals(e.offset, offset)
+            self.assertEquals(e.text, text)
+        
+    def testSourceCodeEncodingsError(self):
+        # Test SyntaxError with encoding definition
+        sjis = "print '\x83\x70\x83\x43\x83\x5c\x83\x93', '\n"
+        ascii = "print '12345678', '\n"
+        encdef = "#! -*- coding: ShiftJIS -*-\n"
+
+        # ascii source without encdef
+        self._testErrEnc(ascii, ascii, 19)
+        
+        # ascii source with encdef
+        self._testErrEnc(encdef+ascii, ascii, 19)
+        
+        # non-ascii source with encdef
+        self._testErrEnc(encdef+sjis, sjis, 19)
+        
+        # ShiftJIS source without encdef
+        self._testErrEnc(sjis, sjis, 19)  
+
+
 NOLINENO = (compiler.ast.Module, compiler.ast.Stmt, compiler.ast.Discard)
 
 ###############################################################################
Index: Parser/parsetok.c
===================================================================
--- Parser/parsetok.c	(revision 43130)
+++ Parser/parsetok.c	(working copy)
@@ -16,6 +16,9 @@
 /* Forward */
 static node *parsetok(struct tok_state *, grammar *, int, perrdetail *, int);
 static void initerr(perrdetail *err_ret, const char* filename);
+#ifdef Py_USING_UNICODE
+static PyObject * dec_utf8(const char *enc, const char *text, size_t len);
+#endif
 
 /* Parse input coming from a string.  Return error code, print some errors. */
 node *
@@ -214,16 +217,48 @@
 			err_ret->error = E_EOF;
 		err_ret->lineno = tok->lineno;
 		if (tok->buf != NULL) {
+			char *text = NULL;
 			size_t len;
 			assert(tok->cur - tok->buf < INT_MAX);
 			err_ret->offset = (int)(tok->cur - tok->buf);
 			len = tok->inp - tok->buf;
-			err_ret->text = (char *) PyObject_MALLOC(len + 1);
-			if (err_ret->text != NULL) {
-				if (len > 0)
-					strncpy(err_ret->text, tok->buf, len);
-				err_ret->text[len] = '\0';
+#ifdef Py_USING_UNICODE
+			if (tok->encoding) {
+				/* convert source to original encondig */
+				PyObject *lineobj = dec_utf8(tok->encoding, tok->buf, len);
+				if (lineobj != NULL) {
+					int linelen = PyString_Size(lineobj);
+					const char *line = PyString_AsString(lineobj);
+					text = PyObject_MALLOC(linelen + 1);
+					if (text != NULL && line != NULL) {
+						if (linelen)
+							strncpy(text, line, linelen);
+						text[linelen] = '\0';
+					}
+					Py_DECREF(lineobj);
+					
+					/* adjust error offset */
+					if (err_ret->offset > 1) {
+						PyObject *offsetobj = dec_utf8(tok->encoding, 
+							tok->buf, err_ret->offset-1);
+						if (offsetobj) {
+							err_ret->offset = PyString_Size(offsetobj) + 1;
+							Py_DECREF(offsetobj);
+						}
+					}
+					
+				}
 			}
+#endif
+			if (text == NULL) {
+				text = (char *) PyObject_MALLOC(len + 1);
+				if (text != NULL) {
+					if (len > 0)
+						strncpy(text, tok->buf, len);
+					text[len] = '\0';
+				}
+			}
+			err_ret->text = text;
 		}
 	} else if (tok->encoding != NULL) {
 		node* r = PyNode_New(encoding_decl);
@@ -250,3 +285,19 @@
 	err_ret->token = -1;
 	err_ret->expected = -1;
 }
+
+#ifdef Py_USING_UNICODE
+static PyObject *
+dec_utf8(const char *enc, const char *text, size_t len) {
+	PyObject *ret = NULL;	
+	PyObject *unicode_text = PyUnicode_DecodeUTF8(text, len, "replace");
+	if (unicode_text) {
+		ret = PyUnicode_AsEncodedString(unicode_text, enc, "replace");
+		Py_DECREF(unicode_text);
+	}
+	if (!ret) {
+		PyErr_Print();
+	}
+	return ret;
+}
+#endif