diff -r 7b41f5e83732 Parser/tokenizer.c
--- a/Parser/tokenizer.c	Thu Nov 07 13:33:36 2013 +0100
+++ b/Parser/tokenizer.c	Thu Nov 07 13:37:00 2013 +0100
@@ -143,6 +143,7 @@ tok_new(void)
     tok->decoding_readline = NULL;
     tok->decoding_buffer = NULL;
 #endif
+    tok->input_is_utf8 = 0;
     return tok;
 }
 
@@ -195,32 +196,39 @@ error_ret(struct tok_state *tok) /* XXX 
 static char *
 get_normal_name(char *s)        /* for utf-8 and latin-1 */
 {
-    char buf[13];
-    int i;
-    for (i = 0; i < 12; i++) {
-        int c = s[i];
-        if (c == '\0')
-            break;
-        else if (c == '_')
-            buf[i] = '-';
-        else
-            buf[i] = tolower(c);
-    }
-    buf[i] = '\0';
-    if (strcmp(buf, "utf-8") == 0 ||
-        strncmp(buf, "utf-8-", 6) == 0)
+    char lower[13];
+    extern int _Py_normalize_encoding(const char *, char *, size_t);
+
+    if (!_Py_normalize_encoding(s, lower, sizeof(lower)))
+        return s;
+
+    if (strcmp(lower, "utf-8") == 0 || strcmp(lower, "utf8") == 0)
         return "utf-8";
-    else if (strcmp(buf, "latin-1") == 0 ||
-             strcmp(buf, "iso-8859-1") == 0 ||
-             strcmp(buf, "iso-latin-1") == 0 ||
-             strncmp(buf, "latin-1-", 8) == 0 ||
-             strncmp(buf, "iso-8859-1-", 11) == 0 ||
-             strncmp(buf, "iso-latin-1-", 12) == 0)
+    else if (strcmp(lower, "ascii") == 0)
+        return "ascii";
+    else if (strcmp(lower, "latin-1") == 0 ||
+             strcmp(lower, "iso-8859-1") == 0 ||
+             strcmp(lower, "iso-latin-1") == 0 ||
+             strncmp(lower, "latin-1-", 8) == 0 ||
+             strncmp(lower, "iso-8859-1-", 11) == 0 ||
+             strncmp(lower, "iso-latin-1-", 12) == 0)
         return "iso-8859-1";
     else
         return s;
 }
 
+static int
+encoding_is_utf8(char *name)
+{
+    char *norm;
+    norm = get_normal_name(name);
+    if (strcmp(norm, "utf-8") == 0)
+        return 1;
+    if (strcmp(norm, "ascii") == 0)
+        return 1;
+    return 0;
+}
+
 /* Return the coding spec in S, or NULL if none is found.  */
 
 static int
@@ -293,9 +301,12 @@ check_coding_spec(const char* line, Py_s
     tok->read_coding_spec = 1;
     if (tok->encoding == NULL) {
         assert(tok->decoding_state == STATE_RAW);
-        if (strcmp(cs, "utf-8") == 0) {
+
+        if (encoding_is_utf8(cs) == 0) {
             tok->encoding = cs;
-        } else {
+            tok->input_is_utf8 = 1;
+        }
+        else {
             r = set_readline(tok, cs);
             if (r) {
                 tok->encoding = cs;
@@ -379,6 +390,7 @@ check_bom(int get_char(struct tok_state 
     tok->encoding = new_string("utf-8", 5, tok);
     if (!tok->encoding)
         return 0;
+    tok->input_is_utf8 = 1;
     /* No need to set_readline: input is already utf-8 */
     return 1;
 }
@@ -814,6 +826,7 @@ PyTokenizer_FromUTF8(const char *str, in
         return NULL;
     }
     strcpy(tok->encoding, "utf-8");
+    tok->input_is_utf8 = 1;
 
     /* XXX: constify members. */
     tok->buf = tok->cur = tok->end = tok->inp = (char*)str;
@@ -826,7 +839,10 @@ struct tok_state *
 PyTokenizer_FromFile(FILE *fp, const char* enc,
                      const char *ps1, const char *ps2)
 {
-    struct tok_state *tok = tok_new();
+    struct tok_state *tok;
+    extern int _Py_normalize_encoding(const char *, char *, size_t);
+
+    tok = tok_new();
     if (tok == NULL)
         return NULL;
     if ((tok->buf = (char *)PyMem_MALLOC(BUFSIZ)) == NULL) {
@@ -848,6 +864,8 @@ PyTokenizer_FromFile(FILE *fp, const cha
         }
         strcpy(tok->encoding, enc);
         tok->decoding_state = STATE_NORMAL;
+
+        tok->input_is_utf8 = encoding_is_utf8(tok->encoding);
     }
     return tok;
 }
@@ -911,11 +929,12 @@ tok_nextc(struct tok_state *tok)
                     return EOF;
                 newtok = translated;
             }
-            if (tok->encoding && newtok && *newtok) {
+            if (!tok->input_is_utf8 && tok->encoding && newtok && *newtok) {
                 /* Recode to UTF-8 */
                 Py_ssize_t buflen;
                 const char* buf;
-                PyObject *u = translate_into_utf8(newtok, tok->encoding);
+                PyObject *u;
+                u = translate_into_utf8(newtok, tok->encoding);
                 PyMem_FREE(newtok);
                 if (!u) {
                     tok->done = E_DECODE;
diff -r 7b41f5e83732 Parser/tokenizer.h
--- a/Parser/tokenizer.h	Thu Nov 07 13:33:36 2013 +0100
+++ b/Parser/tokenizer.h	Thu Nov 07 13:37:00 2013 +0100
@@ -65,6 +65,7 @@ struct tok_state {
     const char* enc;        /* Encoding for the current str. */
     const char* str;
     const char* input; /* Tokenizer's newline translated copy of the string. */
+    int input_is_utf8;   /* input byte string is encoded to UTF-8? */
 };
 
 extern struct tok_state *PyTokenizer_FromString(const char *, int);