diff --git a/Lib/encodings/__init__.py b/Lib/encodings/__init__.py
index b189bd9..469c5cf 100644
--- a/Lib/encodings/__init__.py
+++ b/Lib/encodings/__init__.py
@@ -55,15 +55,12 @@ def normalize_encoding(encoding):
     if isinstance(encoding, bytes):
         encoding = str(encoding, "ascii")
     chars = []
-    punct = False
     for c in encoding:
-        if c.isalnum() or c == '.':
-            if punct and chars:
-                chars.append('_')
+        if '0' <= c <= '9' or 'a' <= c <= 'z':
             chars.append(c)
-            punct = False
-        else:
-            punct = True
+        elif 'A' <= c <= 'Z':
+            chars.append(c.lower())
+        # else: skip characters
     return ''.join(chars)
 
 def search_function(encoding):
diff --git a/Objects/unicodeobject.c b/Objects/unicodeobject.c
index e7bbd80..84b257c 100644
--- a/Objects/unicodeobject.c
+++ b/Objects/unicodeobject.c
@@ -1419,34 +1419,32 @@ PyObject *PyUnicode_FromEncodedObject(register PyObject *obj,
     return v;
 }
 
-/* Convert encoding to lower case and replace '_' with '-' in order to
-   catch e.g. UTF_8. Return 0 on error (encoding is longer than lower_len-1),
-   1 on success. */
+/* Convert encoding to lower case and skip characters different than a-z and
+   0-9. Return 0 on error (encoding is longer than lower_len-1), 1 on success.
+
+   See also normalizestring() in codecs and normalize_encoding() in encodings.
+ */
 static int
 normalize_encoding(const char *encoding,
                    char *lower,
                    size_t lower_len)
 {
     const char *e;
-    char *l;
+    char *l, c;
     char *l_end;
 
     e = encoding;
     l = lower;
     l_end = &lower[lower_len - 1];
-    while (*e) {
+    for (; *e; e++) {
         if (l == l_end)
             return 0;
-        if (Py_ISUPPER(*e)) {
-            *l++ = Py_TOLOWER(*e++);
-        }
-        else if (*e == '_') {
-            *l++ = '-';
-            e++;
-        }
-        else {
-            *l++ = *e++;
-        }
+        c = *e;
+        if (('a' <= c && c <= 'z') || ('0' <= c && c <= '9'))
+            *l++ = c;
+        else if ('A' <= c && c <= 'Z')
+            *l++ = c + ('a' - 'A');
+        /* else: ignore the character */
     }
     *l = '\0';
     return 1;
@@ -1466,10 +1464,10 @@ PyObject *PyUnicode_Decode(const char *s,
 
     /* Shortcuts for common default encodings */
     if (normalize_encoding(encoding, lower, sizeof(lower))) {
-        if (strcmp(lower, "utf-8") == 0)
+        if (strcmp(lower, "utf8") == 0)
             return PyUnicode_DecodeUTF8(s, size, errors);
-        else if ((strcmp(lower, "latin-1") == 0) ||
-                 (strcmp(lower, "iso-8859-1") == 0))
+        else if ((strcmp(lower, "latin1") == 0) ||
+                 (strcmp(lower, "iso88591") == 0))
             return PyUnicode_DecodeLatin1(s, size, errors);
 #if defined(MS_WINDOWS) && defined(HAVE_USABLE_WCHAR_T)
         else if (strcmp(lower, "mbcs") == 0)
@@ -1477,9 +1475,9 @@ PyObject *PyUnicode_Decode(const char *s,
 #endif
         else if (strcmp(lower, "ascii") == 0)
             return PyUnicode_DecodeASCII(s, size, errors);
-        else if (strcmp(lower, "utf-16") == 0)
+        else if (strcmp(lower, "utf16") == 0)
             return PyUnicode_DecodeUTF16(s, size, errors, 0);
-        else if (strcmp(lower, "utf-32") == 0)
+        else if (strcmp(lower, "utf32") == 0)
             return PyUnicode_DecodeUTF32(s, size, errors, 0);
     }
 
@@ -1674,12 +1672,12 @@ PyObject *PyUnicode_AsEncodedString(PyObject *unicode,
 
     /* Shortcuts for common default encodings */
     if (normalize_encoding(encoding, lower, sizeof(lower))) {
-        if (strcmp(lower, "utf-8") == 0)
+        if (strcmp(lower, "utf8") == 0)
             return PyUnicode_EncodeUTF8(PyUnicode_AS_UNICODE(unicode),
                                         PyUnicode_GET_SIZE(unicode),
                                         errors);
-        else if ((strcmp(lower, "latin-1") == 0) ||
-                 (strcmp(lower, "iso-8859-1") == 0))
+        else if ((strcmp(lower, "latin1") == 0) ||
+                 (strcmp(lower, "iso88591") == 0))
             return PyUnicode_EncodeLatin1(PyUnicode_AS_UNICODE(unicode),
                                           PyUnicode_GET_SIZE(unicode),
                                           errors);
diff --git a/Python/codecs.c b/Python/codecs.c
index 45d9929..856200c 100644
--- a/Python/codecs.c
+++ b/Python/codecs.c
@@ -45,15 +45,18 @@ int PyCodec_Register(PyObject *search_function)
     return -1;
 }
 
-/* Convert a string to a normalized Python string: all characters are
-   converted to lower case, spaces are replaced with underscores. */
+/* Convert encoding to lower case and skip characters different than a-z and
+   0-9.
+
+   See also normalizestring() in codecs and normalize_encoding() in encodings.
+ */
 
 static
 PyObject *normalizestring(const char *string)
 {
     register size_t i;
     size_t len = strlen(string);
-    char *p;
+    char *buffer, *p;
     PyObject *v;
 
     if (len > PY_SSIZE_T_MAX) {
@@ -61,22 +64,22 @@ PyObject *normalizestring(const char *string)
         return NULL;
     }
 
-    p = PyMem_Malloc(len + 1);
-    if (p == NULL)
+    buffer = PyMem_Malloc(len + 1);
+    if (buffer == NULL)
         return NULL;
+    p = buffer;
     for (i = 0; i < len; i++) {
         register char ch = string[i];
-        if (ch == ' ')
-            ch = '-';
-        else
-            ch = tolower(Py_CHARMASK(ch));
-        p[i] = ch;
+
+        if (('a' <= ch && ch <= 'z') || ('0' <= ch && ch <= '9'))
+            *p++ = ch;
+        else if ('A' <= ch && ch <= 'Z')
+            *p++ = ch + ('a' - 'A');
     }
-    p[i] = '\0';
-    v = PyUnicode_FromString(p);
+    v = PyUnicode_FromStringAndSize(buffer, p - buffer);
     if (v == NULL)
         return NULL;
-    PyMem_Free(p);
+    PyMem_Free(buffer);
     return v;
 }
 
diff --git a/Lib/test/test_encodings.py b/Lib/test/test_encodings.py
new file mode 100644
index 0000000..eb9b823
--- /dev/null
+++ b/Lib/test/test_encodings.py
@@ -0,0 +1,28 @@
+from test import support
+import unittest
+import encodings
+import sys, _testcapi, io
+
+class EncodingTests(unittest.TestCase):
+
+    def test_normalize(self):
+        def test(name, expected):
+            self.assertEqual(encodings.normalize_encoding(name), expected)
+
+        test(b'utf8', 'utf8')
+        test('utf8', 'utf8')
+        test('utf-8', 'utf8')
+        test('UTF-8', 'utf8')
+        test('fr_FR.UTF-8', 'frfrutf8')
+        test('UTF-8\xe9', 'utf8')
+        test('[ utf-8 ]', 'utf8')
+
+
+def test_main():
+    support.run_unittest(
+        EncodingTests,
+    )
+
+
+if __name__ == "__main__":
+    test_main()