Use table, replace with 'ascii'

Author: StanFromIreland

Lib/email/_header_value_parser.py

@@ -796,14 +796,14 @@ def params(self):
                         value = urllib.parse.unquote(value, encoding='latin-1')
                     else:
                         try:
-                            charset = utils._sanitize_charset_name(charset, 'us-ascii')
+                            charset = utils._sanitize_charset_name(charset, 'ascii')
                             value = value.decode(charset, 'surrogateescape')
                         except (LookupError, UnicodeEncodeError):
                             # XXX: there should really be a custom defect for
                             # unknown character set to make it easy to find,
                             # because otherwise unknown charset is a silent
                             # failure.
-                            value = value.decode('us-ascii', 'surrogateescape')
+                            value = value.decode('ascii', 'surrogateescape')
                         if utils._has_surrogates(value):
                             param.defects.append(errors.UndecodableBytesDefect())
                 value_parts.append(value)

Lib/email/utils.py

@@ -446,17 +446,16 @@ def decode_params(params):
                 new_params.append((name, '"%s"' % value))
     return new_params
 
+_SANITIZE_TABLE = str.maketrans({i: None for i in range(128, 65536)})
+
 def _sanitize_charset_name(charset, fallback_charset):
     if not charset:
         return charset
-    sanitized = ''.join(
-        c for c in charset
-        if (ord(c) < 0xDC80 or ord(c) > 0xDCFF) and c.isascii()
-    )
+    sanitized = charset.translate(_SANITIZE_TABLE)
     return sanitized if sanitized else fallback_charset
 
 def collapse_rfc2231_value(value, errors='replace',
-                           fallback_charset='us-ascii'):
+                           fallback_charset='ascii'):
     if not isinstance(value, tuple) or len(value) != 3:
         return unquote(value)
     # While value comes to us as a unicode string, we need it to be a bytes