#19063: partially fix set_payload handling of non-ASCII string input.

bitdancer · bitdancer · commit d5c4c7411af6 · 2013-12-11T16:34:34.000-05:00
This is a backward compatible partial fix, the complete fix requires raising
an error instead of accepting the invalid input, so the real fix is only
suitable for 3.4.
diff --git a/Lib/email/charset.py b/Lib/email/charset.py
@@ -386,7 +386,8 @@ def body_encode(self, string):
         string using the ascii codec produces the correct string version
         of the content.
         """
-        # 7bit/8bit encodings return the string unchanged (module conversions)
+        if not string:
+            return string
         if self.body_encoding is BASE64:
             if isinstance(string, str):
                 string = string.encode(self.output_charset)
@@ -398,13 +399,9 @@ def body_encode(self, string):
             # character set, then, we must turn it into pseudo bytes via the
             # latin1 charset, which will encode any byte as a single code point
             # between 0 and 255, which is what body_encode is expecting.
-            #
-            # Note that this clause doesn't handle the case of a _payload that
-            # is already bytes.  It never did, and the semantics of _payload
-            # being bytes has never been nailed down, so fixing that is a
-            # longer term TODO.
             if isinstance(string, str):
-                string = string.encode(self.output_charset).decode('latin1')
+                string = string.encode(self.output_charset)
+            string = string.decode('latin1')
             return email.quoprimime.body_encode(string)
         else:
             if isinstance(string, str):
diff --git a/Lib/email/message.py b/Lib/email/message.py
@@ -275,9 +275,19 @@ def set_payload(self, payload, charset=None):
         Optional charset sets the message's default character set.  See
         set_charset() for details.
         """
-        if isinstance(payload, bytes):
-            payload = payload.decode('ascii', 'surrogateescape')
-        self._payload = payload
+        if hasattr(payload, 'encode'):
+            if charset is None:
+                # We should check for ASCII-only here, but we can't do that
+                # for backward compatibility reasons.  Fixed in 3.4.
+                self._payload = payload
+                return
+            if not isinstance(charset, Charset):
+                charset = Charset(charset)
+            payload = payload.encode(charset.output_charset)
+        if hasattr(payload, 'decode'):
+            self._payload = payload.decode('ascii', 'surrogateescape')
+        else:
+            self._payload = payload
         if charset is not None:
             self.set_charset(charset)
 
@@ -316,7 +326,15 @@ def set_charset(self, charset):
             try:
                 cte(self)
             except TypeError:
-                self._payload = charset.body_encode(self._payload)
+                # This if is for backward compatibility and will be removed
+                # in 3.4 when the ascii check is added to set_payload.
+                payload = self._payload
+                if payload:
+                    try:
+                        payload = payload.encode('ascii', 'surrogateescape')
+                    except UnicodeError:
+                        payload = payload.encode(charset.output_charset)
+                self._payload = charset.body_encode(payload)
                 self.add_header('Content-Transfer-Encoding', cte)
 
     def get_charset(self):
diff --git a/Lib/test/test_email/test_email.py b/Lib/test/test_email/test_email.py
@@ -92,6 +92,38 @@ def test_set_payload_with_charset(self):
         msg.set_payload('This is a string payload', charset)
         self.assertEqual(msg.get_charset().input_charset, 'iso-8859-1')
 
+    def test_set_payload_with_8bit_data_and_charset(self):
+        data = b'\xd0\x90\xd0\x91\xd0\x92'
+        charset = Charset('utf-8')
+        msg = Message()
+        msg.set_payload(data, charset)
+        self.assertEqual(msg['content-transfer-encoding'], 'base64')
+        self.assertEqual(msg.get_payload(decode=True), data)
+        self.assertEqual(msg.get_payload(), '0JDQkdCS\n')
+
+    def test_set_payload_with_non_ascii_and_charset_body_encoding_none(self):
+        data = b'\xd0\x90\xd0\x91\xd0\x92'
+        charset = Charset('utf-8')
+        charset.body_encoding = None # Disable base64 encoding
+        msg = Message()
+        msg.set_payload(data.decode('utf-8'), charset)
+        self.assertEqual(msg['content-transfer-encoding'], '8bit')
+        self.assertEqual(msg.get_payload(decode=True), data)
+
+    def test_set_payload_with_8bit_data_and_charset_body_encoding_none(self):
+        data = b'\xd0\x90\xd0\x91\xd0\x92'
+        charset = Charset('utf-8')
+        charset.body_encoding = None # Disable base64 encoding
+        msg = Message()
+        msg.set_payload(data, charset)
+        self.assertEqual(msg['content-transfer-encoding'], '8bit')
+        self.assertEqual(msg.get_payload(decode=True), data)
+
+    def test_set_payload_to_list(self):
+        msg = Message()
+        msg.set_payload([])
+        self.assertEqual(msg.get_payload(), [])
+
     def test_get_charsets(self):
         eq = self.assertEqual
 
diff --git a/Misc/NEWS b/Misc/NEWS
@@ -23,6 +23,11 @@ Core and Builtins
 Library
 -------
 
+- Issue #19063: if a Charset's body_encoding was set to None, the email
+  package would generate a message claiming the Content-Transfer-Encoding
+  was 7bit, and produce garbage output for the content.  This now works.
+  A couple of other set_payload mishandlings of non-ASCII are also fixed.
+
 - Issue #17200: telnetlib's read_until and expect timeout was broken by the
   fix to Issue #14635 in Python 3.3.0 to be interpreted as milliseconds
   instead of seconds when the platform supports select.poll (ie: everywhere).
