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

This version of the fix raises an error instead of accepting the invalid
input (ie: if a non-ASCII string is used but no charset is specified).

2 files changed, 42 insertions, 15 deletions

diff --git a/Lib/test/test_email/test_contentmanager.py b/Lib/test/test_email/test_contentmanager.py
index 1629e2d..fa2d3d2 100644
--- a/Lib/test/test_email/test_contentmanager.py
+++ b/Lib/test/test_email/test_contentmanager.py
@@ -208,12 +208,11 @@ class TestRawDataManager(TestEmailBase):
                          "Basìc tëxt.\n")
 
     def test_get_text_plain_utf8_base64_recoverable_bad_CTE_data(self):
-        m = self._str_msg(textwrap.dedent("""\
+        m = self._bytes_msg(textwrap.dedent("""\
             Content-Type: text/plain; charset="utf8"
             Content-Transfer-Encoding: base64
 
-            QmFzw6xjIHTDq3h0Lgo\xFF=
-            """))
+            QmFzw6xjIHTDq3h0Lgo""").encode('ascii') + b'\xFF=\n')
         self.assertEqual(raw_data_manager.get_content(m, errors='ignore'),
                          "Basìc tëxt.\n")
 
diff --git a/Lib/test/test_email/test_email.py b/Lib/test/test_email/test_email.py
index 57f12ae..51a9438 100644
--- a/Lib/test/test_email/test_email.py
+++ b/Lib/test/test_email/test_email.py
@@ -92,6 +92,44 @@ class TestMessageAPI(TestEmailBase):
         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_set_payload_with_non_ascii_and_no_charset_raises(self):
+        data = b'\xd0\x90\xd0\x91\xd0\x92'.decode('utf-8')
+        msg = Message()
+        with self.assertRaises(TypeError):
+            msg.set_payload(data)
+
     def test_get_charsets(self):
         eq = self.assertEqual
 
@@ -558,20 +596,10 @@ class TestMessageAPI(TestEmailBase):
         self.assertIsInstance(msg.defects[0],
                               errors.InvalidBase64CharactersDefect)
 
-    def test_broken_unicode_payload(self):
-        # This test improves coverage but is not a compliance test.
-        # The behavior in this situation is currently undefined by the API.
-        x = 'this is a br\xf6ken thing to do'
-        msg = Message()
-        msg['content-type'] = 'text/plain'
-        msg['content-transfer-encoding'] = '8bit'
-        msg.set_payload(x)
-        self.assertEqual(msg.get_payload(decode=True),
-                         bytes(x, 'raw-unicode-escape'))
-
     def test_questionable_bytes_payload(self):
         # This test improves coverage but is not a compliance test,
-        # since it involves poking inside the black box.
+        # since it involves poking inside the black box in a way
+        # that actually breaks the model invariants.
         x = 'this is a quéstionable thing to do'.encode('utf-8')
         msg = Message()
         msg['content-type'] = 'text/plain; charset="utf-8"'