#1368247: make set_charset/MIMEText automatically encode unicode _payload.

Fixes (mysterious, to the end user) UnicodeErrors when using utf-8 as
the charset and unicode as the _text argument.  Also makes the way in
which unicode gets encoded to quoted printable for other charsets more
sane (it only worked by accident previously).  The _payload now is encoded
to the charset.output_charset if it is unicode.

5 files changed, 39 insertions, 6 deletions

diff --git a/Doc/library/email.message.rst b/Doc/library/email.message.rst
index 5ebc960..fa1df88 100644
--- a/Doc/library/email.message.rst
+++ b/Doc/library/email.message.rst
@@ -136,9 +136,10 @@ Here are the methods of the :class:`Message` class:
       :mailheader:`Content-Type` header. Anything else will generate a
       :exc:`TypeError`.
 
-      The message will be assumed to be of type :mimetype:`text/\*` encoded with
-      *charset.input_charset*.  It will be converted to *charset.output_charset*
-      and encoded properly, if needed, when generating the plain text
+      The message will be assumed to be of type :mimetype:`text/\*`, with the
+      payload either in unicode or encoded with *charset.input_charset*.
+      It will be encoded or converted to *charset.output_charset*
+      and transfer encoded properly, if needed, when generating the plain text
       representation of the message.  MIME headers (:mailheader:`MIME-Version`,
       :mailheader:`Content-Type`, :mailheader:`Content-Transfer-Encoding`) will
       be added as needed.
diff --git a/Doc/library/email.mime.rst b/Doc/library/email.mime.rst
index 10f3e37..a092feb 100644
--- a/Doc/library/email.mime.rst
+++ b/Doc/library/email.mime.rst
@@ -191,9 +191,11 @@ Here are the classes:
    minor type and defaults to :mimetype:`plain`.  *_charset* is the character
    set of the text and is passed as a parameter to the
    :class:`~email.mime.nonmultipart.MIMENonMultipart` constructor; it defaults
-   to ``us-ascii``.  No guessing or encoding is performed on the text data.
+   to ``us-ascii``.  If *_text* is unicode, it is encoded using the
+   *output_charset* of *_charset*, otherwise it is used as-is.
 
    .. versionchanged:: 2.4
-      The previously deprecated *_encoding* argument has been removed.  Encoding
-      happens implicitly based on the *_charset* argument.
+      The previously deprecated *_encoding* argument has been removed.  Content
+      Transfer Encoding now happens happens implicitly based on the *_charset*
+      argument.
 
diff --git a/Lib/email/message.py b/Lib/email/message.py
index 993a1ac..08423cd 100644
--- a/Lib/email/message.py
+++ b/Lib/email/message.py
@@ -256,6 +256,8 @@ class Message:
                             charset=charset.get_output_charset())
         else:
             self.set_param('charset', charset.get_output_charset())
+        if isinstance(self._payload, unicode):
+            self._payload = self._payload.encode(charset.output_charset)
         if str(charset) != charset.get_output_charset():
             self._payload = charset.body_encode(self._payload)
         if 'Content-Transfer-Encoding' not in self:
diff --git a/Lib/email/test/test_email.py b/Lib/email/test/test_email.py
index bf41be7..7d01079 100644
--- a/Lib/email/test/test_email.py
+++ b/Lib/email/test/test_email.py
@@ -1045,6 +1045,31 @@ class TestMIMEText(unittest.TestCase):
         eq(msg.get_charset().input_charset, 'us-ascii')
         eq(msg['content-type'], 'text/plain; charset="us-ascii"')
 
+    def test_7bit_unicode_input(self):
+        eq = self.assertEqual
+        msg = MIMEText(u'hello there', _charset='us-ascii')
+        eq(msg.get_charset().input_charset, 'us-ascii')
+        eq(msg['content-type'], 'text/plain; charset="us-ascii"')
+
+    def test_7bit_unicode_input_no_charset(self):
+        eq = self.assertEqual
+        msg = MIMEText(u'hello there')
+        eq(msg.get_charset(), 'us-ascii')
+        eq(msg['content-type'], 'text/plain; charset="us-ascii"')
+        self.assertTrue('hello there' in msg.as_string())
+
+    def test_8bit_unicode_input(self):
+        teststr = u'\u043a\u0438\u0440\u0438\u043b\u0438\u0446\u0430'
+        eq = self.assertEqual
+        msg = MIMEText(teststr, _charset='utf-8')
+        eq(msg.get_charset().output_charset, 'utf-8')
+        eq(msg['content-type'], 'text/plain; charset="utf-8"')
+        eq(msg.get_payload(decode=True), teststr.encode('utf-8'))
+
+    def test_8bit_unicode_input_no_charset(self):
+        teststr = u'\u043a\u0438\u0440\u0438\u043b\u0438\u0446\u0430'
+        self.assertRaises(UnicodeEncodeError, MIMEText, teststr)
+
 
 
 # Test complicated multipart/* messages
diff --git a/Misc/NEWS b/Misc/NEWS
index 4e0b0d2..cd9da37 100644
--- a/Misc/NEWS
+++ b/Misc/NEWS
@@ -46,6 +46,9 @@ C-API
 Library
 -------
 
+- Issue #1368247: set_charset (and therefore MIMEText) now automatically
+  encodes a unicode _payload to the output_charset.
+
 - Issue #7150: Raise OverflowError if the result of adding or subtracting
   timedelta from date or datetime falls outside of the MINYEAR:MAXYEAR range.