- Python 3 compatibility

CHANGES.rst

@@ -4,6 +4,10 @@ Changelog
 4.0 (unreleased)
 ----------------
 
+- Python 3 compatibility
+
+- add test coverage reporting
+
 - Use `@implementer` class decorator.
 
 - Drop long-deprecated support for uuencoded emails.

setup.py

@@ -40,6 +40,7 @@
     ],
     install_requires=[
         'setuptools',
+        'six',
         'AccessControl',
         'Acquisition',
         'DateTime',

src/Products/MailHost/MailHost.py

@@ -24,7 +24,7 @@
 import logging
 from os.path import realpath
 import re
-import sys
+import six
 from threading import Lock
 import time
 
@@ -51,9 +51,6 @@
 from Products.MailHost.interfaces import IMailHost
 from Products.MailHost.decorator import synchronized
 
-if sys.version_info > (3, 0):
-    basestring = str
-    unicode = str
 
 queue_threads = {}  # maps MailHost path -> queue processor threada
 
@@ -361,20 +358,25 @@ def _encode(body, encode=None):
             mo['Mime-Version'] = '1.0'
     return mo.as_string()
 
+def _string_transform(text, charset=None):
+    if six.PY2 and isinstance(text, six.text_type):
+        # Unicode under Python 2
+        return _try_encode(text, charset)
+
+    if six.PY3 and isinstance(text, bytes):
+        # Already-encoded byte strings which the email module does not like
+        return text.decode(charset)
+
+    return text
 
 def _mungeHeaders(messageText, mto=None, mfrom=None, subject=None,
                   charset=None, msg_type=None):
     """Sets missing message headers, and deletes Bcc.
        returns fixed message, fixed mto and fixed mfrom"""
-    # If we have been given unicode fields, attempt to encode them
-    if isinstance(messageText, unicode):
-        messageText = _try_encode(messageText, charset)
-    if isinstance(mto, unicode):
-        mto = _try_encode(mto, charset)
-    if isinstance(mfrom, unicode):
-        mfrom = _try_encode(mfrom, charset)
-    if isinstance(subject, unicode):
-        subject = _try_encode(subject, charset)
+    messageText = _string_transform(messageText, charset)
+    mto = _string_transform(mto, charset)
+    mfrom = _string_transform(mfrom, charset)
+    subject = _string_transform(subject, charset)
 
     if isinstance(messageText, Message):
         # We already have a message, make a copy to operate on
@@ -403,7 +405,7 @@ def _mungeHeaders(messageText, mto=None, mfrom=None, subject=None,
         mo['Subject'] = '[No Subject]'
 
     if mto:
-        if isinstance(mto, basestring):
+        if isinstance(mto, six.string_types):
             mto = [formataddr(addr) for addr in getaddresses((mto, ))]
         if not mo.get('To'):
             mo['To'] = ', '.join(str(_encode_address_string(e, charset))
@@ -488,12 +490,13 @@ def _encode_address_string(text, charset):
     parts should be encoded appropriately."""
     header = Header()
     name, addr = parseaddr(text)
-    try:
-        name.decode('us-ascii')
-    except UnicodeDecodeError:
-        if charset:
-            charset = Charset(charset)
-            name = charset.header_encode(name)
+    if isinstance(name, six.binary_type):
+        try:
+            name.decode('us-ascii')
+        except UnicodeDecodeError:
+            if charset:
+                charset = Charset(charset)
+                name = charset.header_encode(name)
     # We again replace rather than raise an error or pass an 8bit string
     header.append(formataddr((name, addr)), errors='replace')
     return header

src/Products/MailHost/tests/testMailHost.py

@@ -14,15 +14,12 @@
 """
 
 from email import message_from_string
-import sys
+import six
 import unittest
 
 from Products.MailHost.MailHost import MailHost
 from Products.MailHost.MailHost import MailHostError, _mungeHeaders
 
-if sys.version_info > (3, 0):
-    unicode = str
-
 
 class DummyMailHost(MailHost):
     meta_type = 'Dummy Mail Host'
@@ -273,7 +270,7 @@ def testSendEncodedBody(self):
         # there is a default transfer encoding for the charset, then
         # the content will be encoded and the transfer encoding header
         # will be set.
-        msg = "Here's some encoded t\xc3\xa9xt."
+        msg = b"Here's some encoded t\xc3\xa9xt."
         mailhost = self._makeOne('MailHost')
         mailhost.send(messageText=msg,
                       mto='"Name, Nick" <recipient@domain.com>, '
@@ -297,9 +294,9 @@ def testSendEncodedBody(self):
     def testEncodedHeaders(self):
         # Headers are encoded automatically, email headers are encoded
         # piece-wise to ensure the adresses remain ASCII
-        mfrom = "Jos\xc3\xa9 Andr\xc3\xa9s <jose@example.com>"
-        mto = "Ferran Adri\xc3\xa0 <ferran@example.com>"
-        subject = "\xc2\xbfEsferificaci\xc3\xb3n?"
+        mfrom = b"Jos\xc3\xa9 Andr\xc3\xa9s <jose@example.com>"
+        mto = b"Ferran Adri\xc3\xa0 <ferran@example.com>"
+        subject = b"\xc2\xbfEsferificaci\xc3\xb3n?"
         mailhost = self._makeOne('MailHost')
         mailhost.send(messageText='A message.', mto=mto, mfrom=mfrom,
                       subject=subject, charset='utf-8')
@@ -377,6 +374,7 @@ def testAlreadyEncodedMessageWithCharset(self):
             out['From'],
             '=?utf-8?q?Jos=C3=A9_Andr=C3=A9s?= <jose@example.com>')
 
+    @unittest.skipIf(six.PY3, 'Test not applicable on Python 3.')
     def testUnicodeMessage(self):
         # unicode messages and headers are decoded using the given charset
         msg = unicode("Here's some unencoded <strong>t\xc3\xa9xt</strong>.",
@@ -402,6 +400,7 @@ def testUnicodeMessage(self):
             out.get_payload(),
             "Here's some unencoded <strong>t=C3=A9xt</strong>.")
 
+    @unittest.skipIf(six.PY3, 'Test not applicable on Python 3.')
     def testUnicodeNoEncodingErrors(self):
         # Unicode messages and headers raise errors if no charset is passed to
         # send
@@ -480,7 +479,9 @@ def testExplicitBase64Encoding(self):
         mailhost.send('Date: Sun, 27 Aug 2006 17:00:00 +0200\n\nA Message',
                       mfrom='sender@domain.com',
                       mto='Foo Bar <foo@domain.com>', encode='base64')
-        self.failUnlessEqual(mailhost.sent, """\
+        # Explicitly stripping the output here since the base64 encoder
+        # in Python 3 adds a line break at the end.
+        self.failUnlessEqual(mailhost.sent.strip(), """\
 Date: Sun, 27 Aug 2006 17:00:00 +0200
 Subject: [No Subject]
 To: Foo Bar <foo@domain.com>