From 97e200007d84cb6d3fc066f2dec43c9eb3aa3af4 Mon Sep 17 00:00:00 2001
From: Arkadiusz Hiler <arek@hiler.eu>
Date: Mon, 11 May 2020 21:49:21 +0300
Subject: [PATCH] bpo-40597: email: Use CTE if lines are longer than
 max_line_length consistently

RFC5322 in section 2.1.1 mandates that the line cannot be longer than
998 characters and should not be longer than 78 characters (excluding
CRLF).

When we use raw_data_manager (default for EmailPolicy, EmailMessage) it
does the correct thing as long as the message contains characters
outside of 7bit US-ASCII set - base64 or qp Content-Transfer-Encoding is
applied if the lines would be too long without it.

However if our message is limited to the characters from the 7bit US-ASCII
set no transfer encoding is applied, and such messages can easily go
beyond 78 or even 998 characters.

Let's fix the CTE heuristic so it doesn't care about 7bit vs 8bit.
---
 Lib/email/contentmanager.py                       | 14 +++++++-------
 Lib/test/test_email/test_contentmanager.py        | 15 +++++++++++++++
 .../2020-05-11-19-17-23.bpo-40597.4SGfgm.rst      |  1 +
 3 files changed, 23 insertions(+), 7 deletions(-)
 create mode 100644 Misc/NEWS.d/next/Library/2020-05-11-19-17-23.bpo-40597.4SGfgm.rst

diff --git a/Lib/email/contentmanager.py b/Lib/email/contentmanager.py
index b904ded94c92ef..2b4b8757f46f62 100644
--- a/Lib/email/contentmanager.py
+++ b/Lib/email/contentmanager.py
@@ -146,13 +146,13 @@ def embedded_body(lines): return linesep.join(lines) + linesep
     def normal_body(lines): return b'\n'.join(lines) + b'\n'
     if cte==None:
         # Use heuristics to decide on the "best" encoding.
-        try:
-            return '7bit', normal_body(lines).decode('ascii')
-        except UnicodeDecodeError:
-            pass
-        if (policy.cte_type == '8bit' and
-                max(len(x) for x in lines) <= policy.max_line_length):
-            return '8bit', normal_body(lines).decode('ascii', 'surrogateescape')
+        if max(len(x) for x in lines) <= policy.max_line_length:
+            try:
+                return '7bit', normal_body(lines).decode('ascii')
+            except UnicodeDecodeError:
+                pass
+            if policy.cte_type == '8bit':
+                return '8bit', normal_body(lines).decode('ascii', 'surrogateescape')
         sniff = embedded_body(lines[:10])
         sniff_qp = quoprimime.body_encode(sniff.decode('latin-1'),
                                           policy.max_line_length)
diff --git a/Lib/test/test_email/test_contentmanager.py b/Lib/test/test_email/test_contentmanager.py
index 169058eac83da3..64dca2d017e629 100644
--- a/Lib/test/test_email/test_contentmanager.py
+++ b/Lib/test/test_email/test_contentmanager.py
@@ -329,6 +329,21 @@ def test_set_text_charset_latin_1(self):
         self.assertEqual(m.get_payload(decode=True).decode('utf-8'), content)
         self.assertEqual(m.get_content(), content)
 
+    def test_set_text_plain_long_line_heuristics(self):
+        m = self._make_message()
+        content = ("Simple but long message that is over 78 characters"
+                   " long to force transfer encoding.\n")
+        raw_data_manager.set_content(m, content)
+        self.assertEqual(str(m), textwrap.dedent("""\
+            Content-Type: text/plain; charset="utf-8"
+            Content-Transfer-Encoding: quoted-printable
+
+            Simple but long message that is over 78 characters long to =
+            force transfer encoding.
+            """))
+        self.assertEqual(m.get_payload(decode=True).decode('utf-8'), content)
+        self.assertEqual(m.get_content(), content)
+
     def test_set_text_short_line_minimal_non_ascii_heuristics(self):
         m = self._make_message()
         content = "et là il est monté sur moi et il commence à m'éto.\n"
diff --git a/Misc/NEWS.d/next/Library/2020-05-11-19-17-23.bpo-40597.4SGfgm.rst b/Misc/NEWS.d/next/Library/2020-05-11-19-17-23.bpo-40597.4SGfgm.rst
new file mode 100644
index 00000000000000..1b9fe609c25b71
--- /dev/null
+++ b/Misc/NEWS.d/next/Library/2020-05-11-19-17-23.bpo-40597.4SGfgm.rst
@@ -0,0 +1 @@
+If text content lines are longer than policy.max_line_length, always use a content-encoding to make sure they are wrapped.