fix exception in MsgInfo.msg_id when email body contains 8 bit char…

…acters

schwarz/mailqueue/message_utils.py

@@ -6,8 +6,8 @@
 import re
 from datetime import datetime as DateTime, timedelta as TimeDelta
 from email.header import decode_header
-from email.parser import FeedParser, HeaderParser
-from io import BytesIO, TextIOWrapper
+from email.parser import BytesHeaderParser, FeedParser
+from io import BytesIO
 from typing import BinaryIO, NamedTuple, Optional, Sequence
 
 from boltons.timeutils import ConstantTZInfo, LocalTZ
@@ -92,14 +92,11 @@ def __new__(cls, from_addr, to_addrs, msg_fp, queue_date=None, last=None, retrie
     def msg_id(self):
         old_pos = self.msg_fp.tell()
         self.msg_fp.seek(0)
-        # Unfortunately Python's TextIOWrapper always closes wrapped files:
-        #    https://bugs.python.org/issue21363
-        msg_str_fp = TextIOWrapper(UnclosableWrapper(self.msg_fp), encoding='ascii')
-        msg_headers = HeaderParser().parse(msg_str_fp, headersonly=True)
-        # message ids are usually enclosed in angle brackets but these do NOT
-        # belong to the message id.
+        msg_headers = BytesHeaderParser().parse(self.msg_fp, headersonly=True)
         msg_id_value = msg_headers['Message-ID']
         self.msg_fp.seek(old_pos)
+        # message ids are usually enclosed in angle brackets but these do NOT
+        # belong to the message id.
         return strip_brackets(msg_id_value)
 
     @property
@@ -111,18 +108,6 @@ def msg_bytes(self):
         return data
 
 
-class UnclosableWrapper(object):
-    def __init__(self, wrapped_instance):
-        self.wrapped_instance = wrapped_instance
-
-    def __getattr__(self, name):
-        return getattr(self.wrapped_instance, name)
-
-    def close(self):
-        pass
-
-
-
 _re_angle_brackets = re.compile(br'^<?(.+?)>?$')
 _re_angle_brackets_str = re.compile('^<?(.+?)>?$')
 _re_header_list = re.compile(r'\s*,\s*')

tests/message_parsing_test.py

@@ -2,6 +2,7 @@
 # SPDX-License-Identifier: MIT
 
 from datetime import datetime as DateTime
+from email.message import Message
 from io import BytesIO
 
 from boltons.timeutils import LocalTZ
@@ -43,6 +44,25 @@ def test_can_parse_encoded_header():
     assert msg_info.from_addr == '[email protected]'
     assert msg_info.to_addrs == ('[email protected]',)
 
+def test_can_parse_message_with_utf8_data():
+    msg = Message()
+    msg['Message-ID'] = '<[email protected]>'
+    msg['Mime-Version'] = '1.0'
+    msg['Content-Transfer-Encoding'] = '8bit'
+    msg['Content-Type'] = 'text/plain; charset=UTF-8'
+    msg.set_payload(b'some \xc2\xbbnon-ascii\xc2\xab text')
+
+    queue_fp = build_queued_message(
+        sender='[email protected]',
+        recipient='[email protected]',
+        msg=msg,
+    )
+    msg_info = parse_message_envelope(queue_fp)
+    assert msg_info.from_addr == '[email protected]'
+    assert msg_info.to_addrs == ('[email protected]',)
+    assert msg_info.msg_id == '[email protected]'
+    assert msg_info.msg_fp.read() == msg.as_bytes()
+
 def test_can_parse_queue_metadata():
     queue_date = DateTime(2020, 10, 1, hour=15, minute=42, second=21, tzinfo=LocalTZ)
     last_attempt = DateTime(2020, 10, 1, hour=16, minute=0, tzinfo=LocalTZ)