Fix for email.generator.Generator with whitespace between encoded words.

email.generator.Generator currently does not handle whitespace between
encoded words correctly when the encoded words span multiple lines.  The
current generator will create an encoded word for each line.  If the end
of the line happens to correspond with the end real word in the
plaintext, the generator will place an unencoded space at the start of
the subsequent lines to represent the whitespace between the plaintext
words.

A compliant decoder will strip all the whitespace from between two
encoded words which leads to missing spaces in the round-tripped
output.

The fix for this is to make sure that whitespace between two encoded
words ends up inside of one or the other of the encoded words.  This
fix places the space inside of the second encoded word.

A second problem happens with continuation lines.  A continuation line that
starts with whitespace and is followed by a non-encoded word is fine because
the newline between such continuation lines is defined as condensing to
a single space character.  When the continuation line starts with whitespace
followed by an encoded word, however, the RFCs specify that the word is run
together with the encoded word on the previous line.  This is because normal
words are filded on syntactic breaks by encoded words are not.

The solution to this is to add the whitespace to the start of the encoded word
on the continuation line.

Test cases are from #92081

Author: abadger

Lib/email/_header_value_parser.py

@@ -2764,10 +2764,14 @@ def _refold_parse_tree(parse_tree, *, policy):
     # max_line_length 0/None means no limit, ie: infinitely long.
     maxlen = policy.max_line_length or sys.maxsize
     encoding = 'utf-8' if policy.utf8 else 'us-ascii'
-    lines = ['']
-    last_ew = None
+    lines = ['']  # Folded lines to be output
+    leading_whitespace = ''  # When we have whitespace between two encoded
+                             # words, we may need to encode the whitespace
+                             # at the beginning of the second word.
+    last_ew = None  # Points to the last encoded character if there's an ew on
+                    # the line
     wrap_as_ew_blocked = 0
-    want_encoding = False
+    want_encoding = False  # This is set to True if we need to encode this part
     end_ew_not_allowed = Terminal('', 'wrap_as_ew_blocked')
     parts = list(parse_tree)
     while parts:
@@ -2791,10 +2795,12 @@ def _refold_parse_tree(parse_tree, *, policy):
                 # 'charset' property on the policy.
                 charset = 'utf-8'
             want_encoding = True
+
         if part.token_type == 'mime-parameters':
             # Mime parameter folding (using RFC2231) is extra special.
             _fold_mime_parameters(part, lines, maxlen, encoding)
             continue
+
         if want_encoding and not wrap_as_ew_blocked:
             if not part.as_ew_allowed:
                 want_encoding = False
@@ -2821,20 +2827,37 @@ def _refold_parse_tree(parse_tree, *, policy):
                 # It's a terminal, wrap it as an encoded word, possibly
                 # combining it with previously encoded words if allowed.
                 last_ew = _fold_as_ew(tstr, lines, maxlen, last_ew,
-                                      part.ew_combine_allowed, charset)
+                                      part.ew_combine_allowed, charset, leading_whitespace)
+                # This whitespace has been added to the lines in _fold_as_ew()
+                # so clear it now.
+                leading_whitespace = ''
             want_encoding = False
             continue
+
         if len(tstr) <= maxlen - len(lines[-1]):
             lines[-1] += tstr
             continue
+
         # This part is too long to fit.  The RFC wants us to break at
         # "major syntactic breaks", so unless we don't consider this
         # to be one, check if it will fit on the next line by itself.
+        leading_whitespace = ''
         if (part.syntactic_break and
                 len(tstr) + 1 <= maxlen):
             newline = _steal_trailing_WSP_if_exists(lines)
             if newline or part.startswith_fws():
+                # We're going to fold the data onto a new line here.  Due to
+                # the way encoded strings handle continuation lines, we need to
+                # be prepared to encode any whitespace if the next line turns
+                # out to start with an encoded word.
                 lines.append(newline + tstr)
+
+                leading_whitespace = []
+                for char in lines[-1]:
+                    if char not in WSP:
+                        break
+                    leading_whitespace.append(char)
+                leading_whitespace = ''.join(leading_whitespace)
                 last_ew = None
                 continue
         if not hasattr(part, 'encode'):
@@ -2858,9 +2881,10 @@ def _refold_parse_tree(parse_tree, *, policy):
         else:
             # We can't fold it onto the next line either...
             lines[-1] += tstr
+
     return policy.linesep.join(lines) + policy.linesep
 
-def _fold_as_ew(to_encode, lines, maxlen, last_ew, ew_combine_allowed, charset):
+def _fold_as_ew(to_encode, lines, maxlen, last_ew, ew_combine_allowed, charset, leading_whitespace):
     """Fold string to_encode into lines as encoded word, combining if allowed.
     Return the new value for last_ew, or None if ew_combine_allowed is False.
 
@@ -2875,14 +2899,15 @@ def _fold_as_ew(to_encode, lines, maxlen, last_ew, ew_combine_allowed, charset):
         to_encode = str(
             get_unstructured(lines[-1][last_ew:] + to_encode))
         lines[-1] = lines[-1][:last_ew]
-    if to_encode[0] in WSP:
+    elif to_encode[0] in WSP:
         # We're joining this to non-encoded text, so don't encode
         # the leading blank.
         leading_wsp = to_encode[0]
         to_encode = to_encode[1:]
         if (len(lines[-1]) == maxlen):
             lines.append(_steal_trailing_WSP_if_exists(lines))
         lines[-1] += leading_wsp
+
     trailing_wsp = ''
     if to_encode[-1] in WSP:
         # Likewise for the trailing space.
@@ -2902,11 +2927,20 @@ def _fold_as_ew(to_encode, lines, maxlen, last_ew, ew_combine_allowed, charset):
 
     while to_encode:
         remaining_space = maxlen - len(lines[-1])
-        text_space = remaining_space - chrome_len
+        text_space = remaining_space - chrome_len - len(leading_whitespace)
         if text_space <= 0:
             lines.append(' ')
             continue
 
+        # If we are at the start of a continuation line, prepend whitespace
+        # (we only want to do this when the line starts with an encoded word
+        # but if we're folding in this helper function, then we know that we
+        # are going to be writing out an encoded word.)
+        if len(lines) > 1 and len(lines[-1]) == 1 and leading_whitespace:
+            encoded_word = _ew.encode(leading_whitespace, charset=encode_as)
+            lines[-1] += encoded_word
+            leading_whitespace = ''
+
         to_encode_word = to_encode[:text_space]
         encoded_word = _ew.encode(to_encode_word, charset=encode_as)
         excess = len(encoded_word) - remaining_space

Lib/test/test_email/test_generator.py

@@ -232,6 +232,41 @@ class TestBytesGenerator(TestGeneratorBase, TestEmailBase):
     ioclass = io.BytesIO
     typ = lambda self, x: x.encode('ascii')
 
+    def test_defaults_handle_spaces_between_encoded_words_when_folded(self):
+        source = ("Уведомление о принятии в работу обращения для"
+                  " подключения услуги")
+        expected = ('Subject: =?utf-8?b?0KPQstC10LTQvtC80LvQtdC90LjQtSDQviDQv9GA0LjQvdGP0YLQuNC4?=\n'
+                    ' =?utf-8?b?INCyINGA0LDQsdC+0YLRgyDQvtCx0YDQsNGJ0LXQvdC40Y8g0LTQu9GPINC/0L4=?=\n'
+                    ' =?utf-8?b?0LTQutC70Y7Rh9C10L3QuNGPINGD0YHQu9GD0LPQuA==?=\n\n').encode('ascii')
+        msg = EmailMessage()
+        msg['Subject'] = source
+        s = io.BytesIO()
+        g = BytesGenerator(s)
+        g.flatten(msg)
+        self.assertEqual(s.getvalue(), expected)
+
+    def test_defaults_handle_spaces_at_start_of_subject(self):
+        source = " Уведомление"
+        expected = b"Subject:  =?utf-8?b?0KPQstC10LTQvtC80LvQtdC90LjQtQ==?=\n\n"
+        msg = EmailMessage()
+        msg['Subject'] = source
+        s = io.BytesIO()
+        g = BytesGenerator(s)
+        g.flatten(msg)
+        self.assertEqual(s.getvalue(), expected)
+
+    def test_defaults_handle_spaces_at_start_of_continuation_line(self):
+        source = " ф ффффффффффффффффффф ф ф"
+        expected = (b"Subject:  "
+                    b"=?utf-8?b?0YQg0YTRhNGE0YTRhNGE0YTRhNGE0YTRhNGE0YTRhNGE0YTRhNGE0YQ=?=\n"
+                    b" =?utf-8?b?INGEINGE?=\n\n")
+        msg = EmailMessage()
+        msg['Subject'] = source
+        s = io.BytesIO()
+        g = BytesGenerator(s)
+        g.flatten(msg)
+        self.assertEqual(s.getvalue(), expected)
+
     def test_cte_type_7bit_handles_unknown_8bit(self):
         source = ("Subject: Maintenant je vous présente mon "
                  "collègue\n\n").encode('utf-8')

Lib/test/test_email/test_headerregistry.py

@@ -7,6 +7,7 @@
 from test.test_email import TestEmailBase, parameterize
 from email import headerregistry
 from email.headerregistry import Address, Group
+from email.header import decode_header
 from test.support import ALWAYS_EQ
 
 
@@ -1628,7 +1629,7 @@ def test_address_display_names(self):
                     'Lôrem ipsum dôlôr sit amet, cônsectetuer adipiscing. '
                     'Suspendisse pôtenti. Aliquam nibh. Suspendisse pôtenti.',
                     '=?utf-8?q?L=C3=B4rem_ipsum_d=C3=B4l=C3=B4r_sit_amet=2C_c'
-                    '=C3=B4nsectetuer?=\n =?utf-8?q?adipiscing=2E_Suspendisse'
+                    '=C3=B4nsectetuer?=\n =?utf-8?q?_adipiscing=2E_Suspendisse'
                     '_p=C3=B4tenti=2E_Aliquam_nibh=2E?=\n Suspendisse =?utf-8'
                     '?q?p=C3=B4tenti=2E?=',
                     ),

Misc/NEWS.d/next/Library/2023-04-26-22-24-17.gh-issue-92081.V8xMot.rst

@@ -0,0 +1 @@
+Fix missing spaces in email headers when the spaces are mixed with encoded 8-bit characters.