headers are getting corrupted, remove surrogates again

distribution/gentoo/mail-filter/pyquarantine-milter/pyquarantine-milter-9999.ebuild

@@ -37,7 +37,7 @@ RDEPEND="
 	lxml? ( dev-python/lxml[${PYTHON_USEDEP}] )
 	dev-python/netaddr[${PYTHON_USEDEP}]
 	dev-python/peewee[${PYTHON_USEDEP}]
-	dev-python/pymilter[${PYTHON_USEDEP}]"
+	>=dev-python/pymilter-1.5[${PYTHON_USEDEP}]"
 
 python_install_all() {
 	distutils-r1_python_install_all

pyquarantine/__init__.py

@@ -27,7 +27,7 @@ __all__ = [
     "whitelist",
     "QuarantineMilter"]
 
-__version__ = "2.0.4"
+__version__ = "2.0.6"
 
 from pyquarantine import _runtime_patches
 
@@ -100,7 +100,7 @@ class QuarantineMilter(Milter.Base):
             self.logger.warning(f"unable to serialize message as bytes: {e}")
             try:
                 self.logger.warning("try to serialize as str and encode")
-                data = self.msg.as_string().encode("ascii", errors="replace")
+                data = self.msg.as_string().encode(errors="replace")
             except Exception as e:
                 self.logger.error(
                     f"unable to serialize message, giving up: {e}")
@@ -208,6 +208,7 @@ class QuarantineMilter(Milter.Base):
 
         return Milter.CONTINUE
 
+    @Milter.decode("replace")
     def envfrom(self, mailfrom, *str):
         try:
             self.mailfrom = "@".join(parse_addr(mailfrom)).lower()
@@ -219,6 +220,7 @@ class QuarantineMilter(Milter.Base):
 
         return Milter.CONTINUE
 
+    @Milter.decode("replace")
     def envrcpt(self, to, *str):
         try:
             self.rcpts.add("@".join(parse_addr(to)).lower())
@@ -243,6 +245,7 @@ class QuarantineMilter(Milter.Base):
 
         return Milter.CONTINUE
 
+    @Milter.decode("replace")
     def header(self, field, value):
         try:
             # remove CR and LF from address fields, otherwise pythons
@@ -262,7 +265,7 @@ class QuarantineMilter(Milter.Base):
             field = field.encode("ascii", errors="replace")
             value = value.encode("ascii", errors="replace")
 
-            self.fp.write(field + b": " + value + b"\r\n")
+            self.fp.write(field.encode() + b": " + value.encode() + b"\r\n")
         except Exception as e:
             self.logger.exception(
                 f"an exception occured in header method: {e}")

pyquarantine/_runtime_patches.py

@@ -12,6 +12,7 @@
 # along with pyquarantine.  If not, see <http://www.gnu.org/licenses/>.
 #
 
+from sys import version_info
 import encodings
 
 
@@ -150,27 +151,45 @@ def get_obs_local_part(value):
 setattr(email._header_value_parser, "get_obs_local_part", get_obs_local_part)
 
 
-# https://bugs.python.org/issue30681
+#######################################
+#  add charset alias for windows-874  #
+#######################################
 #
-# fix: https://github.com/python/cpython/pull/22090
+# https://bugs.python.org/issue17254
+#
+# fix: https://github.com/python/cpython/pull/10237
 
-import email.errors
-from email.errors import HeaderDefect
+aliases = encodings.aliases.aliases
+
+for alias in ["windows-874", "windows_874"]:
+    if alias not in aliases:
+        aliases[alias] = "cp874"
+
+setattr(encodings.aliases, "aliases", aliases)
 
 
-class InvalidDateDefect(HeaderDefect):
+if version_info.major == 3 and version_info.minor < 10:
+    # https://bugs.python.org/issue30681
+    #
+    # fix: https://github.com/python/cpython/pull/22090
+
+    import email.errors
+    from email.errors import HeaderDefect
+
+
+    class InvalidDateDefect(HeaderDefect):
         """Header has unparseable or invalid date"""
 
 
-setattr(email.errors, "InvalidDateDefect", InvalidDateDefect)
+    setattr(email.errors, "InvalidDateDefect", InvalidDateDefect)
 
 
-import email.utils
-from email.utils import _parsedate_tz
-import datetime
+    import email.utils
+    from email.utils import _parsedate_tz
+    import datetime
 
 
-def parsedate_to_datetime(data):
+    def parsedate_to_datetime(data):
         parsed_date_tz = _parsedate_tz(data)
         if parsed_date_tz is None:
             raise ValueError('Invalid date value or format "%s"' % str(data))
@@ -181,14 +200,14 @@ def parsedate_to_datetime(data):
                 tzinfo=datetime.timezone(datetime.timedelta(seconds=tz)))
 
 
-setattr(email.utils, "parsedate_to_datetime", parsedate_to_datetime)
+    setattr(email.utils, "parsedate_to_datetime", parsedate_to_datetime)
 
 
-import email.headerregistry
-from email import utils, _header_value_parser as parser
+    import email.headerregistry
+    from email import utils, _header_value_parser as parser
 
-@classmethod
-def parse(cls, value, kwds):
+    @classmethod
+    def parse(cls, value, kwds):
         if not value:
             kwds['defects'].append(errors.HeaderMissingRequiredValue())
             kwds['datetime'] = None
@@ -209,21 +228,4 @@ def parse(cls, value, kwds):
         kwds['parse_tree'] = cls.value_parser(kwds['decoded'])
 
 
-setattr(email.headerregistry.DateHeader, "parse", parse)
-
-
-#######################################
-#  add charset alias for windows-874  #
-#######################################
-#
-# https://bugs.python.org/issue17254
-#
-# fix: https://github.com/python/cpython/pull/10237
-
-aliases = encodings.aliases.aliases
-
-for alias in ["windows-874", "windows_874"]:
-    if alias not in aliases:
-        aliases[alias] = "cp874"
-
-setattr(encodings.aliases, "aliases", aliases)
+    setattr(email.headerregistry.DateHeader, "parse", parse)

pyquarantine/run.py

@@ -135,7 +135,7 @@ def main():
     sysloghandler = logging.handlers.SysLogHandler(
         address="/dev/log", facility=logging.handlers.SysLogHandler.LOG_MAIL)
     sysloghandler.setFormatter(
-        logging.Formatter("pyquarantine: %(message)s"))
+        logging.Formatter(f"{name}[%(process)d]: %(message)s"))
     root_logger.addHandler(sysloghandler)
 
     logger.info("milter starting")

setup.py

@@ -48,6 +48,6 @@ setup(name = "pyquarantine",
             ]
         )
     ],
-    install_requires = ["pymilter", "jsonschema", "netaddr", "beautifulsoup4[lxml]", "peewee"],
+    install_requires = ["pymilter >= 1.5", "jsonschema", "netaddr", "beautifulsoup4[lxml]", "peewee"],
     python_requires = ">=3.9"
 )