diff options
| -rw-r--r-- | changes/better_charset_handling | 2 | ||||
| -rw-r--r-- | src/leap/mail/imap/server.py | 28 |
2 files changed, 28 insertions, 2 deletions
diff --git a/changes/better_charset_handling b/changes/better_charset_handling new file mode 100644 index 0000000..832f7b2 --- /dev/null +++ b/changes/better_charset_handling @@ -0,0 +1,2 @@ + o Improve charset handling when exposing mails to the mail client. Related to + #3660. diff --git a/src/leap/mail/imap/server.py b/src/leap/mail/imap/server.py index ae76833..10d338a 100644 --- a/src/leap/mail/imap/server.py +++ b/src/leap/mail/imap/server.py @@ -18,7 +18,9 @@ Soledad-backed IMAP Server. """ import copy +import email import logging +import re import StringIO import cStringIO import time @@ -693,6 +695,26 @@ class LeapMessage(WithMsgFields): the more complex MIME-based interface. """ + def _get_charset(self, content): + """ + Mini parser to retrieve the charset of an email + + :param content: mail contents + :type content: unicode + + :returns: the charset as parsed from the contents + :rtype: str + """ + charset = "UTF-8" + try: + em = email.message_from_string(content.encode("utf-8")) + # Miniparser for: Content-Type: <something>; charset=<charset> + charset_re = r'''charset=(?P<charset>[\w|\d|-]*)''' + charset = re.findall(charset_re, em["Content-Type"])[0] + except Exception: + pass + return charset + def open(self): """ Return an file-like object opened for reading. @@ -704,7 +726,8 @@ class LeapMessage(WithMsgFields): :rtype: StringIO """ fd = cStringIO.StringIO() - fd.write(str(self._doc.content.get(self.RAW_KEY, ''))) + charset = self._get_charset(self._doc.content.get(self.RAW_KEY, '')) + fd.write(self._doc.content.get(self.RAW_KEY, '').encode(charset)) fd.seek(0) return fd @@ -723,7 +746,8 @@ class LeapMessage(WithMsgFields): :rtype: StringIO """ fd = StringIO.StringIO() - fd.write(str(self._doc.content.get(self.RAW_KEY, ''))) + charset = self._get_charset(self._doc.content.get(self.RAW_KEY, '')) + fd.write(self._doc.content.get(self.RAW_KEY, '').encode(charset)) # SHOULD use a separate BODY FIELD ... fd.seek(0) return fd |