utils: fix decoding of invalid headers

For some mails, the following exception was thrown:

Traceback (most recent call last):
  File "/usr/local/bin/imapami", line 9, in <module>
    load_entry_point('imapami==0.1.0', 'console_scripts', 'imapami')()
  File "build/bdist.linux-x86_64/egg/imapami/__init__.py", line 321, in main
  File "build/bdist.linux-x86_64/egg/imapami/__init__.py", line 221, in process_rules
  File "build/bdist.linux-x86_64/egg/imapami/rules.py", line 251, in process
  File "build/bdist.linux-x86_64/egg/imapami/actions.py", line 149, in process
  File "build/bdist.linux-x86_64/egg/imapami/actions.py", line 317, in process
  File "build/bdist.linux-x86_64/egg/imapami/actions.py", line 98, in evaluate
  File "build/bdist.linux-x86_64/egg/imapami/utils.py", line 65, in headers_to_unicode
  File "/usr/lib/python2.7/email/header.py", line 108, in decode_header
    raise HeaderParseError
email.errors.HeaderParseError

It can be reproduced manually with:

>>> import email.header
>>> email.header.decode_header('=?UTF-8?B?UG91ciB2b3MgYWNoYXRzIGRlIE5vw6tsLCBwcm9maXRleiBkJ3VuIGRv?==?UTF-8?B?dWJsZSBhdmFudGFnZSAh?=')

According to:
http://stackoverflow.com/questions/7384006/email-header-decode-headers-throws-an-headerparseerror
A workaround is to fix the header manually.

Signed-off-by: Olivier Matz <zer0@droids-corp.org>

diff --git a/imapami/utils.py b/imapami/utils.py
index abf79b0a79a2028c0092949b65203d69bad776e9..291b71fdba881bceecb23af780f0c76835d99516 100644 (file)
--- a/imapami/utils.py
+++ b/imapami/utils.py
@@ -28,6 +28,7 @@
 #
 
 import email.header
+import re
 
 # pylint: disable=deprecated-module
 # see https://www.logilab.org/ticket/2481
@@ -62,7 +63,16 @@ def headers_to_unicode(headers):
     """
     unicode_headers = {}
     for key, hdr in headers.items():
-        value, encoding = email.header.decode_header(hdr)[0]
+        try:
+            value, encoding = email.header.decode_header(hdr)[0]
+        except email.header.HeaderParseError:
+            try:
+                # try to workaround badly formatted RFC2047 tokens
+                hdr = re.sub(r"(==)(?!$)", u"= =", hdr)
+                value, encoding = email.header.decode_header(hdr)[0]
+            except email.header.HeaderParseError:
+                # fallback to wrong decoding
+                value, encoding = hdr, 'utf-8'
         if encoding is None:
             value = unicode(value)
         else: