Broken test now check if headers are correctly encoded.

* 'broken' test, in addition to testing that the mail was
  correctly parsed by email python module, now also test
  that every header can be decoded correctly (that mean
  that for example, a subject header with 8 bits contents
  will be marked as broken.)
git-archimport-id: frederic@jolliton.com--2004-private/mail-filter--main--0.1.2--patch-4

diff --git a/mailfilter.py b/mailfilter.py
index b46ca71..4d916f2 100644 (file)
--- a/mailfilter.py
+++ b/mailfilter.py
@@ -601,6 +601,22 @@ def decodeHeader( s ) :
                logMessage( 'Error decoding %r' % s )
                return s
 
+def isMailBroken() :
+
+       if g_mail == None :
+               return True
+       brokenHeaders = []
+       for headerName , headerContents in g_mail.items() :
+               try :
+                       for text , charset in email.Header.decode_header( headerContents ) :
+                               text.decode( charset or 'ascii' )
+               except UnicodeDecodeError , e :
+                       brokenHeaders.append( headerName )
+       if brokenHeaders :
+               logMessage( 'The following headers are broken: %r' % brokenHeaders )
+               return True
+       return False
+
 #
 # Test rule 'rule' against the mail.
 #
@@ -676,7 +692,7 @@ def testRule( rule ) :
        # Broken mail
        #
        if cmd == 'broken' :
-               return g_mail == None
+               return g_mail == None or isMailBroken()
 
        #
        # Infected mail