--- CookHeaders.py.orig	Tue Apr  1 06:49:42 2003
+++ CookHeaders.py	Wed Apr 30 13:33:02 2003
@@ -15,6 +15,7 @@
 # Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
 
 """Cook a message's Subject header.
+   (sequence version)
 """
 
 from __future__ import nested_scopes
@@ -29,6 +30,7 @@
 from Mailman import Utils
 from Mailman.i18n import _
 from Mailman.Logging.Syslog import syslog
+from Mailman.Handlers.ToDigest import oneline
 
 CONTINUATION = ',\n\t'
 COMMASPACE = ', '
@@ -72,7 +74,12 @@
     # VirginRunner sets _fasttrack for internally crafted messages.
     fasttrack = msgdata.get('_fasttrack')
     if not msgdata.get('isdigest') and not fasttrack:
-        prefix_subject(mlist, msg, msgdata)
+        try:
+            prefix_subject(mlist, msg, msgdata)
+        except (UnicodeError, ValueError):
+            # TK: Sometimes subject header is not MIME encoded for 8bit
+            # simply abort prefixing.
+            pass
     # Mark message so we know we've been here, but leave any existing
     # X-BeenThere's intact.
     msg['X-BeenThere'] = mlist.GetListEmail()
@@ -218,7 +225,9 @@
     # Add the subject prefix unless the message is a digest or is being fast
     # tracked (e.g. internally crafted, delivered to a single user such as the
     # list admin).
-    prefix = mlist.subject_prefix
+    prefix = mlist.subject_prefix.strip()
+    if not prefix: 
+        return
     subject = msg.get('subject', '')
     # Try to figure out what the continuation_ws is for the header
     if isinstance(subject, Header):
@@ -229,35 +238,68 @@
     if len(lines) > 1 and lines[1] and lines[1][0] in ' \t':
         ws = lines[1][0]
     msgdata['origsubj'] = subject
-    if not subject:
-        subject = _('(no subject)')
     # The header may be multilingual; decode it from base64/quopri and search
     # each chunk for the prefix.  BAW: Note that if the prefix contains spaces
     # and each word of the prefix is encoded in a different chunk in the
     # header, we won't find it.  I think in practice that's unlikely though.
-    headerbits = decode_header(subject)
-    if prefix and subject:
-        pattern = re.escape(prefix.strip())
-        for decodedsubj, charset in headerbits:
-            if re.search(pattern, decodedsubj, re.IGNORECASE):
-                # The subject's already got the prefix, so don't change it
-                return
-    del msg['subject']
+    # TK: We assume the subject is mono-lingual in sequence version
+    #     and search first non-None charset
+    cmatch = re.search(r'=\?([^\?]+)\?[bq]\?[^\?]+\?=', subject, re.I)
+    if cmatch:
+        cset = cmatch.group(1).lower()
+    else:
+        cset = 'us-ascii'
+    subject = oneline(subject, cset)
+    # Note: searching prefix in subject is REMOVED. (seq version)
+    # If the subject_prefix contains '%d', it is replaced with the
+    # mailing list sequential number. Also, if the prefix is closed with
+    # [],(), or {}, the prefix in the responding post subject will be cared.
+    # sequential number format allows '%05d' like pattern.
+    p = re.compile('%\d*d')
+    if p.search(prefix,1):
+        # prefix have number, so we should search prefix w/number
+        # in subject.
+        prefix_pattern = p.sub(r'\s*\d+\s*', prefix)
+    else:
+        prefix_pattern = prefix
+    prefix_pattern = re.sub('([\[\(\{])', '\\\\\g<1>', prefix_pattern)
+    subject = re.sub(prefix_pattern, '', subject)
+    subject = re.compile('(RE:\s*)+', re.I).sub('Re: ', subject, 1)
+    # At this point, subject may become null if someone post mail with
+    # subject: [subject prefix]
+    if subject.strip() == '':
+        subject = _('(no subject)')
+        cset = Utils.GetCharSet(mlist.preferred_language)
+    # and substitute %d in prefix with post_id
+    try:
+        prefix = prefix % mlist.post_id
+    except TypeError:
+        pass
+    # If charset is 'us-ascii', try to concatnate as string because there
+    # is some weirdness in Header module (TK)
+    if cset == 'us-ascii':
+        try:
+            h = prefix + ' ' + subject
+            if type(h) == UnicodeType:
+                h = h.encode('us-ascii')
+            else:
+                h = unicode(h, 'us-ascii').encode('us-ascii')
+            del msg['subject']
+            msg['Subject'] = h
+            return
+        except UnicodeError:
+            pass
     # Get the header as a Header instance, with proper unicode conversion
     h = uheader(mlist, prefix, 'Subject', continuation_ws=ws)
-    for s, c in headerbits:
-        # Once again, convert the string to unicode.
-        if c is None:
-            c = Charset('iso-8859-1')
-        if not isinstance(c, Charset):
-            c = Charset(c)
-        if not _isunicode(s):
-            codec = c.input_codec or 'ascii'
-            try:
-                s = unicode(s, codec, 'replace')
-            except LookupError:
-                # Unknown codec, is this default reasonable?
-                s = unicode(s, Utils.GetCharSet(mlist.preferred_language),
-                            'replace')
-        h.append(s, c)
+    # in seq version, subject header is already concatnated
+    if not _isunicode(subject):
+        try:
+            subject = unicode(subject, cset, 'replace')
+        except (LookupError, TypeError):
+            # unknown codec
+            cset = Utils.GetCharSet(mlist.preferred_language)
+            subject = unicode(subject, cset, 'replace')
+    subject = subject.encode(cset,'replace')
+    h.append(subject, cset)
+    del msg['subject']
     msg['Subject'] = h