# vim: ft=python
IMAP_SERVER = 'imap://username:******@localhost/'
pop('localhost', None, 'username', '******')
pop('pop3.isp.net', None, 'username', '******')

def deliver(message):

	MAXSPAMCSIZE = 200000

	def spam():
		if size() < MAXSPAMCSIZE:
			cc("|spamc -L spam", 1)
		to('INBOX.Junk')

	def spam2():
		if size() < MAXSPAMCSIZE:
			to("|spamc -L spam", 1)
		else:
			kill()

	def ham():
		if size() < MAXSPAMCSIZE:
			cc("|spamc -L ham", 1)
			pass
	
	def ml(list):
		ham()
		to('INBOX.' + list)

	if (match('^From: news <news@realh.co.uk>') \
			and match('^To: usenet@realh.co.uk')) \
			or match('^Subject: Cron <mail@realh.co.uk>') \
			or match('^From: Anacron <root@realh.co.uk>'):
		to("INBOX.NoisyLogs")

	if size() < MAXSPAMCSIZE:
		xfilter("spamc")

	# Automatically deal with newsstar spam
	if match(r'^Subject: Newsstar-general post from .* requires approval'):
		print "newsstar post"
		if match(r'Reason:.*Message has implicit destination', match = 'body'):
			m = match(r'Subject: confirm ([0-9a-f]+)', match = 'body')
			if m and m.groups()[0]:
				print "Group"
				import email
				import subprocess
				import sys
				reply = "From h@realh.co.uk " + email.Utils.formatdate()+"\n"+\
						"From: h@realh.co.uk\n" + \
						"To: newsstar-general-request@lists.sourceforge.net\n"+\
						"Subject: Re: confirm " + m.groups()[0] + "\n" + \
						"Date: " + email.Utils.formatdate() + "\n\n"
				try:
					sendmail = subprocess.Popen(['sendmail', '-t'],
							stdin = subprocess.PIPE, stdout = subprocess.PIPE,
							stderr = subprocess.PIPE)
					sendmail.stdin.write(reply)
					sendmail.stdin.close()
					o1 = sendmail.stdout.readlines()
					o2 = sendmail.stderr.readlines()
					if o1:
						print "sendmail stdout:"
						print o1
					if o2:
						print >>sys.stderr, "sendmail stderr:"
						print >>sys.stderr, o2
				except:
					print >>sys.stderr, "Failed to send newsstar spam " + \
							"confirmation message"
				else:
					to("INBOX.Junk")
		to("INBOX")

	# Mails with a score of 15 or higher are almost certainly spam (with 0.05%
	# false positives according to rules/STATISTICS.txt). Let's put them in a
	# different mbox. (This one is optional.)
	if match(r'^X-Spam-Level: \*\*\*\*\*\*\*\*\*\*\*\*\*\*\*'):
		kill()

	# Not sent to a real address
	if match(r'^X-Original-To: unknown@realh\.ukfsn\.org'):
		if not match(r'for <?postmaster@realh\.co\.uk'):
			spam2()

	# Asian junk
	if match('^Content-Type:.*charset="?(big5|ks_c_5601|iso-2022-jp|iso-2022-kr|euc-kr|gb2312)"?') \
			or match('^[ \t]*charset="?(big5|ks_c_5601|iso-2022-jp|iso-2022-kr|euc-kr|gb2312)"?') \
			or (match('^Content-Type: multipart') and \
				match('^(Content-Type:.*|[ \t]*)charset="?(big5|ks_c_5601|iso-2022-jp|iso-2022-kr|euc-kr|gb2312)"?', False, 'body')):
		spam2()

	# Not quite as bad
	if (match(r'^(Received: from|From:|To:) .*\.tw\>') and not \
			match(r'^(Received: from|From:|To:) .*[.@]asus\.com\.tw\>'):
		spam()

	# Not exactly spam, but not wanted
	if match('^From: .*sales@heartinternet'):
		kill()

	# Common spamming addresses
	if match(r'^From: .*(kazaa|k-lite|allcoolmusic\.com|koffi.wilson|crazy-offer\.com|lastminoffers\.com|clubpenguin\.com)'):
		spam2()

	if match('^Subject: Re: Hi'):
		# If it matches all of these it's definitely med spam
		if match('^V[a-zA-Z]{1,2}A[a-zA-Z]{1,2}GRA', False, 'body') \
				and match('^AMB[a-zA-Z]{1,2}EN', False, 'body') \
				and match('^VAL[a-zA-Z]{1,2}UM', False, 'body') \
				and match('^C[a-zA-Z]{1,2}A[a-zA-Z]{1,2}LIS', False, 'body'):
			spam2()

	if match('^Subject: Re: ED[0-9][0-9]*$'):
		spam2()

	if match('^Subject: Re: (your )?(PHA[a-z]*RMACY|VIA[a-z]*GRA)$'):
		spam2()

	# This detects stuff like:
	# From: Xxx Yyy <iii@jjj.com>
	# Subject: Xxx wrote
	# or
	# Subject: hi it's Yyy
	m = match('^From: "?([a-z]+) ([a-z]+)')
	if m:
		m1, m2 = m.group(1, 2)
		# Filter through false positives
		if (match('^Subject:.*[ :]' + m1 + r'\b') \
				or match('^Subject:.*[ :]' + m2 + r'\b')) \
				and m1 != "eBay" and m1 != "Friends" and m1 != "Vodafone" \
				and not (m1 == "Richer" and m2 == "Sounds") \
				and not (m1 == "Superior" and m2 == "Interactive") \
				and not (m1 == "Ebuyer") \
				and not match('^List-Id: ') \
				and not match('^X-Trac-Project: ' + m1 + ' ' + m2):
			if match('^From:.* <debora[a-z]*@') \
					or match('^Subject: Greetings ' + m1 + '$'):
				spam2()
			else:
				spam()

	# spamassassin nearly always gets this right
	if match('^X-Spam-Flag: YES'):
		to("INBOX.Junk")

	# Other highly suspect subjects
	if match('^Subject:.* PH[A-Za-z]*R[a-z]*MA') \
			or match('^Subject: .*VI[A-Za-z]*GRA') \
			or match('^Subject: EQSE News') \
			or match('^Subject: RE: ?Report$') \
			or match('^Subject: .*([0O]EM|[Ll]ow [Cc][0o]st|[Nn]eed|[Cc]heap) [$Ss][o0]ft') \
			or match('^Subject: .*[Ss$][o0]ftware [Yy][0o]u [Nn]eed') \
			or match('^Subject: .*[Nn]eed [Mm]ed') \
			or match('^Subject: .*([Nn][0o]rt[o0]n|[Ss]ystemworks|[Ss]ymantec)') \
			or match('^Subject: .*C[o0]rel') \
			or match('^Subject: .*[Bb]est ([Pp]r[o0]duct|[Pp]r[i1]ce|[Mm]eds)') \
			or (match('^Subject: .*([S5s$][o0]ftware|XP|Ad[o0]be|M[1i]cr[o0][Ss5$][o0]ft)') \
				and match(r'^Subject: .*(\$|[Pp]r[i1]c[e3]|[Dd][i1][5s]c[o0]unt|% [Oo0]ff)')) \
			or (match('^Subject: .*([Pp]h[o0]t[o0]sh[o0]p|[o0]ffice)') \
				and match('^Subject: .*(M[1i]cr[o0][Ss5$][o0]ft|Wind[o0]ws)')) \
			or (match('^Subject: .*([Tt][i1]tles|[O0]EM|[Ss][o0]ftware)') \
				and match(r'^Subject: .*[0-9]\.[0-9][0-9]\>')):
		spam()

	# Stuff that's not addressed correctly to me
	#m = match(r'for <?([a-z.]+)@realh\.co\.uk')
	m = match(r'([a-z0-9.]+)@realh\.co\.uk')
	if m:
		unm = m.group(1)
		if unm == "<" or not unm:
			unm = m.group(2)
		if unm != "h" and unm != "me" and unm != "do.not.edit.nospam" \
				and unm != "postmaster" and unm != "root":
			spam()

	if match(r'junk@realh\.ukfsn\.org'):
		spam()

	# Does To: have my address but someone else's name?
	m = match(r'^(To|Cc): .*\<([a-z]+)"? <[a-z.]+@realh\.co\.uk>')
	if m:
		m = m.group(2)
		if m != "Houghton" and m != "Tony" and m != "h" and m != "for" \
				and m != "Anthony":
			spam()
	
	# Mailing lists
	if match('^List-Id: .*linux-dvb'):
		ml("DVB")

	if match('^List-Id: .*mythtv'):
		ml("mythtv")

	if match('^List-Id: .*VDR'):
		ml("VDR")

	if match('^List-Id: .*UKFSN'):
		ml("UKFSN")

	if match('^List-Id: .*livtech'):
		ml("LWT")

	if match('^(To|List-Id|Cc): .*newsstar'):
		ml("newsstar")

	if match('^List-Id: .*pkg-rox-devel'):
		ml("RoxDebPkg")

	if match('^List-Id: .*HantsME'):
		ml("HantsME")

	if match('^List-Id: .*MPlayer'):
		ml("MPlayer")

	if match('^List-Id: .*dbus'):
		ml("DBus")

	if match('^List-Id: .*gtk'):
		ml("Gtk")

	if match('^List-Id: .*xorg'):
		ml("xorg")

	if match(r'^(To|Cc): .*pkg-rox-devel@lists\.alioth\.debian\.org'):
		ml("RoxDebPkg")

	# This has to come after the MPlayer and pkg-rox rules to avoid catching the
	# "rox" in them.
	if match('^List-Id: .*rox'):
		ml("Rox")

	if match(r'^Sender: owner-mechat-l@LISTSERV\.AOL\.COM'):
		ml("MECHAT")

	if size() < MAXSPAMCSIZE:
		ham()