1#!/bin/bash
2#
3# Copyright (C) 2002 Laird Breyer
4#
5# This program is free software; you can redistribute it and/or modify
6# it under the terms of the GNU General Public License as published by
7# the Free Software Foundation; either version 3 of the License, or
8# (at your option) any later version.
9#
10# This program is distributed in the hope that it will be useful,
11# but WITHOUT ANY WARRANTY; without even the implied warranty of
12# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
13# GNU General Public License for more details.
14#
15# You should have received a copy of the GNU General Public License
16# along with this program; if not, write to the Free Software
17# Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
18#
19# Author:   Laird Breyer <laird@lbreyer.com>
20#
21# IMPLEMENTATION NOTES
22#
23# This script follows the mailcross testsuite interface
24# requirements. Type man mailcross for details.
25#
26# The script accepts one of more commands on the command line,
27# and may read STDIN and write STDOUT as follows:
28#
29# If $1 == "filter":
30# In this case, a single email is expected on STDIN,
31# and a list of category filenames is expected in $2, $3, etc.
32# The script writes the category name corresponding to the
33# input email on STDOUT.
34#
35# If $1 == "learn":
36# In this case, a standard mbox stream is expected on STDIN,
37# while a suitable category file name is expected in $2. No output
38# is written to STDOUT.
39#
40# If $1 == "clean":
41# In this case, a directory is expected in $2, which is examined
42# for old database information. If any old databases are found, they
43# are purged or reset. No output is written to STDOUT.
44#
45# If $1 == "describe":
46# In this case, STDIN and the command line are ignored. A single
47# line is written on STDOUT, describing the filter functionality.
48#
49# If $1 == "bootstrap":
50# In this case, the current script is copied to the directory $2,
51# provided the classifier we're wrapping exists on the system.
52#
53
54# besides the standard interface, we also define a "mutilate" command,
55# this cannot be a bash function, as it will be called by formail.
56
57CRM="crm"
58MIMEDEC=`which mimencode`
59MUTILATE="$0 mutilate"
60
61case "$1" in
62    filter)
63	shift
64	ALLCATS=`for f in $@; do echo -n "$f.css " ; done`
65	$MUTILATE | $CRM "-{
66isolate (:lcr:)
67alter (:lcr:) /[[:graph:]][-.,:[:alnum:]]*[[:graph:]]?/
68isolate (:stats:)
69{
70        match (:data:) /.*/
71        classify ($ALLCATS) ( :stats: ) [:data:] /:*:lcr:/
72	match <nomultiline> (:x: :y: :z:) [:stats:] /\\\((.*\/)*(.*)\\.css\\\)/
73	output /:*:z:/
74}
75}"
76	;;
77    learn)
78	shift
79	CATEGORY=`basename $1`
80	DBPATH=${1/$CATEGORY/}
81	formail -s $MUTILATE | $CRM "-{
82isolate (:lcr:)
83alter (:lcr:) /[[:graph:]][-.,:[:alnum:]]*[[:graph:]]?/
84{
85        match (:data:) /.*/
86        learn <microgroom> (${DBPATH}${CATEGORY}.css) [:data:] /:*:lcr:/
87        exit /0/
88}
89}"
90	;;
91    clean)
92	shift
93	find "$1" -name "*.css" -exec rm {} \;
94	find "$1" -name "*.toe" -exec rm -f {} \;
95	find "$1" -name "*.foot" -exec rm -f {} \;
96	find "$1" -name "*.tmp" -exec rm -f {} \;
97	;;
98    describe)
99	VER="(unavailable?)"
100	if [ -n "`which $CRM`" ] ; then
101	    VER=`$CRM -v 2>&1 | head -1 | sed 's/^.*version //'`
102	fi
103	echo "crm114 $VER with default settings"
104	;;
105    bootstrap)
106	if [ -d "$2" ] ; then
107            if [ -n "`which $CRM`" -a -n "$MIMEDEC" ] ; then
108		echo "selecting $0"
109		cp "$0" "$2"
110	    else
111		echo "crm114 appears to be missing"
112            fi
113	else
114	    echo "bad target directory $2"
115	fi
116	;;
117
118    toe)
119	ME="$0"
120	shift
121	TRUECAT="$1"
122	shift
123	CATS=`for f in "$@"; do echo -n "-c $f " ; done`
124	cat > "$TEMPDIR/mailtoe.tmp"
125	VERDICT=`cat $TEMPDIR/mailtoe.tmp | $ME filter "$@"`
126	if [ "x$VERDICT" != "x`basename $TRUECAT`" ] ; then
127	    cat "$TEMPDIR/mailtoe.tmp" >> $TRUECAT.toe
128	    cat $TRUECAT.toe | $ME learn $TRUECAT
129	fi
130	echo -ne "$VERDICT"
131	;;
132
133    foot)
134	ME="$0"
135	shift
136	TRUECAT="$1"
137	shift
138	CATS=`for f in "$@"; do echo -n "-c $f " ; done`
139	cat > "$TEMPDIR/mailfoot.tmp"
140	VERDICT=`cat "$TEMPDIR/mailfoot.tmp" | $ME filter "$@"`
141	cat "$TEMPDIR/mailfoot.tmp" >> "$TRUECAT.foot"
142	cat "$TRUECAT.foot" | $ME learn $TRUECAT
143	echo -ne "$VERDICT"
144	;;
145
146    mutilate)
147	# This function cleans up the email attachments
148	# The code was taken verbatim from mailfilter.crm, version crm114-20031111-RC8
149	if [ -z "$MIMEDEC" ] ; then
150	    cat
151	else
152	    MIMEDEC="mimencode -u"
153	    $CRM "-{
154isolate (:do_base64:) /yes/
155isolate (:undo_interruptus:) /yes/
156isolate (:mime_decoder:) /$MIMEDEC/
157isolate (:m_text:) //
158isolate (:b_text:) /:*:_dw:/
159isolate (:i_text:) /:*:_dw:/
160#
161#       do we do any expansions?
162{
163        #   expansion 1: - do we perform base64 expansions?
164        {
165                {
166                        match [:do_base64:] /yes/
167                        {
168                            #  yes, expand base64's if there are any
169                            #
170                            #    Note: some spams don't even bother to use
171                            #    a 'Content-Transfer-Encoding' marker,
172                            #    and even fewer use Content-Type: text/whatever
173                            #    so we have to sort of wing it, when to expand
174                            #    what _might_ be base64 and when to ignore it.
175                            #    For now, if it says it's a base64, it gets
176                            #    expanded, no matter what the type.  Maybe
177                            #    someday someone will put in a lockout for
178                            #    things like .jpg files, .doc files, etc.
179                            #
180                            isolate (:exp_text:)
181                            match [:b_text:] (:a: :h: :b:) /(Content-Transfer-Encoding): base64(.*)/
182                            match (:c:) [:b:] /([a-zA-Z0-9+=!\/]+:*:_nl:)\{2,200\}/
183                            #
184                            syscall (:*:c:) (:exp_text:) /:*:mime_decoder: /
185                            #   and stuff the result back into b_text for
186                            #   classification right in context.
187                            alter (:c:) /:*:exp_text:/
188                            #   and mark this piece of mime as \"prior\".
189                            alter (:h:) /Content-Transfer-Prior-Encoding/
190                            #   repeat till no more Mime base64 encodings
191                            liaf
192                        }
193                }
194                alius
195                {
196                        #   if no base64 expansions enabled, empty out :b_text:
197                        #
198                        alter (:b_text:) //
199                }
200        }
201        #
202        #   If we had expansions, bust the html contents out of them, otherwise
203        #   ignore b_text as it's redundant
204        {
205                {
206                        match [:b_text:] /Content-Transfer-Prior-Encoding/
207                        alter (:i_text:) /:*:b_text:/
208                }
209                alius
210                {
211                        #   if :b_text: _didn't_ have a base64, it's useless
212                        alter (:b_text:) //
213                }
214        }
215        #   expansion 2 :  do we bust HTML comments ( a.k.a.
216        #    hypertextus interruptus) out?
217        {
218                match [:undo_interruptus:] /yes/
219                isolate (:commentbin:) //
220                {
221                    match [:i_text:] (:comment:) /<!--([^-]|-[^-]|--[^>])*-->/
222                    alter (:commentbin:) /:*:commentbin: :*:comment:/
223                    alter (:comment:) //
224                    liaf
225                }
226                #     if we had at least 80 characters worth of comments, then
227                #     it's worth using the decommented text, else not.
228                #     (this my personal judgement call)
229                {
230                   {
231                       match [:commentbin:] /(.)\{80,\}/
232                   }
233                   alius
234                   {
235                       alter (:i_text:) //
236                   }
237                }
238        }
239}
240#    and reassemble the mucked-over text into the :m_text: var, always
241#    with the base64's expanded, then a second decommented copy
242#
243{
244       alter (:m_text:) /:*:_dw: :*:_nl: :*:b_text: :*:_nl: :*:i_text: :*:_nl:/
245}
246
247    output /:*:m_text:/
248}"
249	fi
250	;;
251esac
252