DKIM: more care over untrustworthy data during verify
[exim.git] / src / src / exigrep.src
1 #! PERL_COMMAND
2
3 use warnings;
4 use strict;
5
6 # Copyright (c) 2007-2015 University of Cambridge.
7 # See the file NOTICE for conditions of use and distribution.
8
9 # Except when they appear in comments, the following placeholders in this
10 # source are replaced when it is turned into a runnable script:
11 #
12 # PERL_COMMAND
13 # ZCAT_COMMAND
14 # COMPRESS_SUFFIX
15
16 # PROCESSED_FLAG
17
18 # This is a perl script which extracts from an Exim log all entries
19 # for all messages that have an entry that matches a given pattern.
20 # If *any* entry for a particular message matches the pattern, *all*
21 # entries for that message are displayed.
22
23 # We buffer up information on a per-message basis. It is done this way rather
24 # than reading the input twice so that the input can be a pipe.
25
26 # There must be one argument, which is the pattern. Subsequent arguments
27 # are the files to scan; if none, the standard input is read. If any file
28 # appears to be compressed, it is passed through zcat. We can't just do this
29 # for all files, because zcat chokes on non-compressed files.
30
31 # Performance optimized in 02/02/2007 by Jori Hamalainen
32 # Typical run time acceleration: 4 times
33
34
35 use Getopt::Std qw(getopts);
36 use POSIX qw(mktime);
37
38
39 # This subroutine converts a time/date string from an Exim log line into
40 # the number of seconds since the epoch. It handles optional timezone
41 # information.
42
43 sub seconds {
44 my($year,$month,$day,$hour,$min,$sec,$tzs,$tzh,$tzm) =
45 $_[0] =~ /^(\d{4})-(\d\d)-(\d\d)\s(\d\d):(\d\d):(\d\d)(?>\s([+-])(\d\d)(\d\d))?/o;
46
47 my $seconds = mktime $sec, $min, $hour, $day, $month - 1, $year - 1900;
48
49 if (defined $tzs)
50 {
51 $seconds -= $tzh * 3600 + $tzm * 60 if $tzs eq "+";
52 $seconds += $tzh * 3600 + $tzm * 60 if $tzs eq "-";
53 }
54
55 return $seconds;
56 }
57
58
59 # This subroutine processes a single line (in $_) from a log file. Program
60 # defensively against short lines finding their way into the log.
61
62 my (%saved, %id_list, $pattern, $queue_time, $insensitive, $invert);
63
64 # If using "related" option, have to track extra message IDs
65 my $related;
66 my $related_re='';
67 my @Mids = ();
68
69 sub do_line {
70
71 # Convert syslog lines to mainlog format, as in eximstats.
72
73 if (!/^\d{4}-/o) { $_ =~ s/^.*? exim\b.*?: //o; }
74
75 return unless
76 my($date,$id) = /^(\d{4}-\d\d-\d\d \d\d:\d\d:\d\d (?:[+-]\d{4} )?)(?:\[\d+\] )?(\w{6}\-\w{6}\-\w{2})?/o;
77
78 # Handle the case when the log line belongs to a specific message. We save
79 # lines for specific messages until the message is complete. Then either print
80 # or discard.
81
82 if (defined $id)
83 {
84 $saved{$id} = '' unless defined($saved{$id});
85
86 # Save up the data for this message in case it becomes interesting later.
87
88 $saved{$id} .= $_;
89
90 # Are we interested in this id ? Short circuit if we already were interested.
91
92 if ($invert)
93 {
94 $id_list{$id} = 1 if (!defined($id_list{$id}));
95 $id_list{$id} = 0 if (($insensitive && /$pattern/io) || /$pattern/o);
96 }
97 else
98 {
99 if (defined $id_list{$id} ||
100 ($insensitive && /$pattern/io) || /$pattern/o)
101 {
102 $id_list{$id} = 1;
103 get_related_ids($id) if $related;
104 }
105 elsif ($related && $related_re)
106 {
107 grep_for_related($_, $id);
108 }
109 }
110
111 # See if this is a completion for some message. If it is interesting,
112 # print it, but in any event, throw away what was saved.
113
114 if (index($_, 'Completed') != -1 ||
115 index($_, 'SMTP data timeout') != -1 ||
116 (index($_, 'rejected') != -1 &&
117 /^(\d{4}-\d\d-\d\d \d\d:\d\d:\d\d (?:[+-]\d{4} )?)(?:\[\d+\] )?\w{6}\-\w{6}\-\w{2} rejected/o))
118 {
119 if ($queue_time != -1 &&
120 $saved{$id} =~ /^(\d{4}-\d\d-\d\d \d\d:\d\d:\d\d ([+-]\d{4} )?)/o)
121 {
122 my $old_sec = &seconds($1);
123 my $sec = &seconds($date);
124 $id_list{$id} = 0 if $id_list{$id} && $sec - $old_sec <= $queue_time;
125 }
126
127 print "$saved{$id}\n" if ($id_list{$id});
128 delete $id_list{$id};
129 delete $saved{$id};
130 }
131 }
132
133 # Handle the case where the log line does not belong to a specific message.
134 # Print it if it is interesting.
135
136 elsif ( ($invert && (($insensitive && !/$pattern/io) || !/$pattern/o)) ||
137 (!$invert && (($insensitive && /$pattern/io) || /$pattern/o)) )
138 { print "$_\n"; }
139 }
140
141 # Rotated log files are frequently compressed and there are a variety of
142 # formats it could be compressed with. Rather than use just one that is
143 # detected and hardcoded at Exim compile time, detect and use what the
144 # logfile is compressed with on the fly.
145 #
146 # List of known compression extensions and their associated commands:
147 my $compressors = {
148 gz => { cmd => 'zcat', args => '' },
149 bz2 => { cmd => 'bzcat', args => '' },
150 xz => { cmd => 'xzcat', args => '' },
151 lzma => { cmd => 'lzma', args => '-dc' }
152 };
153 my $csearch = 0;
154
155 sub detect_compressor_bin
156 {
157 my $ext = shift();
158 my $c = $compressors->{$ext}->{cmd};
159 $compressors->{$ext}->{bin} = `which $c 2>/dev/null`;
160 chomp($compressors->{$ext}->{bin});
161 }
162
163 sub detect_compressor_capable
164 {
165 my $filename = shift();
166 map { &detect_compressor_bin($_) } keys %$compressors
167 if (!$csearch);
168 $csearch = 1;
169 return undef
170 unless (grep {$filename =~ /\.(?:$_)$/} keys %$compressors);
171 # Loop through them, figure out which one it detected,
172 # and build the commandline.
173 my $cmdline = undef;
174 foreach my $ext (keys %$compressors)
175 {
176 if ($filename =~ /\.(?:$ext)$/)
177 {
178 # Just die if compressor not found; if this occurs in the middle of
179 # two valid files with a lot of matches, error could easily be missed.
180 die("Didn't find $ext decompressor for $filename\n")
181 if ($compressors->{$ext}->{bin} eq '');
182 $cmdline = $compressors->{$ext}->{bin} ." ".
183 $compressors->{$ext}->{args};
184 last;
185 }
186 }
187 return $cmdline;
188 }
189
190 sub grep_for_related {
191 my ($line,$id) = @_;
192 $id_list{$id} = 1 if $line =~ m/$related_re/;
193 }
194
195 sub get_related_ids {
196 my ($id) = @_;
197 push @Mids, $id unless grep /\b$id\b/, @Mids;
198 my $re = join '|', @Mids;
199 $related_re = qr/$re/;
200 }
201
202 # The main program. Extract the pattern and make sure any relevant characters
203 # are quoted if the -l flag is given. The -t flag gives a time-on-queue value
204 # which is an additional condition. The -M flag will also display "related"
205 # loglines (msgid from matched lines is searched in following lines).
206
207 getopts('Ilvt:M',\my %args);
208 $queue_time = $args{'t'}? $args{'t'} : -1;
209 $insensitive = $args{'I'}? 0 : 1;
210 $invert = $args{'v'}? 1 : 0;
211 $related = $args{'M'}? 1 : 0;
212
213 die "usage: exigrep [-I] [-l] [-M] [-t <seconds>] [-v] <pattern> [<log file>]...\n"
214 if ($#ARGV < 0);
215
216 $pattern = shift @ARGV;
217 $pattern = quotemeta $pattern if $args{l};
218
219
220 # If file arguments are given, open each one and process according as it is
221 # is compressed or not.
222
223 if (@ARGV)
224 {
225 foreach (@ARGV)
226 {
227 my $filename = $_;
228 if (-x 'ZCAT_COMMAND' && $filename =~ /\.(?:COMPRESS_SUFFIX)$/o)
229 {
230 open(LOG, "ZCAT_COMMAND $filename |") ||
231 die "Unable to zcat $filename: $!\n";
232 }
233 elsif (my $cmdline = &detect_compressor_capable($filename))
234 {
235 open(LOG, "$cmdline $filename |") ||
236 die "Unable to decompress $filename: $!\n";
237 }
238 else
239 {
240 open(LOG, "<$filename") || die "Unable to open $filename: $!\n";
241 }
242 do_line() while (<LOG>);
243 close(LOG);
244 }
245 }
246
247 # If no files are named, process STDIN only
248
249 else { do_line() while (<STDIN>); }
250
251 # At the end of processing all the input, print any uncompleted messages.
252
253 for (keys %id_list)
254 {
255 print "+++ $_ has not completed +++\n$saved{$_}\n";
256 }
257
258 # End of exigrep