Commit | Line | Data |
---|---|---|
2813c06e | 1 | #! PERL_COMMAND |
420a0d19 | 2 | |
2813c06e | 3 | use warnings; |
420a0d19 | 4 | use strict; |
2813c06e | 5 | BEGIN { pop @INC if $INC[-1] eq '.' }; |
420a0d19 | 6 | |
2813c06e | 7 | # Copyright (c) 2007-2015 University of Cambridge. |
420a0d19 CE |
8 | # See the file NOTICE for conditions of use and distribution. |
9 | ||
10 | # Except when they appear in comments, the following placeholders in this | |
11 | # source are replaced when it is turned into a runnable script: | |
12 | # | |
13 | # PERL_COMMAND | |
14 | # ZCAT_COMMAND | |
15 | # COMPRESS_SUFFIX | |
16 | ||
17 | # PROCESSED_FLAG | |
18 | ||
19 | # This is a perl script which extracts from an Exim log all entries | |
20 | # for all messages that have an entry that matches a given pattern. | |
21 | # If *any* entry for a particular message matches the pattern, *all* | |
22 | # entries for that message are displayed. | |
23 | ||
24 | # We buffer up information on a per-message basis. It is done this way rather | |
25 | # than reading the input twice so that the input can be a pipe. | |
26 | ||
27 | # There must be one argument, which is the pattern. Subsequent arguments | |
28 | # are the files to scan; if none, the standard input is read. If any file | |
29 | # appears to be compressed, it is passed through zcat. We can't just do this | |
30 | # for all files, because zcat chokes on non-compressed files. | |
31 | ||
32 | # Performance optimized in 02/02/2007 by Jori Hamalainen | |
33 | # Typical run time acceleration: 4 times | |
34 | ||
35 | ||
36 | use Getopt::Std qw(getopts); | |
37 | use POSIX qw(mktime); | |
38 | ||
39 | ||
40 | # This subroutine converts a time/date string from an Exim log line into | |
41 | # the number of seconds since the epoch. It handles optional timezone | |
42 | # information. | |
43 | ||
44 | sub seconds { | |
45 | my($year,$month,$day,$hour,$min,$sec,$tzs,$tzh,$tzm) = | |
46 | $_[0] =~ /^(\d{4})-(\d\d)-(\d\d)\s(\d\d):(\d\d):(\d\d)(?>\s([+-])(\d\d)(\d\d))?/o; | |
47 | ||
48 | my $seconds = mktime $sec, $min, $hour, $day, $month - 1, $year - 1900; | |
49 | ||
50 | if (defined $tzs) | |
51 | { | |
52 | $seconds -= $tzh * 3600 + $tzm * 60 if $tzs eq "+"; | |
53 | $seconds += $tzh * 3600 + $tzm * 60 if $tzs eq "-"; | |
54 | } | |
55 | ||
56 | return $seconds; | |
57 | } | |
58 | ||
59 | ||
60 | # This subroutine processes a single line (in $_) from a log file. Program | |
61 | # defensively against short lines finding their way into the log. | |
62 | ||
63 | my (%saved, %id_list, $pattern, $queue_time, $insensitive, $invert); | |
64 | ||
2813c06e CE |
65 | # If using "related" option, have to track extra message IDs |
66 | my $related; | |
67 | my $related_re=''; | |
68 | my @Mids = (); | |
69 | ||
420a0d19 CE |
70 | sub do_line { |
71 | ||
72 | # Convert syslog lines to mainlog format, as in eximstats. | |
73 | ||
74 | if (!/^\d{4}-/o) { $_ =~ s/^.*? exim\b.*?: //o; } | |
75 | ||
76 | return unless | |
77 | my($date,$id) = /^(\d{4}-\d\d-\d\d \d\d:\d\d:\d\d (?:[+-]\d{4} )?)(?:\[\d+\] )?(\w{6}\-\w{6}\-\w{2})?/o; | |
78 | ||
79 | # Handle the case when the log line belongs to a specific message. We save | |
80 | # lines for specific messages until the message is complete. Then either print | |
81 | # or discard. | |
82 | ||
83 | if (defined $id) | |
84 | { | |
85 | $saved{$id} = '' unless defined($saved{$id}); | |
86 | ||
87 | # Save up the data for this message in case it becomes interesting later. | |
88 | ||
89 | $saved{$id} .= $_; | |
90 | ||
91 | # Are we interested in this id ? Short circuit if we already were interested. | |
92 | ||
93 | if ($invert) | |
94 | { | |
95 | $id_list{$id} = 1 if (!defined($id_list{$id})); | |
96 | $id_list{$id} = 0 if (($insensitive && /$pattern/io) || /$pattern/o); | |
97 | } | |
98 | else | |
99 | { | |
2813c06e CE |
100 | if (defined $id_list{$id} || |
101 | ($insensitive && /$pattern/io) || /$pattern/o) | |
102 | { | |
103 | $id_list{$id} = 1; | |
104 | get_related_ids($id) if $related; | |
105 | } | |
106 | elsif ($related && $related_re) | |
107 | { | |
108 | grep_for_related($_, $id); | |
109 | } | |
420a0d19 CE |
110 | } |
111 | ||
112 | # See if this is a completion for some message. If it is interesting, | |
113 | # print it, but in any event, throw away what was saved. | |
114 | ||
115 | if (index($_, 'Completed') != -1 || | |
116 | index($_, 'SMTP data timeout') != -1 || | |
117 | (index($_, 'rejected') != -1 && | |
118 | /^(\d{4}-\d\d-\d\d \d\d:\d\d:\d\d (?:[+-]\d{4} )?)(?:\[\d+\] )?\w{6}\-\w{6}\-\w{2} rejected/o)) | |
119 | { | |
120 | if ($queue_time != -1 && | |
121 | $saved{$id} =~ /^(\d{4}-\d\d-\d\d \d\d:\d\d:\d\d ([+-]\d{4} )?)/o) | |
122 | { | |
123 | my $old_sec = &seconds($1); | |
124 | my $sec = &seconds($date); | |
125 | $id_list{$id} = 0 if $id_list{$id} && $sec - $old_sec <= $queue_time; | |
126 | } | |
127 | ||
128 | print "$saved{$id}\n" if ($id_list{$id}); | |
129 | delete $id_list{$id}; | |
130 | delete $saved{$id}; | |
131 | } | |
132 | } | |
133 | ||
134 | # Handle the case where the log line does not belong to a specific message. | |
135 | # Print it if it is interesting. | |
136 | ||
137 | elsif ( ($invert && (($insensitive && !/$pattern/io) || !/$pattern/o)) || | |
138 | (!$invert && (($insensitive && /$pattern/io) || /$pattern/o)) ) | |
139 | { print "$_\n"; } | |
140 | } | |
141 | ||
142 | # Rotated log files are frequently compressed and there are a variety of | |
143 | # formats it could be compressed with. Rather than use just one that is | |
144 | # detected and hardcoded at Exim compile time, detect and use what the | |
145 | # logfile is compressed with on the fly. | |
146 | # | |
147 | # List of known compression extensions and their associated commands: | |
148 | my $compressors = { | |
149 | gz => { cmd => 'zcat', args => '' }, | |
150 | bz2 => { cmd => 'bzcat', args => '' }, | |
151 | xz => { cmd => 'xzcat', args => '' }, | |
152 | lzma => { cmd => 'lzma', args => '-dc' } | |
153 | }; | |
154 | my $csearch = 0; | |
155 | ||
156 | sub detect_compressor_bin | |
157 | { | |
158 | my $ext = shift(); | |
159 | my $c = $compressors->{$ext}->{cmd}; | |
160 | $compressors->{$ext}->{bin} = `which $c 2>/dev/null`; | |
161 | chomp($compressors->{$ext}->{bin}); | |
162 | } | |
163 | ||
164 | sub detect_compressor_capable | |
165 | { | |
166 | my $filename = shift(); | |
167 | map { &detect_compressor_bin($_) } keys %$compressors | |
168 | if (!$csearch); | |
169 | $csearch = 1; | |
170 | return undef | |
171 | unless (grep {$filename =~ /\.(?:$_)$/} keys %$compressors); | |
172 | # Loop through them, figure out which one it detected, | |
173 | # and build the commandline. | |
174 | my $cmdline = undef; | |
175 | foreach my $ext (keys %$compressors) | |
176 | { | |
177 | if ($filename =~ /\.(?:$ext)$/) | |
178 | { | |
2813c06e | 179 | # Just die if compressor not found; if this occurs in the middle of |
420a0d19 CE |
180 | # two valid files with a lot of matches, error could easily be missed. |
181 | die("Didn't find $ext decompressor for $filename\n") | |
182 | if ($compressors->{$ext}->{bin} eq ''); | |
183 | $cmdline = $compressors->{$ext}->{bin} ." ". | |
184 | $compressors->{$ext}->{args}; | |
185 | last; | |
186 | } | |
187 | } | |
188 | return $cmdline; | |
189 | } | |
190 | ||
2813c06e CE |
191 | sub grep_for_related { |
192 | my ($line,$id) = @_; | |
193 | $id_list{$id} = 1 if $line =~ m/$related_re/; | |
194 | } | |
195 | ||
196 | sub get_related_ids { | |
197 | my ($id) = @_; | |
198 | push @Mids, $id unless grep /\b$id\b/, @Mids; | |
199 | my $re = join '|', @Mids; | |
200 | $related_re = qr/$re/; | |
201 | } | |
202 | ||
420a0d19 CE |
203 | # The main program. Extract the pattern and make sure any relevant characters |
204 | # are quoted if the -l flag is given. The -t flag gives a time-on-queue value | |
2813c06e CE |
205 | # which is an additional condition. The -M flag will also display "related" |
206 | # loglines (msgid from matched lines is searched in following lines). | |
420a0d19 | 207 | |
2813c06e | 208 | getopts('Ilvt:M',\my %args); |
420a0d19 CE |
209 | $queue_time = $args{'t'}? $args{'t'} : -1; |
210 | $insensitive = $args{'I'}? 0 : 1; | |
211 | $invert = $args{'v'}? 1 : 0; | |
2813c06e | 212 | $related = $args{'M'}? 1 : 0; |
420a0d19 | 213 | |
2813c06e | 214 | die "usage: exigrep [-I] [-l] [-M] [-t <seconds>] [-v] <pattern> [<log file>]...\n" |
420a0d19 CE |
215 | if ($#ARGV < 0); |
216 | ||
217 | $pattern = shift @ARGV; | |
218 | $pattern = quotemeta $pattern if $args{l}; | |
219 | ||
220 | ||
221 | # If file arguments are given, open each one and process according as it is | |
222 | # is compressed or not. | |
223 | ||
224 | if (@ARGV) | |
225 | { | |
226 | foreach (@ARGV) | |
227 | { | |
228 | my $filename = $_; | |
2813c06e | 229 | if (-x 'ZCAT_COMMAND' && $filename =~ /\.(?:COMPRESS_SUFFIX)$/o) |
420a0d19 CE |
230 | { |
231 | open(LOG, "ZCAT_COMMAND $filename |") || | |
232 | die "Unable to zcat $filename: $!\n"; | |
233 | } | |
234 | elsif (my $cmdline = &detect_compressor_capable($filename)) | |
235 | { | |
236 | open(LOG, "$cmdline $filename |") || | |
237 | die "Unable to decompress $filename: $!\n"; | |
238 | } | |
239 | else | |
240 | { | |
241 | open(LOG, "<$filename") || die "Unable to open $filename: $!\n"; | |
242 | } | |
243 | do_line() while (<LOG>); | |
244 | close(LOG); | |
245 | } | |
246 | } | |
247 | ||
248 | # If no files are named, process STDIN only | |
249 | ||
250 | else { do_line() while (<STDIN>); } | |
251 | ||
252 | # At the end of processing all the input, print any uncompleted messages. | |
253 | ||
254 | for (keys %id_list) | |
255 | { | |
256 | print "+++ $_ has not completed +++\n$saved{$_}\n"; | |
257 | } | |
258 | ||
259 | # End of exigrep |