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