| 1 | #! PERL_COMMAND -w |
| 2 | |
| 3 | use strict; |
| 4 | |
| 5 | # Copyright (c) 2007-2014 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 | sub do_line { |
| 64 | |
| 65 | # Convert syslog lines to mainlog format, as in eximstats. |
| 66 | |
| 67 | if (!/^\d{4}-/o) { $_ =~ s/^.*? exim\b.*?: //o; } |
| 68 | |
| 69 | return unless |
| 70 | my($date,$id) = /^(\d{4}-\d\d-\d\d \d\d:\d\d:\d\d (?:[+-]\d{4} )?)(?:\[\d+\] )?(\w{6}\-\w{6}\-\w{2})?/o; |
| 71 | |
| 72 | # Handle the case when the log line belongs to a specific message. We save |
| 73 | # lines for specific messages until the message is complete. Then either print |
| 74 | # or discard. |
| 75 | |
| 76 | if (defined $id) |
| 77 | { |
| 78 | $saved{$id} = '' unless defined($saved{$id}); |
| 79 | |
| 80 | # Save up the data for this message in case it becomes interesting later. |
| 81 | |
| 82 | $saved{$id} .= $_; |
| 83 | |
| 84 | # Are we interested in this id ? Short circuit if we already were interested. |
| 85 | |
| 86 | if ($invert) |
| 87 | { |
| 88 | $id_list{$id} = 1 if (!defined($id_list{$id})); |
| 89 | $id_list{$id} = 0 if (($insensitive && /$pattern/io) || /$pattern/o); |
| 90 | } |
| 91 | else |
| 92 | { |
| 93 | $id_list{$id} = 1 if defined $id_list{$id} || |
| 94 | ($insensitive && /$pattern/io) || /$pattern/o; |
| 95 | } |
| 96 | |
| 97 | # See if this is a completion for some message. If it is interesting, |
| 98 | # print it, but in any event, throw away what was saved. |
| 99 | |
| 100 | if (index($_, 'Completed') != -1 || |
| 101 | index($_, 'SMTP data timeout') != -1 || |
| 102 | (index($_, 'rejected') != -1 && |
| 103 | /^(\d{4}-\d\d-\d\d \d\d:\d\d:\d\d (?:[+-]\d{4} )?)(?:\[\d+\] )?\w{6}\-\w{6}\-\w{2} rejected/o)) |
| 104 | { |
| 105 | if ($queue_time != -1 && |
| 106 | $saved{$id} =~ /^(\d{4}-\d\d-\d\d \d\d:\d\d:\d\d ([+-]\d{4} )?)/o) |
| 107 | { |
| 108 | my $old_sec = &seconds($1); |
| 109 | my $sec = &seconds($date); |
| 110 | $id_list{$id} = 0 if $id_list{$id} && $sec - $old_sec <= $queue_time; |
| 111 | } |
| 112 | |
| 113 | print "$saved{$id}\n" if ($id_list{$id}); |
| 114 | delete $id_list{$id}; |
| 115 | delete $saved{$id}; |
| 116 | } |
| 117 | } |
| 118 | |
| 119 | # Handle the case where the log line does not belong to a specific message. |
| 120 | # Print it if it is interesting. |
| 121 | |
| 122 | elsif ( ($invert && (($insensitive && !/$pattern/io) || !/$pattern/o)) || |
| 123 | (!$invert && (($insensitive && /$pattern/io) || /$pattern/o)) ) |
| 124 | { print "$_\n"; } |
| 125 | } |
| 126 | |
| 127 | # Rotated log files are frequently compressed and there are a variety of |
| 128 | # formats it could be compressed with. Rather than use just one that is |
| 129 | # detected and hardcoded at Exim compile time, detect and use what the |
| 130 | # logfile is compressed with on the fly. |
| 131 | # |
| 132 | # List of known compression extensions and their associated commands: |
| 133 | my $compressors = { |
| 134 | gz => { cmd => 'zcat', args => '' }, |
| 135 | bz2 => { cmd => 'bzcat', args => '' }, |
| 136 | xz => { cmd => 'xzcat', args => '' }, |
| 137 | lzma => { cmd => 'lzma', args => '-dc' } |
| 138 | }; |
| 139 | my $csearch = 0; |
| 140 | |
| 141 | sub detect_compressor_bin |
| 142 | { |
| 143 | my $ext = shift(); |
| 144 | my $c = $compressors->{$ext}->{cmd}; |
| 145 | $compressors->{$ext}->{bin} = `which $c 2>/dev/null`; |
| 146 | chomp($compressors->{$ext}->{bin}); |
| 147 | } |
| 148 | |
| 149 | sub detect_compressor_capable |
| 150 | { |
| 151 | my $filename = shift(); |
| 152 | map { &detect_compressor_bin($_) } keys %$compressors |
| 153 | if (!$csearch); |
| 154 | $csearch = 1; |
| 155 | return undef |
| 156 | unless (grep {$filename =~ /\.(?:$_)$/} keys %$compressors); |
| 157 | # Loop through them, figure out which one it detected, |
| 158 | # and build the commandline. |
| 159 | my $cmdline = undef; |
| 160 | foreach my $ext (keys %$compressors) |
| 161 | { |
| 162 | if ($filename =~ /\.(?:$ext)$/) |
| 163 | { |
| 164 | # Just die if compressor not found; if this occurrs in the middle of |
| 165 | # two valid files with a lot of matches, error could easily be missed. |
| 166 | die("Didn't find $ext decompressor for $filename\n") |
| 167 | if ($compressors->{$ext}->{bin} eq ''); |
| 168 | $cmdline = $compressors->{$ext}->{bin} ." ". |
| 169 | $compressors->{$ext}->{args}; |
| 170 | last; |
| 171 | } |
| 172 | } |
| 173 | return $cmdline; |
| 174 | } |
| 175 | |
| 176 | # The main program. Extract the pattern and make sure any relevant characters |
| 177 | # are quoted if the -l flag is given. The -t flag gives a time-on-queue value |
| 178 | # which is an additional condition. |
| 179 | |
| 180 | getopts('Ilvt:',\my %args); |
| 181 | $queue_time = $args{'t'}? $args{'t'} : -1; |
| 182 | $insensitive = $args{'I'}? 0 : 1; |
| 183 | $invert = $args{'v'}? 1 : 0; |
| 184 | |
| 185 | die "usage: exigrep [-I] [-l] [-t <seconds>] [-v] <pattern> [<log file>]...\n" |
| 186 | if ($#ARGV < 0); |
| 187 | |
| 188 | $pattern = shift @ARGV; |
| 189 | $pattern = quotemeta $pattern if $args{l}; |
| 190 | |
| 191 | |
| 192 | # If file arguments are given, open each one and process according as it is |
| 193 | # is compressed or not. |
| 194 | |
| 195 | if (@ARGV) |
| 196 | { |
| 197 | foreach (@ARGV) |
| 198 | { |
| 199 | my $filename = $_; |
| 200 | if ($filename =~ /\.(?:COMPRESS_SUFFIX)$/o) |
| 201 | { |
| 202 | open(LOG, "ZCAT_COMMAND $filename |") || |
| 203 | die "Unable to zcat $filename: $!\n"; |
| 204 | } |
| 205 | elsif (my $cmdline = &detect_compressor_capable($filename)) |
| 206 | { |
| 207 | open(LOG, "$cmdline $filename |") || |
| 208 | die "Unable to decompress $filename: $!\n"; |
| 209 | } |
| 210 | else |
| 211 | { |
| 212 | open(LOG, "<$filename") || die "Unable to open $filename: $!\n"; |
| 213 | } |
| 214 | do_line() while (<LOG>); |
| 215 | close(LOG); |
| 216 | } |
| 217 | } |
| 218 | |
| 219 | # If no files are named, process STDIN only |
| 220 | |
| 221 | else { do_line() while (<STDIN>); } |
| 222 | |
| 223 | # At the end of processing all the input, print any uncompleted messages. |
| 224 | |
| 225 | for (keys %id_list) |
| 226 | { |
| 227 | print "+++ $_ has not completed +++\n$saved{$_}\n"; |
| 228 | } |
| 229 | |
| 230 | # End of exigrep |