Commit | Line | Data |
---|---|---|
059ec3d9 | 1 | #! PERL_COMMAND -w |
75b1493f | 2 | # $Cambridge: exim/src/src/exigrep.src,v 1.5 2007/02/07 12:23:35 ph10 Exp $ |
059ec3d9 PH |
3 | |
4 | use strict; | |
5 | ||
f3f065bb | 6 | # Copyright (c) 2007 University of Cambridge. |
059ec3d9 PH |
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 | ||
75b1493f PH |
31 | # Performance optimized in 02/02/2007 by Jori Hamalainen |
32 | # Typical run time acceleration: 4 times | |
33 | ||
34 | ||
059ec3d9 PH |
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) = | |
75b1493f | 45 | $_[0] =~ /^(\d{4})-(\d\d)-(\d\d)\s(\d\d):(\d\d):(\d\d)(?>\s([+-])(\d\d)(\d\d))?/o; |
059ec3d9 PH |
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 | ||
75b1493f | 62 | my (%saved, %id_list, $pattern, $queue_time, $insensitive); |
059ec3d9 PH |
63 | |
64 | sub do_line { | |
395ff96d PH |
65 | |
66 | # Convert syslog lines to mainlog format, as in eximstats. | |
67 | ||
75b1493f | 68 | if (!/^\d{4}-/o) { $_ =~ s/^.*? exim\b.*?: //o; } |
395ff96d | 69 | |
059ec3d9 | 70 | return unless |
75b1493f | 71 | my($date,$id) = /^(\d{4}-\d\d-\d\d \d\d:\d\d:\d\d (?:[+-]\d{4} )?)(\w{6}\-\w{6}\-\w{2})?/o; |
059ec3d9 PH |
72 | |
73 | # Handle the case when the log line belongs to a specific message. We save | |
74 | # lines for specific messages until the message is complete. Then either print | |
f3f065bb | 75 | # or discard. |
059ec3d9 | 76 | |
75b1493f | 77 | if (defined $id) |
059ec3d9 PH |
78 | { |
79 | $saved{$id} = '' unless defined($saved{$id}); | |
80 | ||
81 | # Save up the data for this message in case it becomes interesting later. | |
82 | ||
83 | $saved{$id} .= $_; | |
84 | ||
75b1493f | 85 | # Are we interested in this id ? Short circuit if we already were interested. |
059ec3d9 | 86 | |
75b1493f PH |
87 | $id_list{$id} = 1 if defined $id_list{$id} || |
88 | ($insensitive && /$pattern/io) || /$pattern/o; | |
059ec3d9 PH |
89 | |
90 | # See if this is a completion for some message. If it is interesting, | |
91 | # print it, but in any event, throw away what was saved. | |
92 | ||
75b1493f PH |
93 | if (index($_, 'Completed') != -1 || |
94 | (index($_, 'rejected') != -1 && | |
95 | /rejected (?:by local_scan|by non-SMTP ACL|after DATA)/o)) | |
059ec3d9 | 96 | { |
75b1493f PH |
97 | if ($queue_time != -1 && |
98 | $saved{$id} =~ /^(\d{4}-\d\d-\d\d \d\d:\d\d:\d\d ([+-]\d{4} )?)/o) | |
059ec3d9 PH |
99 | { |
100 | my $old_sec = &seconds($1); | |
101 | my $sec = &seconds($date); | |
102 | delete $id_list{$id} if $id_list{$id} && $sec - $old_sec <= $queue_time; | |
103 | } | |
104 | ||
105 | if ($id_list{$id}) | |
106 | { | |
107 | delete $id_list{$id}; | |
108 | print "$saved{$id}\n"; | |
109 | } | |
110 | ||
111 | delete $saved{$id}; | |
112 | } | |
113 | } | |
114 | ||
115 | # Handle the case where the log line does not belong to a specific message. | |
116 | # Print it if it is interesting. | |
117 | ||
75b1493f PH |
118 | elsif (($insensitive && $_ =~ /$pattern/io) || $_ =~ /$pattern/o) |
119 | { print "$_\n"; } | |
059ec3d9 PH |
120 | } |
121 | ||
122 | ||
123 | # The main program. Extract the pattern and make sure any relevant characters | |
124 | # are quoted if the -l flag is given. The -t flag gives a time-on-queue value | |
125 | # which is an additional condition. | |
126 | ||
75b1493f | 127 | getopts('Ilt:',\my %args); |
059ec3d9 | 128 | $queue_time = $args{'t'}? $args{'t'} : -1; |
75b1493f | 129 | $insensitive = $args{'I'}? 0 : 1; |
059ec3d9 | 130 | |
75b1493f | 131 | die "usage: exigrep [-I] [-l] [-t <seconds>] <pattern> [<log file>]...\n" |
059ec3d9 PH |
132 | if ($#ARGV < 0); |
133 | ||
134 | $pattern = shift @ARGV; | |
135 | $pattern = quotemeta $pattern if $args{l}; | |
136 | ||
137 | ||
138 | # If file arguments are given, open each one and process according as it is | |
139 | # is compressed or not. | |
140 | ||
141 | if (@ARGV) | |
142 | { | |
143 | foreach (@ARGV) | |
144 | { | |
145 | my $filename = $_; | |
75b1493f | 146 | if ($filename =~ /\.(?:COMPRESS_SUFFIX)$/o) |
059ec3d9 PH |
147 | { |
148 | open(LOG, "ZCAT_COMMAND $filename |") || | |
149 | die "Unable to zcat $filename: $!\n"; | |
150 | } | |
151 | else | |
152 | { | |
153 | open(LOG, "<$filename") || die "Unable to open $filename: $!\n"; | |
154 | } | |
155 | do_line() while (<LOG>); | |
156 | close(LOG); | |
157 | } | |
158 | } | |
159 | ||
160 | # If no files are named, process STDIN only | |
161 | ||
162 | else { do_line() while (<STDIN>); } | |
163 | ||
164 | # At the end of processing all the input, print any uncompleted data | |
165 | ||
75b1493f | 166 | for (keys %id_list) { print "+++ $_ not completed +++\n$saved{$_}\n"; } |
059ec3d9 PH |
167 | |
168 | # End of exigrep |