Commit | Line | Data |
---|---|---|
059ec3d9 | 1 | #! PERL_COMMAND -w |
55ee9ee3 | 2 | # $Cambridge: exim/src/src/exigrep.src,v 1.2 2004/12/21 11:28:38 ph10 Exp $ |
059ec3d9 PH |
3 | |
4 | use strict; | |
5 | ||
6 | # Copyright (c) 2004 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 | use Getopt::Std qw(getopts); | |
32 | use POSIX qw(mktime); | |
33 | ||
34 | ||
35 | # This subroutine converts a time/date string from an Exim log line into | |
36 | # the number of seconds since the epoch. It handles optional timezone | |
37 | # information. | |
38 | ||
39 | sub seconds { | |
40 | my($year,$month,$day,$hour,$min,$sec,$tzs,$tzh,$tzm) = | |
41 | $_[0] =~ /^(\d{4})-(\d\d)-(\d\d)\s(\d\d):(\d\d):(\d\d)(?>\s([+-])(\d\d)(\d\d))?/; | |
42 | ||
43 | my $seconds = mktime $sec, $min, $hour, $day, $month - 1, $year - 1900; | |
44 | ||
45 | if (defined $tzs) | |
46 | { | |
47 | $seconds -= $tzh * 3600 + $tzm * 60 if $tzs eq "+"; | |
48 | $seconds += $tzh * 3600 + $tzm * 60 if $tzs eq "-"; | |
49 | } | |
50 | ||
51 | return $seconds; | |
52 | } | |
53 | ||
54 | ||
55 | # This subroutine processes a single line (in $_) from a log file. Program | |
56 | # defensively against short lines finding their way into the log. | |
57 | ||
58 | my (%saved, %id_list, $pattern, $queue_time); | |
59 | ||
60 | sub do_line { | |
61 | return unless | |
62 | my($date,$entry) = /^(\d{4}-\d\d-\d\d \d\d:\d\d:\d\d (?:[+-]\d{4} )?)(.*)/; | |
63 | ||
64 | # Handle the case when the log line belongs to a specific message. We save | |
65 | # lines for specific messages until the message is complete. Then either print | |
66 | # discard. | |
67 | ||
68 | if (my($id) = $entry =~ /^(\w{6}\-\w{6}\-\w{2})/) | |
69 | { | |
70 | $saved{$id} = '' unless defined($saved{$id}); | |
71 | ||
72 | # Save up the data for this message in case it becomes interesting later. | |
73 | ||
74 | $saved{$id} .= $_; | |
75 | ||
76 | # Are we interested in this id ? | |
77 | ||
78 | $id_list{$id} = 1 if /$pattern/io; | |
79 | ||
80 | # See if this is a completion for some message. If it is interesting, | |
81 | # print it, but in any event, throw away what was saved. | |
82 | ||
83 | if ($entry =~ | |
55ee9ee3 | 84 | /(?:Completed|rejected (?:by local_scan|by non-SMTP ACL|after DATA))/) |
059ec3d9 PH |
85 | { |
86 | if ($saved{$id} =~ /^(\d{4}-\d\d-\d\d \d\d:\d\d:\d\d ([+-]\d{4} )?)(\w{6}\-\w{6}\-\w{2})/) | |
87 | { | |
88 | my $old_sec = &seconds($1); | |
89 | my $sec = &seconds($date); | |
90 | delete $id_list{$id} if $id_list{$id} && $sec - $old_sec <= $queue_time; | |
91 | } | |
92 | ||
93 | if ($id_list{$id}) | |
94 | { | |
95 | delete $id_list{$id}; | |
96 | print "$saved{$id}\n"; | |
97 | } | |
98 | ||
99 | delete $saved{$id}; | |
100 | } | |
101 | } | |
102 | ||
103 | # Handle the case where the log line does not belong to a specific message. | |
104 | # Print it if it is interesting. | |
105 | ||
106 | elsif ($entry =~ /$pattern/io) { print "$_\n"; } | |
107 | } | |
108 | ||
109 | ||
110 | # The main program. Extract the pattern and make sure any relevant characters | |
111 | # are quoted if the -l flag is given. The -t flag gives a time-on-queue value | |
112 | # which is an additional condition. | |
113 | ||
114 | getopts('lt:',\my %args); | |
115 | $queue_time = $args{'t'}? $args{'t'} : -1; | |
116 | ||
117 | die "usage: exigrep [-l] [-t <seconds>] <pattern> [<log file>]...\n" | |
118 | if ($#ARGV < 0); | |
119 | ||
120 | $pattern = shift @ARGV; | |
121 | $pattern = quotemeta $pattern if $args{l}; | |
122 | ||
123 | ||
124 | # If file arguments are given, open each one and process according as it is | |
125 | # is compressed or not. | |
126 | ||
127 | if (@ARGV) | |
128 | { | |
129 | foreach (@ARGV) | |
130 | { | |
131 | my $filename = $_; | |
132 | if ($filename =~ /\.(?:COMPRESS_SUFFIX)$/) | |
133 | { | |
134 | open(LOG, "ZCAT_COMMAND $filename |") || | |
135 | die "Unable to zcat $filename: $!\n"; | |
136 | } | |
137 | else | |
138 | { | |
139 | open(LOG, "<$filename") || die "Unable to open $filename: $!\n"; | |
140 | } | |
141 | do_line() while (<LOG>); | |
142 | close(LOG); | |
143 | } | |
144 | } | |
145 | ||
146 | # If no files are named, process STDIN only | |
147 | ||
148 | else { do_line() while (<STDIN>); } | |
149 | ||
150 | # At the end of processing all the input, print any uncompleted data | |
151 | ||
152 | for (keys %id_list) { print "+++ $_ not completed +++\n$saved{$_}\n;" } | |
153 | ||
154 | # End of exigrep |