find_unsleeper.pl

#!/usr/bin/perl -w
use strict;
use warnings;
use Time::Local 'timelocal';
$| = 1;

my $config = {
    # ログの検索条件(ユーザIDを指定することになるはず)
    criterion => undef,
    # 探索するApacheログの格納パス
    logs => '~/logs',
    # ここで指定した秒数を超えてアクセスがない場合は、サービスから離れたとみなす
    threshold_sec => 300,   # デフォルト5分(300sec)
};
my $MONTH = {Jan => 0, Feb => 1, Mar => 2, Apr => 3, May => 4, Jun => 5, Jul => 6, Aug => 7, Sep => 8, Oct => 9, Nov => 10, Dec => 11};

die "usage: perl ./who_is_unsleeper.pl --target=<target-viewer-id>\nexample)117965337" unless ($ARGV[0]);
$config->{criterion} = $ARGV[0];
main();

sub main {
    my $stat = {
        criterion => $config->{criterion},
        date_logs => {},
        intervals => []
    };
    my @files = list_apache_accesslog_files($config->{logs});

#    my $file = pop @files;
#    check_logfile($stat, $file);
    foreach my $file (@files) {
        check_logfile($stat, $file);
    }

    output_report($stat);
}

sub check_logfile {
    my $stat = shift or die;
    my $file = shift or die;
    print "Reading... '$file' \n";
    if ($file =~ m/\.gz$/) {
        check_gz_logfile($stat, $file);
    } elsif ($file =~ m/\.log$/) {
        check_txt_logfile($stat, $file);
    } else {
        die "Unknown logfile format: $file";
    }
}

sub check_gz_logfile {
    my $stat = shift or die;
    my $file = shift or die;

    open my $fh, "zcat $file 2>/dev/null |" or die "Cannot zcat '$file' for reading: $!";
    while (<$fh>) {
        next unless (m/$stat->{criterion}/);

        # ログが残された日時を取得
        die unless (m/(\d{2})\/([A-Z][a-z]{2})\/(\d{4}):(\d{2}):(\d{2}):(\d{2})\s\+\d{4}/);
        my ($sec, $min, $hour, $mday, $month, $year) = (int($6), int($5), int($4), int($1), $MONTH->{$2}, int($3));
        my $time = timelocal($sec, $min, $hour, $mday, $month, $year - 1900);
        my $date = sprintf("%04d-%02d-%02d", $year, $month + 1, $mday);

        # ログ日付の統計情報がなければ、この時点で作成する
        unless (exists $stat->{'date_logs'}->{$date}) {
            $stat->{'date_logs'}->{$date} = {
                access_count => 0,
                access_list => [],
                lasttime => 0,
            };
            foreach (0..23) {
                push @{$stat->{'date_logs'}->{$date}->{access_list}},  { access_count => 0, interval_sec => [] };
            }
        }

        # 前日の日付を取得
        my $prev_time = timelocal(0, 0, 0, $mday, $month, $year - 1900) - (24 * 60 * 60);
        my ($prev_day, $prev_month, $prev_year) = (localtime($prev_time))[3..5];
    	my $lastdate = sprintf("%04d-%02d-%02d", $prev_year + 1900, $prev_month + 1, $prev_day);

        # 当日のアクセスログがある場合→それと比較
        # 当日のアクセスログがない場合→前日分をみて、あれば比較
        # アクセス時間差を取る
        my $lasttime = undef;
        if ($stat->{'date_logs'}->{$date}->{'lasttime'} > 0) {
            $lasttime = $stat->{'date_logs'}->{$date}->{'lasttime'};
        } elsif (exists $stat->{'date_logs'}->{$lastdate} && $stat->{'date_logs'}->{$lastdate}->{'lasttime'} > 0) {
            $lasttime = $stat->{'date_logs'}->{$lastdate}->{'lasttime'};
        }

        if ($lasttime) {
            my $interval_sec = $time - $lasttime;
            if ($interval_sec > $config->{threshold_sec}) {
                # 休息をカウント
                push @{$stat->{'date_logs'}->{$date}->{access_list}[$hour]->{interval_sec}}, $interval_sec;
                push @{$stat->{intervals}}, {
                    rest_start => $lasttime,
                    rest_end => $time
                };
            }
        }
        #TODO: アクセス数をカウント
        $stat->{'date_logs'}->{$date}->{access_count}++;
        $stat->{'date_logs'}->{$date}->{access_list}[$hour]->{access_count}++;

        if ($stat->{'date_logs'}->{$date}->{'lasttime'} < $time) {
            $stat->{'date_logs'}->{$date}->{'lasttime'} = $time;
        }

        #print $_;
    }
    close $fh;
}

# 指定したディレクトリ配下のapacheログファイル名をリストアップして返す
sub list_apache_accesslog_files {
    my $dir = shift or die;
    my @files = ();

    opendir DIR, $dir or die "Cannot open '$dir' accesslog dir: $!";
    foreach (readdir(DIR)) {
        next unless (m/apache\.(?:sp\.)?accesslog\.\d{8}\.log\.gz/);
        push @files, $config->{logs}.'/'.$_;
    }
    closedir DIR;

    return sort {
        my $a_date = $1 if ($a =~ m/(\d{8})/);
        my $b_date = $1 if ($b =~ m/(\d{8})/);
        return  1 if ($a_date > $b_date);
        return -1 if ($a_date < $b_date);
        return 0;
    } @files;
}

# Apacheログの日付形式をエポック秒に変換して返す
# 例) '09/May/2016:00:00:02 +0900' => 1462719602
sub str_to_epochtime {
    die unless ($_[0] =~ m/(\d{2})\/([A-Z][a-z]{2})\/(\d{4}):(\d{2}):(\d{2}):(\d{2})\s\+\d{4}/);
    return timelocal(int($6), int($5), int($4), int($1), $MONTH->{$2}, (int($3) - 1900));
}

sub epochtime_to_str {
    my ($sec, $min, $hour, $mday, $mon, $year) = localtime($_[0]);
    return sprintf("%04d-%02d-%02d %02d:%02d:%02d", $year + 1900, $mon + 1, $mday, $hour, $min, $sec);
}

#sub get_yesterday {
#    my $date = shift or die;
#    die unless ($date =~ m/^(\d{4})(\d{2})(\d{2})$/);
#    my $time = timelocal(0, 0, 0, $3, $2 - 1, $1 - 1900) - (24 * 60 * 60);
#    my ($day, $month, $year) = (localtime($time))[3..5];
#	return sprintf("%04d%02d%02d", $year + 1900, $month + 1, $day);
#}

sub output_report {
    my $stat = shift or die;
    my $date_logs = $stat->{date_logs};
    foreach my $date (sort keys %$date_logs) {
        print "[$date]\n";
        print "\tアクセス数: $date_logs->{$date}->{access_count}\n";
        print "\t最終アクセス: ".epochtime_to_str($date_logs->{$date}->{lasttime})."\n";
        foreach my $hour (0..23) {
            print "\t\t".sprintf("%02d", $hour).": ".sprintf("%4d", $date_logs->{$date}->{access_list}[$hour]->{access_count});
            #print ", 5分以上の休憩:".($#{$date_logs->{$date}->{access_list}[$hour]->{interval_sec}} + 1);
            print "\n";
        }
    }
    print "\n";
    print "[休憩時間]\n";
    foreach my $interval (@{$stat->{intervals}}) {
        my $diff = ($interval->{rest_end} - $interval->{rest_start});
        my $hour = int($diff / 60 / 60);
        $diff -= ($hour * 60 * 60);
        my $min = int($diff / 60);
        my $sec = $diff - ($min * 60);

        print sprintf("\t%d時間%02d分%02d秒", $hour, $min, $sec), "\t(", epochtime_to_str($interval->{rest_start})." 〜 ".epochtime_to_str($interval->{rest_end}), ")\n";
    }
}