#!/usr/local/bin/perl

use Time::Local;
require 'getopts.pl';

###
### default values
###

$ath = 64;
$tth = $ath;
$tunit = 3600;  # 1 hour
$sth = 8;
$sunit = 1024;  # 1 K
$pth = $ath;


sub usage {
    print <<EOM;
log reducer for httpd common log format.
        by k-chinen\@is.aist-nara.ac.jp, 2000, 2001.

options:                                            : current
    -t num      threshould of scale shift for time  : $tth
    -T num      unit of time.  use seconds          : $tunit
    -s num      threshould of scale shift for size  : $sth
    -S num      unit of size.  use seconds          : $sunit
    -p num      threshould of scale shift for peer  : $pth

time items are shown by range, if the number of time items is lager than $tth .
size items are shown by range, if the number of size items is lager than $sth .
peer items are not shown, if the number of peer is lager than $pth .

EOM
}


&Getopts('t:T:s:S:p:h');
if($opt_t) {
    $tth = $opt_t;
}
if($opt_T) {    ### time unit
    $tunit = $opt_T;
}
if($opt_s) {
    $sth = $opt_s;
}
if($opt_S) {
    $sunit = $opt_S;
}
if($opt_p) {
    $pth = $opt_p;
}
if($opt_h) {
    &usage;
    exit 0;
}


%monthconv = (
    "jan" => 0, "feb" => 1, "mar" => 2, "apr" => 3,
    "may" => 4, "jun" => 5, "jul" => 6, "aug" => 7,
    "sep" => 8, "oct" => 9, "nov" =>10, "dec" =>11,
);



sub report_normal
{
    my($tempalist) = @_;
    my($n, $v);

    while( ($n, $v) = each %{ $tempalist } ) {
        print "\t\t$n $v\n";
    }
}


sub report_timeband
{
    my($tempalist) = @_;
    my(%alist) = %{ $tempalist } ;
    my(@tlist) = sort {$a<=>$b} keys %alist;
    my($n, $v);
    my($t0, $t1, $diff, $be, $en, $sum);
    my(%band);

 if(0) {
    $sum = 0;
    while( ($n, $v) = each %alist ) {
        print "\t\t$n $v\n";
        $sum += $v;
    }
    print "\t\ttmp-sum $sum\n";
 }

    @nlist = keys %alist;
    if($#nlist<=$tth) {
        &report_normal( $tempalist );
        return 0;
    }

    while($#tlist>-1) {
        $t0  = shift @tlist;
        $be  = $t0;
        $en  = $t0;
        $sum = $alist{$t0};

#       print "\tt0 $t0 $alist{$t0}\n";

        while(1) {
            $t1  = shift @tlist;
            if($t1 eq '') {
                last;
            }
            $diff= $t1 - $en;

#           print "\t  t1 $t1 $alist{$t1} diff $diff\n";

            if($diff<=$tunit) {
                $en = $t1;
                $sum += $alist{$t1};
            }
            else {
                unshift @tlist, $t1;
                last;
            }
        }

        $band{$be ."-". ($en+$tunit-1) } = $sum;
#print "\t\t band{$be-$en}=$sum\n";
    }

    $sum = 0;
    foreach $n (sort {$a<=>$b} keys %band) {
        print "\t\t$n $band{$n}\n";
        $sum += $band{$n};
    }
#   print "\t\tsum $sum\n";
    undef %band;

}


sub report_sizeband
{
    my($tempalist) = @_;
    my(%alist) = %{ $tempalist } ;
    my(%nalist);
    my($n, $be, $en, $v);
    my(@nlist);

    @nlist = keys %alist;
    if($#nlist<=$sth) {
        &report_normal( $tempalist );
        return 0;
    }

    foreach $n (@nlist) {
        $nalist{ int($n/$sunit) } += $alist{$n};
    }

    foreach $n (sort {$a<=>$b} keys %nalist) {
        $be = $n*$sunit;
        $en = $be+$sunit -1;
        print "\t\t$be-$en $nalist{$n}\n";
    }
}



while(<>) {
    if(/^(\S+)\s+(\S+)\s+(\S+)\s+\[(\S+)\s+(\+?-?\d+)\]\s+"(\w+)\s+(\S+)\s+(HTTP\/\d+\.\d+)"\s+(\d+)\s+(\d+)$/) {
        $peer   = $1;
        $xdate  = $4;
        $method = $6;
        $uri    = $7;
        $xver   = $8;
        $rcode  = $9;
        $size   = $10;


        # typical value
        #   29/Dec/2000:00:00:03
        #
        $xdate =~ /(\d+)\/(\w+)\/(\d+):(\d+):(\d+):(\d+)/;
        $xda = $1; $xxmo = $2; $xye = $3; $xho = $4; $xmi = $5; $xse = $6;
        $xxmo =~ tr/A-Z/a-z/;
        $xmo = $monthconv{$xxmo};
        $xye -= 1900;
        $t = timelocal($xse, $xmi, $xho, $xda, $xmo, $xye);

        $ucount{$uri}++;

        $count{$uri}{"time"}{int($t/$tunit)*$tunit}++;
        $count{$uri}{"peer"}{$peer}++;
        $count{$uri}{"method"}{$method}++;
        $count{$uri}{"rcode"}{$rcode}++;
        $count{$uri}{"size"}{$size}++;

    }
}



foreach $name (keys %count) {
    print "$name\n";
    print "\tcount\n\t\t$ucount{$name}\n";
    while(($a, $av) = each %{ $count{$name} } ) {
        print "\t$a\n";
        if($a eq 'time') {
            &report_timeband( $av );
        }
        elsif($a eq 'size') {
            &report_sizeband( $av );
        }
        elsif($a eq 'peer' && $ucount{$name} > $pth ) {
            print "\t\t-\n";
        }
        else {
            &report_normal( $av );
        }
    }
}