#!/usr/bin/perl 

# History
# 1.0.1 - 07/09/1997 Updated script to handle cookie log
#                    created by Apache 1.2 as well as Apache 1.1
#
# 1.0 - 07/10/1996 Initial version. 8-)
#

$version="1.0.1";

# 'ctime.pl' and 'localtime.pl' are standard parts
# of the perl library.
# In Other Words: Don't write me asking about them
# - they should *already* be on your system.

require 'ctime.pl';
require 'timelocal.pl';

# Strip '?' parameters (0=don't strip it, 1= do strip it)
$STRIPTARGETPARMS=1;

# Strip '#' anchors (0=don't strip it, 1= do strip it)
$STRIPANCHORS=1;

# Exclusion patterns (uncomment to use)
# This excludes any URL matching this pattern
# from the report.
$EXCLUDEREFSTO='\.(gif|png|xbm|jpg)$';

# Inclusion patterns (uncomment to use)
# This includes *ONLY* URLs matching the pattern
# in the report.
# $INCLUDEONLYREFSTO='(\.html?|/)$';

# Blank out any URL seqment matchin this pattern
# (Good for compacting virtual URL spaces)
$BLANKOUT='(index.html?|/part\d+|/\d+)$';

# default path to the cookie_log
$CookieLog="/wwwlogs/netimages/cookie_log";

# Name of server
$HTTPDSERVER='www.netimages.com';

# Log type (apache)
$LogType="apache";

&Initialize;

@ARGV = ("$CookieLog") if ($#ARGV == -1);

foreach $LogFile (@ARGV) {
	if (! open(COOKIESLOG,$LogFile) ) {
		print STDERR "Can't open $LogFile. Skipped.\n $!";
		next;
	}
	while($line=<COOKIESLOG>) {
		chop $line;

		# Count the number of hits before doing anything
		$rawhitscounter++;

	if ($LogType eq 'apache') {
		($cookie,$method,$target,$protocal,$day,$month,$year,$hour,$minute,$second,$time,$zone) 
		= $line =~ m#^(\S+)\s\"?(\S+)\s(\S*)\s(\S*)\"?\s\[(\d\d?)/(\w\w\w)/(\d\d\d\d):(\d\d?):(\d\d):(\d\d)\s+(\S+)\]#o;
			$day  = "0$day" if  (length($day) == 1);
			$hour = "0$hour" if (length($day) == 1);
			$datestamp = "${year}$NameToMonth{$month}${day}${hour}${minute}${second}";
	}

		next if (! ($target));
		$target=~ s/\%7[eE]/~/o;	# Caniconalize %7E and %7e as ~
		$target=~ s#//#/#go;		# Remove any extra slashes
		$target=~ s#^ *$#/#o;		# fix root ref if needed
		$target=~ s/#.+$//o;		# combine #anchor refs with root doc
		$target=~ s/\?.*$//o if $STRIPTARGETPARMS;	# strip '?' parameters
		$target=~ s/\#.*$//o if $STRIPANCHORS;	# strip '#' anchors
		$target=~ s#${BLANKOUT}##o if $BLANKOUT;

		# Dump anything we have specifically excluded 
		#if we are using an exclusion pattern.
		next if ($EXCLUDEREFSTO && ($target=~m#$EXCLUDEREFSTO#io));

		# Dump anything not matching our inclusion patter
		# if we are using one.
		next if ($INCLUDEONLYREFSTO && (!( $target=~m#$INCLUDEONLYREFSTO#io)));

		# Count the total number of lines analyzed
		$refscounter++;

		# Count total number of hits on each target
		$TotalTargetCounter{$target}++;

		# Count total number of cookies
		$CookiesCounter++ if (! defined($TotalCookieCounter{$cookie}));

		# Count total number of hits by each cookie
		$TotalCookieCounter{$cookie}++;

		$EndDate = $datestamp if ($EndDate < $datestamp);
		$StartDate = $datestamp if ($StartDate > $datestamp);

		# Keep track of the first hit on this target
		if (defined $TargetFirstDate{$target}) {
			if ($TargetFirstDate{$target} > $datestamp) {
				$TargetFirstDate{$target}=$datestamp;
			}
		} else {
			$TargetFirstDate{$target}=$datestamp;
		}

		# Keep track of the last hit on this target
		if (defined $TargetLastDate{$target}) {
			if ($TargetLastDate{$target} < $datestamp) {
				$TargetLastDate{$target}=$datestamp;
			}
		} else {
			$TargetLastDate{$target}=$datestamp;
		}

		# Keep track of the first hit by this cookie
		if (defined $CookieFirstDate{$cookie}) {
			if ($CookieFirstDate{$cookie} < $datestamp) {
				$CookieFirstDate{$cookie}=$datestamp;
			}
		} else {
			$CookieFirstDate{$cookie}=$datestamp;
		}
		# Keep track of the last hit by this cookie
		if (defined $CookieLastDate{$cookie}) {
			if ($CookieLastDate{$cookie} > $datestamp) {
				$CookieLastDate{$cookie}=$datestamp;
			}
		} else {
			$CookieLastDate{$cookie}=$datestamp;
		}

		# Keep track of unique URLs visited for each cookie
		# and unique cookies for each URL visited
		$key = "$target $cookie";
		if (! defined($TargetCookieTuple{$key})) {
			$TargetUniqueCookies{$target}++;
			$CookieUniqueTargets{$cookie}++;
			$TargetCookieTuple{$key}++;

		}
	}
	close(COOKIESLOG);
}

if ($OutputFile) {
	open (OUTPUTFILE,">$OutputFile") || 
		die ("Could not open $OutputFile for writing.\n$!");
	select(OUTPUTFILE);
}

if ($CookiesCounter == 0) {
	&NoCookies;
	exit;
}

$date=&ctime(time);
$avhits=int(10 * $refscounter/$CookiesCounter) /10;

$End=&ConvertDate($EndDate);
$Start=&ConvertDate($StartDate);

foreach $cookie (keys(%TotalCookieCounter)) {
	$TimeOnSite=&DiffDates($CookieFirstDate{$cookie},$CookieLastDate{$cookie});
	$TotalTime+=$TimeOnSite;
}
$avtime=int($TotalTime/$CookiesCounter);

print <<"EOF";
<html>
<head>
<title>Cookie Statistics for ${HTTPDSERVER}</title>

<!-- Got style? You will. -->
<link rel="stylesheet" href="/stylesheets/default.css">
</head>

<body>
<h1 align=center>Cookie Statistics for ${HTTPDSERVER}</h1>
<p>
$rawhitscounter raw references examined;<br>
$refscounter references used in final statistics generation.

</p>
<table>
<tr>
<td valign=top align=left>
<strong>Start Date of Report:</strong>
</td>
<td valign=top align=left>
$Start
</td>
</tr>
<tr>
<td valign=top align=left>
<strong>End Date of Report:</strong>
</td>
<td valign=top align=left>
$End
</td>
</tr>
<tr>
<td valign=top align=left>
<strong>Number of Cookies Assigned:</strong>
</td>
<td valign=top align=left>
$CookiesCounter
</td>
</tr>
<tr>
<td valign=top align=left>
<strong>Average Number of Hits Per Assigned Cookie:</strong>
</td>
<td valign=top align=left>
$avhits
</td>
</tr>
<tr>
<td valign=top align=left>
<strong>Average Time Spent on Site:</strong>
</td>
<td valign=top align=left>
$avtime seconds
</td>
</tr>
</table>
<hr>
<h2 align=left>Cookie breakdown by file accessed</h2>
<pre>
Last Accessed                  Cookies  File
-------------------------------------------------------------------------------
EOF
foreach $target (sort keys(%TargetUniqueCookies)) {

	$date=&ConvertDate($TargetLastDate{$target});
	chop $date;
	$cookies=$TargetUniqueCookies{$target};
	$file=&htmlize($target);
	printf "%-25s %12s  %-s\n",$date,$cookies,$file;
}



print <<"EOF";
-------------------------------------------------------------------------------
</pre>
<address>
<a href="http://www.netimages.com/~snowhare/utilities/">CookieStats 
$version</a> / 
<a 
href="http://www.netimages.com/~snowhare/email/">snowhare\@netimages.com</a>
</address>
</body>
</html>
EOF
select(STDOUT);

sub htmlize {

        local($line)=$_[0];

        $line=~s/&/\&amp;/go;
        $line=~s/>/\&gt;/go;
        $line=~s/</\&lt;/go;
        $line=~s/\"/\&quot;/go;
        $line;
}
sub NoCookies {

}

sub ConvertDate {
	local($datestring)=$_[0];
	local($cyear,$cmonth,$cday,$chour,$cminute,$csecond,$cjulian);

	($cyear,$cmonth,$cday,$chour,$cminute,$csecond) = $datestring =~
	m#^\d\d(\d\d)(\d\d)(\d\d)(\d\d)(\d\d)(\d\d)$#o;
	$julian=&timelocal($csecond,$cminute,$chour,$cday,$cmonth,$cyear);
	&ctime($julian);
}

sub DiffDates {
	local($datestring1,$datestring2)=($_[0],$_[1]);
	local($cyear,$cmonth,$cday,$chour,$cminute,$csecond,$cjulian);
	local($difftime);

	($cyear,$cmonth,$cday,$chour,$cminute,$csecond) = $datestring1 =~
	m#^\d\d(\d\d)(\d\d)(\d\d)(\d\d)(\d\d)(\d\d)$#o;
	$julian1=&timelocal($csecond,$cminute,$chour,$cday,$cmonth,$cyear);

	($cyear,$cmonth,$cday,$chour,$cminute,$csecond) = $datestring2 =~
	m#^\d\d(\d\d)(\d\d)(\d\d)(\d\d)(\d\d)(\d\d)$#o;
	$julian2=&timelocal($csecond,$cminute,$chour,$cday,$cmonth,$cyear);
	
	$difftime=$julian2 - $julian1;
	$difftime = -$difftime if ($difftime < 0);
	$difftime;
}

sub Initialize {

	%NameToMonth=(
		'Jan','00','Feb','01','Mar','02','Apr','03','May','04',
		'Jun','05','Jul','06','Aug','07','Sep','08','Oct','09',
		'Nov','10','Dec','11',);
	@MonthToName=(Jan,Feb,Mar,Apr,May,Jun,Jul,Aug,Sep,Oct,Nov,Dec);

	&ReadCommandLine('output:name:logtype:include:exclude:stripref:striptarget');

	# Output file (optional - goes to STDOUT if not specified)
	$OutputFile = $opt{'output'} if ($opt{'output'});

	# Type of log (legal: combined, referer)
	$LogType = $opt{'logtype'} if ($opt{'logtype'});

	# Exclusion patterns
	$EXCLUDEREFSTO= $opt{'exclude'} if ($opt{'exclude'});

	# Exclusion patterns
	$INCLUDEONLYREFSTO= $opt{'include'} if ($opt{'include'});

	# blankout patterns
	$BLANKOUT= $opt{'blank'} if ($opt{'blank'});

	# '?' parameter stripping options
	$STRIPTARGETPARMS = $opt{'striptarget'} if (defined($opt{'striptarget'}));

	# '#' anchor stripping options
	$STRIPANCHORS = $opt{'stripanchors'} if (defined($opt{'stripanchors'}));

	# Name of server
	$HTTPDSERVER=$opt{'name'} if ($opt{'name'});

	# Artificial boundaries for date stuff
	$StartDate="99999999999999";	
	$EndDate  ="00000000000000";	
}

sub ReadCommandLine {

        # parse list has the form 'a:b:c'
        # flags with parse list entries must take values

        local($parselist)=$_[0];
        local(@CommandLine)=@ARGV;
        local(@ParseList,%ParseRules,@GenericList);

        (@ParseList)=split(/:/,$parselist);

        foreach $item (@ParseList) {
                $ParseRules{$item}=1;
        }

        while ($parm=shift(@CommandLine)) {
                if ($parm =~ m#^\-([a-zA-Z]+)$#o) {
                        $parm=$1;
                        $opt{$parm}=1;
                        if ($ParseRules{$parm}) {
                                $value=shift(@CommandLine);
                                if ($value eq "") {
                                        die ("Invalid comand line switch usage, '-$parm' requires value\n");
                                }
                                $opt{$parm}=$value;
                        }
                        next;
                }
                push(@GenericList,$parm);
        }
        @ARGV=@GenericList;
}