#!/usr/local/bin/perl

#Name: stats: develop stats page from log files.
$baseurl = "/home/grad23/doyle/.www/cgi-bin";
$| = 1;
$logname= "$baseurl/logs/counter.log";
$totalfile= "$baseurl/logs/counter.txt";
$histdir = "$baseurl/../logs";
$spacing = 10;


if (!(-e "$baseurl/countrys.txt"))
{
    print "missing the countries file\n\n";
    exit;
}

# get country codes for different hostnames
open (CTRY, "$baseurl/countrys.txt")|| die "$0: can\'t open countrys.txt:
$!\n";
while (<CTRY>)
{
	$line = $_;
	$line =~ /\A\s*(\w+)\s+(.*)/;
	push @ends, $1;
	push @ctrys, $2;
}
close (CTRY);


#get total number of hits
open (TOTAL, "$totalfile")|| die "$0: can\'t $totalfile: $!\n";
$total = <TOTAL>;
close (TOTAL);

&view_current if ($ARGV[0] eq 'view_current');
&roll_history if ($ARGV[0] eq 'roll_hist') ;
&view_current;

sub get_date {
#generate date string mm/dd/yy  ($mdy)
    @time = localtime(time);
    $mm = $time[4] + 1;
    $mm = "0$mm" if (length($mm) == 1);
    $dd = $time[3];
    $dd = "0$dd" if (length($dd) == 1);
    $yy = $time[5];
    $mdy = "$mm/$dd/$yy";
    return $mdy;
}


sub get_month {
#generate month mm-yy ($my)
    @time = localtime(time);
    $mm = $time[4] + 1;
    $mm = "0$mm" if (length($mm) == 1);
    $yy = $time[5];
    $my = "$mm-$yy";
    return $my;
}


sub gen_page {
    $today = &get_date;
    %servertypes=();
    %doctypes=();
    %referertypes=();
    %browsertypes=();
    %ostypes=();
    %datetypes=();
    %timetypes=();
    $maxsize = 300;
    $maxdom =1;
    $maxdoc =1;
    $maxref =1;
    $maxtime=1;
    $maxdate=1;

    &bad_name unless (-e $logname);

    open(LOG_FILE,"$logname");
    while(<LOG_FILE>){
	$line=$_;
	
	( $tdate, $document, $referer, $hostname, $browser, $os, $user)
	    = split(/\|/,$line);
	($idate, $itime) = split(/\s+/,$tdate);

	#split browser and OS stuff, if needed.
	if ($os =~ /\A\s*\Z/)                           # if os type missing
	{
	    $browser =~ /(\S+)\s+(\S+.*)/;
	    $browser = $1;
	    $os = $2;
	    $os =~ s/\[.*\]\s+//;
	}
	else
	{
	    $browser =~ s/\s+\S+.*//;
	}

	#get hostname
	if ($hostname =~ /([A-Za-z]+)$/) {  	  #takes last word
	    $type = $1;	}
	else { #convert all numeric only addresses
	    $type = "unknown (numeric)"; }

	     
	if (defined $servertypes{$type}){
	    $servertypes{$type}++;
	    if($servertypes{$type} > $maxdom){
		$maxdom = $servertypes{$type}; }
	}
	else {
	    $servertypes{$type} = 1; }

	if ($hostname =~/\.aol\.com/i)
	{
	    if (defined $servertypes{"aol (America On-Line)"}) {
		$servertypes{"aol (America On-Line)"}++; }
	    else {
		$servertypes{"aol (America On-Line)"} = 1; }
	}

	if ($hostname =~/\.umn\.edu/i)
	{
	    if (defined $servertypes{"umn (U of MN)"}) {
		$servertypes{"umn (U of MN)"}++; }
	    else {
		$servertypes{"umn (U of MN)"} = 1; }
	}

	#get document
	if (defined $doctypes{$document}){      #document
	    $doctypes{$document}++;
	    if($doctypes{$document} > $maxdoc){
		$maxdoc = $doctypes{$document}; }
	}
	else {
	    $doctypes{$document}=1; }

	#get referer
	if (($referer =~ /~doyle/) || ($referer =~ /\%7Edoyle/)) {
	    $referer =~ s/\A.*?doyle//; }
	if ($referer =~ /\A\s*\Z/) {
	    $referer = "unknown"; }
	if (defined $referertypes{$referer}){        #referer
	    $referertypes{$referer}++; }
	else {
	    $referertypes{$referer}=1; }

	#find browser
	$browser =~ s/(Mozilla\/\d+\.\d+).+/$1/;
	if ($browser =~ /\A\s*\Z/) {
	    $browser = "unknown"; }

	if (defined $browsertypes{$browser}){    #browser
	    $browsertypes{$browser}++; }
	else {
	    $browsertypes{$browser}=1; }

	if ($os =~ /MSIE/) {
	    $xtra= $os;
	    $xtra =~ /(MSIE[\s\d\.]*)/;
	    $xtra = $1;
	    if (defined $browsertypes{$xtra}){    #browser
		$browsertypes{$xtra}++; }
	    else {
		$browsertypes{$xtra}=1; }
	}


	$os =~ s/(.*?)via.*/$1/;
	$os =~ s/\A\((.*?)\)(.*)/$1$2/;
	$os =~ s/\;\s+[IU]\s*\Z//;
	$os =~ s/\;\s+[IU]\s*\;/;/;
	$os =~ s/X11\s*\;\s*//;
	$os =~ s/\A\s*//;
	$os =~ s/\s*\Z//;
	
	if (($os =~ /win98/i) || ($os =~ /windows\s98/i)) {
	    $os = "Windows 98";
	}
	if (($os =~ /win95/i) || ($os =~ /windows\s95/i)) {
	    $os = "Windows 95";
	}
	elsif (($os =~ /macintosh/i)  || ($os =~ /macos/i)  || ($os =~ /\sMac_/)){
	    $os = "Macintosh";
	}
	elsif (($os =~ /windows\snt/i) || ($os =~ /winnt/i))  {
	    $os = "Windows NT";
	}
	elsif (($os =~ /windows\s*\;/i) ||  ($os =~ /windows\s3\./i)) {
	    $os = "Windows 3.1";
	}
	elsif (($os =~ /windows 16/i) || ($os =~ /win\s?16/i)) {
	    $os = "Windows 16";
	}
	elsif ($os =~ /webtv/i) {
	    $os =~ s/([^\(]*).*/$1/;
	}
	elsif ($os =~ /linux/i) {
	    $os = "Linux";
	}
	elsif ($os =~ /sunos/i) {
	    $os = "SunOS";
	}
	elsif ($os =~ /HP-UX/) {
	    $os = "HP-UX";
	}
	elsif ($os =~ /IRIX/) {
	    $os = "IRIX";
	}
	elsif ($os =~ /OSF1/) {
	    $os = "OSF1";
	}
	elsif ($os =~ /AIX/) {
	    $os = "AIX";
	}
	elsif (($os =~ /\A\s*default\s*\Z/i) || ($os =~ /\A\s*undefined\s*\Z/i) || ($os =~ /\A\s*unknown\s*\Z/i) || ($os =~ /\A\s*\Z/i)) {
	    $os = "Unknown";
	}
	$os =~ s/\A\s*compatible;?\s*(.*)/$1/i;

	if (defined $ostypes{$os}){              #os
	    $ostypes{$os}++; }
	else {
	    $ostypes{$os}=1; }


	if (defined $datetypes{$idate}){         #datetypes
	    $datetypes{$idate}++;
	    if($datetypes{$idate}>$maxdate){
		$maxdate = $datetypes{$idate};
	    }
	}
	else {
	    $datetypes{$idate}=1; }

	($ihh,$imm,$iss)= split (/:/,$itime);
	if (defined $timetypes{$ihh}){           #timetypes
	    $timetypes{$ihh}++;
	    if($timetypes{$ihh}>$maxtime){
		$maxtime = $timetypes{$ihh};
	    }
	}
	else {
	    $timetypes{$ihh} = 1; }
	$monthtotal++;
    }
    close(<LOG_FILE>);
    print OUTPAGE "<HTML><HEAD><TITLE>Monthly Access Stats</TITLE></HEAD></HTML>\n";
    print OUTPAGE "<BODY BGCOLOR=\"#ffffff\">\n";
    print OUTPAGE "<B>This month, $monthtotal users have hit your webpage. Giving a total of $total.</B>\n\n<p>";

# connect to history

    if ($show_hist == "yes") {
	print OUTPAGE "To see historical data, <a href=\"logs/history\">go here</a><P>";
}
#-----------------------------   SERVERTYPES    ----------------------------
 
    print OUTPAGE "<a name=\"domain\">\n";
    print OUTPAGE "<B><U>Domain Report:</U></B><p>\n\n";
    $unitsize = $maxsize/$maxdom;  
    $xsize = $maxdom/10;
    $xsize = int $xsize;
    if($xsize==0){
	$xsize =1;
    }
    $xunitsizes = int $xsize*$unitsize+1;
    print OUTPAGE "Each <IMG WIDTH=\"$xunitsizes\" HEIGHT=\"3\" SRC=\"bar1.gif\" BORDER=\"0\"> ";
    print OUTPAGE "Represents $xsize users.<p>\n\n";
    
    print OUTPAGE "<table cellspacing=$spacing>\n";

    print OUTPAGE "<tr align="left"><th>domain:</th><th>#logins:</th></tr>\n";
    print OUTPAGE "<tr><td>\n";

    foreach $server (sort { $servertypes{$b} <=> $servertypes{$a}; } keys %servertypes)
    { 
	for ($i=0; $i< $#ends; $i++)
	{
	    if ($server =~ /\A$ends[$i]\Z/i) {
		$label = "\L$ends[$i]\E ($ctrys[$i])"; }
	}
	if ($label eq "") {
	    print OUTPAGE "<nobr>$server</nobr><br>\n";
	}
	else {
	    print OUTPAGE "<nobr>$label</nobr><br>\n"; 
	}
	$label = "";
    }
    print OUTPAGE "</td><td>\n";
    foreach $server (sort { $servertypes{$b} <=> $servertypes{$a}; } keys %servertypes)
    { 
	$numb = int $servertypes{$server}*$unitsize+1;
	
	print OUTPAGE "$servertypes{$server} <IMG WIDTH=\"$numb\" HEIGHT=\"3\" SRC=\"bar1.gif\" BORDER=\"0\"><br>\n";
    }
    print OUTPAGE "</td></tr></table>\n";

#-----------------------------   DOCTYPES    ----------------------------
    print OUTPAGE "<HR><p>";

    print OUTPAGE "<a name=\"Document\">\n";
    print OUTPAGE "<B><U>Document Request Summary:</U></B><p>\n\n";
    $unitsize = $maxsize/$maxdoc;
    
    $xsize = $maxdate/10;
    $xsize = int $xsize;
    if($xsize==0){
	$xsize =1;
    }
    $xunitsizes = int $xsize*$unitsize+1;
    print OUTPAGE "Each <IMG WIDTH=\"$xunitsizes\" HEIGHT=\"3\" SRC=\"bar1.gif\" BORDER=\"0\"> ";
    print OUTPAGE "Represents $xsize users.<p>\n\n";
    
    print OUTPAGE "<table cellspacing=$spacing>\n";
    print OUTPAGE "<tr align="left"><th>document:</th><th>#hits:</th></tr>\n";
    print OUTPAGE "<tr><td>\n";
    foreach $doc (sort keys %doctypes){
	print OUTPAGE "<a href=\"$doc\">$doc</a><br>\n";
    }  
    print OUTPAGE "</td><td>\n";

    foreach $doc (sort keys %doctypes){
	$numb =  int $doctypes{$doc}*$unitsize+1;
	print OUTPAGE "$doctypes{$doc}  <IMG WIDTH=\"$numb\" HEIGHT=\"3\" SRC=\"bar1.gif\" BORDER=\"0\"><br>\n";
    }  
    print OUTPAGE "</td></tr></table>\n";

#-----------------------------   REFTYPES    ----------------------------
    print OUTPAGE "<HR><p>";

    print OUTPAGE "<a name=\"Referer\">\n";
    print OUTPAGE "<B><U>Referer Summary:</U></B><p>\n\n";

    print OUTPAGE "<table cellspacing=$spacing>\n";
    print OUTPAGE "<tr align="left"><th>referer:</th><th>#referred:</th></tr>\n";
    print OUTPAGE "<tr><td>\n";
    foreach $ref (sort keys %referertypes){
	print OUTPAGE "$referertypes{$ref}<br>\n";
    }
    print OUTPAGE "</td><td>\n";
    foreach $ref (sort keys %referertypes){
#	$numb =  int $referertypes{$ref}*$unitsize;
	if (($ref =~ /\Ainternal\Z/) || ($ref =~ /\Aunknown\Z/)){
	    print OUTPAGE "$ref<br>\n";
	} 
	else {
	    print OUTPAGE "<a href=\"$ref\">$ref</a><br>\n";
	}
    }
    print OUTPAGE "</td></tr></table>\n";


#-----------------------------   DATETYPES    ----------------------------

    print OUTPAGE "<HR><p>";

    print OUTPAGE "<a name=\"date\">\n";
    print OUTPAGE "<B><U>Daily Access Summary:</U></B><p>\n\n";
    $unitsize = $maxsize/$maxdate;
    
    $xsize = $maxdate/10;
    $xsize = int $xsize;
    if($xsize==0){
	$xsize =1;
    }
    $xunitsizes = int $xsize*$unitsize+1;
    print OUTPAGE "Each <IMG WIDTH=\"$xunitsizes\" HEIGHT=\"3\" SRC=\"bar1.gif\" BORDER=\"0\"> ";
    print OUTPAGE "Represents $xsize users.<p>\n\n";
    
    print OUTPAGE "<table cellspacing=$spacing>\n";
    print OUTPAGE "<tr align="left"><th>date:</th><th>#logins:</th></tr>\n";
    print OUTPAGE "<tr><td>\n";
    foreach $datet (sort keys %datetypes){
	$numb =  int $datetypes{$datet}*$unitsize;
	print OUTPAGE "$datet<br>\n";
    }  
    print OUTPAGE "</td><td>\n";

    foreach $datet (sort keys %datetypes){
	$numb =  int $datetypes{$datet}*$unitsize+1;
	print OUTPAGE "$datetypes{$datet} <IMG WIDTH=\"$numb\" HEIGHT=\"3\" SRC=\"bar1.gif\" BORDER=\"0\"><br>\n";
    }  
    print OUTPAGE "</td></tr></table>\n";

#-----------------------------   HOURTYPES    ----------------------------

    print OUTPAGE "<HR><p>";
    print OUTPAGE "<a name=\"hour\">\n";
    print OUTPAGE "<B><U>Hourly Summary:</U></B><p>\n\n";
    $unitsize = $maxsize/$maxtime;
    
    $xsize = $maxtime/10;
    $xsize = int $xsize;
    if($xsize==0){
	$xsize =1;
    }
    $xunitsizes = int $xsize*$unitsize+1;
    print OUTPAGE "Each <IMG WIDTH=\"$xunitsizes\" HEIGHT=\"3\" SRC=\"bar1.gif\" BORDER=\"0\"> ";
    print OUTPAGE "Represents $xsize users.<p>\n\n";
    
    print OUTPAGE "<table cellspacing=$spacing>\n";
    print OUTPAGE "<tr align="left"><th>hr:</th><th>#logins:</th></tr>\n";
    print OUTPAGE "<tr><td>\n";
    foreach $hourt (sort keys %timetypes){
	$numb = int $timetypes{$hourt}*$unitsize;
	print OUTPAGE "$hourt<br>\n";
    }
    print OUTPAGE "</td><td>\n";

    foreach $hourt (sort keys %timetypes){
	$numb = int $timetypes{$hourt}*$unitsize+1;
	print OUTPAGE "$timetypes{$hourt}  <IMG WIDTH=\"$numb\" HEIGHT=\"3\" SRC=\"bar1.gif\" BORDER=\"0\"><br>\n";
    }
    print OUTPAGE "</td></tr></table>\n";


#-----------------------------   BROSWERTYPES    ----------------------------

    print OUTPAGE "<HR><p>";
    print OUTPAGE "<a name=\"browser\">\n";
    print OUTPAGE "<B><U>Browser Version Report:</U></B><p>\n\n";
    print OUTPAGE "<table cellspacing=$spacing>\n";
    print OUTPAGE "<tr align="left"><th>#browsers:</th><th>browser version:</th></tr>\n";
    print OUTPAGE "<tr><td>\n";
    foreach $browsert (sort { $browsertypes{$b} <=> $browsertypes{$a} }keys %browsertypes){
	print OUTPAGE "$browsertypes{$browsert}<br>\n";
    }
    print OUTPAGE "</td><td>\n";

    foreach $browsert (sort { $browsertypes{$b} <=> $browsertypes{$a} }keys %browsertypes){
	print OUTPAGE "$browsert<br>\n";
    }
    print OUTPAGE "</td></tr></table>\n";


#-----------------------------   OSTYPES    ----------------------------

    print OUTPAGE "<HR><p>";
    print OUTPAGE "<a name=\"os\">\n";
    print OUTPAGE "<B><U>Operating System Report:</U></B><p>\n\n";
    print OUTPAGE "<table cellspacing=$spacing>\n";
    print OUTPAGE "<tr align="left"><th>#os's:</th><th>operating system:</th></tr>\n";
    print OUTPAGE "<tr><td>\n";
    foreach $osty (sort { $ostypes{$b} <=> $ostypes{$a} } keys %ostypes){
	print OUTPAGE "$ostypes{$osty}<br>\n";
    }
    print OUTPAGE "</td><td>\n";

    foreach $osty (sort { $ostypes{$b} <=> $ostypes{$a} } keys %ostypes){
	print OUTPAGE "$osty<br>\n";
    }

    print OUTPAGE "</td></tr></table>\n";
    print OUTPAGE "</BODY></HTML>\n";
}

sub view_current { 
    print "Content-type: text/html\n\n";

    open OUTPAGE, ">&STDOUT";
    $show_hist = "yes";
    &gen_page;
    close OUTPAGE;
}

sub roll_history {
    $month = &get_month;
#check existence of the directory and/or file
    $file = "$baseurl/logs/history/$month.html";
    open OUTPAGE, ">$file" || die "$0: can\'t open $file: $!\n";
    $show_hist = "no";
    &gen_page;
    close OUTPAGE;
    chmod (0644, "$file");
    truncate $logname,0;
    exit;
}


sub bad_name {
    print "Sorry, log $logname does not exist.\n";
    exit;
}