#!/usr/bin/perl
#
# Name : Matthew Reeves
#
# This program will analyze a web log using perl
#
#use diagnostics;
use Modern::Perl;
use Net::hostent;

#checks if file entered at cmd line, exits w/ usage stmnt if not
if ($ARGV[0] ~~ undef)
{
	print "\nUsage: $0 <filename/s>\n\n";
	exit;
}

#declare/initialize variables
my $filecount = 0;
my $total = 0;
my $count = 0;
my $totalbytes = 0;

my %hostnames = ();
my %domains = ();
my %dates = ();
my %hours = ();
my %status = ();
my %url = ();
my %filetype = ();
my %browsers = ();
my %browserfamily = ();
my %referer = ();
my %referdomain = ();
my %operatingsystem = ();

#opens output file for write access or die
open (OUTPUT, ">", "mdreeves.report")
	or die "\nUnable to open file: mdreeves.report: $!\n";

#while that executes functionality of program once per file entered
while ($ARGV[$filecount])
{
	#opens file entered as cmd line arguement, or die
	my $filename = $ARGV[$filecount];
	open (FILE, $filename) or die "\nUnable to open file: $filename: $!\n";

	#while loop to read in data from current file, and extract desired data
	while (<FILE>)
	{
		$_ =~ /(.+?)\s.+?\[(.+?):(\d\d):.+?\s.+?"(.+?)\s(.+?)\s(.+?)"\s(.+?)\s(.+?)\s"(.+?)"\s"(.+?)"/;
	
		my $tmp = $1;

		my $name = $1;

		#runs reverse lookup
		if (my $h = gethost($tmp))
		{
			$name = $h->name();
			$tmp = reverse($name);

			$tmp =~ /(.+?\..+?)\..+/;
			$tmp = $1;
			$tmp = reverse($tmp);
		} 	
		else
		{
			$tmp = "OTHER";
		}

		#stores value as key, occurences as value. 1 hash per report
		$hostnames{$name}++;
		$domains{$tmp}++;
		$dates{$2}++;
		$hours{$3}++;
		$status{$7}++;
		$url{$5}++;

		$tmp = $5;
		
		#given statement to identify file types
		given ($tmp)
		{
			when (/.+?\.cgi/i) { $filetype{"CGI Program"}++; }
			when (/.+?\.css/i) { $filetype{"Style Sheet"}++; }
			when (/.+?\.htm/i) { $filetype{"Web Pages"}++; }
			when (/.+?\.gif/i) { $filetype{"Image"}++; }
			when (/.+?\.jpg/i) { $filetype{"Image"}++; }
			when (/.+?\.jpeg/i) { $filetype{"Image"}++; }
			when (/.+?\.png/i) { $filetype{"Image"}++; }
			when (/.+?\.ico/i) { $filetype{"Image"}++; }
			default { $filetype{"Other Request"}++; }
		}

		$tmp = $10;

		#checks if browser ID was present
		$tmp = "NO BROWSER ID" if ($10 ~~ "-");
		$browsers{$tmp}++;

		#given statement to id browser families
		given ($tmp)
		{
			when (/Chrome/) { $browserfamily{"Chrome"}++; }
			when (/Firefox/) { $browserfamily{"Firefox"}++; }
			when (/MSIE/) { $browserfamily{"MSIE"}++; }
			when (/Opera/) { $browserfamily{"Opera"}++; }
			when (/Safari/) { $browserfamily{"Safari"}++; }
			default { $browserfamily{"Unknown"}++; }
		}

		#given statement to id OS's
		given ($tmp)
		{
			when (/Windows/) { $operatingsystem{"Windows"}++; } 
			when (/Macintosh/) { $operatingsystem{"Macintosh"}++; } 
			default { $operatingsystem{"Other"}++; } 
		}

		my $bytes = $8;

		$tmp = $9;

		#given statement to ID presence of referer
		given ($tmp)
		{
			when (/http/) { $referer{$tmp}++; }
			default { $referer{"NO REFERER"}++; }
		}

		#if/else block to ID presence of referer domain
		if ($tmp =~ /http.+?\.(\w+\.\w+)\//)
		{
			$referdomain{$1}++;
		}
		else
		{
			$referdomain{"NONE"}++;
		}
			
		#track total number of bytes from each log file
		$totalbytes += $bytes if ($bytes =~ /\d+/);

		#increments count to track entries per file
		$count++;
	}

	#increments filecount to track number of files entered	
	$filecount++;

	#closes input file
	close FILE;
}

################################################################################

#prints formatted summary of analyzer results for all files
print OUTPUT "########################################################\n";
print OUTPUT "Web Server Log Analyzer - SUMMARY\n";
print OUTPUT "########################################################\n";

print OUTPUT "\n";
print OUTPUT "Processed $count entries from the following $filecount files: ";

my $x = 0;

while ($x < $filecount)
{
	print OUTPUT "\n\t- $ARGV[$x]";
	$x++;
}


print OUTPUT "\n";

################################################################################

	#prints formatted results for hostnames
	print OUTPUT "\n";
	print OUTPUT "========================================================\n";
	print OUTPUT "HOSTNAMES\n";
	print OUTPUT "========================================================\n";

	print OUTPUT "\n";
	print OUTPUT "  Hits   %-age   Resource\n";
	print OUTPUT "  ----   -----   --------\n";	

	my @hash = sort {$a cmp $b} keys %hostnames;

	foreach my $x (@hash)
	{
		printf OUTPUT ("%6d   %5.2f   %s\n",
			$hostnames{$x}, ($hostnames{$x}/$count)*100, $x);
	}
	print OUTPUT "  ----\n";  
	printf OUTPUT ("%6d entries displayed\n", $count);

################################################################################

	#prints formatted results for domains
	print OUTPUT "\n";
	print OUTPUT "========================================================\n";
	print OUTPUT "DOMAINS\n";
	print OUTPUT "========================================================\n";

	print OUTPUT "\n";
	print OUTPUT "  Hits   %-age   Resource\n";
	print OUTPUT "  ----   -----   --------\n";	

	@hash = sort {$a cmp $b} keys %domains;

	foreach my $x (@hash)
	{
		printf OUTPUT ("%6d   %5.2f   %s\n",
			$domains{$x}, ($domains{$x}/$count)*100, $x);
	}
	print OUTPUT "  ----\n";  
	printf OUTPUT ("%6d entries displayed\n", $count);

################################################################################

	#prints formatted results for dates
	print OUTPUT "\n";
	print OUTPUT "========================================================\n";
	print OUTPUT "DATES\n";
	print OUTPUT "========================================================\n";

	print OUTPUT "\n";
	print OUTPUT "  Hits   %-age   Resource\n";
	print OUTPUT "  ----   -----   --------\n";	

	@hash = sort {$a cmp $b} keys %dates;

	foreach my $x (@hash)
	{
		printf OUTPUT ("%6d   %5.2f   %s\n",
			$dates{$x}, ($dates{$x}/$count)*100, $x);
	}
	print OUTPUT "  ----\n";  
	printf OUTPUT ("%6d entries displayed\n", $count);

################################################################################

	#prints formatted results for hours
	print OUTPUT "\n";
	print OUTPUT "========================================================\n";
	print OUTPUT "HOURS\n";
	print OUTPUT "========================================================\n";

	print OUTPUT "\n";
	print OUTPUT "  Hits   %-age   Resource\n";
	print OUTPUT "  ----   -----   --------\n";	

	@hash = sort {$a <=> $b} keys %hours;

	foreach my $x (@hash)
	{
		printf OUTPUT ("%6d   %5.2f   %s\n",
			$hours{$x}, ($hours{$x}/$count)*100, $x);
	}
	print OUTPUT "  ----\n";  
	printf OUTPUT ("%6d entries displayed\n", $count);

################################################################################

	#prints formatted results for status codes
	print OUTPUT "\n";
	print OUTPUT "========================================================\n";
	print OUTPUT "STATUS CODES\n";
	print OUTPUT "========================================================\n";

	print OUTPUT "\n";
	print OUTPUT "  Hits   %-age   Resource\n";
	print OUTPUT "  ----   -----   --------\n";	

	@hash = sort {$a <=> $b} keys %status;

	foreach my $x (@hash)
	{
		printf OUTPUT ("%6d   %5.2f   %s\n",
			$status{$x}, ($status{$x}/$count)*100, $x);
	}
	print OUTPUT "  ----\n";  
	printf OUTPUT ("%6d entries displayed\n", $count);

################################################################################

	#prints formatted results for urls
	print OUTPUT "\n";
	print OUTPUT "========================================================\n";
	print OUTPUT "URLS\n";
	print OUTPUT "========================================================\n";

	print OUTPUT "\n";
	print OUTPUT "  Hits   %-age   Resource\n";
	print OUTPUT "  ----   -----   --------\n";	

	@hash = sort {$a cmp $b} keys %url;

	foreach my $x (@hash)
	{
		printf OUTPUT ("%6d   %5.2f   %s\n",
			$url{$x}, ($url{$x}/$count)*100, $x);
	}
	print OUTPUT "  ----\n";  
	printf OUTPUT ("%6d entries displayed\n", $count);

################################################################################

	#prints formatted results for filestypes
	print OUTPUT "\n";
	print OUTPUT "========================================================\n";
	print OUTPUT "FILETYPES\n";
	print OUTPUT "========================================================\n";

	print OUTPUT "\n";
	print OUTPUT "  Hits   %-age   Resource\n";
	print OUTPUT "  ----   -----   --------\n";	

	while ( my ($value,$occur) = each %filetype)
	{
		printf OUTPUT ("%6d   %5.2f   %s\n",
			$occur, ($occur/$count)*100, $value);
	}
	print OUTPUT "  ----\n";  
	printf OUTPUT ("%6d entries displayed\n", $count);

################################################################################

	#prints formatted results for browsers
	print OUTPUT "\n";
	print OUTPUT "========================================================\n";
	print OUTPUT "BROWSERS\n";
	print OUTPUT "========================================================\n";

	print OUTPUT "\n";
	print OUTPUT "  Hits   %-age   Resource\n";
	print OUTPUT "  ----   -----   --------\n";	

	@hash = sort {$a cmp $b} keys %browsers;

	foreach my $x (@hash)
	{
		printf OUTPUT ("%6d   %5.2f   %s\n",
			$browsers{$x}, ($browsers{$x}/$count)*100, $x);
	}
	print OUTPUT "  ----\n";  
	printf OUTPUT ("%6d entries displayed\n", $count);

################################################################################
	
	#prints formatted results for browser families
	print OUTPUT "\n";
	print OUTPUT "========================================================\n";
	print OUTPUT "BROWSER FAMILIES\n";
	print OUTPUT "========================================================\n";

	print OUTPUT "\n";
	print OUTPUT "  Hits   %-age   Resource\n";
	print OUTPUT "  ----   -----   --------\n";	

	@hash = sort {$a cmp $b} keys %browserfamily;

	foreach my $x (@hash)
	{
		printf OUTPUT ("%6d   %5.2f   %s\n",
			$browserfamily{$x}, ($browserfamily{$x}/$count)*100, $x);
	}
	print OUTPUT "  ----\n";  
	printf OUTPUT ("%6d entries displayed\n", $count);

################################################################################

	#prints formatted results for referer
	print OUTPUT "\n";
	print OUTPUT "========================================================\n";
	print OUTPUT "REFERER\n";
	print OUTPUT "========================================================\n";

	print OUTPUT "\n";
	print OUTPUT "  Hits   %-age   Resource\n";
	print OUTPUT "  ----   -----   --------\n";	


	@hash = sort {$a cmp $b} keys %referer;

	foreach my $x (@hash)
	{
		printf OUTPUT ("%6d   %5.2f   %s\n",
			$referer{$x}, ($referer{$x}/$count)*100, $x);
	}
	print OUTPUT "  ----\n";  
	printf OUTPUT ("%6d entries displayed\n", $count);

################################################################################

	#prints formatted results for referer domain
	print OUTPUT "\n";
	print OUTPUT "========================================================\n";
	print OUTPUT "REFERER DOMAIN\n";
	print OUTPUT "========================================================\n";

	print OUTPUT "\n";
	print OUTPUT "  Hits   %-age   Resource\n";
	print OUTPUT "  ----   -----   --------\n";	

	@hash = sort {$a cmp $b} keys %referdomain;

	foreach my $x (@hash)
	{
		printf OUTPUT ("%6d   %5.2f   %s\n",
			$referdomain{$x}, ($referdomain{$x}/$count)*100, $x);
	}
	print OUTPUT "  ----\n";  
	printf OUTPUT ("%6d entries displayed\n", $count);

################################################################################
	
	#prints formatted results for operating systems
	print OUTPUT "\n";
	print OUTPUT "========================================================\n";
	print OUTPUT "OPERATING SYSTEMS\n";
	print OUTPUT "========================================================\n";

	print OUTPUT "\n";
	print OUTPUT "  Hits   %-age   Resource\n";
	print OUTPUT "  ----   -----   --------\n";	

	while ( my ($value,$occur) = each %operatingsystem)
	{
		printf OUTPUT ("%6d   %5.2f   %s\n",
			$occur, ($occur/$count)*100, $value);
	}
	print OUTPUT "  ----\n";  
	printf OUTPUT ("%6d entries displayed\n", $count);

################################################################################

	#prints formatted results for total bytes
print OUTPUT "\n########################################################\n";
print OUTPUT "Total bytes served: $totalbytes bytes\n";
print OUTPUT "########################################################\n\n";

################################################################################



#closes output file
close OUTPUT;

#prints to console status message indicating successful completion
print "\nSuccessfully Processed $filecount files.";
print "\nAnalyzed results stored: mdreeves.report\n\n";