t2_access.pl

#!/usr/bin/perl
#
# Collecting data from access.log and
# presenting data as top 10 lists.
#
# The lists are; IP adresses, domains,
# files, month, hours and browsers.
#
# When domains does not exists IP is printed.
#
# Author Anton Lindström
# me@antonlindstrom.com

use strict;
use warnings;

use HTML::ParseBrowser;

# Logfile
my $accessfile = "access.log";

# Declare global variables.
my %ipadresses;
my %efiles;
my %pmonth;
my %phours;
my %browsers;
my %pdomains;

# Open Apache logfile
open(ACCESS, $accessfile);
my @access = <ACCESS>;
close(ACCESS);

print "Collecting data, hold on..\n\n";

foreach (@access) {
  # Parse logfile.
  $_ =~ m/(^[0-9\.]+)[\s\-]+\[(.+)\]\s\"[a-z]+\s(.+)\sHTTP\/[\d.]+\"[\s\d]+\"\-\"\s\"(.+)\"\s(\".*)?/gi;
  
  # List for better readability.
  my ($ip, $datetime, $file, $browser) = ($1, $2, $3, $4);

  # Split date for use in peaktimes.
  my ($day, $month, $next) = split(/\//, $datetime);
  my ($year, $hour, $minute, $second) = split(/:/, $next);

  # Parse Browser.
  my $ua = HTML::ParseBrowser->new($ENV{HTTP_USER_AGENT});
  $ua->Parse($browser);
  my $browsername = $ua->name;

  # For every occurance of IP add +1.
  if (exists $ipadresses{$ip}) {$ipadresses{$ip} += 1;} 
  else {$ipadresses{$ip} = 1;}
  # For every occurance of file add +1.
  if (exists $efiles{$file}) {$efiles{$file} += 1;} 
  else {$efiles{$file} = 1;}
  # Peak months.
  if (exists $pmonth{$month}) {$pmonth{$month} += 1;} 
  else {$pmonth{$month} = 1;}
  # Peak hours.
  if (exists $phours{"$hour:00"}) {$phours{"$hour:00"} += 1;} 
  else {$phours{"$hour:00"} = 1;}
  # Peak hours.
  if (exists $browsers{$browsername}) {$browsers{$browsername} += 1;} 
  else {$browsers{$browsername} = 1;}
  
  # Domain lookup.
  if($ip) {
    my $dom = `host $ip`;
    $dom =~ m/pointer\s(.+)/g;
    $dom = $1;

    # For every occurance of domains add +1.
    if (exists $pdomains{$dom}) {$pdomains{$dom} += 1;} 
    else {$pdomains{$dom} = 1;}
  }
}

# Print most frequent domains.
print "Most frequent visitors (domain):\n";
my $i=0;
# Print visits and Domains, sorted.
foreach my $key (sort {$pdomains{$b} <=> $pdomains{$a} } keys %pdomains) {
  next if ($i == 10);
  print "\t$pdomains{$key}\t $key\n";
  $i++;
}
# Print most frequent IP addresses.
print "Most frequent visitors (IP):\n";
$i=0;
# Print visits and IP addresses, sorted.
foreach my $key (sort {$ipadresses{$b} <=> $ipadresses{$a} } keys %ipadresses) {
  next if ($i == 10);
  print "\t$ipadresses{$key}\t $key\n";
  $i++;
}
# Print most frequent files accessed.
print "Most popular files:\n";
$i=0;
# Print amount of files and filenames, sorted.
foreach my $key (sort {$efiles{$b} <=> $efiles{$a} } keys %efiles) {
  next if ($i == 10);
  print "\t$efiles{$key}\t $key\n";
  $i++;
}
# Print peak month.
print "Most popular month to access site:\n";
$i=0;
# Print amount of visits and month, sorted.
foreach my $key (sort {$pmonth{$b} <=> $pmonth{$a} } keys %pmonth) {
  next if ($i == 10);
  print "\t$pmonth{$key}\t $key\n";
  $i++;
}
# Print peak hours.
print "Most popular hour to access site:\n";
$i=0;
# Print amount of visits and hours, sorted.
foreach my $key (sort {$phours{$b} <=> $phours{$a} } keys %phours) {
  next if ($i == 10);
  print "\t$phours{$key}\t $key\n";
  $i++;
}
# Print most used browser.
print "Top 10 Browsers:\n";
$i=0;
# Print usage of browsers, sorted.
foreach my $key (sort {$browsers{$b} <=> $browsers{$a} } keys %browsers) {
  next if ($i == 10);
  print "\t$browsers{$key}\t $key\n";
  $i++;
}