#!/usr/bin/perl

use URI::Escape;

open(LOGS, $ARGV[0] || "-") or die $!;

my %urls;
my %counts;
while (<LOGS>) {
  if (m/([0-9.]+) .*?"(?:GET|POST|HEAD)([^"]+)".*?"([^"]+)"/) { 
    ($ip,$a,$b) = ($1,$2,$3); 
    $a =~ s,HTTP/1.[01]$,,;
    if ($b ne "-" && $b !~ m/http:\/\/(www\.)?semicomplete.com/i) { 
    #if(1) {
      $b = "Google: $1" if ($b =~ m/google.*?q=([^&]+)/);
      $b = "Yahoo: $1" if ($b =~ m/search\.yahoo.*?p=([^&]+)/);
      $b = "MSN: $1" if ($b =~ m/search\.msn.*?q=([^&]+)/);
      $b = uri_unescape($b);
      $b =~ s/\+/ /g;
      $a =~ s/^\s+//;
      $a =~ s,^/blog,,;
      #next if ($b !~ /^http/);
      next if ($b =~ /(bloglines|livejournal|google.com\/reader)/);
      push(@{$urls{$a}}, $b);
      $counts{$a}{$b}++;
      $counts{$a}{":root:"}++;
    }
  }
}

@urls = sort { $counts{$b}{":root:"} <=> $counts{$a}{":root:"} || $a cmp $b } keys(%urls);
for my $x (@urls) {
  print "\n";
  print "$x [".$counts{$x}{":root:"}."]\n";
  my %uniq;
  map( $uniq{$_}++, @{$urls{$x}});
  my @k = sort { $counts{$x}{$b} <=> $counts{$x}{$a} || $a cmp $b } keys(%uniq);
  for my $y (@k) {
    print "      [".$counts{$x}{$y}."] $y\n";
  }
}