#!/usr/bin/perl
# Figure 12.11: fig12_11.pl
# Analyzes the Apache access log file

use warnings;
use strict;
use English;

# System settings changed to make output appear correctly
$FORMAT_LINES_PER_PAGE = 25;
$FORMAT_FORMFEED = '';
my $access='C:\Program Files\Apache Group\Apache\logs\access.log';
my $interval = 3;

( $~, $^, $-, $% ) = ( 'PRINTACCESS', 'PRINTACCESS_TOP', 0, 0 );
my ( @entry, %most, %hourly );
   
# loop that prints each entry and gathers the statistics
open( FILE, $access ) or die( "Cannot open $access" );

while ( <FILE> ) {
   next if ( $_ eq "\n" );           # skip blank lines
   @entry = parseAccessEntry( $_ );
   write();

   prompt() if ( $FORMAT_LINES_LEFT == 1 );
   $most{ $entry[ 3 ] }++;
   
   my $hour = ( split( /:/, $entry[ 1 ], 3 ) )[ 1 ];
   $hourly{ $hour.'requests' }++;
   $hourly{ $hour.'transferred' } += $entry[ 6 ] 
                  unless ( $entry[ 6 ] eq '-' );
}

close( FILE ) or die( "Cannot close: $!" );
print( "\n" x ( $- - 1 ) );
prompt();
   
# printing most frequently accessed statistics
( $~, $^, $-, $% ) = ( 'MOST', 'MOST_TOP', 0, 0 );
my $key;

foreach $key ( sort { $most{ $b } <=> $most{ $a } } 
             ( keys( %most ) ) ) {
   write();
   prompt() if ( $FORMAT_LINES_LEFT == 1 );
}

print( "\n" x ( $- - 1 ) );
prompt();

# printing hourly access rate and bytes transferred.
( $~, $^, $-, $% ) = ( 'TIME', 'TIME_TOP', 0, 0 );
my ( $requests, $transferred, $startTime );

for ( $startTime = 0; $startTime < 24; 
      $startTime += $interval ) {
   ( $requests, $transferred ) = ( 0, 0 );

   for ( my $hour = $startTime; 
         $hour < $startTime + $interval && $hour < 24; 
         $hour++ ) {
      $requests += $hourly{ $hour.'requests' } 
                if exists( $hourly{ $hour.'requests' } );
      $transferred += $hourly{ $hour.'transferred' } 
                if exists( $hourly{ $hour.'transferred' } );
   }

   write();
   prompt() if ( $- == 1 );
}
print( "\n" x ( $- - 2 ) );

sub parseAccessEntry 
{
   my $entry = shift;
  
   if ( $entry =~ /^( \d+\.\d+\.\d+.\d+ )\s-\s-\s\[ #IP Address
                    ( [\d\w:\/]+ )\s-\d{4}\]\s\"  #time
                    ( \w* )\s                     #Request type
                    ( [^\s]* )\s                  #File name
                    ( [^\"]* )\"\s                #Protocol
                    ( \d+ )\s                     #Request result
                    ( \d+|\- )\s*\n$/x ) {        #Total bytes
      return ( $1, $2, $3, $4, $5, $6, $7 );
   }

   warn "Not a normal access log entry.\n$entry\n";
   return undef;
}

sub prompt 
{
   print( "--- Press Enter to Continue ---" );
   <STDIN>;
   $FORMAT_LINES_LEFT--;
}

# format definitions   
format PRINTACCESS_TOP =
access.log                                                Page @<<
                                                               $% 
Requester    Date and time        Type File              Size Code
------------ -------------------- ---- ----              ---- ----
.

format PRINTACCESS =
@<<<<<<<<<<< @<<<<<<<<<<<<<<<<<<< @<<< @<<<<<<<<<<<<<... @<<<<<@<<
$entry[0], $entry[1],   $entry[2], $entry[3], $entry[6], $entry[5]
.
   
format MOST_TOP =
Most Accessed Files              Page @<<
                                       $%
File                             Requests
--------------------             --------
.

format MOST =
@<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< @|||||||
$key,                            $most{ $key }
.

format TIME_TOP =
                                               Page @<<
                                                    $%
Times                   Requests        Total bytes Transferred
-----------             --------        ----------------------
.

format TIME =
@#:00 - @#:00           @|||||||        @|||||||||||||||||||||
{ hour( $startTime ), hour( $startTime + $interval ), 
                       $requests,       $transferred }
.
  
sub hour
{
   return ( $_[ 0 ] % 12 == 0 )? 12 : ( $_[ 0 ] % 12 );
}


###########################################################################
#  (C) Copyright 2001 by Deitel & Associates, Inc. and Prentice Hall.     #
#  All Rights Reserved.                                                   #
#                                                                         #
#  DISCLAIMER: The authors and publisher of this book have used their     #
#  best efforts in preparing the book. These efforts include the          #
#  development, research, and testing of the theories and programs        #
#  to determine their effectiveness. The authors and publisher make       #
#  no warranty of any kind, expressed or implied, with regard to these    #
#  programs or to the documentation contained in these books. The authors #
#  and publisher shall not be liable in any event for incidental or       #
#  consequential damages in connection with, or arising out of, the       #
#  furnishing, performance, or use of these programs.                     #
###########################################################################