#!/usr/bin/perl
# NAME:      logparse
# AUTHOR:    Jeffrey Fulmer
# EMAIL:     jdfulmer@armstrong.com
# DATE:      Thu Mar 14 15:11:05 EST 2002
# This program builds URLs from apache style
# access logs and populates a urls.txt file.
#
use Getopt::Std;
use vars qw($opt_V $opt_h $opt_v $opt_H $opt_f $VERSION);
use strict;

my  $LOCK_SH = 1;                # shared lock (not used)
my  $LOCK_EX = 2;                # exclusive lock
my  $LOCK_NB = 4;                # non-blocking lock (not used)
my  $LOCK_UN = 8;                # unlock
my ($host, $file);               # host to set, file to read
my ($verbose);                   # boolean, verbose mode

getopts( 'VHvh:f:' );            # get cmd line options

$VERSION = sprintf(              # specify version number
  "%1.2f", 1.00                  #
);  
sub Version { $VERSION; }        #

if( $opt_V == 1 ){               # user wants version number
  printf(                        #
    "logparse %s\n", &Version    # print it and exit
  );                             #
  exit 0;                        #
}                                #

if( $opt_H == 1 ){               # user wants assistance
  display_help();                # display help section
  exit( 0 );                     # and exit nicely
}                                #

if( $opt_v == 1 ){               # user wants verbose mode
  $verbose = 1;                  # set boolean true
}                                #

if( $opt_h ne "" ){               # user wants to change date
  $host = $opt_h;                # set the local date value
}                                #

if( $opt_f ne "" ){              # user wants to change month
  $file = $opt_f;                # set the local month value
}                                #

if(( $host eq "" )||( $file eq "" )){
  display_help();
  exit( 0 );
}
  
parse_it( $host, $file );

exit( 0 );

sub
parse_it()
{
  my  $lines = "";
  my( $host, $file ) = @_;
  my( @LINES );
  my $out = "urls.txt";
  
  $host =~ s/(\/)$//; 

  if( open( FILE, "<" . $file )){
    flock( FILE, $LOCK_EX );
    @LINES = <FILE>;
    flock( FILE, $LOCK_UN );
    close( FILE );
    if( open ( FILE, ">" . $out )){
      flock( FILE, $LOCK_EX );
      for( my $x = 0; $x <= $#LINES; $x++ ){
        $LINES[$x] =~ s/^.*GET //;
        $LINES[$x] =~ s/^.*HEAD //;
        $LINES[$x] =~ s/HTTP.*$//;
        if( $host =~ m/^http/ ){
          $LINES[$x] =~ s/^/$host/;
        }
        else{
          $LINES[$x] =~ s/^/http:\/\/$host/;
        }
        if( $verbose ){ print $LINES[$x]; }
        print FILE $LINES[$x];
      }
      flock( FILE, $LOCK_UN );
      close( FILE );
    }
    else{ die "couldn't open: $file"; }
  }
  else{ die "couldn't open: $file"; }
}

sub
display_help
{
  print <<END_OF_HELP;
logparse v$VERSION
Usage: logparse [options]
       logparse -h http://www.whoohoo.com -f access_log
Options:
  -V           VERSION, print version number to screen.
  -H           HELP, display this section.
  -v           VERBOSE, print progress info to screen.
  -h hostname  HOST, set the URL's hostname.
  -f filename  FILE, the access_log file to parse for URLs.
END_OF_HELP
  return;
}

1;