#!/usr/bin/perl

#perl script to format httpd log into delimited database load file
#

# first parameter is the name of the log file

$flname=shift ;

# copyright 2000 Martin P. Lurie and Informix Software
# NOT supported. Best of luck.
# 

# use associative array to translate the date into
# database format
$monthtbl {"Jan"}="01";
$monthtbl {"Feb"}="02";
$monthtbl {"Mar"}="03";
$monthtbl {"Apr"}="04";
$monthtbl {"May"}="05";
$monthtbl {"Jun"}="06";
$monthtbl {"Jul"}="07";
$monthtbl {"Aug"}="08";
$monthtbl {"Sep"}="09";
$monthtbl {"Oct"}="10";
$monthtbl {"Nov"}="11";
$monthtbl {"Dec"}="12";

open (WEBLOG, $flname ) || die "can not open weblog  file: $!\n";

while (<WEBLOG>) {
chop;
##Don't you just love these!
$_ =~ /(^.*) (.) (.) (\[)(.*2000\:)(.{8}).*\"GET \/([a-z]*[0-9]*)(.*) HTTP.*\" ([0-9]*) ([0-9]*).*/;

$CDATE = $5;
chop ($CDATE );
#print "web2datetime ",  web2datetime ( $CDATE ) ;
$URL="/$7$8";

print "$1|$2|$3|".web2datetime ($CDATE)." $6|$7|/$7$8|$9|$10|".filetype($URL)."|".linkdepth($URL)."|\n";
# debugging tree depth print $URL."|".linkdepth($URL)."\n";

# set up the delimiter string to use shorthand on output printing
$Q="\",\"";
#print "\"",$lname,$Q,$fname,$Q,$phone,$Q,$extn,$Q,$uid,$Q,$loc,$Q,$dept,"\"\n";
#
}

sub web2datetime {
           my($dstr) = @_;

	$CDATE =~ /(.{2})\/(.{3})\/(.{4})/;
	return "$3-$monthtbl{$2}-$1";
	}


sub linkdepth {
           my($inputurl) = @_ ;
           my($lndepth) ;
           my(@spliturl) ;
	# use translate  tr function to find all /
	$_ =  $inputurl ;
	#print "splitting $inputurl \n";
	# @spliturl  = split ( /\//, $inputurl );
	$lndepth = tr /\//\T/;
	#print "in sub with  $_ \n" ;

	return ( $lndepth ) ;
	}

sub filetype {
           my($inputurl) = @_ ;
	
	#$_ = $inputurl;
	#print "$_ \n";

	$result = $inputurl =~ /(.*)\.(.*)/;
	return ( $2 ) ;
	}