#! /opt/bin/perl
#	________________________________________________________________
#	  /\ /\		The Web Developer's Virtual Library
#	-{-<*>-}-	World-Wide Web	   <URL:http://WDVL.com/>
#	__\/_\/_________________________________________________________
#	Author	:	Lucy Richmond
#	Purpose	:	To read the access log and create an output file
#			of site web pages and their number of successfull page views.
#			The output file is used to create the Top 100.
#	Usage	:	log.pl fn
#				where fn is the name of the output file, fn.log
#	Comment	:	Because this is run against a daily log, only URLs with more than
#			10 page views are included in the output file (specified in the
#			print command near the end of the program.
#	Disclaimer:	This software is provided freely on the understanding
#			that the Author will not be held responsible for any
#		problems arising from it's use, and that there is no support.
#	________________________________________________________________

	$tothits = 0;

	$log	= $ARGV[0];

	open	(IN, "access_log")||die$!;
	open	(OUT, ">$log.log")||die$!;

	while   (<IN>)    {
		/"\S+ (\S+)/;
#		/access to (\S+)/;		# Error log
		next	if / 404 /;
		$_	= $1;

#		If the file name ends with a slash, add 'index.html' to the end
		s/^$/\/index.html/;
		s/\/$/\/index.html/;

#		There are many old links to Vlib/Providers/Images_and_Icons which go to Multimedia now
		s/\/Vlib\/Providers\/Images/\/Vlib\/Multimedia\/Images/;
		next	if /Images_and_Icons.html\/index.html/;

		next	unless /\.html$/ or /\.wrl$/;
		next	if /\/menubar.\.html/;
		$tothits++;

		s/^\/\//\//;	


		$file{$_}++;
		}

	foreach (keys %file)    {

		$count	= $file{$_};
		print OUT	"$count	$_\n"	if $count > 10;
		}

	print "total hits 	$tothits\n";