#!/usr/local/bin/perl

#	wnhits - summarize hits on wn server.

#	Assumes common log format.

#	21 Sep 95 Paul DuBois, dubois@primate.wisc.edu

# get script name for messages

($prog = $0) =~ s|.*/||;

# BEGIN SITE-CONFIGURABLE STUFF (change as necessary)

# default log file if no arguments given
$defaultlog = "/usr/local/log/wnlog";
# requests ending in / have this added to them
$defaultpage = "index.html";

# END SITE-CONFIGURABLE STUFF

push (@ARGV, $defaultlog) if @ARGV == 0;

$summarycmd = "sort | uniq -c | sort -rn";

open (FH, "|$summarycmd") || die "$prog: cannot open summary command\n";

while (<>)
{
	($host, $rfc931, $authuser, $timestamp, $request, $status, $bytes) =
	/^(\S+) +(\S+) +(\S+) +\[(.+)\] +\"(.+)\" +(\S+) +(\S+)/;
	# no longer exclude bytes == - lines; I put that in to exclude aborted
	# connections, but it also excludes CGI requests that don't send a
	# byte count in the output stream
	#next unless $status == 200 && $bytes ne "-";
	next unless $status == 200;
	($method, $path, $httpvers) = split (' ', $request, 3);
	# convert some common hex chars to ascii equivalents
	$path =~ s/%3B/;/g;
	#$path =~ s/%3F/?/g;
	# fix paths that have // or /./ in them
	$path =~ s|//+|/|g;
	1 while s|/\./|/|;
	$path .= $defaultpage if $path =~ /\/$/;
	# strip trailing query information
	$path =~ s/;.*//;
	$path =~ s/\?.*//;
	print FH "$path\n";
}

close (FH);

exit (0);