#!/usr/bin/perl

# findnewfeed
# Run the news spool reading the paths of articles.  Track the sites by
# number of hops and by which of our feeds the article arrived on.
# Find the sites which are, on average, farthest from us and would therefore
# do us (and them) the most good to hook up to.
# This program runs for a LONG time.

print "Findnewfeed:\n";

$root = shift @ARGV;
chop($root = `pwd`) unless defined $root;
print "\troot is $root\n";

&rundirectory($root);	# recursively run the directory

&dostats;		# find the answer

sub rundirectory{
    local($root) = @_;

    push(@dirlist,$root);
    while($dir = pop(@dirlist)){
	chdir($dir) || do {
	    print "CHDIR failed: $!\n";
	    next;
	};

	opendir(DIR, ".") || do {
	    print "OPENDIR failed: $!\n";
	    next;
	};
	foreach $f (readdir(DIR)){
	    ($f =~m/^\./) && next;	# skip . ..
	    ($links,$inode)=(lstat($f))[3,1];
	    if(-l _){
		next;		# punt on symlinks so we don't need to
				# track devno as well as inodes
	    };

	    if(-d _){
		push(@dirlist,"$dir/$f");
		next;
	    }

	    if($links > 1){
		if($links{$inode}){
		    next;	# already done this one
		} else {
		    $link{$inode} = 1;	# don't do it again later
		}
	    }

	    &doart($f);		# do it, relative to current directory
	}
	closedir DIR;
    }
}

sub doart{
    local($file) = @_;
    open(ART, $file) || do{
	print "Can't open $file: $!\n";
	return;
    };

    while(<ART>){
	$path = $_;
	next unless m/^Path: (.*)/i;
	$section = $1;
	last;
    }
    close ART;

    $dist = 1;
    @path = split(/!/, $section);
    if((shift @path) ne 'navair2.nalda.navy.mil'){
	print "INTERNAL ERROR: lost self: $f\n";
	return;
    }
    $feed = shift @path;
    $feeds{$feed} = 1;			# list of feeds
    grep(
	do{++$sites{$_},		# hit count for this site and list of sites
	++$sitec{$feed,$_},	    # hit count for this site from this feed
	$sitew{$feed,$_}+= $dist,   # total length to this site from this feed
	++$dist,0}
    ,@path);

}

sub dostats{
	@sitelist = keys %sites;
		# pre-calculate the messy bits
	foreach $x (@sitelist){
	    $fcount = 0;	# number of feeds that got to this site
	    $totav = 0;		# total average
	    foreach $f (keys %feeds){
		$bot = $sitec{$f,$x};
		if($bot){
		    $top = $sitew{$f,$x};
		    $fcount++;
		    $totav += ($top/$bot);
		}
	    }
	    if($fcount == 0){
		print "INTERNAL ERROR: /0: $x\n";
		next;
	    }
	    $sitecalc{$x} = $totav/$fcount;
	}
	@res = sort MESSYSORT @sitelist;

	print "Results:\t";
	foreach $f (keys %feeds){
		print "$f\t";
	}
	print "\n";
	foreach $x (@res){
	    printf "%6.2f %30s ",$sitecalc{$x},$x;
	    foreach $f (keys %feeds){
		if($sitec{$f,$x}){
			printf "%6.2f\t", $sitew{$f,$x}/$sitec{$f,$x};
		} else {
			print "\t\t";
		}
	    }
	    print "\n";
	}
}

sub MESSYSORT{
	$sitecalc{$a} <=> $sitecalc{$b}
}