#!/usr/bin/perl
# Reverse lookup US phone numbers using whitepages.com
# Takes numbers on STDIN or from file
# You will need to 'touch rlookup.cache' to get things started.
# WHOS YOUR DADDY?? - MadCamel

use LWP::Simple;
$|++;

load_rcache("rlookup.cache");
while (<ARGV>) {
	chomp;
	$num = $_;
	$foo = rlookup($_);
	print("$num|$foo\n");
}

sub rlookup {
	my $number = $_[0];
	foreach (keys %rcache) {
		return("$rcache{$_} (c)") if ($_ == $number);
	}

	my $info = rlookup_remote($number);
	$rcache{$number} = $info;

	return($info);
}

sub rlookup_remote {
        my $content = get("http://www.whitepages.com/search/Reverse_Phone?phone=$_[0]");
        return("Error") unless defined $content;

	$content =~ s/[\r\n\|]//g;
	$content =~ s/\s+/ /g;
	$_ = $content;

	# Case 1: Info fully available
	if (/City=(.+?)&State=(.+?)&/i) {
		my ($city, $state) = ($1, $2);
		my $first = $1 if (/_RM_HTML_FIRST_ESC_=([\w\d\%\s\,\"]+?)&/);
		my $last = $1  if (/_RM_HTML_LAST_ESC_=([\w\d\%\s\,\"]+?)&/);

		foreach(($city, $state, $first, $last)) {
			$_ = urldecode($_);
		}

		return("$first $last, $city, $state") if ($first);
		return("$last, $city, $state") if ($last);
		return("$city, $state");
	}

	# Case 2: Location available
	if (/quot; is a (.+?) based/) {
		my $location = $1;
		my $carrier = $1 if (/carrier is (.+) However/);
		return("$location ($carrier)") if ($carrier);
		return("$location");
	}

	# Case 3: Unknown
	return("Unknown");
}   

sub urldecode {
	$_[0] =~ s/%(..)/pack("c",hex($1))/eg;
	return($_[0]);
}

sub load_rcache {
	open(CACHE, $_[0]) or die("Can't load $_[0] $!\n");
	while(<CACHE>) {
		chomp;
		my ($num, $info) = split(/\|/, $_, 2);
		$rcache{$num} = $info;
	}
	close(CACHE);
}

sub save_rcache {
	open(CACHE, ">$_[0]") or die("Can't save $_[0]: $!\n");
	foreach (keys %rcache) {
		print CACHE ("$_|$rcache{$_}\n");
	}
	close(CACHE);
}

END {
	save_rcache("rlookup.cache");
}