#####################################################################
# yellowpag3r.pl - simple email grabber                             #
#   [[[  by pr0misc (pr0misc@gmail.com) ]]]                         #
#####################################################################
#                                                                   #
# - version 0.5: (july.07)                                          #
#   ============                                                    #
#                                                                   #
#   simple email grabber from a webpage (needs libwww-perl)         #
#                                                                   #
# usage: perl yellowpag3r.pl <webpage> <file> <output>              #
#                                                                   # 
#  ex: (http://www.uspto.gov/main/contacts.htm)                     #
#                                                                   #
# perl yellowpag3r.05.pl www.uspto.gov /main/contacts.htm mails.txt #
#                                                                   #
#####################################################################

#/usr/local/bin/perl

$input_address = $ARGV[0];
$output_file = $ARGV[2];
$web_file = $ARGV[1];

print "[[ yellowpag3r 0.1 . by pr0misc ]]\n";

unless ($input_address) {
 die "__ usage: yellowpag3r.pl <webpage> <file> <output_file>\n";
}

unless ($web_file) {
 die "__ usage: yellowpag3r.pl <webpage> <file> <output_file>\n";
}

unless ($output_file) {
 die "__ usage: yellowpag3r.pl <webpage> <file> <output_file>\n";
}

 use Net::HTTP;
 my $s = Net::HTTP->new(Host => $input_address) || die $@;
 $s->write_request(GET => $web_file, 'User-Agent' => "Mozilla/5.0");
 my($code, $mess, %h) = $s->read_response_headers;

 print "yP: getting $web_file from $input_address...\n";

 open(OUTPUT, ">temp");

 while (1) {
    my $buf;
    my $n = $s->read_entity_body($buf, 1024);
    die "__ p4nic: read failed: $!" unless defined $n;
    last unless $n;
    print OUTPUT $buf;
 }

 close(OUTPUT);

 print "yP: gattering e-mails...\n";

open (INPUT, "temp") or die " cannot open the input file ($input_file)...\n";
open (OUTPUT, ">>$output_file") or die "__ p4nic: cannot open the output file ($output_file), check your permissions...n";

$email_count = 0;

while ($line = <INPUT>) {

  @cenas = split(/</, $line);
  foreach $item (@cenas)
 {
   	if($item =~ /mailto:/ ) {
	$email_count++;
	$email = substr($item, index($item, ":")+1, rindex($item, "\"")-index($item, ":")-1);
	print OUTPUT "$email\n";
	}
 }

 }

close (INPUT);
close (OUTPUT);

print "yP: wrote $email_count mail adresses to [$output_file]....\n";