#!/usr/bin/perl


use Digest::MD5 qw(md5_hex);
use Email::Address;
use JSON;
use strict;

# There are 3 sections in this example you should edit
# to integrate with your Customer Management System (CMS)

my $searchhost = 'ws.contactclean.com'; # the load balanced search cluster 

my $json = new JSON;


# EDIT 1
# config for the control files and to get the results back
my $bulk_name = 's_' . time(); # a unique name for each search set
my $return_userhost = 'user@myserver.domain';
my $notify_email = 'my.email@domain';

my $account = 'acc name'; # this will be given to you by ContactClean admin
my $key = '0123456789abcdef0123456789abcdef'; # from https://www.contactclean.com/user/license_key.php?gdet=yes
# end EDIT 1


# EDIT 2
# generate list of emails
# this is the section to integrate into your CMS...
my @emails = (
	'Arthur Pewty <arthur.pewty@example.com>',
	'TestOld@ContactClean.com',
);
# end EDIT 2


# generate data and control files
my $fnamedata = "${bulk_name}_data.gz";
my $fnamecontrol = "${bulk_name}_control.json";

open(DATA, "| gzip > $fnamedata") || die "Failed: | gzip > $fnamedata";
open(CONTROL, "> $fnamecontrol") || die "Failed: > $fnamecontrol";

my %md_map = ();
Email::Address->disable_cache;
for my $email (@emails) {
	my @addr = Email::Address->parse($email);
	my $addr = lc $addr[0]->address; 
	# print "$email cleaned up to: $addr \n";
	my $md = md5_hex($addr); 
	$md_map{$md} = $email; # possible EDIT for large lists this should be stored in a database
	
	print DATA "$md\n";
}
close DATA;

print CONTROL $json->objToJson({
	key				=> $key,
	return_userhost	=> $return_userhost,
	notify_email	=> $notify_email,
});
close CONTROL;


# transfer data and control
my $remote = "${account}\@${searchhost}:";
`scp $fnamedata $fnamecontrol $remote`;

# set the search going
`touch ${bulk_name}_ready`;
`scp ${bulk_name}_ready $remote`;

# now poll $return_dir until the file ${bulk_name}_done appears...
print "Looking for ${bulk_name}_done \n";
while (1) {
	if (-f "${bulk_name}_done") {
		print "Results are here in ${bulk_name}_results.gz \n";
		last;
	}
	print "Results not appeared yet\n";
	sleep(10);
}

open(R, "${bulk_name}_report.json") || die "Report file ${bulk_name}_report.json missing";
my $report = $json->jsonToObj(join('', <R>));
close(R);

if ($report->{'error'}) {
	print join("\n", $report->{'error'}), "\n";
	exit 1; 
}

# EDIT 3
# process $return_dir/${bulk_name}_results.gz
open(RES, "zcat ${bulk_name}_results.gz |") || die "Results file ${bulk_name}_results.gz missing";
while (<RES>) {
	chomp;
	next if m/norecord$/;
	if (! m/,/) {
		print "ERROR: malformed result line: $_ \n";
	}
	
	my ($md, $update) = split ',';
	my $old = $md_map{$md};
	
	# this is the second section to integrate into your CMS...
	if ($update =~ /@/) {
		print "UPDATE LIST: $old -> $update \n";
	} else {
		print "INFO: $old - $update \n";
	} 
}