# Simple Brand Monitoring using Perl
# Proof of concept/sample code for
#	- Phishing Sites	
#	- Web References

# Load required modules
use LWP::UserAgent; 		# Used for Web Page Query
use Crypt::SSLeay; 		# Used for SSL Queries
use URI::Escape;
use Time::localtime;

$tm = localtime;
$now = sprintf("%04d-%02d-%02d", $tm->year+1900,($tm->mon)+1, $tm->mday) . " " . sprintf(" %02d:%02d:%02d", $tm->hour, $tm->min, $tm->sec);

#Read in the ini file
$inifile = $ARGV[0];

if (! $inifile)
{
	print "brandmonitor.pl Monitor Google search engine for hits matching a specific string\n";
	print "To use, create an inifile, and then run \"perl brandmonitor.pl inifilename\"\n";
	print "\nSample inifile contents:\n";
	print "  querystring=\"KEVIN BONG\"\n  smtpserver=localhost.localdomain.com\n  mailfrom=webmaster@petsits.com\n  mailto=info@petsits.com\n  sendmail=no\n";
}

if (! -e $inifile)
{
	print "INI File $inifile not found\n";
	exit();
}
else
{
	open INI, $inifile or die "Could not open INI file $inifile, $!\n";
	while (<INI>)
	{
		chomp;
		if (m/^(.*)\=(.*)$/)
		{
			$ini{$1} = $2;
		}
	}
}

# Pick an output file name
$datafile = $inifile;
$datafile =~ s/\..*$//;
$reportfile = $datafile;
$datafile .= ".data";
$reportfile .= ".out.html";


if (! $ini{"querystring"} )
{
	print ("Querystring not found in ini file $inifile\n");
	exit;
}

$searchstring = $ini{"querystring"};

if (open LASTRUN, $datafile) #ok if it fails, no last run
{
	while (<LASTRUN>)
	{
		if (m/^METADATA(.*)\t(.*)$/)
		{
			$lastrunini{$1} = $2;
		}
		elsif ( m/(.*)\t(.*)$/)
		{
			$lastrunini{$1} = {$2};
		}
	}
}

$onmoreresultsstart = 0;

# &filter=0 causes it not to omit search results that are similar to those already displayed.
$searchurl = "http://www.google.com/search?q=" . uri_escape($searchstring) . "&filter=0&btnG=Search";
print "Searching using $searchurl\n";

$allcontents = &geturlcontents($searchurl);
print "Fetching first 10 results\n";

while ($allcontents) 
{
	
	@searchresults = ($allcontents =~ m/\<h2\sclass\=r\>(.*?)\<\/h2\>/igs);

	foreach $resultmatch (@searchresults)
	{
		#print "************\n$match\n*********\n\n";
		if ($resultmatch =~ m/\<a\shref\=\"(..*?)\"..*?\>(.*)\<\/a\>/)
		{
			$searchresultshash{$1} = $2;
			if (! $lastrunini{$1})
			{
				$newresultshash{$1} = $2;
				#print "Description: $2";
			}
			
		}
		else
		{
			throwerror ("Match $match does not match format");
		}
	}
	if ($allcontents =~ m/\<div\sid\=nn\>\<\/div\>Next\<\/a\>/)
	{
		$onmoreresultsstart += 10;
		print "Fetching results $onmoreresultsstart through " . ($onmoreresultsstart + 10) . "\n";
		$allcontents = &geturlcontents($searchurl . "&start=" . $onmoreresultsstart);
	}
	else
	{
		$allcontents = 0;
	}
	
}

# print the output to the screen, or email it as appropriate.
open REPORTFILE, ">$reportfile" or die "Could not write reportfile $reportfile $!\n";

print REPORTFILE "Perl Brand Monitor Run " ."<BR>\n";
print REPORTFILE "Search for:" . $searchstring . "<BR>\n";
print REPORTFILE "Report Created $now<BR>\n";
print REPORTFILE "Last Run on: " . $lastrunini{"LastRunDate"} .  "<BR>\n\n";

$totalmatches = keys %searchresultshash;
print REPORTFILE "Total Search Results: $totalmatches<BR>&nbsp;<br>\n";
$totalmatches = keys %newresultshash;
print REPORTFILE "Total New Search Results: $totalmatches<BR>&nbsp;<br>\n";

foreach $matchurl (sort keys %newresultshash)
{
	if (!($lastrun{$matchurl}))
	{
		print REPORTFILE "$newresultshash{$matchurl} <br><a href=\"" . $matchurl . "\">" . $matchurl . "</a><BR>&nbsp;<br>\n";
	}
}

close REPORTFILE;
print "\nReport complete.  Output saved in $reportfile\n\n";


# print the results out to the lastrun data file
open DATAFILE, ">$datafile" or die "Could not write last run datafile $datafile $!\n";

print DATAFILE "METADATALastRunDate\t$now\n";
foreach $matchurl (sort keys %searchresultshash)
{
	if (!($lastrun{$matchurl}))
	{
		print DATAFILE "$matchurl\t" . $searchresultshash{$matchurl} . "\n";
	}
}
close DATAFILE;

exit();

sub throwerror
{
	my ($errmsg, @rest) = @_;
	
	print $errmsg;
}

sub geturlcontents
{
	my ($url, @rest) = @_;

	my $ua = new LWP::UserAgent(agent => 'User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.8.1.4) Gecko/20070515 Firefox/2.0.0.4');

	my $req = new HTTP::Request('GET', $url);

	print "\nRetrieve web page $url ?\n(y/n)";
	$proceed = <STDIN>;
	if ( !($proceed =~ m/y/i))
	{
		print "\n\nQuitting\n";
		exit;
	}
	
	my $res = $ua->request($req);

	if ($res->is_success) 	
	{
		return $res->content;
	}
	else
	{
		return 0;
	}
}