Perl模拟浏览器（cookie）抓Email

时间：2009-03-18 来源：snowtty

Perl模拟浏览器（cookie）抓Email

关键字: email perl Perl代码

#!/usr/bin/perl -w
#use CGI qw(:standard);
use LWP::UserAgent;
use HTTP::Request::Common;
use HTTP::Request::Common qw(POST);
#$e=exp(1);
open(OPENFILE,"keyword.txt") or die("open file failure!");
#$/=undef;
#$cgi=new CGI;
#print $cgi->header;
$email="email.txt";
$k=1;
while($line=<OPENFILE>){
#if($line=~s/##[0-9a-zA-Z]+##/$e/){
# print $line;
# }
#else
#{
# print $line;
#}
#print $line;
if($line){
$line=~s/\n|\r/ /ig;
$line=~s/\s+$//ig;
if( -e $email )
{
open(SRC,">>$email");
}
else{
open(SRC,">$email");
}
$email1=getEmail($line);
if($email1){
print SRC "$line\t$email1\r\n";
}
else{
printf SRC "$line\r\n";
}
close(SRC);
sleep(10);
if($k%8==1){
`wget http://www.lead411.com`;
}
print "$k\n";
$k++;
}
}
print "Total Record:\$";
close(OPENFILE);
#$slurp=<OPENFILE>;
sub getEmail{
my $agent=LWP::UserAgent->new;
my @header=(
'User-Agent'=>'Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1; iCafeMedia; .NET CLR 2.0.50727; CIBA)',
'Accept' => '*/*',
'Accept-Charset' => 'gzip, deflate',
'Accept-Language' => 'zh-cn',
'Cookie'=> 'Tango_UserReference=38D8FF1624305B16496E9808; MTCCK=1; _csuid=48feeef505683659; cookmcnt=999; CID=1459382; cookMemberName=YunFan; cookMemberID=61448; [email protected]; DLDExec=OK; __utma=232384002.1655516880.1231991960.1231994793.1232000250.3; __utmb=232384002; __utmc=232384002; __utmz=232384002.1231991960.1.1.utmccn=(direct)|utmcsr=(direct)|utmcmd=(none)'
);
my $url="http://www.lead411.com/allsearch.taf?_function=headersearch&SearchParam=$_[0]&SearchType=Company";
#print $_[0]."\n";
#my $arg={'SearchParam'=>"$_[0]",'SearchType'=>'Company'};
#print $arg->{SearchParam}."\n";
my $temp2="";
my $temp="",$temp1="";
#my $req=HTTP::Request->new(GET=>"$url",@header);
my $res =$agent->get($url,@header);
my $content="";
#my $res =$agent->post($url,$arg,'Content_Type' => 'form-data');
#request($req);get($url,@header);
if($res->is_success){
$content= $res->content;
#print "test";
#$content=~s/\s+|>|<|\///ig;
#print $content;
while($content=~/company.taf\?_function=detail&Company_uid1=(\d+)&_UserReference=(\w+)/ig)
{
$temp=$1;$temp1=$2;
if($temp and $temp1){
$url='http://www.lead411.com/company.taf?_function=detail&Company_uid1='.$temp.'&_UserReference='.$temp1;
$content=~s/http:\/\/www\.lead411\.com\/company\.taf\?_function=detail\&Company_uid1=$temp\&_UserReference=$temp1//ig;
print $url."\n";
$res =$agent->get($url,@header);
if( $res->is_success)
{
$content=$res->content;
$content=~s/\s+|>|<|\///ig;
$content=~s/\s+//ig;
$i=0;$temp="";
do{
$i=index($content,'CEO',$i);
$temp=substr($content,$i,150);
#print $temp;
if($temp=~/"mailto:(.*)"font/ig){
$temp2.=$1;
print $_[0]."---------------".$temp2."\n";
}
if($i!=-1){
$i=$i+150;
}
else{
$i=0;
}
}while($i)
#print $content;
#if($content=~/(<TD>.*?CEO.*?<\/TD> <TD>.*?mailto:"[A-Z0-9._%-]+\@[A-Z0-9._%-]+\.[A-Z]{2,4}"<\/TD>)/ig)
#{
# print "test $1";
#}
#print $content;
#$temp=~/mailto:"[A-Z0-9._%-]+\@[A-Z0-9._%-]+\.[A-Z]{2,4}"/ig;
}
}
}
}
return $temp2;
}
exit;