百度收录信息抓取

#!/usr/bin/perl
=pod
前段时间朋友让我帮他写个小工具用来从这个网站http://www.baidu.com/s?wd=site%3A
上来抓取一些信息比如:域名,www.example.com,百度权重,0,站内链接,域名IP,x.x.x.x,同IP网站,0,域名年龄,2月19天,索引量-百度收录,7等这些信息，抽时间就写了一下，我主要是通
LWP::Simple和正则来扑捉关键字。
=cut
#!/usr/bin/perl
use strict;

use utf8;
use Encode;
use Encode::CN;
use File::Find;
use LWP::Simple;
use POSIX qw(strftime);
binmode(STDOUT, ":utf8");

#时间
my ($sec,$min,$hour,$day,$mon,$year,$wday,$yday,$isdst)=localtime(time());
$year = $year+1900;
$mon = $mon + 1;
my $date="$year-$mon-$day";
my $count;

mkdir ("D:\\seo_web",0775) if ! -e  "D:\\seo_web";
if (! -e "D:\\seo_web\\url.txt" ) {
print "Please create 'D:\\seo_web\\url.txt' file\n and write info to 'D:\\seo_web\\url.txt'\n
example:www.example.com in url.txt\n";
sleep(15);
print "\n";
}

open(my $FL, ">D:\\seo_web\\$date.csv.txt") or die "Cant write 'D:\\seo_web\\$date.csv.txt':$!\n";

open(BQURL, "<D:\\seo_web\\url.txt")
or die "Can't open 'D:\\seo_web\\url.txt':$!\n";
while (<BQURL>) {
my $val = "$_";
my $url = "http://seo.chinaz.com/?host=$val/";
my $content  = get encode("utf-8",$url);
die  "Couldn't get $url" unless defined $content;

for($content)
{
$count++;
print $FL "$count,";
print "$count,";
if (/(请输入网站地址.*\/")/)
{
my $n = "$1";
if ($n =~/(\w+\.\w+\.\w+)/) {
print $FL encode("utf8","域名,$1,");
print "$1,";
}
}
if (/(?:<span\s+style=\"\s+margin-left\:\d+px\;\">)(百度权重)/) {
print $FL encode("utf8","$1,");
my $n = "$1";
}
if (/(?:images\/baiduapp\/)(\d+)(?:\.gif)/) {
print $FL encode("utf8","$1,");
print "$1,";
}

if (/(出站链接.*<\/a>)/) {
my $n = "$1\n";
if ($n =~/(\d+)/) {
print $FL encode("utf8","出站链接,$1,");
print "$1,";
}else{
print $FL encode("utf8","出站链接,none,");
print "none,";
}
}
if (/(站内链接.*<\/span>)/) {
my $n = "$1";
if ($n =~ /(站内链接)/) {
print $FL encode("utf8","$1,");
if ($n =~ /(\d+)/) {
$1 = "none" if (!$1);
print $FL encode("utf8","$1,");
print "$1,";
}
}
}
if (/(域名IP.*  )/) {
my $n = "$1";
if ($n =~/(\d+\.\d+\.\d+\.\d+)/) {
$1 = "none" if (!$1);
print $FL encode("utf8","域名IP,$1,");
print "$1,";
}
}
if (/(同IP网站.*个)/) {
my $n = "$1";
if ($n =~ /(\d+)/) {
print $FL encode("utf8","同IP网站,$1,");
print "$1,";
}
}
if (/(域名年龄.*<\/font>)/) {
my $n = "$1";
if ($n = ~ /(\d+月\d+天)/) {
$1 = "none" if (!$1);
print $FL encode("utf8","域名年龄,$1,");
my $v = "$1";
$v=~ s/月/ Months /;
$v=~ s/天/ Days/;
print "$v,";
}

}
my $url = get("http://www.baidu.com/s?wd=site%3A$val");
for($url)
{
if(/(该网站共有.*<\/b>)/)
{
my $n = "$1";
if($n =~ /(\d+)(?:<\/b>)/)
{
$1 = "none" if (!$1);
print $FL encode("utf8","索引量-百度收录,$1");
print "$1";
}
}
}
}
print $FL "\n";
print "\n";
}
print "\n";
print " File path is:'D:\\seo_web\\$date.csv.txt'\n";
print " Author\@Laomeng\nEmail:18682093512\@163.com\n";
sleep(60);