特定爬虫程序备份代码,PHP代码库,德仔网

代码语言
代码分类
【PHP】特定爬虫程序备份
作者:韩凌飞 / 发布于2015/8/19/ 578
用于爬取一个网站上面的黄页信息

	<?PHP
	error_reporting(E_ALL^E_NOTICE);
	$conn = @ mysql_connect("localhost", "ruzhouren", "ruzhourenmysql");
	mysql_select_db("ruzhouren", $conn);
	mysql_query("set names 'utf8'");
	date_default_timezone_set("ETC/GMT-8");
	$day_date=date('Y-m-d');
	$time_date=date('Y-m-d H:i:s');
	$time_dateline=time();
	//$start_num_query=mysql_query("SELECT * FROM `rzr_wsq_rzbst_num` where `id`= 1 ");
	$start_num=mysql_fetch_array($start_num_query);
	
	function DeleteHtml($str) {
	 $str = trim($str);
	 $str = preg_replace("/\t/","",$str);
	 $str = preg_replace("/\r\n/","",$str);
	 $str = preg_replace("/\r/","",$str);
	 $str = preg_replace("/\n/","",$str);
	 $str = preg_replace("/ /","",$str);
	 $str = preg_replace("/ /","",$str);
	 return trim($str);
	}
	
	for($a=170;$a<1000000;$a++){
	 $url = "http://www.rz0375.com/phonebook/all/pn{$a}/";
	 $fp = @fopen($url, "r") or die("timeout");
	 $fcontents = file_get_contents($url);
	 $fcontents = iconv("GB2312//IGNORE", "UTF-8",trim($fcontents));
	 @preg_match("/target=\"_blank\" >(.*)<\/a><\/span>/ms", $fcontents, $li_span);
	 $exp_li_span = explode("<li>",$li_span[0]);
	 foreach($exp_li_span as $span){
	 $exp_http_1 = explode("http://",$span);
	 if(count($exp_http_1) <3 ){
	 $exp_http_2 = explode("\">",$exp_http_1[1]);
	 }else{
	 $exp_http_2 = explode("\" target=\"_blank\"",$exp_http_1[1]);
	 }
	 $url_con = "http://".$exp_http_2[0];
	 $http = preg_replace('/www.rz0375.com\/phonebook\//','',$exp_http_2[0]);
	 $fp = @fopen($url_con, "r");
	 if(! $fp){
	 continue;
	 }
	 $fcontent_url = file_get_contents($url_con);
	 $fcontent02 = iconv("GB2312//IGNORE", "UTF-8",trim($fcontent_url));
	 @preg_match("/<div class=\"con f14 fa\">(.*)<\/div>/mUs", $fcontent02, $url_contron);
	 $contron_str = preg_replace('/<span class="ewm">(.*?)<\/span>/','',$url_contron[1]);
	 $contron_str = preg_replace('/<span class="yp">(.*?)<\/span>/','',$contron_str);
	 $contron_str_exp = explode("",$contron_str);
	 foreach($contron_str_exp as $li_exp_span){
	 $li_exp_span = DeleteHtml($li_exp_span);
	 $li_exp_sub = mb_substr($li_exp_span,0,2,'utf-8');
	 if($li_exp_sub == "名称"){
	 $li_span = strip_tags($li_exp_span);
	 $name = explode("：",$li_span);
	 }
	 if($li_exp_sub == "电话"){
	 $strip_str_phone = strip_tags($li_exp_span);
	 $phone = explode("：",$strip_str_phone);
	 }
	 if($li_exp_sub == "手机"){
	 $strip_str_tel = strip_tags($li_exp_span);
	 $tel = explode("：",$strip_str_tel);
	 }
	 if($li_exp_sub == "地址"){
	 $strip_str_addres = strip_tags($li_exp_span);
	 $addres = explode("：",$strip_str_addres);
	 }
	 if($li_exp_sub == "标签"){
	 $strip_str_tag = strip_tags($li_exp_span);
	 $tag = explode("：",$strip_str_tag);
	 }
	 }
	 $insert_rs = mysql_query("INSERT INTO `ruzhouren`.`rzrmh_tel` (`id`, `name`, `status`, `order`, `url`, `vip`, `tel`, `tel2`, `addr`, `add_uid`, `own_uid`, `add_dateline`, `type`, `http`) VALUES (NULL, '$name[1]', '1', '1', '0', '1', '$phone[1]', '$tel[1]', '$addres[1]', '1', '1', '$time_dateline', '0', '$http');");
	 if($insert_rs){
	 echo "insert-----$name[1] Succeed!"."\n";
	 }
	 //print_r($contron_str_exp);
	 }
	 sleep(5);
	
	 //print_r($pre_li);
	 /* v1.0
	 foreach($exp_li_span as $span){
	 $ex_url = "<span class='yp'>";
	 $tow_span = explode("$ex_url",$span);
	 $string = preg_replace('/<span class="comp"><a(.*?)href="(.*?)" target="_blank" >/','',$tow_span[0]);
	 $string = preg_replace('/<\/a><\/span>/','',$string);
	 $string_ok = preg_replace('/target="_blank" >/','',$string);
	 $span_2 = explode(".html",$tow_span[1]);
	 if(count($span_2) >2 ){
	 $span_exp = $span_2[2];
	 }else{
	 $span_exp = $span_2[1];
	 }
	 $string2 = preg_replace('/<\/a><\/span><\/li>/','',$span_exp);
	 $string2 = preg_replace('/<\/a><\/span>/','',$string2);
	 $string2_ok = preg_replace('/">/','',$string2 );
	 print $string_ok."------".$string2_ok;
	 }*/
	 sleep(10);
	}
	?>
试试其它关键字
　爬虫程序　
同语言下
可能有用的
韩凌飞贡献的其它代码(1)
.特定爬虫程序备份