ecshop中采集各大站商品详细页面信息
2009-09-18 17:46 来源：www.chinab4c.com 作者：ecshop专家
其实很多时候，很多朋友，需要我帮他们抓各个站的商品信息，于是就写了几个主要的B2C的站，来进行整合。代码如下。
<?php
function to_info($url){
$array = array();
$array['msg'] = '';
//taobao
if(strpos($url,"taobao.")){
$str = file($url);
if($str){
$string = join('',$str);
preg_match("/<img id=\"J_ImgBooth\"\s*src=\"(.*?)\"/i",$string,$img);
preg_match("/<li>价格：<em>(.*?)<\/em>元<\/li>/i",$string,$price);
preg_match("/<div\s*class=\"detail-hd\">(.*?)<\/div>/is",$string,$title);
preg_match("/<a\s*class=\"hCard fn\"[^>]*>(.*?)<\/a>/",$string,$name);
$array['img'] = $img[1];
$array['price'] = $price[1];
$array['title'] = trim(strip_tags($title[1]));
$array['name'] = trim(($name[1]));
} else{
$array['msg'] = 'false';
}
}
//paipai
if(strpos($url,"paipai.")){
$str = file($url);
if($str){
$string = join('',$str);
//<em id="commodityCurrentPrice" defaultVal="135.00">135.00</em>
preg_match("/<em\s*id=\"commodityCurrentPrice\"\s*[^>]*>(.*?)<\/em>/i",$string,$price);
preg_match("/<h1>(.*?)<\/h1>/i",$string,$title);
preg_match("/<li id=\"userNickname\">(.*?)<\/li>/is",$string,$name);
preg_match("/<span\sid=\"shipCost2\">(.*?)<\/span>/i",$string,$shipping1);
//print_r($shipping1[1]);
if($shipping1[1]){
preg_match_all("/(\d|\.)*元\(快递\)/i",iconv('gbk','gb2312',$shipping1[1]),$shipping);
if($shipping[0][0]){
preg_match("/(\d|\.)*/",$shipping[0][0],$sp);
$array[sp] = $sp['0'];
}
}
$name1 = strip_tags($name[1],"<strong>");
preg_match("/<strong>(.*?)<\/strong>/i",$name1,$name);
$array['price'] = $price[1];
$array['name'] = $name[1];
$array['title'] = trim(strip_tags($title[1]));
}else{
$array['msg'] = 'false';
}
}
if(strpos($url,'eachnet.')){
$str = file($url);
if($str){
$string = join('',$str);
preg_match("/<h1>(.*?)<\/h1>/is",$string,$title);
preg_match('/formatPrice\(\'([\d\.]*)/i',$string,$price);
$array['title'] = trim((iconv("utf-8","gbk",$title[1])));
$array['price'] = $price[1];
}else{
$array['msg'] = 'false';
}
}
if(strpos($url,'amazon.')){
$str = file($url);
if($str){
$string = join('',$str);
preg_match("/<h1\s*class=\"DetailTitle\">(.*?)<\/h1>/is",$string,$title);
preg_match("/<span\s*class=\"OurPrice\">([\d\.]*)/",$string,$price);
preg_match("/<img\s*id=\"ImageShow\"(.*?)*src=\"(.*?)\"/i",iconv("utf-8","gbk",$string),$img);
print_r($img);
$array['title'] = trim((iconv("utf-8","gbk",$title[1])));
$array['price'] = $price[1];
$array['img'] = $img[2];
}else{
$array['msg'] = 'false';
}
}
if(strpos($url,'dangdang.')){
$str = file($url);
if($str){
$string = join('',$str);
preg_match("/<span\s*class=\"black000\">(.*?)<\/span>/is",$string,$title);
preg_match("/<span\s*class=\"redc30\">(.*?)<b>([\d\.]*)(.*?)<\/b><\/span>/i",$string,$price);
preg_match("/<img\s*src=\"(.*?)\"\s*id=\"img_show_prd\"/",$string,$img);
$array['title'] = strip_tags($title[1]);
$array['price'] = $price[2];
$array['img'] = $img[1];
if(!$array['title']){
preg_match("/<div\s*class=\"mall_goods_title_text\">(.*?)<\/div>/i",$string,$title1);
$array['title'] = strip_tags($title1[1]);
preg_match("/<b\s*id=\"salePriceTag\">(.*?)<\/b>/i",$string,$price);
if($price[1]){
$array['price'] = str_replace("￥","",$price[1]);
}
preg_match("/<img\s*id='largePic'\s*src=\"(.*?)\"[^>]*\/>/i",$string,$img);
if($img){
$array['img'] = $img[1];
}
}
}else{
$array['msg'] = 'false';
}
}
if(strpos($url,'baidu.')){
$str = file($url);
if($str){
$string = join('',$str);
preg_match("/<h1>(.*?)<\/h1>/is",$string,$title);
//<li class="uname"><a href="/shop/88fc04edf20bcd36a5ab21fa/credit" target='_blank'>3c数码城</a>
preg_match("/<span class=\"price\">(.*?)<\/span>/",$string,$price);
if($price[1]){
preg_match_all('/(.*?)<small>(.*?)<\/small>/i',$price[1],$p);
}
preg_match("/<li\s*class=\"uname\"><a[^>]*>(.*?)<\/a>/i",$string,$name);
$array['price'] = $p[1][0].$p[2][0];
$array['title'] = $title[1];
$array['name'] = $name[1];
}else{
$array['msg'] = 'false';
}
}
if(strpos($url,'china-pub.')){
$file = file($url);
if($file){
$string = join("",$file);
preg_match("/<h1\s*class=\"black15c\">(.*?)<\/h1>/is",$string,$title);
preg_match("/<span\s*class=\"red\">(.*?)<\/span>/is", $string,$price);
preg_match_all("/<img\s*src=\'(.*?)\'\s/i",$string,$img);
if($price[1]){
$array['price'] = str_replace("￥",'',$price[1]);
}
$array[title] = trim(strip_tags($title[1]));
$array['img'] = $img[1][1];
}else{
$array['msg'] = 'false';
}
}
if(strpos($url,'360buy.')){
$file = file($url);
if($file){
$string = join("",$file);
preg_match("/<h1>(.*?)<\/h1>/i",$string,$title);
preg_match_all("/jqimg=\"(.*?)\"/is", $string, $img);
if($img){
$array['img'] = $img[1][1];
}
$array[title] = $title[1];
}else{
$array['msg'] = 'false';
}
}
if(strpos($url,'139shop.')){
$file = file($url);
if($file){
$string = join("",$file);//<font class='black9'>499元</font>
preg_match("/<h1>(.*?)<\/h1>/i",$string,$title);
preg_match("/<font\sclass='black9'>([\d\.]*)元<\/font>/i",$string,$price);
$array[title] = $title[1];
$array[price] = $price[1];
}else{
$array['msg'] = 'false';
}
}
if(strpos($url,'m18.')){
$file = file($url);
if($file){
$string = join("",$file);
preg_match("/<span\sclass=\"price\">(.*?)<\/span>/i",$string,$price);
preg_match("/<div\s*class=\"bigpic fl\"><img\s*src=\"(.*?)\"[^>]*><\/div>/i",$string,$img);
$array[title] = trim(iconv("utf-8","gbk",$title[1]));
$array[price] = $price[1];
$array[img] = $img[1];
}else{
$array['msg'] = 'false';
}
}
if(strpos($url,'e-giordano.')){
$file = file($url);
if($file){
$string = join("",$file);
preg_match("/<span\s*id=\"ctl00_ctl00_ContentPlaceMain_ContentPlaceMain_ProductDetail1_labelproductname\"[^>]*>(.*?)<\/span>/is",$string,$title);
preg_match("/<span\s*id=\"ctl00_ctl00_ContentPlaceMain_ContentPlaceMain_ProductDetail1_TabContainer2_TabPanel1_labNPric\"[^>]*>￥(.*?)<\/span>/i",iconv("utf-8","gbk",$string),$price);
//onmouseover="PreviewImg(
preg_match("/onmouseover=\"PreviewImg\(\'(.*?)\'\)\"/is",$string,$img);
$array[title] = trim(iconv("utf-8","gbk",$title[1]));
$array[price] = $price[1];
if($img[1]){
$array[img] = substr($img[1],0,strpos($img[1],',')-1);
}
} else{
$array['msg'] = 'false';
}
}
if(strpos($url,'818shyf.')){
$file = file($url);
if($file){
$string = join("",$file);
preg_match("/<h1[^>]*>(.*?)<\/h1>/is",$string,$title);
$array[title] =strip_tags($title[1]);
preg_match("/<span\s*class=\"style1\">￥(.*?)<\/span>/i", $string, $price);
$array[price] = $price[1];
preg_match("/<td height=\"180\" align=\"center\">(.*?)<\/td>/is", $string, $img);
if($img[1]){
preg_match("/<img\s*src=\"(.*?)\"[^>]*\/>/is",$img[1],$img1);
$array[img] = "http://www.818shyf.com/".$img1[1];
}
}else{
$array['msg'] = 'false';
}
}
if(strpos($url,'redbaby.')){
$file = file($url);
if($file){
$string = join("",$file);
preg_match("/<h1><img[^>]*>(.*?)<\/h1>/is",$string,$title);
$array[title] =trim($title[1]);
preg_match("/<span\sid=\"price\"\s*class=\"jiage\">(.*?)<\/span>/i",$string,$price);
$array[price] = $price[1];
preg_match("/<img\s*class=\"curr\"\s*src=\"(.*?)\"[^>]*>/is",$string,$img);
$array[img] = $img[1];
}else{
$array['msg'] = 'false';
}
}
if(strpos($url,'.no5.')){
$file = file($url);
if($file){
$string = join("",$file);
//<a href="javascript:viewphoto('20090911170305265004157.jpg');">
preg_match("/<td\sid=\"no5price[\d]*\_\d*\"[^>]*>(.*?)元<\/td>/i", $string, $price);
preg_match("/<td\svalign=\"top\">(.*?)<\/td>/i",$string,$title);
preg_match("/<img height=\"250\" width=\"250\" border=\"0\" alt=\".*?\" src=\"(.*?)\"\/>/is",$string,$img);
$array[title] =$title[1];
$array[price] = $price[1];
//no image
}else{
$array['msg'] = 'false';
}
}
if(strpos($url,'4inlook.')){
$file = file($url);
if($file){
$string = join("",$file);
preg_match("/<img id=\"big\"\s*src=\"(.*?)\"/i",$string,$img);
preg_match("/<span\s*id=\"lblname\">(.*?)<\/span>/i", $string, $title);
//<span class="font400"><span id="lblSalePrice">173</span>
preg_match("/<span\s*class=\"font400\">(.*?)<\/span>/is",$string,$price);
$array[title] =$title[1];
$array[img] = "http://www.4inlook.com/".$img[1];
$array[price] = strip_tags($price[1]);
}else{
$array['msg'] = 'false';
}
}
if(strpos($url,'.x.')){
$file = file($url);
if($file){
$string = join("",$file);
preg_match("/<div\s*class=\"pro_name\">(.*?)<\/div>/i",$string,$title);
preg_match("/<span\s*class=\"text_red jiage\">(.*?)<\/span>/i",$string,$price);
//<img src="http://help.x.com.cn/product_img/pic_mid/N4749.jpg" alt="夜色恋人[201190]" class="pro_img" />
$array[title] =$title[1];
if($price[1]){
$array[price] = str_replace("[桔 色 价]：￥","",$price[1]);
}
preg_match("/<img\s*src=\"(.*?)\"\s*alt=\"(.*?)\"\s*class=\"pro_img\"[^>]*\/>/i",$string,$img);
$array[img] = $img[1];
}else{
$array['msg'] = 'false';
}
}
if(strpos($url,'.7shop24.')){
$file = file($url);
if($file){
$string = join("",$file);
preg_match("/<div\s*class=\"sj_font_a\">￥(.*?)<\/div>/i",$string,$price);
preg_match("/<div\s*align=\"center\"\s*class=\"sjname_font\"[^>]*>(.*?)<\/div>/is",$string,$title);
preg_match("/<IMG\s*src=\"(.*?)\"\s*width=349\s*jqimg=\"(.*?)\">/i",$string,$img);
$array[title] =trim($title[1]);
$array[price] = $price[1];
$array['img'] = "http://www.7shop24.com/".$img[1];
}else{
$array['msg'] = 'false';
}
}
if(strpos($url,'.blemall.')){
$file = file($url);
if($file){
$string = join("",$file);
//id="f_pic" //style="background:url(http://services.blemall.com/pic_show.php?valueKey_Type=U1BfX19fMDAxNzg5MzI==&picType=U1BfX19fMDAxNzg5MzI=&sno=0&temp=1);cursor:pointer;"
preg_match("/<div\s*class=\"odd_right_pmtxt\"\s*style=\"color:#023bbd\">(.*?)<\/div>/i",$string,$title);
preg_match("/id=\"f_pic\"\s*style=\"background:url\((.*?)\);cursor:pointer;\"/i",$string,$img);
$array[title] =trim($title[1]);
$array[img] = $img[1];
} else{
$array['msg'] = 'false';
}
}
if(strpos($url,'.blemall.')){
$file = file($url);
if($file){
$string = join("",$file);
//id="f_pic" //style="background:url(http://services.blemall.com/pic_show.php?valueKey_Type=U1BfX19fMDAxNzg5MzI==&picType=U1BfX19fMDAxNzg5MzI=&sno=0&temp=1);cursor:pointer;"
preg_match("/<div\s*class=\"odd_right_pmtxt\"\s*style=\"color:#023bbd\">(.*?)<\/div>/i",$string,$title);
preg_match("/id=\"f_pic\"\s*style=\"background:url\((.*?)\);cursor:pointer;\"/i",$string,$img);
$array[title] =trim($title[1]);
$array[img] = $img[1];
}else{
$array['msg'] = 'false';
}
}
if(strpos($url,'.newegg.')){
$file = file($url);
if($file){
$string = join("",$file);
preg_match("/<h1\s*[^>]*>(.*?)<\/h1>/i",$string,$title);
preg_match("/<a id=\"bigImg\" href=\"(.*?)\" class=\"jqzoom\"/i",$string,$img);
$array[title] =trim($title[1]);
$array[img] = $img[1];
}else{
$array['msg'] = 'false';
}
}
if(strpos($url,'.vancl.')){
$file = file($url);
if($file){
$string = join("",$file);
/*
<span id="SellPrice" style="color:#333333;">89</span>*/
preg_match("/<h2\s*[^>]*>(.*?)<\/h2>/is",$string,$title);
preg_match("/<a\s*href=\"(.*?)\"\s*class=\"jqzoom\"/i",$string,$img);
preg_match("/<span\sid=\"SellPrice\"\s*style=\"color:#333333;\">(.*?)<\/span>/i",$string,$price);
$array[title] =trim(iconv("utf-8","gbk",$title[1]));
$array[img] = $img[1];
$array[price] = $price[1];
} else{
$array['msg'] = 'false';
}
}
if(strpos($url,'.wangshanghai.')){
$ch = curl_init($url);
curl_setopt($ch,CURLOPT_USERAGENT,$_SERVER["HTTP_USER_AGENT"]);
ob_start();
curl_exec($ch);
curl_close($ch);
$str = ob_get_contents();
ob_end_clean();
if($str){
preg_match("/<td\s*class=\"titgd\"\s*colspan=\"2\"\s*valign=\"center\">(.*?)<\/td>/i",$str,$title);
$array[title] =trim(iconv("utf-8","gbk",$title[1]));
preg_match_all("/<font\s*color=\"#cc0000\">(.*?)<\/font>/i",$str,$price);
/*<img id='pic_02' src="/SHP00001/frontend/shopdata/SHP00001/images/prod/100117461_02.jpg"
border="0" width="200" height="200" alt="点击显示大图">*/
$s = iconv("utf-8","gbk",$price[1][1]);;
$array[price] = str_replace("￥","",$s);
preg_match("/<img\s*id=\'pic_02\'\s*src=\"(.*?)\"[^>]*>/is",$str,$img);
$array['img'] = "http://www.wangshanghai.com".$img[1];
}else{
$array['msg'] = 'false';
}
}
return $array;
}
?>
最近更新
常用插件
- ecshop降价通知登记插件
ecshop降价通知登记插件，主要是为了方便某些客户，对商品价格要求比...
- ecshop分类树中统计商品数
最近忙于开发其他项目,在不少朋友不断要求和催促的情况下,做出了该小...
- ecshop购物车功能改进[插件
ecshop购物车功能改进[插件套餐]主要是我们最近开发工作和开发项目中。...
- ecshop通用红包编码
很多时候，为了结合促销，必须扩展一下ecshop的红包功能。ecshop的红包...
- ecshop红包修改成满多少减
我们在长期使用ecshop的时候，我们可以发现。ecshop的红包是一个非常强...