最近做了一個網(wǎng)站用到了從網(wǎng)址爬取天貓和淘寶的商品信息,首先看了下手機(jī)端的網(wǎng)頁發(fā)現(xiàn)用的react,不太了解沒法搞,所以就考慮從PC入口爬取數(shù)據(jù),但是當(dāng)爬取URL獲取數(shù)據(jù)時并沒有獲取價格,庫存等的信息,仔細(xì)研究了下發(fā)現(xiàn)是異步請求了另一個接口,但是接口要使用refer才能獲取數(shù)據(jù),于是就通過以下方式寫了一個簡單的爬蟲,用于爬取商品預(yù)覽圖和商品的第一個分類的價格、庫存等。
function crawlUrl($url){
import('PhpQuery.Curl');
$curl=new \Curl();
$result = $curl->read($url);
$content = mb_convert_encoding( $result['content'], 'UTF-8', 'UTF-8,GBK,GB2312,BIG5' );
$myres=array();
if(strrpos($url,'taobao.com')!=false) {
//匹配是否下架
if(strpos($content,'此寶貝已下架')!==false){
return false;
}
preg_match("|itemId : '(.*)'|isU", $content, $match);
$item_id=$match[1];
preg_match("|sellerId : '(.*)'|isU", $content, $match);
$sellet_id=$match[1];
preg_match("|title>(.*)/title>|isU",$content,$match);
$title=$match[1];
//價格庫存信息
$ch = curl_init();
curl_setopt ($ch, CURLOPT_URL, 'https://detailskip.taobao.com/service/getData/1/p1/item/detail/sib.htm?itemId='.$item_id.'sellerId='.$sellet_id.'modules=dynStock,qrcode,viewer,price,duty,xmpPromotion,delivery,upp,activity,fqg,zjys,amountRestriction,couponActivity,soldQuantity,originalPrice,tradeContractcallback=onSibRequestSuccess');
$opt[CURLOPT_HEADER]=false;
$opt[CURLOPT_CONNECTTIMEOUT]=15;
$opt[CURLOPT_TIMEOUT]=300;
$opt[CURLOPT_AUTOREFERER]=true;
$opt[CURLOPT_USERAGENT]='Mozilla/5.0 (Windows NT 6.1) AppleWebKit/536.11 (KHTML, like Gecko) Chrome/20.0.1132.47 Safari/536.11';
curl_setopt_array($ch,$opt);
curl_setopt ($ch, CURLOPT_RETURNTRANSFER, 1);
curl_setopt ($ch,CURLOPT_REFERER,$url);
curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false);
$out_put=curl_exec ($ch);
curl_close ($ch);
$res=str_replace('onSibRequestSuccess(',"",$out_put);
$res=rtrim($res,');1');
$result=json_decode($res,true);
//查詢出圖片信息
preg_match('|ul id="J_UlThumb" class="tb-thumb tb-clearfix">(.*)/ul>|isU', $content, $match);
preg_match_all('/img data-src="(.*?)" \//', $match[1], $images);
$myres['title']=str_replace('-淘寶網(wǎng)','',$title);
$myres['price']=current($result['data']['originalPrice']);
$myres['act_price']=current($result['data']['promotion']['promoData']);
$myres['stock']=$result['data']['dynStock']['stock'];
$myres['banners']=$images[1];
}else{
//匹配是否下架
if(strpos($content,'此寶貝已下架')!==false){
return false;
}
$start=strpos($url,'id=');
$item_id=substr($url,$start+4,12);
if(!is_numeric($item_id)){
$start=strpos($url,'?id=');
$end=strpos($url,'spm');
$item_id=substr($url,$start+4,$end-$start-4);
}
preg_match("|title>(.*)/title>|isU",$content,$match);
$title=$match[1];
$myurl='https://mdskip.taobao.com/core/initItemDetail.htm?cachedTimestamp=1500562177777queryMemberRight=truecartEnable=trueofflineShop=falseaddressLevel=2itemId='.$item_id.'tryBeforeBuy=falseisAreaSell=falsetmallBuySupport=trueisPurchaseMallPage=falsehousehold=falseisForbidBuyItem=falseservice3C=falseisRegionLevel=falseshowShopProm=falseisSecKill=falsesellerPreview=falseisUseInventoryCenter=falseisApparel=truecallback=setMdskiptimestamp=1500562172109isg=AiUlDZFWmP/sMgVurQSILU3Ytet/Zdisisg2=Ajk51JIhRFqKzxmiNPP6dkYxSKXT7iySkzSTeVtu9WDf4ll0o5Y9yKdyEtHu';
//價格庫存信息
$ch = curl_init();
curl_setopt ($ch, CURLOPT_URL, $myurl);
$opt[CURLOPT_HEADER]=false;
$opt[CURLOPT_CONNECTTIMEOUT]=15;
$opt[CURLOPT_TIMEOUT]=300;
$opt[CURLOPT_AUTOREFERER]=true;
$opt[CURLOPT_USERAGENT]='Mozilla/5.0 (Windows NT 6.1) AppleWebKit/536.11 (KHTML, like Gecko) Chrome/20.0.1132.47 Safari/536.11';
curl_setopt_array($ch,$opt);
curl_setopt ($ch, CURLOPT_RETURNTRANSFER, 1);
curl_setopt ($ch,CURLOPT_REFERER,$url);
curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false);
$out_put=curl_exec ($ch);
curl_close ($ch);
$res = mb_convert_encoding( $out_put, 'UTF-8', 'UTF-8,GBK,GB2312,BIG5' );
$res=str_replace('setMdskip',"",$res);
$res=str_replace('(',"",$res);
$res=str_replace(')',"",$res);
$result=json_decode($res,true);
$nowk="";
$nowstore="";
foreach($result['defaultModel']['inventoryDO']['skuQuantity'] as $k=>$val){
$nowk=$k;
$nowstore=$val;
break;
}
$myres['title']=str_replace('-tmall.com天貓','',$title);
$myres['price']=$result['defaultModel']['itemPriceResultDO']['priceInfo'][$nowk]['price'];
$myres['act_price']=isset($result['defaultModel']['itemPriceResultDO']['priceInfo'][$nowk]['suggestivePromotionList'])?$result['defaultModel']['itemPriceResultDO']['priceInfo'][$nowk]['suggestivePromotionList']:$result['defaultModel']['itemPriceResultDO']['priceInfo'][$nowk];
$myres['stock']=$result['defaultModel']['inventoryDO']['totalQuantity']?$result['defaultModel']['inventoryDO']['totalQuantity']:$nowstore['quantity'];
//查詢出圖片信息
preg_match('|ul id="J_UlThumb" class="tb-thumb tm-clear">(.*)/ul>|isU',$content, $match);
preg_match_all('/img src="(.*?)" \//',$match[1],$images);
$myres['banners']=$images[1];
}
return $myres;
}
上述代碼用到phpquery的庫,但是其實(shí)沒啥用,直接用Curl就行,具體爬取的數(shù)據(jù)可以穿參查看結(jié)果,方法不區(qū)分淘寶和天貓鏈接,但是前提是必須是PC端鏈接,另外正則寫的不規(guī)范,所以可以自己重寫正則來匹配數(shù)據(jù)。