当前位置： 旋风说(旋风PHPer分享网)> 技术文章> 正文
thinkphp 5的代码写法很容易就能改成其它
<?php
/**
* Created by PhpStorm.
* User: lyq
* Date: 2019/05/24
* Time: 13:48
*/
namespace app\common\model;
use think\Db;
use think\Exception;
class Copy extends Common
{
//实例化后 直接调用 copyUrl 即可
static public function copyUrl($url){
if(strpos($url, 'https://item.taobao.com')===0){
//淘宝
return self::taobao($url);
}elseif(strpos($url, 'https://detail.tmall.com')===0){
//天猫
return self::hejiangTm($url);
}elseif(strpos($url, 'https://item.jd.com')===0){
//京东
return self::jd($url);
}else{
throw new Exception('不支持此链接');
}
}
/**
* 禾匠一个微擎应用团队的api
* @param $url
* @return array
*/
static public function hejiangTm($url){
$id = self::preg_substr('/(\?id=|&id=)/', '/&/', $url);
$id = $id[0];
$api = "https://auth.zjhejiang.com/mall/copy/index?vid=".$id;
$res = file_get_contents($api);
$res = json_decode($res, true);
if ($res && isset($res['code'])) {
if ($res['code'] != 0) {
return self::tm($url);
}
$html = $res['data'];
$coding = mb_detect_encoding($html, array("ASCII", 'UTF-8', "GB2312", "GBK", 'BIG5', 'ISO-8859-1'));
$html = mb_convert_encoding($html, 'UTF-8', $coding);
$html = json_decode($html, true);
return self::hejiangData($html,$url);
} else {
return self::tm($url);
}
}
/**
* 数据处理
* @param null $html
* @param $url
* @return array
*/
static private function hejiangData($html = null,$url)
{
$content_arr = $html;
if ($content_arr['ret'][0] != 'SUCCESS::调用成功') {
return self::tm($url);
}
$info = $content_arr['data']['itemInfoModel'];
//标题
$title = $info['title'];
//缩略图
$picsPath = $info['picsPath'];
$pro_detail = json_decode($content_arr['data']['apiStack']['0']['value'], true);
$pro_data = $pro_detail['data'];
$pro_detail_info = $pro_data['itemInfoModel'];
//商品价格
$price = isset($pro_detail_info['priceUnits'][1]) ? $pro_detail_info['priceUnits'][1]['price'] : $pro_detail_info['priceUnits'][0]['price'];
$price_arr = explode('-', $price);
$goods_price = $price_arr[0];//原价
$sale_price = $pro_detail_info['priceUnits'][0]['price'];
$sale_price_arr = explode('-', $sale_price);
$goods_price_sale = $sale_price_arr[0];//售价
//图文详情
$detail_info = isset($content_arr['data']['descInfo']) ? $content_arr['data']['descInfo'] : "";
return [
'title' => $title,//标题
'thumb'=>$picsPath[0],
'images' => $picsPath,//组图
'price' => $goods_price_sale,//售价
'desc' => $detail_info//图文详情
];
}
/**
* @param $start
* @param $end
* @param $str
* @return array
* 正则截取函数
*/
static public function preg_substr($start, $end, $str){
$temp = preg_split($start, $str);
$result = [];
foreach ($temp as $index => $value) {
if ($index == 0) {
continue;
}
$content = preg_split($end, $value);
array_push($result, $content[0]);
}
return $result;
}
/**
* 淘宝
* @param $url
* @return array
* @throws Exception
*/
static public function taobao($url){
$text = file_get_contents($url);
if(!$text){
throw new Exception('未抓取到内容，请检查链接是否支持');
}
//商品名称
preg_match('/<title>([^<>]*)<\/title>/', $text, $title);
$title = str_replace('-淘宝网','',iconv("GBK","UTF-8",$title[1]));
$title = trim($title);
//缩略图
preg_match('/<img[^>]*id="J_ImgBooth"[^r]*rc=\"([^"]*)\"[^>]*>/', $text, $img);
$thumb = 'http:'.$img[1];
//价格
preg_match('/<em class="tb-rmb-num">([^<>]*)<\/em>/', $text, $price);
$price = explode('-',$price[1]);
$price = $price[0];
//组图
$needle = "auctionImages :";
$images = strstr($text, $needle);
$images = strstr($images, "},",true);
$images = str_replace($needle,'',$images);
$images = explode(',',$images);
foreach ($images as $key=>$val){
$val = str_replace(" ",'',$val);
$val = str_replace("[",'',$val);
$val = str_replace("]",'',$val);
$val = str_replace('"','',$val);
$images[$key] = 'http:'.$val;
}
//详情
$needle = "descUrl : location.protocol==='http:' ? '";
$descUrl = strstr($text, $needle);
$descUrl = strstr($descUrl, "' : ",true);
$descUrl = str_replace($needle,'',$descUrl);
$desc = file_get_contents('http:'.$descUrl);
$desc = iconv("gbk","UTF-8",$desc);
$desc = str_replace("var desc='",'',$desc);
$desc = str_replace("https:",'http:',$desc);
$desc = str_replace("';",'',$desc);
$desc = str_replace('\\','',$desc);
$desc = preg_replace('/<a .*?href="(.*?)".*?>/is',"<a href='javascript:void(0)'>",$desc);
return [
'title'=>$title,
'thumb'=>$thumb,
'price'=>$price,
'images'=>$images,
'desc'=>$desc
];
}
/**
* 天猫
* @param $url
* @return array
* @throws Exception
*/
static public function tm($url){
$text = file_get_contents($url);
if(!$text){
throw new Exception('未抓取到内容，请检查链接是否支持');
}
//商品名称
preg_match('/<title>([^<>]*)<\/title>/', $text, $title);
$title = str_replace('-tmall.com天猫','',iconv("GBK","UTF-8",$title[1]));
$title = trim($title);
//缩略图
preg_match('/<img[^>]*id="J_ImgBooth"[^r]*rc=\"([^"]*)\"[^>]*>/', $text, $img);
$thumb = 'http:'.$img[1];
//价格
// preg_match('/<em class="tb-rmb-num">([^<>]*)<\/em>/', $text, $price);
// $price = explode('-',$price[1]);
// $price = $price[0];
$price = 0;
//组图
preg_match('|<ul id="J_UlThumb" class="tb-thumb tm-clear">(.*)</ul>|isU',$text, $match);
preg_match_all('/<img src="(.*?)" \//',$match[1],$images);
$images = $images[1];
foreach ($images as $key=>$val){
$val = str_replace('_60x60q90.jpg','',$val);
$images[$key]='http:'.$val;
}
//详情
preg_match_all('/TShop.Setup[(].*[)]?/is', $text, $content);//页面js脚本
$res = $content[0][0];
$pattern = '/\{.*\}/';
preg_match_all($pattern,$res,$mat);
$ap = $mat[0][0];
$pattern1 = '/dsc.taobaocdn.com\/.*fetchDcUrl/';
preg_match($pattern1,$ap,$mat1);
$mat2 = $mat1[0];
$mat3 = substr($mat2,0,-13);
$url = "http://".$mat3;
$goods_detail = file_get_contents($url);
$desc = str_replace("var desc='",'',$goods_detail);
$desc = str_replace("https:",'http:',$desc);
$desc = str_replace("';",'',$desc);
$desc = preg_replace('/<a .*?href="(.*?)".*?>/is',"<a href='javascript:void(0)'>",$desc);
return [
'title'=>$title,
'thumb'=>$thumb,
'price'=>$price,
'images'=>$images,
'desc'=>$desc
];
}
/**
* 抓取京东
* @param $url
* @return array
* @throws Exception
*/
static public function jd($url){
$text = file_get_contents($url);
if(!$text){
throw new Exception('未抓取到内容，请检查链接是否支持');
}
//商品名称
$needle = '<div class="sku-name">';
$title = strstr($text, $needle);
$title = str_replace($needle,'',$title);
$title = strstr(iconv('gbk','UTF-8',$title), "</div>",true);
$title = trim($title);
//缩略图
$needle = 'data-origin="';
$thumb = strstr($text, $needle);
$thumb = str_replace($needle,'',$thumb);
$thumb = strstr($thumb, '"',true);
$thumb = 'http:'.$thumb;
//价格
preg_match("/\/(\d+)*\./i", $url, $match_id);
if(isset($match_id[1]) && $match_id[1]){
$id = $match_id[1];
//获取价格
$price_url = "http://p.3.cn/prices/mgets?skuIds=J_$id,J_&type=1";
$price_content = file_get_contents($price_url);
if(!empty($price_content)){
$price_content_arr = json_decode($price_content,true);
$price = $price_content_arr[0]['p'];
}else{
$price = 0;
}
}else{
throw new Exception('京东商品id获取失败');
}
//组图
$imgHead = explode('_',$thumb);
$imgHead = $imgHead[0];
$needle = "imageList:";
$images = strstr($text, $needle);
$images = strstr($images, "],",true);
$images = str_replace($needle,'',$images);
$images = explode(',',$images);
foreach ($images as $key=>$val){
$val = str_replace(" ",'',$val);
$val = str_replace("[",'',$val);
$val = str_replace("]",'',$val);
$val = str_replace('"','',$val);
$images[$key] = $imgHead.'_'.$val;
}
//详情
$desc = '';
return [
'title'=>$title,
'thumb'=>$thumb,
'price'=>$price,
'images'=>$images,
'desc'=>$desc
];
}
}当您发现内容错误或代码bug，以及下载链接无法使用等，请点击屏幕右下角的上报错误来进行提交，我们会尽快修正。
本程序所有源码和工具完全免费，当本网站内容如果侵犯了您的权益，请联系我们，我们会尽快处理，感谢您的合作。
相关文章
评论：
文明上网理性发言，请遵守 新闻评论服务协议
上报错误
