微信真实链接相关测试及获取snuid

src/main/java/com/zhiwei/wechat/search/WechatAritcleSearch.java

@@ -473,16 +473,11 @@ public class WechatAritcleSearch {
 			headerMap.put("Sec-Fetch-User", "?1");
 			headerMap.put("Accept", "text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3");
 			headerMap.put("Sec-Fetch-Site", "same-origin");
-			headerMap.put("Upgrade-Insecure-Requests", "1");
-			headerMap.put("Connection", "keep-alive");
 			headerMap.put("Host", "weixin.sogou.com");
-			headerMap.put("Accept-Encoding", "gzip, deflate, br");
-			headerMap.put("Accept-Language", "zh-CN,zh;q=0.9");
-			headerMap.put("Cache-Control", "no-cache");
 			headerMap.put("Referer", searchUrl);
 			headerMap.put("Cookie", cookie);
 
-			String htmlBody = httpBoot.syncCall(RequestUtils.wrapGet(originalUrl, headerMap)).body().string();
+			String htmlBody = httpBoot.syncCall(RequestUtils.wrapGet(originalUrl, headerMap), ProxyHolder.NAT_HEAVY_PROXY).body().string();
 			if(StringUtils.isNotBlank(htmlBody) ){
 				StringBuilder furl = new StringBuilder();
 				Pattern pa1 = Pattern.compile("url \\+= \'(.*?)\';");
@@ -496,6 +491,31 @@ public class WechatAritcleSearch {
 		return null;
 	}
 
+
+
+	public static String getRealUrlCookie(String originalUrl) throws Exception{
+		String word = originalUrl.split("query=")[1];
+		String searchUrl = "https://weixin.sogou.com/weixin?type=2&s_from=input&query="+ word +"&ie=utf8&_sug_=n&_sug_type_=&page=1";
+		Map<String,String> headerMap = HeaderTool.getCommonHead();
+		headerMap.put("Host", "weixin.sogou.com");
+		headerMap.put("Referer", searchUrl);
+		String cookie = "";
+		String htmlBody = httpBoot.syncCall(RequestUtils.wrapGet(searchUrl, headerMap),ProxyHolder.NAT_HEAVY_PROXY, false).body().string();
+		if(StringUtils.isNotBlank(htmlBody) &&  htmlBody.contains("snuid")){
+			StringBuilder furl = new StringBuilder();
+			Pattern pa1 = Pattern.compile("\"snuid\"         : \"(.*?)\",");
+			Matcher ma1 = pa1.matcher(htmlBody);
+			while (ma1.find()) {
+				furl.append(ma1.group(1));
+			}
+			return  "SNUID=" + furl.toString();
+		}
+		return null;
+	}
+
+
+
+
 	/**
 	 * @Title: getOpenId
 	 * @Description: 获取微信wxID

src/test/java/com/zhiwei/wechat/example/WechatSearchExample.java

@@ -14,6 +14,7 @@ import java.util.regex.Pattern;
 
 import com.zhiwei.crawler.core.HttpBoot;
 import com.zhiwei.crawler.utils.RequestUtils;
+import com.zhiwei.tools.tools.ZhiWeiTools;
 import org.apache.commons.lang3.StringUtils;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
@@ -41,12 +42,41 @@ public class WechatSearchExample{
 	public static void main(String[] args) {
 		ProxyFactory.init(registry, group, GroupType.PROVIDER,10000018);
         proxy = ProxyHolder.SOUGOU_INNER_PROXY.getProxy();
-
-		try {
-			WechatSearchExample.wechatSearchExample();
-		} catch (UnknownHostException e) {
+		String url = "https://weixin.sogou.com/link?url=dn9a_-gY295K0Rci_xozVXfdMkSQTLW6cwJThYulHEtVjXrGTiVgS8CMDfv9wh9qo5s-_tRRSYjdmlThuLl1UVqXa8Fplpd9fM3bn57YTm93DADHLmE53r3LNleAx90O6EdWFlMiLgABVb5FRuhnTbO_GzJrBhvROBdUYdPJ-HwpjtEvi_VZoCFXsP2Y8IMUHXuJCM5s6KSDiXUXG84daahQ5c0PemEIN_1vJiPn8w7tlTjPfiB-Z3QxFXEDiCN7KNRVfuxuX7N535pjGOOjYg..&type=2&query=%E8%85%BE%E8%AE%AF";
+		try{
+			String cookie = WechatAritcleSearch.getRealUrlCookie(url);
+			if(StringUtils.isNotBlank(cookie)){
+				System.out.println("cookie============="+cookie);
+				boolean f = true;
+				int i = 0;
+				while(f){
+					try{
+						String link = WechatAritcleSearch.getRealLink(url, cookie);
+						if(StringUtils.isNotBlank(link) && link.contains("s?src=")){
+							System.out.println(i+++"=========="+link);
+						}else{
+							System.out.println(i+++"=========="+link);
+						}
+						ZhiWeiTools.sleep(3000);
+					}catch (Exception e){
+						ZhiWeiTools.sleep(50);
+						e.printStackTrace();
+					}
+				}
+			}else {
+				System.out.println("cookie============="+cookie);
+			}
+		}catch (Exception e){
 			e.printStackTrace();
 		}
+
+
+
+//		try {
+//			WechatSearchExample.wechatSearchExample();
+//		} catch (UnknownHostException e) {
+//			e.printStackTrace();
+//		}
 	}
 
 
@@ -57,7 +87,7 @@ public class WechatSearchExample{
 		for(String word : wordList)
 		{
 			try {
-				List<WechatAricle> list = WechatAritcleSearch.wechatKeywordSearch(word, 5, null,"2019-10-28", "2019-10-28",proxy, 2);
+				List<WechatAricle> list = WechatAritcleSearch.wechatKeywordSearch(word, 5, null,"2019-10-28", "2019-10-28",proxy, 3);
 				System.out.println("======"+list.size());
 				for(WechatAricle wechat : list){
 					System.out.println(wechat.getId());