搜狗微信热搜采集程序更新

src/main/java/com/zhiwei/searchhotcrawler/crawler/SougoHotSearchCrawler.java

@@ -40,10 +40,10 @@ public class SougoHotSearchCrawler {
 	 * @return void 返回类型
 	 */
 	public static List<HotSearchList> sougoHotSearch(Date date) {
-		String url = "https://weixin.sogou.com";
+		//String url = "https://weixin.sogou.com";
+		String url = "https://weixin.sogou.com/pcindex/pc/web/web.js?";
 		List<HotSearchList> list = new ArrayList<>();
-		Map<String,String> headMap = HeaderTool.getCommonHead();
-		Request request = RequestUtils.wrapGet(url, headMap);
+		Request request = RequestUtils.wrapGet(url);
 		for (int i = 0; i < 3; i++) {
 			String htmlBody = null;
 			Response response = httpBoot.syncCall(request, ProxySupplier.NAT_HEAVY_PROXY);
@@ -55,35 +55,15 @@ public class SougoHotSearchCrawler {
 			}
 			if (htmlBody != null && htmlBody.contains("topwords")) {
 				try {
-					Document document = Jsoup.parse(htmlBody);
-					Elements elements = document.select("ol#topwords").select("li");
-					for (Element element : elements) {
-						try {
-							// 获取排名rank
-							String rankStr = null;
-							if (!element.select("li").select("i").isEmpty()) {
-								rankStr = element.select("li").select("i").text();
-							}
-							Integer rank = null;
-							if (StringUtils.isNoneBlank(rankStr)) {
-								rank = Integer.valueOf(rankStr);
-							}
-
-							// 获取关键词(String)
-							String kw = element.select("li").select("a").attr("title");
-//								log.info("关键词:{}", kw);
-
-							String everurl = element.select("li").select("a").attr("href");
-
-							HotSearchList hotSearch = new HotSearchList(everurl, kw, null, rank, HotSearchType.搜狗微信热搜.name(),date);
-
-							if (Objects.nonNull(rank)) {
-								list.add(hotSearch);
-							}
-
-						} catch (Exception e) {
-							log.error("解析搜狗微信时出现解析错误", e);
-						}
+					int rank = 0;
+					JSONArray jsonArray = JSONObject.parseObject(htmlBody).getJSONArray("topwords");
+					for (Object object : jsonArray) {
+						rank++;
+						JSONObject json = (JSONObject)JSONObject.toJSON(object);
+						String kw = json.getString("word");
+						String everurl="https://weixin.sogou.com/weixin?type=2&query="+kw;
+						HotSearchList hotSearch = new HotSearchList(everurl, kw, null, rank, HotSearchType.搜狗微信热搜.name(),date);
+						list.add(hotSearch);
 					}
 					break;
 				} catch (Exception e) {