2019/2/26 backup定向平媒添加+事件规则过滤修改

src/main/java/com/zhiwei/messageflow/DES4RedisStart.java

@@ -32,6 +32,9 @@ public class DES4RedisStart{
 		// 遍历项目
 		for (Project project : projects) {
 			
+//			if(!project.getProjectName().equals("测试"))
+//				continue;
+
 			/**
 			 * 项目是否开启定向监测，并获取定向渠道组
 			 */

src/main/java/com/zhiwei/messageflow/DirectES4RedisTask.java

@@ -50,14 +50,10 @@ public class DirectES4RedisTask {
 	 * 
 	 * @Title: directTask
 	 * @Description: TODO(定向监测任务)
-	 * @param @param
-	 *            project
-	 * @param @param
-	 *            dgList
-	 * @param @param
-	 *            count
-	 * @param @return
-	 *            设定文件
+	 * @param @param project
+	 * @param @param dgList
+	 * @param @param count
+	 * @param @return 设定文件
 	 * @return boolean 返回类型
 	 */
 	@SuppressWarnings("unchecked")
@@ -87,8 +83,8 @@ public class DirectES4RedisTask {
 			List<KeywordNew> keywordNews = keywordNewDao.getDirectKeywordNewByProject(project.getProjectName());
 
 			for (DirectGroup directGroup : dgList) {
-				//判断渠道组是否需要读取
-				if(!directGroup.getIsUsed()||(null==directGroup.getMemberList())) 
+				// 判断渠道组是否需要读取
+				if (!directGroup.getIsUsed() || (null == directGroup.getMemberList()))
 					continue;
 				if (directGroup.getMemberList().isEmpty())
 					continue;
@@ -101,8 +97,10 @@ public class DirectES4RedisTask {
 							directGroup.getName(), pt);
 
 					// 获取 平台-项目-关键词组 对应的rsid作为查询范围的起始点
-					Long startrsid = rsidMap.get(directRedisKey) == null ? -1L
-							: Long.valueOf(rsidMap.get(directRedisKey));
+					Long startrsid = (rsidMap.get(directRedisKey) == null ? -1L
+							: Long.valueOf(rsidMap.get(directRedisKey))) >= ESGetCommonId.START_BACKUPID
+									? Long.valueOf(rsidMap.get(directRedisKey))
+									: ESGetCommonId.START_BACKUPID;
 
 					// 用于存储数据获取后新的rsid
 					Long keyrsid = -1L;

src/main/java/com/zhiwei/messageflow/DirectES4RedisThread.java

@@ -25,7 +25,7 @@ public class DirectES4RedisThread extends Thread {
 	private List<DirectGroup> dgList;
 	// 单个平台单个关键词组每次查询数量
 //	private static final int count = 20;
-	private static final int count = 50;
+	private static final int count = 300;
 
 	private DirectES4RedisTask directES4RedisTask;
 

src/main/java/com/zhiwei/messageflow/ES4RedisRunner.java

@@ -36,6 +36,7 @@ public class ES4RedisRunner implements ApplicationRunner {
 			ESGetCommonId esGetCommonId = ApplicationContextProvider.getBean("ESGetCommonId", ESGetCommonId.class);
 			esGetCommonId.getCommonId();
 			esGetCommonId.getbackupStartId();
+//			esGetCommonId.getStartId();
 
 			// 启动时更新事件等待采集列表中的前十个任务状态为采集完毕
 			boolean isSuccess = esGetCommonId.updateTopTenCollection();
@@ -45,7 +46,7 @@ public class ES4RedisRunner implements ApplicationRunner {
 			 */
 			// 手动注入bean ES4RedisStart
 			ES4RedisStart start = ApplicationContextProvider.getBean("ES4RedisStart", ES4RedisStart.class);
-//			DES4RedisStart directstart = ApplicationContextProvider.getBean("DES4RedisStart", DES4RedisStart.class);
+			DES4RedisStart directstart = ApplicationContextProvider.getBean("DES4RedisStart", DES4RedisStart.class);
 			EVENTAutoMarkStart eventStart = ApplicationContextProvider.getBean("EVENTAutoMarkStart",
 					EVENTAutoMarkStart.class);
 
@@ -57,8 +58,8 @@ public class ES4RedisRunner implements ApplicationRunner {
 						public void run() {
 							try {
 								start.startThread();
-//								directstart.startThread();
-//								eventStart.startThread();
+								directstart.startThread();
+								eventStart.startThread();
 							} catch (Exception e) {
 								log.error("主定时器异常{}{}", e.getMessage(), e.getStackTrace());
 							}

src/main/java/com/zhiwei/messageflow/ES4RedisThreadNew.java

@@ -29,7 +29,7 @@ public class ES4RedisThreadNew extends Thread {
 	private ES4RedisTask es4RedisTask;
 
 	// 单个平台单个关键词组每次查询数量
-	private static final int count = 50;
+	private static final int count = 300;
 //	private static final int count = 10;
 
 	// private static final int max_Thread_num = 40;

src/main/java/com/zhiwei/messageflow/ESGetCommonId.java

@@ -81,8 +81,22 @@ public class ESGetCommonId {
 		int zhihu = esDao.getCommonidByTime("zhihu", nowtime);
 		cidList.add(zhihu);
 		Collections.sort(cidList);
-		// System.err.println(weibo + " " + media + " " + video + " " + zhihu + "\n" +
-		// cidList);
+		START_BACKUPID = cidList.get(0);
+		System.err.println("START_BACKUPID:" + START_BACKUPID + "\tNOW:" + nowtime);
+	}
+
+	public void getStartId() {
+		long nowtime = System.currentTimeMillis() - 12 * 60 * 60 * 1000L;
+		List<Integer> cidList = new ArrayList<>();
+		int weibo = esDao.getCommonidByTime("weibo", nowtime);
+		cidList.add(weibo);
+		int media = esDao.getCommonidByTime("media", nowtime);
+		cidList.add(media);
+		int video = esDao.getCommonidByTime("video", nowtime);
+		cidList.add(video);
+		int zhihu = esDao.getCommonidByTime("zhihu", nowtime);
+		cidList.add(zhihu);
+		Collections.sort(cidList);
 		START_BACKUPID = cidList.get(0);
 		System.err.println("START_BACKUPID:" + START_BACKUPID + "\tNOW:" + nowtime);
 	}

src/main/java/com/zhiwei/messageflow/es/dao/impl/ESDaoImpl.java

@@ -902,7 +902,7 @@ public class ESDaoImpl implements ESDao {
 			sonQueryBuilder = ESQueryUtil.assembleShouldNotFieldsQuery(sonQueryBuilder, specTypeList, null, "user_id");
 		} else {
 			sonQueryBuilder = ESQueryUtil.assembleShouldNotFieldsQuery(sonQueryBuilder, sourceList, null, "source");
-			if (pt.equals("微信")) {
+			if (pt.equals("微信")||pt.equals("平媒")) {
 				// tsuQueryBuilder = ESQueryUtil.assembleShouldNotFieldsQuery(tsuQueryBuilder,
 				// specTypeList, null, "source");
 			} else if (pt.equals("今日头条")) {

src/main/java/com/zhiwei/messageflow/mongo/dao/impl/DirectGroupDaoImpl.java

@@ -2,6 +2,7 @@ package com.zhiwei.messageflow.mongo.dao.impl;
 
 import java.util.ArrayList;
 import java.util.List;
+import java.util.stream.Collectors;
 
 import org.springframework.beans.factory.annotation.Autowired;
 import org.springframework.beans.factory.annotation.Qualifier;
@@ -32,14 +33,15 @@ public class DirectGroupDaoImpl implements DirectGroupDao {
 		List<String> res = new ArrayList<>();
 		try {
 			List<String> idlist = directGroup.getMemberList();
-			for (String id : idlist) {
-				String pt = id.split("_")[0];
-				if(res.isEmpty()) {
-					res.add(pt);
-				}else if(!res.contains(pt)){
-					res.add(pt);
-				}
-			}
+//			for (String id : idlist) {
+//				String pt = id.split("_")[0];
+//				if(res.isEmpty()) {
+//					res.add(pt);
+//				}else if(!res.contains(pt)){
+//					res.add(pt);
+//				}
+//			}
+			res = idlist.stream().map(id->id.split("_")[0]).distinct().collect(Collectors.toList());
 		} catch (Exception e) {
 			return null;
 		}

src/main/java/com/zhiwei/messageflow/service/impl/EventServiceImpl.java

@@ -4,6 +4,7 @@ import java.io.IOException;
 import java.net.MalformedURLException;
 import java.net.URL;
 import java.util.ArrayList;
+import java.util.Arrays;
 import java.util.Date;
 import java.util.HashSet;
 import java.util.List;
@@ -49,6 +50,11 @@ public class EventServiceImpl implements EventService {
 	private static AutomaticMarkClient client = AutomaticMarkClient.getClient(MiddlewareConfig.zookeeperIp);
 	/** 自动标注每次进聚合条数 **/
 	private static final int AUTOMARKPAGELIMIT = 500;
+	/** 事件采集source过滤条件 **/
+	private static final List<String> SOURCENOTRULE = Arrays.asList("主页", "网页", "首页", "平台首页", "早报", "手机版", "null");
+	/** 事件采集url过滤条件 **/
+	private static final List<String> URLNOTRULE = Arrays.asList("zhihu.com", "ali.71zs.com", "v.qq.com", "v.youku.com",
+			"iqiyi.com", "v.pptv.com", "zuanke8.com", "weibo.com", "socialbeta.com", "www.baidu.com/link?url=", "null");
 
 	@Autowired
 	@Qualifier(value = "primaryMongoTemplate")
@@ -75,7 +81,7 @@ public class EventServiceImpl implements EventService {
 			}
 		} else {
 			// 放入缓存
-			if (CanMsgInsert(msgJsonOb)) {
+			if (canMsgInsert(msgJsonOb)) {
 				if (!isUrlIterate(msgJsonOb)) {
 					// url不重复
 					msgJsonOb.put("offset", offset);
@@ -91,10 +97,8 @@ public class EventServiceImpl implements EventService {
 	 * 
 	 * @Title: dropUrlIterateList
 	 * @Description: 采集结束删除urlHash集
-	 * @param @param
-	 *            msgJsonOb
-	 * @param @return
-	 *            设定文件
+	 * @param @param msgJsonOb
+	 * @param @return 设定文件
 	 * @return boolean 返回类型
 	 */
 	private boolean dropUrlIterateList(JSONObject msgJsonOb) {
@@ -112,12 +116,9 @@ public class EventServiceImpl implements EventService {
 	 * 
 	 * @Title: isUrlIterate
 	 * @Description: 判断重复url
-	 * @param @param
-	 *            string
-	 * @param @param
-	 *            id
-	 * @param @return
-	 *            设定文件
+	 * @param @param string
+	 * @param @param id
+	 * @param @return 设定文件
 	 * @return boolean 返回类型
 	 */
 	private boolean isUrlIterate(JSONObject msgJsonOb) {
@@ -133,7 +134,7 @@ public class EventServiceImpl implements EventService {
 				return false;
 			}
 		} catch (Exception e) {
-			log.error("判断重复urlError:",e);
+			log.error("判断重复urlError:", e);
 			return false;
 		}
 	}
@@ -143,16 +144,14 @@ public class EventServiceImpl implements EventService {
 	 * 
 	 * @Title: formatUrl
 	 * @Description: 格式化url
-	 * @param @param
-	 *            string
-	 * @param @return
-	 *            设定文件
+	 * @param @param string
+	 * @param @return 设定文件
 	 * @return String 返回类型
 	 * @throws Exception
 	 */
 	private String formatUrl(String urlStr) throws Exception {
 		URL url = new URL(urlStr);
-		if("http".equals(url.getProtocol())) {
+		if ("http".equals(url.getProtocol())) {
 			urlStr = urlStr.replaceFirst("http://", "https://");
 		}
 		return urlStr;
@@ -163,22 +162,24 @@ public class EventServiceImpl implements EventService {
 	 * 
 	 * @Title: CanMsgInsert
 	 * @Description: 判断是否能进缓存
-	 * @param @param
-	 *            msgJsonOb
-	 * @param @return
-	 *            设定文件
+	 * @param @param msgJsonOb
+	 * @param @return 设定文件
 	 * @return boolean 返回类型
 	 */
-	private boolean CanMsgInsert(JSONObject msgJsonOb) {
+	private boolean canMsgInsert(JSONObject msgJsonOb) {
 		if (msgJsonOb.containsKey("time")) {
 			if (null == msgJsonOb.getLong("time") || -1 == msgJsonOb.getLong("time")) {
 				return false;
 			}
 		}
 		if (msgJsonOb.containsKey("source")) {
-			if (null == msgJsonOb.getString("source") || "百度网页".equals(msgJsonOb.getString("source"))) {
+			if (conformSourceNotRule(msgJsonOb)) {
+				log.info("MSG_source_Error:{}", msgJsonOb);
 				return false;
 			}
+//			if (null == msgJsonOb.getString("source") || "百度网页".equals(msgJsonOb.getString("source"))) {
+//				return false;
+//			}
 		}
 		if (msgJsonOb.containsKey("title")) {
 			if (null == msgJsonOb.getString("title")) {
@@ -193,8 +194,8 @@ public class EventServiceImpl implements EventService {
 			} else if (!isValidUrl(msgJsonOb.getString("url"))) {
 				log.info("MSG_url_Error_notValid:{}", msgJsonOb);
 				return false;
-			} else if (Tools.approximateStringMatching(msgJsonOb.getString("url"),"www.baidu.com/link?url=")) {
-				log.info("MSG_url_Error_www.baidu.com/link?url=:{}", msgJsonOb);
+			} else if (conformUrlNotRule(msgJsonOb)) {
+				log.info("MSG_url_Error:{}", msgJsonOb);
 				return false;
 			}
 		}
@@ -207,6 +208,21 @@ public class EventServiceImpl implements EventService {
 		return true;
 	}
 
+	private boolean conformUrlNotRule(JSONObject msgJsonOb) {
+		return URLNOTRULE.stream().anyMatch(rule -> msgJsonOb.getString("url").indexOf(rule) != -1);
+	}
+
+	private boolean conformSourceNotRule(JSONObject msgJsonOb) {
+		String source = null == msgJsonOb.getString("source") ? "null" : msgJsonOb.getString("source");
+		if (SOURCENOTRULE.contains(source) || source.startsWith("www.") || source.endsWith(".com")
+				|| source.endsWith(".cn") || -1 != source.indexOf("<img class=") || -1 != source.indexOf("【快资讯】")
+				|| -1 != source.indexOf("「热点新闻」")) {
+			return true;
+		} else {
+			return false;
+		}
+	}
+
 	/**
 	 * 校验url合法性
 	 *

src/main/java/com/zhiwei/messageflow/util/ESQueryUtil.java

@@ -895,6 +895,9 @@ public class ESQueryUtil {
 			list.add("微信");
 			list.add("微信*");
 			break;
+		case "平媒":
+			list.add("平媒");
+			break;
 		default:
 			list.add("网媒");
 			list.add("*新闻");

src/main/resources/redis.properties

@@ -5,7 +5,8 @@ redis.testOnBorrow=true
 redis.testOnReturn=true
 redis.ip = 192.168.0.202
 redis.port=6380
-#redis.ip = 202.107.192.94
+#redis.ip=202.107.192.94
+#redis.port=6479
 #redis.ip=127.0.0.1
 #redis.port=6379
 #redis.ip=192.168.1.74