提交 019f8b31 作者: liuweigang

采集代码更新6

上级 fb9e509a
package com.zzsn.search; package com.zzsn.search;
import cn.hutool.core.date.DateUtil;
import com.fasterxml.jackson.databind.ObjectMapper; import com.fasterxml.jackson.databind.ObjectMapper;
import com.zzsn.cache.JedisUtil; import com.zzsn.cache.JedisUtil;
import com.zzsn.cache.MemcachedUtils; import com.zzsn.cache.MemcachedUtils;
...@@ -649,6 +650,8 @@ public class MetaBaiduSearchThread implements Runnable { ...@@ -649,6 +650,8 @@ public class MetaBaiduSearchThread implements Runnable {
clbAnsProcessitem.setOrigin(docInfo.getOrigin()); clbAnsProcessitem.setOrigin(docInfo.getOrigin());
clbAnsProcessitem.setPublishDate(docInfo.getPublishDate()); clbAnsProcessitem.setPublishDate(docInfo.getPublishDate());
clbAnsProcessitem.setSourceAddress(docInfo.getSourceaddress()); clbAnsProcessitem.setSourceAddress(docInfo.getSourceaddress());
String now = DateUtil.now();
clbAnsProcessitem.setCreateDate(now);
clbAnsProcessitem.setSource("3"); clbAnsProcessitem.setSource("3");
return clbAnsProcessitem; return clbAnsProcessitem;
} }
......
...@@ -337,7 +337,7 @@ public class RecorderUtil { ...@@ -337,7 +337,7 @@ public class RecorderUtil {
// docstr = seleniumTime.getScopehtmlByProxy(urlList.get(i)); // docstr = seleniumTime.getScopehtmlByProxy(urlList.get(i));
if(StringUtils.isEmpty(docstr)){ if(StringUtils.isEmpty(docstr)){
try { try {
Thread.sleep(1000*5); Thread.sleep(1000*2);
SeleniumTime seleniumTime2=new SeleniumTime(); SeleniumTime seleniumTime2=new SeleniumTime();
docstr=seleniumTime2.getScopehtml(urlList.get(i)); docstr=seleniumTime2.getScopehtml(urlList.get(i));
} catch (InterruptedException e) { } catch (InterruptedException e) {
......
...@@ -90,14 +90,14 @@ public class MetaSouGouSearchThread implements Runnable { ...@@ -90,14 +90,14 @@ public class MetaSouGouSearchThread implements Runnable {
return; return;
} }
for (String kWord :keyWords) { for (String kWord :keyWords) {
try { // try {
boolean sismember = JedisUtil.sismember(cache_key, kWord); // boolean sismember = JedisUtil.sismember(cache_key, kWord);
if (sismember) { // if (sismember) {
continue; // continue;
} // }
}catch (Exception e){ // }catch (Exception e){
log.info("缓存出问题"); // log.info("缓存出问题");
} // }
String url1= Constants.META_SEARCH_URL; String url1= Constants.META_SEARCH_URL;
// String[] kwords=kWord.split("\\+"); // String[] kwords=kWord.split("\\+");
String url=""; String url="";
...@@ -362,6 +362,7 @@ public class MetaSouGouSearchThread implements Runnable { ...@@ -362,6 +362,7 @@ public class MetaSouGouSearchThread implements Runnable {
// 加入缓存池中 // 加入缓存池中
System.out.println("加入缓存池"); System.out.println("加入缓存池");
// JedisUtil.sadd(rediskey, cwbm.getSourceaddress());
// addressList.add(docInfo.getSourceaddress()); // addressList.add(docInfo.getSourceaddress());
// MemcachedUtils.set(key,addressList,60*60*24); // MemcachedUtils.set(key,addressList,60*60*24);
JedisUtil.setString(Constants.SOURCEADDRESS+"_"+orgId+"_"+cwbm.getSourceaddress(),"1",-1); JedisUtil.setString(Constants.SOURCEADDRESS+"_"+orgId+"_"+cwbm.getSourceaddress(),"1",-1);
......
Markdown 格式
0%
您添加了 0 到此讨论。请谨慎行事。
请先完成此评论的编辑!
注册 或者 后发表评论