数据库表描述:
Table Name:NewsFromWeb
Column Name 1:_id
Column Name 2:url
Column Name 3:title
Other Columns ....
....
功能实现的前提:如果数据库表当中有相同的url,则认定是重复的记录.
功能实现的思路:通过Group by一样的查询,找到有相同的记录,做出列表,并且把记录的第二条以后的数据,进行删除,只保留第一条的记录.
从SQL实现上类似:select sum(url) as urlCount from NewsFromWeb group by url.
如下是实现方法:
@Test
public void checkRepeat(){
//first group by key
BasicDBObject key = new BasicDBObject();
key.put("url", true);
BasicDBObject initial = new BasicDBObject();
initial.put("urlCount", 0);
BasicDBObject cond = new BasicDBObject();
cond.put("url", 1);
String reduce = "function(obj,prev) { prev.urlCount++; }";
DBObject objMap = mongoDAO.getCollection().group(key, new BasicDBObject(), initial, reduce);
Set set = objMap.keySet();
Iterator it = set.iterator();
while(it.hasNext()){
String str_key = (String) it.next();
String recordStr = objMap.get(str_key).toString();
DBObject doc = (DBObject) JSON.parse(recordStr);
if (Double.valueOf(doc.get("urlCount").toString())>1){
BasicDBObject condDel = new BasicDBObject();
cond.put("url", doc.get("url").toString());
DBCursor map = mongoDAO.getCollection().find(cond);
int j=0;
while(map.hasNext()){
DBObject obj = map.next();
if (j>0){
System.out.println("正在删除:"+obj);
mongoDAO.getCollection().remove(obj);
}
j++;
}
}
}
}
本人通过测试10万条数据处理大约2分钟左右,只是一个大概的数据,其实跟硬件和重复记录有很大关系,感觉速度还能接受(CPU老赛扬1.6,Window环境)