About es aggregation query index filtering, and the number of accurate aggregation results returned

[NiFeng]

 First of all, I want to accurately control the amount of large data and the number of results under multidimensional aggregation statistics. Second, I would also like to filter the results of the polymerization, similar to the having in the mysql.
After reading the official document, I found that composite can control the amount of data returned by multi-dimensional aggregate statistics. But after setting it up the aggregation filter (bucket_selector), the amount of data returned by the query is wrong! Originally, it is to aggregate query results according to the size return, and then index aggregation results filtering, this is not right!
I haven't been able to find the right implementation for a long time. I' m using the 6.3.2 version es. Hope someone can help! Thank you!

[NiFeng]

First of all, I want to accurately control the amount of large data and the number of results under multidimensional aggregation statistics.
Second, I would also like to filter the results of the polymerization, similar to the having in the mysql.
After reading the official document, I found that composite can control the amount of data returned by multi-dimensional aggregate statistics.
But after setting it up the aggregation filter (bucket_selector), the amount of data returned by the query is wrong!
Originally, it is to aggregate query results according to the size return, and then index aggregation results filtering, this is not right!
I haven't been able to find the right implementation for a long time. I' m using the 6.3.2 version es.
Hope someone can help! Thank you!