How can I combine text2sql with python code generation to reduce memory usage

[XJTU-JP]

🚀 The feature

Great job! I find that in the case of a large amount of data in the database, it doesn't seem like a good idea to select all the data directly to the agent. If you combine sql and python code generation. Let some of the aggregation and filtering be done in sql, and then send the filtered data to python code as a dataframe. If you do this, you can certainly save a lot of memory (in the case of large amounts of data) :

Motivation, pitch

I am studying the use of text2sql and python code generation for data analysis under massive data

Alternatives

No response

Additional context

No response