suggest add selectExpr method to DataFrame

[lyhue1991]

In pyspark, selectExpr of DataFrame is very convenient and readable. But there is no such method in odps DataFrame.

There is only select method.

I wrote a selectExpr by first persist the input DataFrame and then use the o.execute_sql to do the select purpose.

But this realization is too slow , Can you give an official fast realization and support Lazy Evaluation？

from odps import ODPS 
from uuid import uuid1 

o = ODPS(xxx)
    
def selectExpr(self, df, *exprs): 
    posfix = str(uuid1()).replace('-','_')
    temp_table = 'temp_data_'+ posfix
    df.persist(temp_table,lifecycle=1)
    sql = 'select ' + ','.join(exprs) + ' from ' + temp_table

    posfix = str(uuid1()).replace('-','_')
    temp_table = 'temp_data_'+ posfix
    o.execute_sql(f'CREATE TABLE {temp_table} lifecycle 1 AS {sql}')  
    dfout = o.get_table(temp_table).to_df()
    return dfout

[wjsi]

1. In the context of pandas DataFrame eco, the equivalent method is df.query;
2. We are now deprecating PyODPS DataFrame by proposing MaxFrame, and we are not adding new features to PyODPS DataFrame.