condition 1.0.0

A pythonic way to construct conditions for pandas dataframe query and sql

Condition

Project Description

This project provides a pythonic way to construct a condition object. The condition object can later used to query pandas Dataframe, filter pyarrow partitions or to generate where conditions in SQL. It takes care of formating and syntax for you.

A Quick Example

>>> from condition import *
>>> import pandas as pd
>>> fl = FieldList('A B C date value'.split())
>>> cond = ((fl.date >= pd.to_datetime('20000101')) 
...             & (fl.date <= pd.to_datetime('20010131'))
...             & (fl.A == 'a1 a3'.split()) \
...             & (fl.B == 'b1')) \
...             | (fl.C != 'c3 c5'.split())
>>> cond # show as sql where condition

        (
                (date >= '2000-01-01 00:00:00'
                and date <= '2001-01-31 00:00:00'
                and A in ('a1','a3')
                and B = 'b1')
        or C not in ('c3','c5'))

>>> cond.to_df_query()
"(((date >= '2000-01-01 00:00:00')&(date <= '2001-01-31 00:00:00')&(A in ('a1','a3'))&(B == 'b1'))|(C not in ('c3','c5')))"
>>> cond.to_pyarrow_filter()
[[('date', '>=', Timestamp('2000-01-01 00:00:00')),
  ('date', '<=', Timestamp('2001-01-31 00:00:00')),
  ('A', 'in', {'a1', 'a3'}),
  ('B', '=', 'b1')],
 [('C', 'not in', {'c3', 'c5'})]]

SQL Generation

>>> from condition.sql import render_sql
>>> sql = """
...     select *
...     from my_table
...     where {{where_condition}}
... """
>>> print(render_sql(sql, cond))

    select *
    from my_table
    where 
        (
                (date >= '2000-01-01 00:00:00'
                and date <= '2001-01-31 00:00:00'
                and A in ('a1','a3')
                and B = 'b1')
        or C not in ('c3','c5'))

Please see Usage Examples for detailed examples.

Installation

This project is distributed via pip. To get started:

pip install condition

To install jinja2 package used by condition.sql, do the following

pip install condition[sql]

To install all packages for development, do the following

pip install condition[dev]