Welcome to Flupy

flupy is a lightweight library and CLI for implementing python data pipelines with a fluent interface.

Under the hood, flupy is built on generators. That means its pipelines evaluate lazily and use a constant amount of memory no matter how much data are being processed. This allows flupy to tackle Petabyte scale data manipulation as easily as it operates on a small list.


import json
from flupy import flu

logs = open('logs.jl', 'r')

error_count = flu(logs).map(lambda x: json.loads(x)) \
                       .filter(lambda x: x['level'] == 'ERROR')

# 14


The flupy library, and python runtime, are also accessible from flu command line utility:

$ cat logs.txt | flu "_.filter(lambda x: x.startswith('ERROR'))"

For more information about the flu command see command line.

Getting Started


Python 3.6+


$ pip install flupy


Since 2008, what domains are our customers comming from?:

from flupy import flu

customers = [
    {'name': 'Jane', 'signup_year': 2018, 'email': 'jane@ibm.com'},
    {'name': 'Fred', 'signup_year': 2011, 'email': 'fred@google.com'},
    {'name': 'Lisa', 'signup_year': 2014, 'email': 'jane@ibm.com'},
    {'name': 'Jack', 'signup_year': 2007, 'email': 'jane@apple.com'},

pipeline = flu(customers).filter(lambda x: x['signup_year'] > 2008) \
                         .map_item('email') \
                         .map(lambda x: x.partition('@')[2]) \
                         .group_by() \ # defaults to identity
                         .map(lambda x: (x[0], x[1].count())) \

# [('google.com', 1), ('ibm.com', 2)]