Written by Trent Hauck.
Provides utilities and functions for managing data projects in python. Requires use of IPython and Pandas.
A quick workflow example::
from gloo import Gloo proj = Gloo('My Project', full_structure, packages=['scipy', ('numpy', 'np')]) proj.create_project() #Now say you've done some work, added some data, and munge files. And come #back the next day, fire up IPython in the project directory. proj = Gloo('My Project') proj.load_project() ##the packages will be available, and any data you put in the data folder #will be available as pandas DataFrames and those packages defined above
Gloo's goal is to tie together a lot of the data analysis actions that happen regularly and make that processes easy. Automatically loading data into the ipython environment, running scripts, making utitlity functions available and more. These are things that have to be done often, but aren't the fun part.
project_name: This is a string that is the name of your project.
Current Config Options:
full_structure If True the folder structure outline below. By default
creates smaller project, i.e. False.
packages A list of strings of python packages to load when
load_project() is called. Defaults to empty. If you want to alias your
package you can pass a tuple to the list.
['scipy', ('numpy', 'np')]
will import scipy as scipy and numpy as np.
logging A boolean to dictate if logging is started when
load_project() is called. Defaults to False.
svn Pass a list or a string to init version control. Currently supports
git and bzr.
svn = ['git', 'bzr'] will init both.
Those options are saved into a pickled file called .gloo at the root of the project directory.
datadirectory is loaded into the environment. This is done recursively so you can have subdirectories. If you do, the parent folder of the data file will be prepended to data file,
folder_file. The plan is to make the prepending optional.
mungedirectory are run. This folder is where you would put files necessary for preprocessing the data.
libdirectory are imported. This folder is where you would put files that you would like to load as a module. So if you have utility.py in the lib directory. When you load the project you'll have utility availble in the namespace.
The full structure is as follows::
data/ : data doc/ : documentation diagnostics/ : automatically check for data issues graphs/ : graph domicile lib/ : utility functions munge/ : preprocessing scripts profiling/ : benchmark performance reports/ : reports you'll produce tests/ : tests
You can update the config. Say you have
packages = ['numpy'] but once
you've worked on the project you realize you need pandas and you want to load
it as pd. It's easy to update this of the future::
> proj.packages ['numpy'] > proj.packages.append(('pandas', 'pd')) > proj.save_project()
So next time you load the project pandas as pd will be available.
pip install Gloois available.
Because this project is in such an early state I would love for anybody and everybody to help contribute. I think this could be very valuable for those working with python for data projets.
This project is a bit of a rip-off or port (however nice you're feeling) of Project Template, which ifyou're using R I would highly recommend. It's fantastic.