crowdflower 0.0.10

CrowdFlower API - Python client

Client library for interacting with the CrowdFlower API with Python.

Installation

Install from PyPI:

easy_install -U crowdflower

Or install the latest version GitHub:

git clone https://github.com/chbrown/crowdflower.git
cd crowdflower
python setup.py develop

Import examples

Import:

import crowdflower

CrowdFlower API keys are 20 characters long; the one below is just random characters.

conn = crowdflower.Connection('LbcxvIlE3x1M8F6TT5hN')

The library will default to an environment variable called CROWDFLOWER_API_KEY if none is specified here:

conn = crowdflower.Connection()

If you want to cache job responses, like judgments, properties, and tags, you can initialize the connection with a cache. cache='filesystem' is the only option currently supported, and serializes JSON files to /tmp/crowdflower/*.

conn = crowdflower.Connection(cache='filesystem')

More examples

Loop through all your jobs and print the titles:

for job in conn.jobs():
    print job.properties['title']

Create a new job with some new units:

data = [
    {'id': '1', 'name': 'Chris Narenz', 'gender_gold': 'male'},
    {'id': '2', 'name': 'George Henckels'},
    {'id': '3', 'name': 'Maisy Ester'},
]
job = conn.upload(data)
job.update({
    'title': 'Gender labels',
    'included_countries': ['US', 'GB'],  # Limit to the USA and United Kingdom
    'payment_cents': 5,
    'judgments_per_unit': 2,
    'instructions': <some instructions html>,
    'cml': <some layout cml>,
    'options': {
        'front_load': 1, # quiz mode = 1; turn off with 0
    }
})
job.gold_add('gender', 'gender_gold')

print job

Motivation

The official Ruby client is hard to use, which is surprising, since the CrowdFlower API is so simple.

Which is not to say the CrowdFlower API is all ponies and rainbows, but all the documentation is there on one page, and it does what it says, for the most part. (Though there’s more that you can do, beyond what’s documented.)

Thus, a thin Python client for the CrowdFlower API.

References

This package uses kennethreitz’s Requests to communicate with the CrowdFlower API over HTTP. Requests is Apache2 licensed.

License

Copyright © 2014 Christopher Brown. MIT Licensed.