Python3 Client for overfitted.io's API

This is a small Python3 client which permits fast interaction with services hosted on overfitted.io (e.g., OCR, document parsing, etc.). It implements all the REST-based communication via the requests package and employs parallel connections via ThreadPoolExecutor.

The module automatically queries overfitted.io for a list of services and their specifications (parameters, endpoint, etc.); this leaves space for modifications or additions of new services on the server's end without requiring changes in the python client.

Installation

The fastest installation method is through pip3 as follows:

pip3 install git+https://github.com/overfitted-io/python-api-client.git

To uninstall:

pip3 uninstall overfitted-io-client

Usage

1. Define the environment variable OVERFITTED_IO_API_KEY and add your API key; if you don't have one, see the Get Started section.

# TODO: replace this
export OVERFITTED_IO_API_KEY=123456789abcdef

2. Import the module and test the connection

import overfitted_io_client as oc

# should display your account information and API key
print(oc.get_account_info([{ 'api_key' : oc.config.get_api_key() }]))

3. If previous command succeeded, you can use the available services. E.g., for performing OCR on a given image:

import overfitted_io_client as oc

img = open('my_img.jpg', 'rb')

result = oc.query_service('glyph', inputs = [{ 'img' : img, 'lang' : 'en-ma', 'api_key' : oc.config.get_api_key()}])

# [({ "text": "Some random text found within the image", ...}, 200)]
print(result)

Explanation: the query_service() method is used to access all hosted services; in this example, OCR is desired, therefore the OCR service (named glyph) is called; the inputs parameter takes a list of 1 element (1 task) in this scenario: to run OCR on the image provided as img using the language code en-ma (English Modern Antiqua) and as api_key the key you supplied in the environment variable at the previous step. the result is also a list which matches the order of the inputs and, in this case, contains 1 element (1 tuple) with the format (json_response, response_status_code).

For a list of parameters names (inside the inputs variable), check each service's documentation on overfitted.io

Note: providing a list of input images to the query_service() function enables the use of multiple parallel connections which facilitates better throughput.

Problems / Bugs

Please open a GitHub issue if there are any problems with this module. Also, take into account that the services on overfitted.io are in active development; implementing feedback from collaborators can introduce new parameters or modifications in the response's format.