Looker SDK

The Looker SDK for Python provides a convenient way to communicate with the Looker API available on your Looker server. The library requires python3.6+ and is annotated using the typing module.

The SDK uses a plug-in architecture (also known as dependency injection) for initializing that supports run-time specific transports (currently only RequestsTransport) and different approaches for managing API authentication (AuthSession and OAuthSession). The methods and models are generated from the Looker API spec by a new code generator developed at Looker.

Please report any issues encountered, and indicate the SDK language in the report.

Basic Usage

import looker_sdk

# For this to work you must either have set environment variables or created a looker.ini as described below in "Configuring the SDK"
sdk = looker_sdk.init40()  # or init31() for the older v3.1 API
my_user = sdk.me()

# output can be treated like a dictionary
print(my_user["first_name"])
# or a model instance (User in this case)
print(my_user.first_name)

# input methods can take either model instances like WriteUser
sdk.create_user(
    body=looker_sdk.models.WriteUser(first_name="Jane", last_name="Doe")
)
# or plain dictionaries
sdk.create_user(body={"first_name": "Jane", "last_name": "Doe"})

Full tutorial

Go from installation all the way to creating a functional micro-application in this 20-30 minute interactive tutorial.

This tutorial is hosted in Google Colaboratory, an interactive online notebook. You can follow along right in the notebook by clicking the button below.

Sample project setup

Install python 3.8. We highly recommend using pyenv to install different versions of python. Mac users should use homebrew to install pyenv:

brew install pyenv

Linux users should use

curl https://pyenv.run | bash

Put this in your shell configuration script:

export PATH=$PATH:$HOME/.pyenv/bin
eval "$(pyenv init --path)"
eval "$(pyenv init -)"
eval "$(pyenv virtualenv-init -)"

Follow the remaining steps 3 - 5 of https://github.com/pyenv/pyenv#basic-github-checkout otherwise your python3.8 installation may break.

Now you're ready to install python 3.8:

pyenv install 3.8.2

We'll use pipenv (fantastic virtualenv manager) to manage project dependencies.

brew install pipenv

Create a project directory

mkdir looker-sdk-example

Set python3.8 as the base interpreter for this directory

cd looker-sdk-example/
pyenv local 3.8.2

Install looker_sdk using pipenv

pipenv --python 3.8.2 install --pre looker_sdk

Configuring the SDK

The SDK supports configuration through

an .ini file on disk
setting environment variables
providing your own implementation of the ApiSettings class

. The latter override the former.

Note: The .ini configuration for the Looker SDK is a sample implementation intended to speed up the initial development of python applications using the Looker API. See this note on Securing your SDK Credentials for warnings about using .ini files that contain your API credentials in a source code repository or production environment.

In order to configure the SDK client, create a "looker.ini" file to reference during client.setup()

example file:

[Looker]
# Base URL for API. Do not include /api/* in the url. If hosted on GCP, remove the :19999 leaving just https://your.cloud.looker.com
base_url=https://your.looker.com:19999
# API 3 client id
client_id=YourClientID
# API 3 client secret
client_secret=YourClientSecret
# Set to false if testing locally against self-signed certs. Otherwise leave True
verify_ssl=True

Note: If the application using the Looker SDK is going to be committed to a version control system, be sure to ignore the looker.ini file so the API credentials aren't unintentionally published.

For any .ini setting you can use an environment variable instead. It takes the form of LOOKERSDK_<UPPERCASE-SETTING-FROM-INI> e.g. LOOKERSDK_CLIENT_SECRET

A final option is to provide your own implementation of the ApiSettings class. It is easiest to subclass api_settings.ApiSettings and override the read_config function (don't forget a call to super().read_config() if appropriate, Example below). However, at a minimum your class must implement the api_settings.PApiSettings protocol.

import os
import looker_sdk
from looker_sdk import api_settings

class MyApiSettings(api_settings.ApiSettings):
    def __init__(self, *args, **kw_args):
        self.my_var = kw_args.pop("my_var")
        super().__init__(*args, **kw_args)

    def read_config(self) -> api_settings.SettingsConfig:
        config = super().read_config()
        # See api_settings.SettingsConfig for required fields
        if self.my_var == "foo":
            config["client_id"] = os.getenv("FOO_CLIENT")
            config["client_secret"] = os.getenv("FOO_SECRET")
        else:
            config["client_id"] = os.getenv("BAR_CLIENT")
            config["client_secret"] = os.getenv("BAR_SECRET")
        return config

sdk = looker_sdk.init40(config_settings=MyApiSettings(my_var="foo"))
...

Code example

See many python sdk examples in our examples repo

Changelog

Located in our github repo