About Domino
Domino Data LabKnowledge BaseData Science BlogTraining

The python-domino library

Python bindings for the Domino API.

Permits interaction with a Domino deployment from Python using the Domino API.

The latest released version is 1.0.2.

Installation

At this time, these Domino Python bindings are not in PyPi. You can install the latest version of this package from our Github master branch with the following:

pip install https://github.com/dominodatalab/python-domino/archive/master.zip

If you are adding install instructions for python-domino to your Domino Environment Dockerfile Instructions field, you must add RUN to the beginning:

RUN pip install https://github.com/dominodatalab/python-domino/archive/master.zip

You can also add python-domino to your requirements.txt file with the following syntax:

-f git+git://github.com/dominodatalab/python-domino.git
Note
 To install lower version of library, for example 0.3.5 use the following command: 
pip install https://github.com/dominodatalab/python-domino/archive/0.3.5.zip

Overview

Setting up the connection

You can set up the connection by creating new instance of Domino.

class Domino(project, api_key=None, host=None, domino_token_file=None)

The parameters are:

  • project: A project identifier (in the form of ownerusername/projectname)

  • api_key: (Optional) An API key to authenticate with. If not provided the library will expect to find one in the DOMINO_USER_API_KEY environment variable.

  • host: (Optional) A host URL. If not provided the library will expect to find one in the DOMINO_API_HOST environment variable.

  • domino_token_file: (Optional) Path to domino token file containing auth token. If not provided the library will expect to find one in the DOMINO_TOKEN_FILE environment variable.

Note:

  1. In case both api_key and domino_token_file are available, then preference will be given to domino_token_file.

  2. By default the log level is set to INFO, to set log level to DEBUG, set DOMINO_LOG_LEVEL environment variable to DEBUG

Methods

project_create(project_name, owner_username=None):

Create a new project with given project name. The parameters are:

  • project_name: The name of the project

  • owner_username: (Optional) The owner username for the project. This parameter is useful in-case project needs to be created under some organization.

runs_list()

List the runs on the selected project.

runs_start(command, isDirect, commitId, title, tier, publishApiEndpoint)

Start a new run on the selected project. The parameters are:

  • command: The command to run as an array of strings where members of the array represent arguments of the command. E.g. ["main.py", "hi mom"]

  • isDirect: (Optional) Whether or not this command should be passed directly to a shell.

  • commitId: (Optional) The commitId to launch from. If not provided, will launch from latest commit.

  • title: (Optional) A title for the run.

  • tier: (Optional) The hardware tier to use for the run. This is the human-readable name of the hardware tier, such as "Free", "Small", or "Medium". Will use project default tier if not provided.

  • publishApiEndpoint: (Optional) Whether or not to publish an API endpoint from the resulting output.

runs_start_blocking(command, isDirect, commitId, title, tier, publishApiEndpoint, poll_freq=5, max_poll_time=6000)

Same as method run_start except make a blocking request that waits until job is finished.

  • command: The command to run as an array of strings where members of the array represent arguments of the command. E.g. ["main.py", "hi mom"]

  • isDirect: (Optional) Whether or not this command should be passed directly to a shell.

  • commitId: (Optional) The commitId to launch from. If not provided, will launch from latest commit.

  • title: (Optional) A title for the run.

  • tier: (Optional) The hardware tier to use for the run. Will use project default tier if not provided.

  • publishApiEndpoint: (Optional) Whether or not to publish an API endpoint from the resulting output.

  • poll_freq: (Optional) Number of seconds in between polling of the Domino server for status of the task that is running.

  • max_poll_time: (Optional) Maximum number of seconds to wait for a task to complete. If this threshold is exceeded, an exception is raised.

  • retry_count: (Optional) Maximum number of retry to do while polling (in-case of transient http errors). If this threshold is exceeded, an exception is raised.

run_stop(*runId, saveChanges=True):

Stop an existing run in the selected project. The parameters are:

  • runId: string that identifies the run

  • saveChanges: (Defaults to True) If false, run results will be discarded.

runs_stdout(runId)

Get stdout emitted by a particular run. The parameters are:

  • runId: string that identifies the run

files_list(commitId, path)

List the files in a folder in the Domino project. The parameters are:

  • commitId: The commitId to list files from.

  • path: (Defaults to "/") The path to list from.

files_upload(path, file)

Upload a Python file object into the specified path inside the project. See examples/upload_file.py for an example. The parameters, both of which are required, are:

  • path: The path to save the file to. For example, /README.md will write to the root directory of the project while /data/numbers.csv will save the file to a subfolder named data (if the data folder does not yet exist, it will be created)

  • file: A Python file object. For example, f = open("authors.txt","rb")

blobs_get(key)

Retrieve a file from the Domino server by blob key. The parameters are:

  • key: The key of the file to fetch from the blob server.

app_publish(unpublishRunningApps=True, hardwareTierId=None)

Publishes an app in the Domino project, or republish an existing app. The parameters are:

  • unpublishRunningApps: (Defaults to True) Will check for any active app instances in the current project and unpublish them before publishing.

  • hardwareTierId: (Optional) Will launch the app on the specified hardware tier. Only applies for Domino 3.4+.

app_unpublish()

Stops all running apps in the Domino project.

Airflow

The python-domino client comes bundled with an Operator for use with airflow as an extra.

To install its dependencies, when installing the package from github add the airflow flag to extras with pip.

pip install -e git+https://github.com/dominodatalab/python-domino.git@master#egg=domino[airflow]

DominoOperator

from domino.airflow import DominoOperator
Note
 This is a python>=3.5 feature due to typing support.

Allows a user to schedule domino runs via airflow. Follows the same function signature as runs_start with two extra arguments:

  • startup_delay: Optional[int] = 10

    Add a startup delay to your job, useful if you want to delay execution until after other work finishes.

  • include_setup_log: Optional[bool] = True

    Determine whether or not to publish the setup log of the job as the log prefix before stdout.

License

This library is made available under the Apache 2.0 License. This is an open-source project of Domino Data Lab.

Visit the python-domino repository

You can find the complete library, with documentation and example code, in the public repository at https://github.com/dominodatalab/python-domino.