polaris-client

Python client for the Imply Polaris hosted Druid offering.

This library provides a Python client to work with the REST API of Imply's Polaris product, which is a hosted version of Apache Druid.

See the Polaris Developer Guide for an overiew of the REST API. The class documentation assumes you are familiar with Polaris concepts and the Polaris API. Each method refers to the Polaris documentation for the details.

The documentation describes JSON objects, which mostly map directly into Python dict objects in this client. In some cases (identified in method comments), this client "unwraps" some Polaris boilerplate.

The Polaris API is at an early stage and is incomplete in some areas. This library works around those flaws where possible. Expect changes to the library as Polaris evolves.

See druid-client for a similar client for Apache (and Imply) Druid.

Getting Started

You should already have a Polaris account (you can create a free trial.) You must also create a custom API client to use below. Remember to give your client the service account roles needed for those APIs you wish to use.

Then connect to your account:

import polaris_client

org = 'your-org'
client_id = 'your-client-id'
secret = 'your-secret'
client = polaris_client.connect(org, client_id, secret)

See Authenticate API requests for details:

• org: the name displayed in the top-left of the home page.
• client_id: the client ID you create as described in the doc referenced abvove.
• secret: The scret associated with the client ID.

The client automatically renews the security ticket as needed.

If you are on a non-standard domain, add the domain as a named argument:

client = polaris_client.connect(org, client_id, secret, domain='test')

Once connected you can use the Polaris APIs. There are two forms. The client itself is a light wrapper over the REST API calls and returns Python objects for use in your application.

client.tables()

>>> [{'name': 'my-table',
  'id': '9b4va3ea-dbdf-4d40-af5c-965450755556',
  'version': 0,
  'lastUpdateDateTime': '2022-02-11T19:07:30Z',
  'createdByUsername': 'bob@example.com',
  'timePartitioning': 'day'}]

If you want to explore the API interactively, we suggest using Jupyter. Then, use the the show() function to display the results as a table:

show = client.show()
show.tables()

Displays:

Table
my-table

Switch to HTML format if you use Jupyter:

show.as_html()
show.tables()

Displays the same information, but formatted as an HTML table.

Working with Tables

The client provides a convenience class to work with tables:

table = client.table_for_name('my-table')
table.name()

>>> my-table

table.id()

>>> 9b4va3ea-dbdf-4d40-af5c-965450755556

table.show_schema()

Results:

Name   | Type
__time | LONG
col1   | STRING
col2   | LONG

To see the "input schema" for pushing events:

table.show_input_schema()

Results:

Name   | Type
col1   | STRING
col2   | LONG

The actual pushed event must include the __time column as well. To add data:

table.insert({'__time': '2022-03-03T22:37:13Z', 'col1': 'value1', 'col2': 10})

Note that the python datetime.isoformat() method will produce the correct format for the __time field.

In Polaris, you create a table with a name, but then reference it in the RST API via the ID. Use client.table_id(name) to get the ID for a table given its name. Or, just use the Table class which handles the ID for you.

Queries

To run a query:

show.sql('SELECT * FROM "my-table"')

>>> __time               | col1   | col2
>>> 2022-03-03T22:37:13Z | value1 | 10

client.sql(stmt) runs the query and returns the results as a list of objects for programmatic use.

Method Documentation

Each method references the API it uses, with a link to the Imply documentation. In Jupyter, use help() to see the available methods given an instance of that object:

help(client)
help(show)
help(table)

To see the detailed documentation, with links:

help(client.projects)

Displays:

Help on method projects in module polaris_client.client:

projects() method of polaris_client.client.Client instance
    Returns the list of projects.
    
    Returns the project list under the `values` key in the REST response.
    
    Calls `/v1/projects`
    
    See https://docs.imply.io/polaris/api-query/#get-project-id

Tracing

You can see the APIs for each call with:

client.trace(True)
client.projects()

>>> GET: https://api.eng.imply.io/v1/projects
>>> [{'metadata': {'uid': '1b460394-3645-404d-a241-7ddbb7845ea6', ...

Tracing also shows the payload for POST calls and the raw details of error messages.