Skip to main content

Client library for accessing RecordKeepers Receiver

Project description

Context

RecordKeeper (abbreviated to RK) is aimed at two broad goals:

  1. Explaining why something happened in your platform. Common example that we want to support is: why event X happened at time T? What Models were used? Who trained them, using training data ingested from which datasources? It achieves it by creating graph of events.

  2. Recreating platform state at that time.

Basics

RKClient library is used to create events (PEMS) and inform RK about them.

You will need a running RecordKeeper Event Receiver to be able to work with it.

Recommended usage:

emitter_id = uuid.UUID('..some static uuid..')
rk_host = os.environ.get('RK_HOST')

rk_client = RKClientFactory.get(rk_host, emitter_id)

By using factory, automatically when RK_MOCK=true env variable will be defined, the returned client will fake the connections and return only success codes.

Details

The Receiver will be configured with max PEM payload size. If your PEM exceeds this limit, it will be rejected (with explanation).

You can query for the PEM limit size in get_info(), in pem_size_limit field.

API

RKClient:

  • prepare_pem
  • prepare_artifact
  • send_pem
  • ping
  • get_info
  • get_tag
  • set_tag

RKAdmin:

  • get_pems
  • get_pems_count
  • get_artifact
  • get_artifacts
  • get_artifacts_count
  • get_taxonomy_file
  • get_tags
  • get_tags_count
  • query_graph
  • clean_dbs

Authenticating

If you want to do anything more than sending a PEM, you will need to pass a RK authentication credential when creating RKClient or RKAdmin.

There are two types:

  1. User auth
  2. PUC auth

First is available in RecordKeeper Dashboard, in your user profile. It's intended to be used in tests.

Machine code on production should rather use PUC-auth, since it's user agnostic. However, you will need to ask RK admin to create the PUC and send you its auth code.

RKClient from Python console

cd rkclient/
python3
>>> from rkclient import RKAdmin
>>> rk = RKAdmin('http://127.0.0.1:8082')
>>> pems, msg, ok = rk.get_pems()
>>> assert ok, msg
>>> for pem in pems:
>>>   print(pem)

Changelog

[Unreleased]

[2.0.0] - 18.11.2022

Make it possible to control GraphBuilder through RKAdmin. Refactor requests code.

[1.9.1] - 20.9.2022

Version updated to match other components

[1.9.0] - 13.9.2022

Added async test.

[1.8.0] - 23.2.2022

Support for PUC auth type. New function get_tags_count.

[1.7.0] - 23.2.2022

RKClient catches timeout exception. Artifact instances in python can be compared using equal operator.

[1.6.1] - 26.01.2022

PEM deduplicates uses artifacts. Fix for function behaviour in mocked RKClient.

[1.6.0] - 17.01.2022

Removed PEM User field, bumped PEM version to 1.0.1

[1.5.0] - 21.12.2021

Version updated to match other components

[1.4.0] - 16.11.2021

Version updated to match other components

[1.3.0] - 6.9.2021

Remove the requests dependency

[1.2.3] - 26.8.2021

Bump the requests version to 2.26.*

[1.2.2] - 17.6.2021

Option to disable SSL cert verification

[1.2.1] - 8.6.2021

Official release on Pypi, added license


RKClient is part of ERST Recordkeeper repository.

RKClient is licensed with GNU General Public License version 3 or later, see LICENSE file for details.

Recordkeeper is ERST's implementation of the Context Cartographer specification.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

RecordKeeper_Client-2.0.0.tar.gz (24.7 kB view hashes)

Uploaded Source

Built Distribution

RecordKeeper_Client-2.0.0-py3-none-any.whl (27.9 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page