Atlan Python Client

9.4.0 · active · verified Tue Apr 14

The `pyatlan` library is the official Python SDK for interacting with Atlan, a data governance and metadata management platform. It provides typed wrappers around the Atlan REST API, enabling users to programmatically search, discover, govern, and manage various data assets like tables, dashboards, and glossary terms. The current version, 9.4.0, includes experimental features for the next-generation `pyatlan_v9` and ongoing quality-of-life improvements. The library maintains an active release cadence with frequent updates.

Warnings

breaking Major architectural shift in version 9.0.0, where the core data layer models transitioned from Pydantic v1 to `msgspec` for improved performance. This is part of the experimental `pyatlan_v9` initiative.
Fix: Review and update code interacting with Atlan models, especially for serialization/deserialization. While functional parity is intended, direct Pydantic v1 model usage may require adaptation. Refer to `pyatlan_v9` documentation for migration guidance.
gotcha In versions 9.2.0 through 9.3.1, the policy duplicate detection logic could inadvertently block legitimate policy *updates*, treating them as duplicate creations.
Fix: Upgrade to `pyatlan` version 9.3.2 or later, which revised the logic to correctly distinguish between retried creations and legitimate update requests, allowing updates to proceed.
gotcha In versions prior to 9.3.0, the `DQCondition.value` field's `Union` type ordering could incorrectly coerce integer values to strings during deserialization, leading to data type mismatches.
Fix: Upgrade to `pyatlan` version 9.3.0 or later, which reordered the `Union` type to prioritize `int` over `str`, resolving the deserialization issue.
gotcha Directly embedding API keys or other sensitive credentials in code is a security risk. API keys should never appear in logs or exception tracebacks.
Fix: Always use environment variables (e.g., `ATLAN_API_KEY`, `ATLAN_BASE_URL`) for credentials. Ensure your logging configuration strips sensitive information from logs.

Install

pip install pyatlan Install latest stable version

Imports

AtlanClient
```
from pyatlan.client import AtlanClient
```

AsyncAtlanClient

from pyatlan.client.aio import AsyncAtlanClient

For asynchronous operations.

FluentSearch

from pyatlan.model.fluent_search import FluentSearch

Table
```
from pyatlan.model.assets import Table
```
Example asset type, many others available under `pyatlan.model.assets`.

Quickstart

Initializes the `AtlanClient` using API key authentication (preferably from environment variables) and performs a basic search for active Table assets.

import os
from pyatlan.client import AtlanClient
from pyatlan.model.fluent_search import FluentSearch
from pyatlan.model.assets import Table

# Configure with environment variables (recommended) or direct parameters
ATLAN_BASE_URL = os.environ.get('ATLAN_BASE_URL', 'https://<your-tenant>.atlan.com')
ATLAN_API_KEY = os.environ.get('ATLAN_API_KEY', 'YOUR_API_KEY') # Assign persona(s) to API token for access

if not ATLAN_API_KEY or 'YOUR_API_KEY' in ATLAN_API_KEY:
    raise ValueError("Please set ATLAN_API_KEY environment variable or replace 'YOUR_API_KEY' with your actual key.")
if not ATLAN_BASE_URL or '<your-tenant>' in ATLAN_BASE_URL:
    raise ValueError("Please set ATLAN_BASE_URL environment variable or replace '<your-tenant>.atlan.com' with your actual Atlan URL.")

try:
    client = AtlanClient(base_url=ATLAN_BASE_URL, api_key=ATLAN_API_KEY)
    
    # Example: Search for active Table assets
    search_request = (
        FluentSearch()
        .where(FluentSearch.asset_type(Table))
        .where(FluentSearch.active_assets())
        .page_size(10)
        .to_request()
    )
    response = client.asset.search(search_request)

    print(f"Found {len(response.assets)} active Tables:")
    for asset in response.assets:
        print(f"  - {asset.name} ({asset.qualified_name})")

except Exception as e:
    print(f"An error occurred: {e}")

view raw JSON →