Collate Python SDK quickstart

Install the OpenMetadata Python SDK, connect to Collate, and create, retrieve, update, and delete a table.

Collate Python SDK quickstart

Collate uses the openmetadata-ingestion Python package for typed metadata operations. Match the package release line to your Collate server release.

Install the SDK

For Collate 1.13.1, install a package from the matching server patch line:

pip install "openmetadata-ingestion~=1.13.1.0"

Configure the connection

Set the Collate API URL and an ingestion bot JSON Web Token (JWT):

export OPENMETADATA_HOST="https://your-company.getcollate.io/api"
export OPENMETADATA_JWT_TOKEN="your-jwt-token"

Call configure() once when the application starts. Entity classes reuse the configured client.

from metadata.sdk import configure

configure()

Pass credentials directly when environment variables aren’t available:

from metadata.sdk import configure

configure(
    host="https://your-company.getcollate.io/api",
    jwt_token="your-jwt-token",
)

Manage a table

The following example creates, retrieves, updates, and soft-deletes a table. Replace sample_data.ecommerce_db.shopify with the fully qualified name of an existing database schema.

from metadata.generated.schema.api.data.createTable import CreateTableRequest
from metadata.generated.schema.entity.data.table import Column, DataType
from metadata.generated.schema.type.basic import Markdown
from metadata.sdk import Tables, configure

configure()

schema_fqn = "sample_data.ecommerce_db.shopify"

# Create or update the table.
created = Tables.create(
    CreateTableRequest(
        name="customers",
        databaseSchema=schema_fqn,
        description="Customer master data",
        columns=[
            Column(name="id", dataType=DataType.BIGINT),
            Column(name="email", dataType=DataType.VARCHAR, dataLength=320),
            Column(name="created_at", dataType=DataType.TIMESTAMP),
        ],
    )
)
print(f"Created: {created.fullyQualifiedName}")

# Retrieve by fully qualified name.
table = Tables.retrieve_by_name(f"{schema_fqn}.customers")
if table is None:
    raise RuntimeError("Table was not found")

# Update the description.
table.description = Markdown("Customer master data managed with the Python SDK")
updated = Tables.update(table)
print(f"Updated: {updated.description}")

# Soft-delete the table. Soft-deleted entities remain recoverable.
Tables.delete(updated.id)

Tables.create() is idempotent. It creates the table when it doesn’t exist and updates it when the fully qualified name already exists.

Avoid hard_delete=True unless permanent removal is intended. Hard deletes can’t be undone.

Next steps

• Review all table creation fields.
• Explore Python SDK entity classes.
• Run the OpenMetadata Python SDK examples.