Connect Databricks Volumes to your preprocessing pipeline, and use the Unstructured Ingest CLI or the Unstructured Ingest Python library to batch process all your documents and store structured outputs locally on your filesystem.

The requirements are as follows.

• A Databricks account on AWS, Azure, or GCP.

• A workspace within the Databricks account for AWS, Azure, or GCP.

• The workspace’s URL. Get the workspace URL for AWS, Azure, or GCP.

  Examples:

  • AWS: https://<workspace-id>.cloud.databricks.com
  • Azure: https://adb-<workspace-id>.<random-number>.azuredatabricks.net
  • GCP: https://<workspace-id>.<random-number>.gcp.databricks.com
  Do not add a trailing slash (/) to the workspace URL.

• The Databricks authentication details. For more information, see the documentation for AWS, Azure, or GCP.

  For the Unstructured UI or the Unstructured API, the following Databricks authentication types are supported:

  • Databricks OAuth machine-to-machine (M2M) authentication for
    AWS, Azure, or GCP.

    You will need the the Client ID (or UUID or Application ID) and OAuth Secret (client secret) values for the corresponding service principal. Note that for Azure, only Databricks managed service principals are supported. Microsoft Entra ID managed service principals are not supported.

    The following video shows how to create a Databricks managed service principal:

  • Databricks personal access token authentication for AWS, Azure, or GCP.

    You will need the personal access token’s value.

    The following video shows how to create a Databricks personal access token:

  For Unstructured Ingest, the following Databricks authentication types are supported:

  • For Databricks personal access token authentication for AWS, Azure, or GCP: The personal access token’s value.

    The following video shows how to create a Databricks personal access token:

  • For username and password (basic) authentication (AWS only): The user’s name and password values.

  • For OAuth machine-to-machine (M2M) authentication (AWS, Azure, and GCP): The client ID and OAuth secret values for the corresponding service principal.

  • For OAuth user-to-machine (U2M) authentication (AWS, Azure, and GCP): No additional values.

  • For Azure managed identities (formerly Managed Service Identities (MSI) authentication) (Azure only): The client ID value for the corresponding managed identity.

  • For Microsoft Entra ID service principal authentication (Azure only): The tenant ID, client ID, and client secret values for the corresponding service principal.

  • For Azure CLI authentication (Azure only): No additional values.

  • For Microsoft Entra ID user authentication (Azure only): The Entra ID token for the corresponding Entra ID user.

  • For Google Cloud Platform credentials authentication (GCP only): The local path to the corresponding Google Cloud service account’s credentials file.

  • For Google Cloud Platform ID authentication (GCP only): The Google Cloud service account’s email address.

• The name of the parent catalog in Unity Catalog for AWS, Azure, or GCP for the volume.

• The name of the parent schema (formerly known as a database) in Unity Catalog for AWS, Azure, or GCP for the volume.

• The name of the volume in Unity Catalog for AWS, Azure, or GCP, and optionally any path in that volume that you want to access directly, beginning with the volume’s root.

• The Databricks workspace user or service principal must have the following minimum set of privileges to read from or write to the existing volume in Unity Catalog:

  • USE CATALOG on the volume’s parent catalog in Unity Catalog.
  • USE SCHEMA on the volume’s parent schema (formerly known as a database) in Unity Catalog.
  • READ VOLUME and WRITE VOLUME on the volume.

  The following videos shows how to create and set privileges for a catalog, schema (formerly known as a database), and volume in Unity Catalog.

  Learn more about how to check and set Unity Catalog privileges for AWS, Azure, or GCP.

The Databricks Volumes connector dependencies:

pip install "unstructured-ingest[databricks-volumes]"

You might also need to install additional dependencies, depending on your needs. Learn more.

The following environment variables:

• DATABRICKS_HOST - The Databricks host URL, represented by --host (CLI) or host (Python).

  Do not add a trailing slash (/) to the host URL.

• DATABRICKS_CATALOG - The Databricks catalog name for the Volume, represented by --catalog (CLI) or catalog (Python).

• DATABRICKS_SCHEMA - The Databricks schema name for the Volume, represented by --schema (CLI) or schema (Python). If not specified, default is used.

• DATABRICKS_VOLUME - The Databricks Volume name, represented by --volume (CLI) or volume (Python).

• DATABRICKS_VOLUME_PATH - Any optional path to access within the volume, specified by --volume-path (CLI) or volume_path (Python).

Environment variables based on your authentication type, depending on your cloud provider:

• For Databricks personal access token authentication (AWS, Azure, and GCP):

  • DATABRICKS_TOKEN - The personal access token, represented by --token (CLI) or token (Python).

• For username and password (basic) authentication (AWS only): The user’s name and password values.

  • DATABRICKS_USERNAME - The user’s name, represented by --username (CLI) or username (Python).
  • DATABRICKS_PASSWORD - The user’s password, represented by --password (CLI) or password (Python).

• For OAuth machine-to-machine (M2M) authentication (AWS, Azure, and GCP): The client ID and OAuth secret values for the corresponding service principal.

  • DATABRICKS_CLIENT_ID - The client ID value for the corresponding service principal, represented by --client-id (CLI) or client_id (Python).
  • DATABRICKS_CLIENT_SECRET - The client ID and OAuth secret values for the corresponding service principal, represented by --client-secret (CLI) or client_secret (Python).

• For OAuth user-to-machine (U2M) authentication (AWS, Azure, and GCP): No additional environment variables.

• For Azure managed identities (MSI) authentication (Azure only):

  • ARM_CLIENT_ID - The client ID value for the corresponding managed identity, represented by --azure-client-id (CLI) or azure_client_id (Python).
  • If the target identity has not already been added to the workspace, then you must also specify the DATABRICKS_AZURE_RESOURCE_ID, represented by --azure-workspace-resource-id (CLI) or azure_workspace_resource_id (Python).

• For Microsoft Entra ID service principal authentication (Azure only):

  • ARM_TENANT_ID - The tenant ID value for the corresponding service principal, represented by --azure-tenant-id (CLI) or azure_tenant_id (Python).
  • ARM_CLIENT_ID - The client ID value for the corresponding service principal, represented by --azure-client-id (CLI) or azure_client_id (Python).
  • ARM_CLIENT_SECRET - The client secret value for the corresponding service principal, represented by --azure-client-secret (CLI) or azure_client_secret (Python).
  • If the service principal has not already been added to the workspace, then you must also specify the DATABRICKS_AZURE_RESOURCE_ID, represented by --azure-workspace-resource-id (CLI) or azure_workspace_resource_id (Python).

• For Azure CLI authentication (Azure only): No additional environment variables.

• For Microsoft Entra ID user authentication (Azure only):

  • DATABRICKS_TOKEN - The Entra ID token for the corresponding Entra ID user, represented by --token (CLI) or token (Python).

• For Google Cloud Platform credentials authentication (GCP only):

  • GOOGLE_CREDENTIALS - The local path to the corresponding Google Cloud service account’s credentials file, represented by --google-credentials (CLI) or google_credentials

• For Google Cloud Platform ID authentication (GCP only):

  • GOOGLE_SERVICE_ACCOUNT - The Google Cloud service account’s email address, represented by --google-service-account (CLI) or google_service_account (Python).

• Alternatively, you can store the preceding settings in a local Databricks configuration profile and then just refer to the profile’s name:

  • DATABRICKS_PROFILE - The name of the Databricks configuration profile, represented by --profile (CLI) or profile (Python).

Now call the Unstructured Ingest CLI or the Unstructured Ingest Python library. The destination connector can be any of the ones supported. This example uses the local destination connector.

This example sends data to Unstructured for processing by default. To process data locally instead, see the instructions at the end of this page.

#!/usr/bin/env bash

# Chunking and embedding are optional.

unstructured-ingest \
  databricks-volumes \
    --profile $DATABRICKS_PROFILE \
    --host $DATABRICKS_HOST \
    --catalog $DATABRICKS_CATALOG \
    --schema $DATABRICKS_SCHEMA \
    --volume $DATABRICKS_VOLUME \
    --volume-path $DATABRICKS_VOLUME_PATH \
    --partition-by-api \
    --api-key $UNSTRUCTURED_API_KEY \
    --partition-endpoint $UNSTRUCTURED_API_URL \
    --strategy hi_res \
    --additional-partition-args="{\"split_pdf_page\":\"true\", \"split_pdf_allow_failed\":\"true\", \"split_pdf_concurrency_level\": 15}" \
    --chunking-strategy by_title \
    --embedding-provider huggingface \
  local \
    --output-dir $LOCAL_FILE_OUTPUT_DIR

For the Unstructured Ingest CLI and the Unstructured Ingest Python library, you can use the --partition-by-api option (CLI) or partition_by_api (Python) parameter to specify where files are processed:

• To do local file processing, omit --partition-by-api (CLI) or partition_by_api (Python), or explicitly specify partition_by_api=False (Python).

  Local file processing does not use an Unstructured API key or API URL, so you can also omit the following, if they appear:

  • --api-key $UNSTRUCTURED_API_KEY (CLI) or api_key=os.getenv("UNSTRUCTURED_API_KEY") (Python)
  • --partition-endpoint $UNSTRUCTURED_API_URL (CLI) or partition_endpoint=os.getenv("UNSTRUCTURED_API_URL") (Python)
  • The environment variables UNSTRUCTURED_API_KEY and UNSTRUCTURED_API_URL

• To send files to the Unstructured Partition Endpoint for processing, specify --partition-by-api (CLI) or partition_by_api=True (Python).

  Unstructured also requires an Unstructured API key and API URL, by adding the following:

  • --api-key $UNSTRUCTURED_API_KEY (CLI) or api_key=os.getenv("UNSTRUCTURED_API_KEY") (Python)
  • --partition-endpoint $UNSTRUCTURED_API_URL (CLI) or partition_endpoint=os.getenv("UNSTRUCTURED_API_URL") (Python)
  • The environment variables UNSTRUCTURED_API_KEY and UNSTRUCTURED_API_URL, representing your API key and API URL, respectively.
  You must specify the API URL only if you are not using the default API URL for Unstructured Ingest, for example, if you are using a self-hosted instance of the Unstructured API.

  The default API URL for Unstructured Ingest is https://api.unstructuredapp.io/general/v0/general, which is the API URL for the Unstructured Partition Endpoint. However, you should always use the URL that was provided to you when your Unstructured account was created. If you do not have this URL, contact Unstructured Sales at sales@unstructured.io.

  If you do not have an API key, get one now.

  If the Unstructured API is self-hosted, the process for generating Unstructured API keys, and the Unstructured API URL that you use, are different. For details, contact Unstructured Sales at sales@unstructured.io.