dbt-loom

dbt-loom is a dbt Core plugin that weaves together multi-project deployments. dbt-loom works by fetching public model definitions from your dbt artifacts, and injecting those models into your dbt project.

flowchart LR

    classDef black fill:#f2f2ebff, stroke:#000, color:#000
    classDef background fill:#f2f2ebff, stroke:#000, color:#000
    classDef hidden fill:#BADC3F, stroke:#BADC3F, color:#BADC3F

   style TOP fill:#BADC3F, stroke:#000

  subgraph TOP[Your Infrastructure]
    direction TB
    dbt_runtime[dbt Core]:::background
    proprietary_plugin[Open Source Metadata Plugin]:::background

    files[Local Files]:::background
    object_storage[Object Storage]:::background
    discovery_api[dbt Cloud APIs]:::background

    discovery_api --> proprietary_plugin
    files --> proprietary_plugin
    object_storage --> proprietary_plugin
    proprietary_plugin --> dbt_runtime
  end

  Project:::black --> TOP --> Warehouse:::black

Loading

dbt-loom currently supports obtaining model definitions from:

• Local manifest files
• dbt Cloud
• GCS
• S3-compatible object storage services
• Azure Storage

⚠️ dbt Core's plugin functionality is still in beta. Please note that this may break in the future as dbt Labs solidifies the dbt plugin API in future versions.

Getting Started

To being, install the dbt-loom python package.

pip install dbt-loom

Next, create a dbt-loom configuration file. This configuration file provides the paths for your upstream project's manifest files.

manifests:
  - name: project_name
    type: file
    config:
      path: path/to/manifest.json

By default, dbt-loom will look for dbt_loom.config.yml in your working directory. You can also set the DBT_LOOM_CONFIG environment variable.

Using dbt Cloud as an artifact source

You can use dbt-loom to fetch model definitions from dbt Cloud by setting up a dbt-cloud manifest in your dbt-loom config, and setting the DBT_CLOUD_API_TOKEN environment variable in your execution environment.

manifests:
  - name: project_name
    type: dbt_cloud
    config:
      account_id: <YOUR DBT CLOUD ACCOUNT ID>

      # Job ID pertains to the job that you'd like to fetch artifacts from.
      job_id: <REFERENCE JOB ID>

      api_endpoint: <DBT CLOUD ENDPOINT>
      # dbt Cloud has multiple regions with different URLs. Update this to
      # your appropriate dbt cloud endpoint.

      step_id: <JOB STEP>
      # If your job generates multiple artifacts, you can set the step from
      # which to fetch artifacts. Defaults to the last step.

Using an S3-compatible object store as an artifact source

You can use dbt-loom to fetch manifest files from S3-compatible object stores by setting up ab s3 manifest in your dbt-loom config. Please note that this approach supports all standard boto3-compatible environment variables and authentication mechanisms. Please see the boto3 documentation for more details.

manifests:
  - name: project_name
    type: s3
    config:
      bucket_name: <YOUR S3 BUCKET NAME>
      # The name of the bucket where your manifest is stored.

      object_name: <YOUR OBJECT NAME>
      # The object name of your manifest file.

Using GCS as an artifact source

You can use dbt-loom to fetch manifest files from Google Cloud Storage by setting up a gcs manifest in your dbt-loom config.

manifests:
  - name: project_name
    type: gcs
    config:
      project_id: <YOUR GCP PROJECT ID>
      # The alphanumeric ID of the GCP project that contains your target bucket.

      bucket_name: <YOUR GCS BUCKET NAME>
      # The name of the bucket where your manifest is stored.

      object_name: <YOUR OBJECT NAME>
      # The object name of your manifest file.

      credentials: <PATH TO YOUR SERVICE ACCOUNT JSON CREDENTIALS>
      # The OAuth2 Credentials to use. If not passed, falls back to the default inferred from the environment.

Using Azure Storage as an artifact source

You can use dbt-loom to fetch manifest files from Azure Storage by setting up an azure manifest in your dbt-loom config. The azure type implements the DefaultAzureCredential class, supporting all environment variables and authentication mechanisms. Alternatively, set the AZURE_STORAGE_CONNECTION_STRING environment variable to authenticate via a connection string.

manifests:
  - name: project_name
    type: azure
    config:
      account_name: <YOUR AZURE STORAGE ACCOUNT NAME> # The name of your Azure Storage account
      container_name: <YOUR AZURE STORAGE CONTAINER NAME> # The name of your Azure Storage container
      object_name: <YOUR OBJECT NAME> # The object name of your manifest file.

Using environment variables

You can easily incorporate your own environment variables into the config file. This allows for dynamic configuration values that can change based on the environment. To specify an environment variable in the dbt-loom config file, use one of the following formats:

${ENV_VAR} or $ENV_VAR

Example:

manifests:
  - name: revenue
    type: gcs
    config:
      project_id: ${GCP_PROJECT}
      bucket_name: ${GCP_BUCKET}
      object_name: ${MANIFEST_PATH}

Gzipped files

dbt-loom natively supports decompressing gzipped manifest files. This is useful to reduce object storage size and to minimize loading times when reading manifests from object storage. Compressed file detection is triggered when the file path for the manifest is suffixed with .gz.

manifests:
  - name: revenue
    type: s3
    config:
      bucket_name: example_bucket_name
      object_name: manifest.json.gz

How does it work?

As of dbt-core 1.6.0-b8, there now exists a dbtPlugin class which defines functions that can be called by dbt-core's PluginManger. During different parts of the dbt-core lifecycle (such as graph linking and manifest writing), the PluginManger will be called and all plugins registered with the appropriate hook will be executed.

dbt-loom implements a get_nodes hook, and uses a configuration file to parse manifests, identify public models, and inject those public models when called by dbt-core.

Known Caveats

Cross-project dependencies are a relatively new development, and dbt-core plugins are still in beta. As such there are a number of caveats to be aware of when using this tool.

1. dbt plugins are only supported in dbt-core version 1.6.0-b8 and newer. This means you must be using a dbt adapter compatible with this version.
2. PluginNodeArgs are not fully-realized dbt ManifestNodes, so documentation generated by dbt docs generate may be sparse when viewing injected models.