Connections

Connect warehouses and object storage, then import external data as Wood Wide datasets.

Connections let you pull data from external systems into Wood Wide without manually downloading and re-uploading files. Create and credential a connection in the console, then browse schemas or objects and import into a dataset for training and inference.

Supported providers include BigQuery, Snowflake, PostgreSQL, Databricks, S3, and GCS.

Add a connection

Create and test a connection in the console.

CLI import

Import from an existing connection with wwai connections.

---

Python SDK workflow

The Python SDK can browse and import from connections created in the console. These examples require woodwide==0.9.0 and read WOODWIDE_API_KEY.

from woodwide import WoodWide

client = WoodWide()

connections = client.data_connections.list()
for connection in connections.items:
    print(
        connection.data_connection_id,
        connection.name,
        connection.provider,
        connection.status,
    )

For a warehouse connection, inspect its schemas and tables, then import a table:

connection_id = "YOUR_CONNECTION_ID"

schemas = client.data_connections.list_schemas(connection_id)
tables = client.data_connections.list_tables(
    connection_id,
    schema="analytics",
)

submission = client.data_connections.import_data(
    connection_id,
    mode="table",
    table_schema="analytics",
    table_name="orders",
    dataset_name="orders snapshot",
)
print(submission.dataset_id, submission.job_id)

For S3-compatible or GCS connections, use list_objects() and import with mode="object" and object_key=.

---

Console workflow

1. Create a connection

   In the console, open Connections → Add connection. Choose a provider and enter credentials (or use a Wood Wide service account where supported).

2. Test the connection

   Run Test connection before browsing or importing. Imports are blocked until the connection passes.

3. Browse and import

   • Warehouses (BigQuery, Snowflake, PostgreSQL, Databricks): browse schemas and tables, or run a SQL query, then import.
   • Object storage (S3, GCS): browse folders and select CSV or Parquet files.

   Import creates a new dataset and an ingestion job. Wait for the job to finish before training.

4. Train and infer

   Use the new dataset like any uploaded dataset — in the console, via the API, or CLI. Imported datasets get the same IDs the Python SDK accepts by dataset_id.

---

CLI workflow

Connections are created in the console. The CLI reuses them to browse and import without re-entering secrets.

wwai connections list
wwai connections tables CONNECTION_ID --schema analytics
wwai connections import CONNECTION_ID \
  --mode table --schema analytics --table orders \
  --name "orders snapshot"
wwai jobs wait JOB_ID
wwai models train --dataset DATASET_ID --type prediction --label-column target --wait

For S3 or GCS, browse with wwai connections objects and import with --mode object.

See the CLI page for the full command set.

---

Related docs

• Datasets API — manage datasets after import
• Data preparation guide — format and schema tips before training