feat: support for Kaggle Notebooks

[filipchristiansen]

Support for Kaggle Notebooks

Invocation

• CLI: gitingest https://www.kaggle.com/code/USER-NAME/NOTEBOOK-NAME
• PyPI package: ingest("https://www.kaggle.com/code/USER-NAME/NOTEBOOK-NAME")

Implementation

• Detect the kaggle.com/code/ host and route to a Kaggle downloader.

• Wrap the Kaggle CLI command:

  kaggle kernels get --path PATH USER-NAME/NOTEBOOK-NAME

Private Kaggle Notebooks

Follow the Kaggle API instructions to create a Kaggle API token.

Options

1. Save the kaggle.json file containing the token in the default path (~/.kaggle/kaggle.json or C:\Users\<username>\.kaggle\kaggle.json).

2. Export the username and token as environment variables:

   export KAGGLE_USERNAME=USER-NAME
   export KAGGLE_KEY=TOKEN

3. Specify the token per run

   • PyPI package: token="TOKEN" (in ingest / ingest_async)

   • CLI: --token TOKEN

   • Web UI: tick Private Repository and paste the token in the Private Access Token field.

   Note: Option 3 temporarily overrides any local KAGGLE_USERNAME / KAGGLE_KEY environment variables.

   Update: Kaggle CLI requires KAGGLE_USERNAME + KAGGLE_KEY even for public notebooks. Therefore, option 3 would require to allow the user to, in some way, enter their username.

Notebook outputs

• Cell outputs are never downloaded by the Kaggle API.

---

References

• Kaggle CLI
• Kaggle API instructions