Feature Request: Allow download competition data into custom path with `competition_download`

[EssamWisam]

def competition_download(handle: str, path: Optional[str] = None, *, force_download: Optional[bool] = False) -> str:
    """Download competition dataset
    Args:
        handle: (string) the competition name
        path: (string) Optional path to a file within a competition dataset
        force_download: (bool) Optional flag to force download a competition dataset, even if it's cached
    Returns:
        A string requesting the path to the requested competition files.
    """

So far, the path argument has a different behaviour compared to that in dataset_download. I suggest having src_path and dest_path or something.

[neshdev]

I don't believe this is the case as seen here. Is your request to have a destination path to save the file on the local FS?

[EssamWisam]

Oh then I got confused when I intended to use it for that purpose (I remember the path on local FS was different). Yes, this is my request. To add a dest_path to both. Many people for instance may want to download it in the same directory and it makes dealing with Colab much easier.

[brendanartley]

@EssamWisam, Does setting the KAGGLEHUB_CACHE environment variable work for you? Something like this.

import os
import kagglehub

os.environ["KAGGLEHUB_CACHE"] = "./tmp"

path = kagglehub.dataset_download("uciml/iris")
print("Path to dataset files:", path)

[EssamWisam]

It works but it is not as convenient. Suppose I aim to download two datasets in different paths then I need to change this environment variable repeatedly.

[Borda]

I may suggest using filter_path instead of the actual path, which may enable even some reg. pattern and add local_path, which would be more explicit that it is the destination location