Manage a package’s sample data

In this section, we’ll use Pooch to manage the download of a Python package’s sample datasets.


The setup will be very similar to what we saw in Fetching files from a registry. It may be helpful to read that first.

The problem

In this example, we’ll work with the follow assumptions:

  • You develop a Python library called plumbus for analysing data emitted by interdimensional portals.

  • You want to distribute sample data so that your users can easily try out the library by copying and pasting from the documentation.

  • You want to have a plumbus.datasets module that defines functions like fetch_c137() that will return the data loaded as a pandas.DataFrame for convenient access.

  • Your sample data are in a folder of your GitHub repository but you don’t want to include the data files with your source and wheel distributions because of their size.

  • You use git tags to mark releases of your project.

  • Your project has a variable that defines the version string.

  • The version string contains an indicator that the current commit is not a release (like 'v1.2.3+12.d908jdl' or 'v0.1+dev').

For now, let’s say that this is the layout of your repository on GitHub:


The sample data are stored in the data folder of your repository.

See also

Pooch can handle different use cases as well, like: FTP/SFTP, authenticated HTTP, multiple URLs, decompressing and unpacking archives, etc. See the tutorials under “Training your Pooch” and the documentation for pooch.create and pooch.Pooch for more options.

Basic setup

This is what the plumbus/ file would look like:

Load sample data.
import pandas
import pooch

from . import version  # The version string of your project

BRIAN = pooch.create(
    # Use the default cache folder for the operating system
    # The remote data is on Github
    # If this is a development version, get the data from the "main" branch
        "c137.csv": "sha256:19uheidhlkjdwhoiwuhc0uhcwljchw9ochwochw89dcgw9dcgwc",
        "cronen.csv": "sha256:1upodh2ioduhw9celdjhlfvhksgdwikdgcowjhcwoduchowjg8w",

def fetch_c137():
    Load the C-137 sample data as a pandas.DataFrame.
    # The file will be downloaded automatically the first time this is run
    # returns the file path to the downloaded file. Afterwards, Pooch finds
    # it in the local cache and doesn't repeat the download.
    fname = BRIAN.fetch("c137.csv")
    # The "fetch" method returns the full path to the downloaded data file.
    # All we need to do now is load it with our standard Python tools.
    data = pandas.read_csv(fname)
    return data

def fetch_cronen():
    Load the Cronenberg sample data as a pandas.DataFrame.
    fname = BRIAN.fetch("cronen.csv")
    data = pandas.read_csv(fname)
    return data

The BRIAN variable captures the value returned by pooch.create, which is an instance of the Pooch class. The class contains the data registry (files, URLs, hashes, etc) and handles downloading files from the registry using the fetch method. When the user calls plumbus.datasets.fetch_c137() for the first time, the data file will be downloaded and stored in the local storage.


We’re using pooch.os_cache to set the local folder to the default cache location for the user’s operating system. You could also provide any other path if you prefer.


The files from different version of your project will be kept in separate folders to make sure they don’t conflict with each other. This way, you can safely update data files while maintaining backward compatibility. For example, if path=".plumbus" and version="v0.1", the data folder will be .plumbus/v0.1.

When your project updates, Pooch will automatically setup a separate folder for the new data files based on the given version string. The remote URL will also be updated. Notice that there is a format specifier {version} in the URL that Pooch substitutes for you.

Versioning is optional and can be ignored by omitting the version and version_dev arguments or setting them to None.

Retry failed downloads

When downloading data repeatedly, like in continuous integration, failures can occur due to sporadic network outages or other factors outside of our control. In these cases, it can be frustrating to have entire jobs fail because a single download was not successful.

Pooch allows you to specify a number of times to retry the download in case of failure by setting retry_if_failed in pooch.create. This setting will be valid for all downloads attempted with pooch.Pooch.fetch. The download can fail because the file hash doesn’t match the known hash (due to a partial download, for example) or because of network errors coming from requests. Other errors (file system permission errors, etc) will still result in a failed download.


Requires Pooch >= 1.3.0.

Where to go from here

Pooch has more features for handling different download protocols, handling large registries, downloading from multiple sources, and more. Check out the tutorials under “Training your Pooch” for more information.

Most users will also benefit from reading at least: