How to find shared-in datasets of a project with Python API?

Haoran

Thanks for your time at the beginning.

I have a project and I want to know which datasets are shared-in from other projects (black icons) with Python API.

I look through the previous Q&A, only find the way to find those which are shared-out to other project:

def find_exposed_datasets(project):

    result = []

    raw = project.get_settings().get_raw()

    exposed_objects = raw['exposedObjects']['objects']

    for obj in exposed_objects:

        if obj['type'] == 'DATASET':

            result.append(obj)

            # # print out share-to projects

            # for item in obj['rules']:

            #     result.append(item['targetProject'])

    return result

Thanks for your help!

Operating system used: Win11 enterprise

Haoran

https://community.dataiku.com/discussion/3372/get-shared-projects-using-dataiku-api

I found solutions. Find Tomas's response from the link. Thanks you Tomas.

FlorentD

https://community.dataiku.com/discussion/45370/how-to-find-shared-in-datasets-of-a-project-with-python-api

There is also another trick, rather than iterating over projects.

You could get the flow from your project, and look if the data contains datasets with a . in the name, something like:

client = dataiku.api_client()
project = client.get_default_project()
flow = project.get_flow()
[dataset for dataset in flow.get_graph().get_source_datasets() if '.' in dataset.dataset_name]

This solution requires the shared dataset as an input, which is usually the case.

Best

Turribeach

I don't believe it's possible to get this at instance level so you will to call client.list_project_keys() and loop through every project exposed objects to collect the whole list so you can then do a lookup per project.

Turribeach

Indeed that does what I said you should do which is to loop through every project. Personally I wouldn't call this in a function since it can take some time to run in large instances. I will build a small flow that refreshes a table every hour or so. Then you can call your function or report using this metadata.