Client get_projects/list_projects does not retrieve all projects

[andrewbapham]

Versions:

$ pip freeze | grep label
label-studio==1.21.0
label-studio-sdk==2.0.4

When calling get_projects/list_projects, the result will only fetch a maximum of 100 projects, even though the docstrings indicate they should list ALL projects in Label Studio. This also seems to affect manual requests to the /api/projects endpoint when trying to up the page_size.

The following snippet prints out the following output:

from label_studio_sdk import Client
def main():
    projects = LS_CLIENT.list_projects()
    print("get_projects length:", len(projects))

    manual = LS_CLIENT.make_request("GET", "/api/projects", params={'page_size': 500})
    res = manual.json()
    print('manual request length', len(res['results']))
    print('count:', res['count'])

~$ python script.py 
get_projects length: 100
manual request length 100
count: 201

There seemed to have been a similar issue previously with tasks (label-studio #1908) that was solved with #31. Perhaps a paginated request solution is needed for projects as well?

[heidi-humansignal]

Hello,

Thank you for contacting Label Studio,
Currently, The /api/projects/ endpoint is paginated and capped by the backend. Nonetheless, we suggest that you try to iterate through pages until you exhaust them by leveraging SDK V2.

https://api.labelstud.io/api-reference/api-reference/projects/list
Example:

from label_studio_sdk import LabelStudiols = LabelStudio( base_url="http://your-host:port", api_key="YOUR_API_KEY",)# Auto-iterates through all pages server-sideresponse = ls.projects.list( page_size=100, # max allowed; safe default)all_projects = []for item in response: all_projects.append(item)print("Total projects fetched:", len(all_projects))

Let us know it has solved your usecase!
Best Regards!

Comment by Oussama Assili
Workflow Run