Connect Jupyter with Odata and Central

Support
1

Hello everyone, I am interested in connecting jupyter with the central ODK data, however, the information in ODATA is not very illustrative, I would greatly appreciate if someone could tell me how to make the connection from python. Thank you

2

Welcome @PedroReina, and thanks for your question. When you have a moment, we'd love to learn a bit about you and your project in the introduction thread!

If you happen to also be comfortable with R, I encourage you to check out ruODK which includes several notebook-style examples.

That said, I know there are many reasons to want to set up a Python-based pipeline. @Lindsay_Stevens_Au and I have just kicked off a project to make that easier and provide some worked examples.

To get a json document, you can use the OData data document:
{url}/v1/projects/{project}/forms/{formid}.svc/Submissions

If you have any repeats, you can replace Submissions with the repeat name to get the json document for each repeat.

You also need to get a session token to be able to make the request. I recommend using a Project Viewer account for analysis.

If you use pandas, you'll likely want to normalize the resulting json with a separator of / to flatten groups.

Putting it all together:

import requests
import json
import pandas as pd

url = "https://"
project = 
formid = ""
username = ""
password = ""

token_response = requests.post(
        f"{url}/v1/sessions",
        data=json.dumps({"email": username, "password": password}),
        headers={"Content-Type": "application/json"},
    )

token = token_response.json()["token"]

response = requests.get(
        f"{url}/v1/projects/{project}/forms/{formid}.svc/Submissions?$filter=__system/reviewState ne 'rejected'",
        headers={"Content-Type": "application/json", "Authorization": f"Bearer {token}"}
    )

df = pd.json_normalize(response.json()['value'], sep='/')

You can also search the forum for various other Python-based examples (e.g. Error in GET request for Central API - #10 by yanokwa).

2 Likes
3

@LN it would indeed be very handy for users to get working examples for a Python-based processing pipeline.

In case this is helpful, I had created a few parameterized Python functions elaborating from the example provided by Yaw. I had given a few examples in this post and I also have a small demo Jupyter Notebook.

2 Likes