Building a Streamlit App in CDF

Prerequisites - Access to a CDF Project. - Some familiarity with Streamlit.

Settings in Streamlit App

Streamlit is a beta feature in CDF. You can use pygen in it by adding cognite-pygen to the installed packages under settings.

pyodide-http==0.2.1
cognite-sdk==7.73.3
pydantic==2.7.0
cognite-pygen==0.99.60

Note that we also set pydantic to a specific version. This is because pygen supports both pydantic v1 and v2, but when we want to use pygen in the CDF streamlit environment, we need to use pydantic v1.

In case you have issues with the installation, check out the troubleshooting in the installation guide.

Minimal Example

The following is a minimal example of a Streamlit app that uses pygen to generate a Python SDK client with the ScenarioInstanceModel from the Working with TimeSeries example.

import streamlit as st
from cognite.client import CogniteClient
from cognite.pygen import generate_sdk_notebook
st.title("An example app in CDF with Pygen")

client = CogniteClient()

@st.cache_data
def get_client():
  return generate_sdk_notebook(
    ("IntegrationTestsImmutable", "ScenarioInstance", "1"), client
  )

@st.cache_data
def get_scenario_instances():
  client = get_client()
  return client.scenario_instance.list().to_pandas()


st.dataframe(get_scenario_instances())

Explanation

The first thing we do is to import the CogniteClient and generate_sdk_notebook from cognite.pygen.

from cognite.client import CogniteClient
from cognite.pygen import generate_sdk_notebook

Then, we create a Streamlit app and a CogniteClient instance.

st.title("An example app in CDF with Pygen")

client = CogniteClient()

Next, we define a function that generates a SDK for the ScenarioInstance model. We use the @st.cache_data decorator to cache the result of the function. This means that the function will only be called once, and the result will be cached for subsequent calls.

@st.cache_data
def get_client():
  return generate_sdk_notebook(
    ("IntegrationTestsImmutable", "ScenarioInstance", "1"), client
  )

We then define a function that uses the generated SDK to list all ScenarioInstance objects in the project. We also cache the result of this function.

@st.cache_data
def get_scenario_instances():
  client = get_client()
  return client.scenario_instance.list().to_pandas()

Finally, we call the get_scenario_instances function and display the result in a dataframe.

st.dataframe(get_scenario_instances())