StreamlitInternalApp

Overview

This is a Python package to turn functions into internal apps powered by Streamlit. It creates forms based on the function's annotation and allows you to run the function with the form's values. Note that this is early-stage project with limited functionality.

Description

This package provides StreamlitInternalApp class that you can use to create internal apps. It provides component method decorator that turns any Python function into ComponentForm class. It is responsible for rendering form with this function arguments as input fields and return type as output. It doesn't infer the runtime types, but rely completely on the typing annotations. The function's docstring is used as a form's description. Once the form is submitted, the function is called with the form's values and the result is displayed in the output section.

StreamlitInternalApp is composed of multiple ComponentForm instances. It is responsible for rendering the sidebar with the list of available forms and the main section with the selected form. It also provides a way to add custom components to the sidebar.

Installation

pip install streamlit-internal-app

This package requires Python 3.11+ and Streamlit 1.29+.

Usage

StreamlitInternalApp is opinionated and designed to be as simple as possible. Custom validation on the UI side is not supported. It is expected that the function will validate the input and raise an exception if the input is invalid. The exception will be displayed in the output section.

Here is a simple example:

from streamlit_internal_app import StreamlitInternalApp

app = StreamlitInternalApp(title="Example app", description="This is an example app.")

@app.component
def add(a: int, b: int) -> int:
    """Adds two numbers."""
    return a + b

app.render()

That's all you need to do to create an internal app. You can the application by executing python -m streamlit run path/to/script.py.

Here is a table of supported types:

Annotation	Input field(s)	Output field	Input widget docs	Output widget docs
`int`	`st.number_input(argument_name, step=1)`	`st.write(result)`	https://docs.streamlit.io/library/api-reference/widgets/st.number_input	https://docs.streamlit.io/library/api-reference/write-magic/st.write
`float`	`st.number_input(argument_name, step=0.01)`	`st.write(result)`	https://docs.streamlit.io/library/api-reference/widgets/st.number_input	https://docs.streamlit.io/library/api-reference/write-magic/st.write
`str`	`st.text_input(argument_name)`	`st.write(result)`	https://docs.streamlit.io/library/api-reference/widgets/st.text_input	https://docs.streamlit.io/library/api-reference/write-magic/st.write
`bool`	`st.checkbox(argument_name)`	`st.write(result)`	https://docs.streamlit.io/library/api-reference/widgets/st.checkbox	https://docs.streamlit.io/library/api-reference/write-magic/st.write
`datetime.date`	`st.date_input(argument_name)`	`st.write(result)`	https://docs.streamlit.io/library/api-reference/widgets/st.date_input	https://docs.streamlit.io/library/api-reference/write-magic/st.write
`datetime.datetime`	`st.date_input(label=f"{argument_name}_date")` and `st.time_input(label=f"{argument_name}_time", step=60)` combined	`st.write(result)`	https://docs.streamlit.io/library/api-reference/widgets/st.date_input https://docs.streamlit.io/library/api-reference/widgets/st.time_input	https://docs.streamlit.io/library/api-reference/write-magic/st.write
`pathlib.Path`	❌	`st.download_button(label="Download")`	❌	https://docs.streamlit.io/library/api-reference/widgets/st.download_button
`typing.Literal`	`st.selectbox(argument_name, literal_values)`	`st.write(result)`	https://docs.streamlit.io/library/api-reference/widgets/st.selectbox	https://docs.streamlit.io/library/api-reference/write-magic/st.write
`list`	`st.text_area(argument_name)` that will be parsed into JSON object (for `list[str]`) and `st.multiselect(argument_name, literal_values)` for `list[Literal[str1,str2,...]]`. Only these two are supported.	`st.json(result)`	https://docs.streamlit.io/library/api-reference/widgets/st.text_area https://docs.streamlit.io/library/api-reference/widgets/st.multiselect	https://docs.streamlit.io/library/api-reference/data/st.json
`dict`	`st.text_area(argument_name)` that will be parsed into JSON object. Only `dict` annotation is supported.	`st.json(result)`	https://docs.streamlit.io/library/api-reference/widgets/st.text_area	https://docs.streamlit.io/library/api-reference/data/st.json
`pandas.DataFrame`	❌	`st.dataframe(result)`	❌	https://docs.streamlit.io/library/api-reference/data/st.dataframe
`dataclasses.dataclass`	All the attributes of dataclass will be treated as separate annotations (exploded). If provided, it needs to be the only annotation - callable can have only one parameter.	`st.write(result)`	❌	https://docs.streamlit.io/library/api-reference/write-magic/st.write

If your annotation is not in the list, the output will be rendered with st.write by default.

Here is a short demo of how the example app looks like:

Refer to /tests/static/example_app.py for a full example. If you have poetry installed you can run the example app with scripts/run_sample_app.sh, or you can do it with development Docker image.

Refer to scripts/ directory on how to run unit tests, formatting and linting.

Logging support