cast-func 0.1.0

Build live notebook-backed presentations from Python functions.

cast

Present from a notebook with two decorators.

import cast, polars as pl, plotly.express as px

cast.serve(port=8000)            # live page in a background thread

@cast.data                        # function returns a polars LazyFrame
def sales():
    return pl.scan_parquet("sales.parquet")

@cast.figure                      # a factory: takes one table, returns a Plotly figure
def trend(tbl: pl.LazyFrame):
    df = tbl.select(["date", "value"]).collect()
    return px.line(df, x="date", y="value")

sales()                           # register the table (data is injected from the browser)

@cast.figure registers a factory at decoration time — you don't pass a table in the notebook. Open the printed URL and use the controls to add the figure to the inventory with a chosen table. The same factory can be added many times with different data, and each instance has its own table dropdown; picking a table re-runs your figure function on the server against it. Calling a @cast.data function again updates its data and the page refreshes live.

See examples/demo.py for a runnable showcase.