Python.Execute

The data frames a Python script leaves behind, as a navigation table: the script runs in this computer's Python.

Group: Python and R. Power Query M has it too: a query that uses it is written as in Power BI.

Syntax

Python.Execute(script as text, optional arguments as nullable record) as table

Parameter	Type	Description
`script`	text	The Python script.
`arguments`	nullable record optional	Tables to hand to the script: each field becomes a pandas data frame of that name, as Power BI's dataset.

Returns

table. Name and Value: a row per pandas data frame the script leaves, Value its table.

Remarks

The script runs in a folder of its own, with pandas already imported as pd. Every variable that holds a data frame when it ends is a row, named after the variable; a frame with an index of its own gets the index as columns.
It needs Python with pandas and pyarrow. OpenBI finds the newest Python installed (python.org's, the Python install manager's, Anaconda's), or the one set in OpenBI Studio's options.
A script runs only once the project allows it: OpenBI Studio asks once per project before its scripts run. Without that, the step stops with a message saying scripts are off.
A script that fails stops the step with the end of what it printed (its traceback). A script may run for 30 minutes at most.
Python and R scripts are kept so that Power BI projects open and refresh. For analysis, OpenBI's notebooks are the place (Why no Python or R, in the docs).

Examples

A data frame made by the script

The result is a navigation table with one row, shops, whose Value is the frame as a table.

Python.Execute("import pandas as pd#(lf)shops = pd.DataFrame({'Shop': ['Oslo', 'Bergen'], 'Sales': [120.5, 95.0]})")

Result: not shown here: the docs' build runs no scripts.

A step's table, changed by the script

Power BI's Run Python script step writes this: the table goes in as dataset and comes back with a new column, Total (120 for the chairs, 46.5 for the lamps).

let
    Source = #table(type table [Product = text, Qty = Int64.Type, Price = number], {{"Chair", 2, 60}, {"Lamp", 3, 15.5}}),
    Run = Python.Execute("dataset['Total'] = dataset['Qty'] * dataset['Price']", [dataset = Source])
in
    Run{[Name = "dataset"]}[Value]