Python Code

Type: python_exec • Category: flow • Tags: python, exec, sandbox

Description

Execute limited Python code (numpy, pandas, math)

Parameters

Name	Type	Description	Required	Default
code	code	Multiline Python code	no
persistent	boolean	Use persistent worker process (faster for multiple calls)	no	false

Help

Overview
This worker provides a sandboxed environment for executing short Python snippets that rely on the core scientific libraries NumPy, pandas, and the standard math module. It is designed for quick calculations on financial market data, such as price series transformations, statistical summaries, or simple model evaluations. The execution is deliberately limited to keep the runtime safe and deterministic, while still offering enough functionality for most data‑analysis tasks in finance.

Inputs

• code (string, required) – A multiline Python script. The script may import numpy, pandas, and math, and should contain the logic you want to run. The final result should be assigned to a variable named result (or printed) so the worker can return it.
• persistent (boolean, optional, default = false) – When set to true, the worker keeps the Python interpreter alive across multiple calls. This reduces start‑up overhead and is useful when you need to run several related snippets in succession (e.g., loading a large DataFrame once and reusing it).

Minimal Example Usage

**Request**  
{
  "code": "import pandas as pd\nprices = pd.Series([101, 102, 98, 105])\nresult = prices.pct_change().fillna(0).tolist()",
  "persistent": false
}

Response

{
  "result": [0.0, 0.009900990099009901, -0.0392156862745098, 0.07142857142857144]
}

In this example, the snippet imports pandas, creates a simple price series, computes the percentage change for each entry, replaces the initial NaN with 0, and returns the list of returns. Because persistent is false, the worker starts a fresh interpreter for this single call and shuts it down afterward.

Additional Notes

• Only the whitelisted libraries (numpy, pandas, math) are available; attempts to import other modules will raise an error.
• The worker runs each script in isolation, so variables defined in one call are not retained unless persistent is enabled.
• For reproducibility, avoid relying on external state (e.g., network access or file I/O) within the code block.

This documentation provides a concise yet complete guide to using the Python execution worker for financial data processing tasks.