Run Python Code

Controlled node

Overview

Execute Python code in a serverless container environment. This node allows you to run custom Python scripts with support for package installation, prebuild scripts, and structured input/output handling. The code executes in an isolated environment with access to a filesystem where inputs are provided via inputs.json and results are captured from output.json.

Execution Environment

• Language: Python 3.x
• Timeout: 10 minutes (600 seconds) maximum execution time
• Package Manager: pip (packages installed fresh per execution)
• Prebuild: Shell commands supported for environment setup
• Storage: Ephemeral filesystem with inputs.json and output.json access

Input/Output Pattern

The node follows a file-based IO pattern:

• Read workflow inputs from inputs.json (automatically populated from the inputs socket)
• Write results to output.json (automatically captured to the result output)
• Use print() for logging (captured in stdout)

Inputs

Input	Type	Description	Default
Run	Event	Triggers the Python code execution	-
Code	Text	Python code to execute. Must read from inputs.json and write to output.json.	See default template below
Packages	Text	Comma-separated list of pip packages to install (e.g., requests==2.28.1, pandas). Can also accept an array of strings via input connection.	-
Prebuild Script	Text	Shell commands to run before code execution (e.g., apt-get update). Useful for system dependencies.	-
Inputs	Data	Dictionary/object passed to the script via inputs.json	{}

Default Code Template

# Read inputs from inputs.json
import json

with open('inputs.json') as f:
    inputs = json.load(f)

# Your code here
result = {
    'message': 'Hello from Python!',
    'data': inputs
}

# Write result to output.json (this becomes the node's output)
with open('output.json', 'w') as f:
    json.dump(result, f)

# stdout is for logging only (visible in logs, not returned as output)
print('Processing completed successfully')

Outputs

Output	Type	Description
Done	Event	Fires when execution completes successfully or with error
Result	Data	Contains the JSON data written to output.json, or an error object if execution failed. Error format: { error: { message, stdout, stderr }, buildLogs?: string }

Error Handling

When execution fails, the result output contains:

• error.message: The error message or exception details
• error.stdout: Standard output from the execution
• error.stderr: Standard error output
• buildLogs: Build logs if package installation failed (only present on build failures)

Runtime Behavior and Defaults

• Package Installation: Packages are installed via pip before code execution. Specify versions using standard pip syntax (package==version). If the input is provided as an array via connection, each element is treated as a package name.
• Prebuild Phase: Runs before package installation. Use for system-level setup (e.g., installing system libraries required by Python packages).
• Input Serialization: The inputs data socket is automatically serialized to inputs.json in the execution environment. Complex objects, arrays, and primitives are supported.
• Output Capture: The contents of output.json are parsed as JSON and emitted from the result output. Ensure your code writes valid JSON.
• Execution Limits:
  • Maximum 10 minutes total execution time (includes prebuild, package installation, and code execution)
  • Maximum 600 polling attempts with 1-2 second intervals
  • Memory and CPU limits apply based on the serverless runtime configuration

Example Usage

Basic Data Processing

Process incoming data and return a transformed result:

import json

# Read inputs
with open('inputs.json') as f:
    data = json.load(f)

# Process data (e.g., calculate statistics)
numbers = data.get('numbers', [])
average = sum(numbers) / len(numbers) if numbers else 0

# Write result
with open('output.json', 'w') as f:
    json.dump({
        'average': average,
        'count': len(numbers),
        'sum': sum(numbers)
    }, f)

Using External Packages

Install and use external libraries like requests or pandas:

Packages input: requests==2.31.0, pandas

import json
import requests
import pandas as pd

with open('inputs.json') as f:
    config = json.load(f)

# Fetch data
response = requests.get(config['url'])
data = response.json()

# Process with pandas
df = pd.DataFrame(data['results'])
summary = df.describe().to_dict()

with open('output.json', 'w') as f:
    json.dump(summary, f)

With Prebuild Script

Install system dependencies before Python packages:

Prebuild Script:

apt-get update && apt-get install -y libxml2-dev libxslt1-dev

Packages: lxml, beautifulsoup4

from bs4 import BeautifulSoup
import json

with open('inputs.json') as f:
    html_content = json.load(f)['html']

soup = BeautifulSoup(html_content, 'lxml')
text = soup.get_text()

with open('output.json', 'w') as f:
    json.dump({'clean_text': text}, f)