The Code Node allows you to build and execute custom Python functions.

Use the Code node instead of the function calling code block in LLM Tools if you want a more deterministic pipeline (In function calling your LLM will reason to determine if it makes sense to call the function).

​
Important Notes

  1. Function must be called main()
    • Only a single root function is allow
      • Nested Functions are allowed
    • main() does not take any arguements
  2. Output from Code Node will be stored as CODE_id
    • a return statement is required to output from the code node
    • return output
  3. Outputs from previous nodes or variables can be passed in using standard variable naming convention
    • <?REPLACE_WITH_REF_NODE_ID?>

​
Allowed functionality with Code Node:

  1. Built-In Python Functions
    • int, str, isinstance, range, len, type, sum, round, ord, float, abs, min, max, all, any, sorted, enumerate, zip, filter, map, reversed, chr, divmod, pow, bin, hex, and oct
  2. Supported Packages
    • json, requests, datetime, time, re, hashlib, bs4, typing, contextlib, decimal
NOTE: in order to use these packages, you will need to import them into your code at the top

​
Following are commonly used examples:

Use Case: Webscraping with BeautifulSoup

import requests
from bs4 import BeautifulSoup
import json
import re

def main():
    url = "https://example.com/product-page"  # Replace with actual product page URL
    response = requests.get(url)
    
    # Parse the HTML content
    soup = BeautifulSoup(response.text, 'html.parser')
    
    # Assume the product data is embedded in a script tag or fetched via AJAX
    script_tag = soup.find('script', text=re.compile('window.__PRODUCT_DATA__'))
    
    if script_tag:
        # Extract JSON string from the script tag
        json_text = re.search(r'window\.__PRODUCT_DATA__\s*=\s*({.*?});', script_tag.string).group(1)
    else:
        # If script tag not found, assume data was fetched via AJAX
        ajax_url = "https://example.com/api/product"  # Replace with actual AJAX endpoint
        ajax_response = requests.get(ajax_url)
        json_text = ajax_response.text
    
    # Instance check to determine if input is a JSON string or a Python dictionary
    if isinstance(json_text, str):
        product_data = json.loads(json_text)  # Parse the JSON string into a Python dictionary
    elif isinstance(json_text, dict):
        product_data = json_text  # Already a dictionary, no need to parse
    else:
        raise ValueError("Input must be a JSON string or a Python dictionary")
    
    # Extract specific product details
    product_name = product_data.get('name', 'Unknown')
    product_price = product_data.get('price', 'Unavailable')
    
    output = {
        "Product Name": product_name,
        "Product Price": product_price
    }
    
    return json.dumps(output, indent=4)