Transform your data using natural language instructions. This powerful node uses AI to generate and execute Python code based on your transformation requirements.

How It Works

1. You provide your data (in any structured format)

2. You describe how you want to transform it in plain English

3. The AI generates and executes the appropriate code

4. If there are any errors, it automatically attempts to fix them

Input Formats

The node accepts various data formats including:

- JSON arrays/objects

- Markdown tables

- HTML tables

- CSV strings

- Any other structured data format

Example Transformations

1. Basic Sorting and Filtering

Input Data:
[
    {"name": "John", "age": 30, "salary": 50000},
    {"name": "Jane", "age": 25, "salary": 60000},
    {"name": "Bob", "age": 35, "salary": 45000}
]

Transformation Request:
"Sort by salary in descending order and only include people over 25"

Output:
[
    {"name": "Jane", "age": 25, "salary": 60000},
    {"name": "John", "age": 30, "salary": 50000},
    {"name": "Bob", "age": 35, "salary": 45000}
]

2. Adding Calculated Columns

Input Data:
| Product | Price |
|---------|-------|
| A       | 100   |
| B       | 200   |
| C       | 150   |

Transformation Request:
"Add a 'Discounted_Price' column that's 90% of the original price and sort by it"

Output:
| Product | Price | Discounted_Price |
|---------|-------|-----------------|
| B       | 200   | 180             |
| C       | 150   | 135             |
| A       | 100   | 90

Tips

1. Be specific in your transformation instructions

2. You can request multiple transformations in one instruction

3. The node will automatically handle data parsing and formatting

4. If you get an error, try rephrasing your transformation request

Common Transformation Types

• Sorting and filtering data

• Adding calculated columns

• Aggregating or grouping data

• Reformatting data structure

• Data type conversions

• Statistical calculations

• Text processing

• Date/time manipulations

Example Instructions

• "Convert all prices to integers and add a 'total' column"

• "Group by category and calculate the average price for each group"

• "Extract all unique values from the 'tags' column and count their frequencies"

• "Convert the markdown table to JSON format and add an 'id' field to each row"

• "Calculate the percentage difference between consecutive rows in the 'value' column"

Working with Multiple Data Sources

You can transform multiple data sources simultaneously by passing a JSON object in the data field. Each key in the object should reference a different data source using Moonlit's double brackets syntax.

Example:

// Data field:
{
    "sales_data": "{{monthly_sales}}",
    "inventory_data": "{{stock_levels}}"
}

// Transformation Request:
"Join the sales and inventory data on product_id, then calculate the sell-through rate (sales quantity / inventory quantity) for each product"

This approach is particularly useful for:

- Joining multiple datasets

- Cross-referencing data from different sources

- Performing comparative analysis

- Creating consolidated reports

Example Multi-Source Transformations

1. Combining Customer and Order Data:

// Data field:
{
    "customers": "{{customer_table}}",
    "orders": "{{orders_table}}"
}

// Transformation:
"Join customers and orders on customer_id, then calculate total spend per customer and sort by highest spenders"

1. Product Performance Analysis:

// Data field:
{
    "sales": "{{sales_data}}",
    "reviews": "{{product_reviews}}",
    "inventory": "{{inventory_levels}}"
}

// Transformation:
"Create a summary table showing each product's total sales, average rating, and current stock level"

Remember:

- Each referenced variable must be properly formatted data

- Use descriptive keys in your JSON object to make the transformation request clearer

- The transformation instructions should explicitly mention which data source you're referring to