GenXL

A tool that converts structured JSON data into professionally formatted Excel reports using LLM-powered layout generation.

How It Works

1. Input: Takes extracted document data as JSON (e.g., bank statements, invoices)
2. LLM Processing: Sends the data to Gemini API, which determines optimal Excel cell layout and styling
3. Excel Generation: Produces a styled .xlsx file with proper formatting, borders, and color coding

Setup

Prerequisites

• Python 3.10+
• Poetry for dependency management
• A Gemini API key

Installation

poetry install

Configuration

Create a .env file in the project root:

GEMINI_API_KEY=your_api_key_here

Usage

Place your input JSON file in the project directory and run:

poetry run python main.py

By default, it reads from testing_jsons/testing_json_1.json and outputs to output.xlsx.

Input Format

The input JSON should contain extracted document data with the following structure:

{
  "file_name": "document.pdf",
  "classified_file_type": "bank_statement",
  "fields": [
    {
      "field_name": "Account Holder",
      "field_key": "account_holder",
      "section": "Account Information",
      "data_type": "String",
      "value": "John Doe"
    },
    {
      "field_name": "Account Balance",
      "field_key": "account_balance",
      "section": "Account Information",
      "data_type": "Number",
      "value": 15250.75
    },
    {
      "field_name": "Statement Date",
      "field_key": "statement_date",
      "section": "Account Information",
      "data_type": "Date",
      "value": "2024-01-15"
    },
    {
      "field_name": "Transactions",
      "field_key": "transactions",
      "section": "Transaction History",
      "data_type": "Table",
      "value": [
        { "Date": "2024-01-01", "Description": "Direct Deposit", "Amount": 3000.00 },
        { "Date": "2024-01-05", "Description": "Grocery Store", "Amount": -85.50 }
      ]
    }
  ]
}

Field Data Types

Data Type	Description	Example Value
String	Plain text values	"John Doe"
Number	Numeric values (integers or decimals)	1500.75
Date	Date strings in any format	"2024-01-15"
Table	Array of objects representing tabular data	[{"col1": "val1", "col2": "val2"}]

Field Schema

Each field object in the fields array supports:

Key	Type	Description
field_name	string	Human-readable label displayed in the Excel output
field_key	string	Machine identifier for the field
section	string	Logical grouping (e.g., "Account Information", "Transaction History")
data_type	string	One of: String, Number, Date, Table
value	any	The extracted value — can be null, a scalar, or an array of objects (for Table type)

Dependencies

• openpyxl - Excel file generation and styling
• google-genai - Gemini API client
• py-toon-format - Compact data encoding for LLM prompts
• python-dotenv - Environment variable management

License

See LICENSE for details.