SIE Parser for LLM Analysis

Open Source Project - Privacy First

IMPORTANT: This SIE parser is open source software intended to be used locally on a private machine to guarantee personal data protection. If you're certain your SIE 4 files don't contain any personal information, you may run it publicly.

Contributions are welcome! The vision is to expand support for more accounting programs and make this a powerful tool for LLMs to provide financial reports and analyses.

A web-based application for processing Swedish SIE 4 files and preparing them for LLM analysis. This tool handles CP437 encoding for proper Swedish character support and provides a user-friendly interface for uploading, processing, and exporting SIE data.

Features

• Upload and parse Swedish SIE 4 files with CP437 encoding support
• View financial data in an organized, visual format
• Generate LLM-friendly JSON output with financial summaries
• Add custom descriptions to enhance LLM context
• Save processed data as JSON files
• Automatic aggregation for large transaction sets

Requirements

• Python 3.7+
• Flask
• Modern web browser

Installation

1. Clone or download this repository
2. Install the required dependencies:

pip install -r requirements.txt

3. Run the application:

python app.py

4. Open your browser and navigate to http://localhost:5000

Usage

1. Click "Choose a SIE file" to select your SIE 4 file (with .sie or .se extension)
2. Click "Process File" to upload and parse the file
3. View the processed data in the various tabs:
   • Summary: Overview of the financial data
   • Balance Sheet: Assets, liabilities, and equity
   • Income Statement: Income and expenses
   • Transactions: Individual transactions
   • JSON Output: LLM-ready JSON data
4. Add a description to provide context for LLM analysis
5. Save the JSON file for use with your preferred LLM (Gemma, Llama, Mistral, Claude, etc.)

SIE Format Support

This application supports the SIE 4 format, including:

• #FLAGGA, #PROGRAM, #FORMAT, #GEN, #SIETYP, #FNAMN, #ORGNR sections
• #RAR (fiscal year) sections
• #KONTO (account) definitions
• #IB, #UB, #RES (opening balance, closing balance, result) sections
• #VER (verification) entries with #TRANS (transaction) records

LLM Optimization

For large transaction sets (over 500 transactions), the application automatically:

• Creates aggregated summaries by account and month
• Provides sample transactions for context
• Generates balance sheet and income statement summaries
• Structures data to fit within typical LLM context windows

Developer Documentation

For developers working on this project, please refer to these important documentation files:

• Data Model Architecture - Defines the standardized data model and how different bookkeeping systems (Bokio, Dooer, Fortnox) are handled
• Testing Guidelines - Procedures for testing consistency across different SIE file formats

Core Architecture Principles

1. Standardized Data Model - All SIE formats (Bokio, Dooer, Fortnox) are converted to a common representation
2. Parser-level Normalization - System-specific quirks are handled in the parser, not the frontend
3. Clean API Boundary - The data model serves as the API between the parser and frontend

License

MIT