PDF Invoice Data Extractor with Gemini AI

🤖 An intelligent PDF invoice analyzer that extracts structured data using Google's Gemini AI.

Features

• PDF Text Extraction: Converts PDF invoices to text using pdfminer
• AI-Powered Analysis: Uses Google Gemini AI to intelligently extract invoice data
• Comprehensive Data Extraction:
  • Invoice number, dates, amounts (TTC, HT, TVA)
  • Company information (name, address, phone, email, ICE)
  • Client information
  • Bank details (IBAN, RIB)
  • Detailed articles/items with quantities and prices
• Multilingual Support: Available in English and French
• Structured Output: JSON format with organized data
• File Organization: Automatic saving to organized folders

Installation

1. Clone the repository:

git clone https://github.com/yourusername/invoice-pdf-extractor.git
cd invoice-pdf-extractor

2. Install dependencies:

pip install -r requirements.txt

3. Set up your Google AI API key:
   • Get your API key from Google AI Studio
   • Set it as an environment variable (recommended):

     export GOOGLE_API_KEY="your_api_key_here"

   • Or edit the script to use your hardcoded key (for testing only)

Usage

English Version

python test_ai.py "path/to/your/invoice.pdf"

French Version

python test_ai_fr.py "path/to/your/invoice.pdf"

Example Output

🤖 ANALYSE DE FACTURE POWERED BY GEMINI AI
============================================================
📄 Numéro de Facture: FAC-2024-001
📅 Date de Facturation: 15/03/2024
💰 Total TTC: 3,876 €
💵 Total HT: 3,230 €
🧾 Montant TVA: 646 €

🏢 INFORMATIONS ENTREPRISE:
  Nom: SARL EXEMPLE SERVICES
  Adresse: 123 Rue de la République, 75001 Paris
  Téléphone: +33 1 42 36 78 90
  Email: [email protected]

🛒 ARTICLES:
  1. Service de consultation IT
     Quantité: 5
     Prix Unitaire: 450 €
     Prix Total: 2,250 €
     Taux TVA: 20%

File Structure

project/
├── test_ai.py              # English version
├── test_ai_fr.py           # French version (Version française)
├── test.py                 # Classic regex-based extraction
├── analysis/               # AI analysis results
├── text save/             # Extracted PDF text
├── requirements.txt       # Python dependencies
└── README.md             # This file

API Key Setup

Option 1: Environment Variable (Recommended)

# Linux/Mac
export GOOGLE_API_KEY="your_api_key_here"

# Windows
set GOOGLE_API_KEY=your_api_key_here

Option 2: Direct in Code (Testing Only)

Edit the script and replace the API key in the ai_extract_invoice_data function.

Dependencies

• pdfminer.six: PDF text extraction
• google-generativeai: Google Gemini AI API
• re, json, os, sys: Standard Python libraries

Supported Invoice Formats

The AI can extract data from various invoice formats including:

• French invoices (factures)
• Multi-language invoices
• Different layouts and structures
• Various currencies (€, $, etc.)

Output Files

The script automatically creates:

• Extracted text: text save/{filename}_extracted.txt
• AI analysis: analysis/{filename}_gemini_analysis.txt (English) or analysis/{filename}_gemini_analysis_fr.txt (French)

Troubleshooting

Common Issues

1. "API key not valid":

   • Ensure your API key is correct
   • Check that Generative AI API is enabled in Google Cloud Console
   • Verify billing is set up (required for Gemini API)

2. "No suitable model found":

   • Check your internet connection
   • Verify API key permissions

License

This project is licensed under the MIT License - see the LICENSE file for details.

Note: Keep your API keys secure and never commit them to public repositories!