This repository hosts a production-grade, containerized data extraction engine designed to parse complex, multi-page financial PDF reports (specifically Centrale Rischi documents). The solution automates the extraction of critical financial indicators, tabular data, and metadata, transforming unstructured PDF content into structured, queryable datasets.

Key Achievement: Engineered to process high volumes of financial documents with precision, this parser handles dynamic table structures, multi-page layouts, and data normalization, significantly reducing manual data entry efforts.

The system is architected as a modular microservice, leveraging Docker for portability and ease of deployment.

1. Parser Engine (src/core):
   • Utilizes Camelot (Lattice mode) for high-fidelity table extraction.
   • Integrates PyMuPDF (Fitz) for layout analysis, metadata extraction (headers, dates), and page orientation detection.
2. API Layer (src/api):
   • A robust Flask API endpoint (/parse_document) handles file uploads and processing requests.
   • Implements asynchronous processing using threading to ensure API responsiveness.
3. Data Pipeline (src/storage):
   • S3 Integration: Securely uploads raw PDF backups to AWS S3.
   • DynamoDB Integration: Stores structured parsing results and processing status in AWS DynamoDB.
4. Infrastructure (docker/):
   • Fully containerized using Docker.
   • Deployed with Gunicorn as the WSGI HTTP server for production performance.
   • Includes Nginx configuration for reverse proxy handling.

• Docker and Docker Compose installed.
• AWS Credentials configured (via Environment Variables).

git clone [https://github.com/Arazmalek/centrale_rischi_parser.git](https://github.com/Arazmalek/centrale_rischi_parser.git)
cd centrale_rischi_parser

Create a .env file in the root directory (based on .env.example) and populate it with your credentials. Note: Never commit real credentials to Git.

# .env example
AWS_ACCESS_KEY_ID=your_access_key
AWS_SECRET_ACCESS_KEY=your_secret_key
AWS_REGION=eu-south-1
S3_BUCKET_NAME=your-bucket-name
DYNAMO_TABLE_NAME=your-table-name

docker-compose up --build

The API will be available at http://localhost:80 (via Nginx) or http://localhost:5000 (direct).

• Method: POST
• Payload: form-data with a file field named file.

Example Request (cURL):

curl -X POST -F "file=@/path/to/sample_report.pdf" http://localhost/parse_document

Response:

{
    "statusCode": 200,
    "request_id": "unique-uuid-v4",
    "message": "File accepted for processing."
}

⚠️ IMPORTANT DISCLAIMER:

• Synthetic Data: The PDF files provided in the data_samples/ directory are synthetically generated dummy documents. No real financial data or Personally Identifiable Information (PII) from real clients is used, stored, or processed in this public repository.
• Credential Safety: All sensitive configuration (AWS keys, DB endpoints) has been abstracted into environment variables. The code provided here is a sanitized version of the production system, designed for portfolio demonstration purposes.

Araz Malekazari | Senior Data Engineer & System Architect