How datpham0412/invoice-processor works
Project type and purpose
• Full-stack web application (“MatchFlow”) for invoice reconciliation
• Purpose: Users upload PDF invoices; the system extracts structured data via Azure Form Recognizer, matches it against stored Purchase Orders, flags discrepancies, and presents audit trails.
File structure
• Root directories:
– InvoiceProcessor.API – .NET 6 Web API (Application, Controllers, Domain, Infrastructure, Migrations)
– InvoiceProcessor.Tests – xUnit tests for repositories and services
– frontend – React + Vite single-page app with shadcn/ui and Tailwind CSS
– redirect-app – NGINX proxy/redirect service
• API layer follows Clean Architecture:
– Application: interfaces (IInvoiceRepository, IFormRecognizer…), DTO models, services (UploadInvoiceService, MatchingService…)
– Controllers: REST endpoints for auth, invoice upload/list/match, PO operations, OCR tests
– Domain: entities (Invoice, PurchaseOrder, LineItem, ExceptionRecord, User), enums, exceptions
– Infrastructure: Azure Blob Storage client, Form Recognizer client, EF Core DbContext & repositories, JWT & Azure settings
• Deployment/config: Dockerfiles (API, frontend, redirect-app), docker-compose.yml, GitHub Actions workflow, Fly.io manifests
README for additional insights
• Describes end-to-end flow: upload → OCR → compare → alert
• Tech stack: React + Vite, ASP NET Core, EF Core, SQL Server, Azure Blob, Cognitive Services, Service Bus, JWT auth, Docker/Fly.io, Nginx
• Mentions CI/CD, monitoring via Application Insights
Elements
• Clear, descriptive labels for each component and interface
• Annotations for protocols (HTTPS, AMQP, SQL) and data formats (JSON, JWT)