How goosebbeoms/sogoo works
Project type and purpose
• This is a full-stack subscription-platform application (“SOGOO”) composed of:
– A React/Vite-based frontend served behind NGINX
– A Spring Boot “config” microservice (Spring Cloud Config Server)
– A Spring Boot “main” microservice (the core API)
– An AI microservice (FastAPI-based KoBERT model for review analysis)
– Supporting infrastructure: PostgreSQL, Redis, Elasticsearch, AWS S3, external payment gateway (TossPayments).
File structure
• Top level:
– ai/ → Jupyter notebooks and a FastAPI app for NLP (review summarization)
– backend/config/ → Spring Cloud Config Server project
– backend/main/ → Spring Boot main application project (domain modules, security, AWS S3, JWT, Quartz schedulers)
– frontend/ → Vite/React UI, Tailwind, React Query, Zustand state management, Axios configs
– docker-compose.*.yml → separate compose files for config service, main service, frontend
– docs/ → API specs, ERD, architecture docs
• Patterns:
– Microservice separation (config vs. main vs. AI)
– Layered MVC within the main service (controllers → services → repositories → entities)
– External integrations managed via dedicated “client” classes (e.g. TossPaymentsClient, ChatClientConfig)
README for additional insights
• Purpose: connect small-business owners with subscription customers
• Key features: subscription design, dashboard analytics, review-based sentiment analysis, search & recommendation
• Tech stack confirms use of Spring Cloud Config, Redis caching, PostgreSQL, Elasticsearch, AWS S3, JWT, FastAPI/KoBERT
a. Main components:
– Users (Buyer, Seller) via browser/mobile
– Frontend (React app behind NGINX)
– API Gateway / Load Balancer (NGINX)
– Config Service (Spring Cloud Config)
– Main Service (Spring Boot)
– AI Service (FastAPI KoBERT)
– Datastores: PostgreSQL, Redis, Elasticsearch
– File storage: AWS S3
– External Payment Gateway (TossPayments)
– CI/CD pipeline: Jenkins → Docker images → AWS EC2
b. Relationships & interactions:
– Browser → NGINX → React frontend
– React → REST calls → Main Service
– Main Service on startup fetches config from Config Service
– Main Service → reads/writes to PostgreSQL, caches to Redis, indexes/searches in Elasticsearch, stores images/files in S3
– Main Service → invokes AI Service (review analysis) via HTTP
– Main Service → invokes TossPayments API for payment processing
– CI/CD: GitLab/Jenkins pipelines build Docker images, push to registry, deploy to EC2 via Docker Compose
c. Architectural patterns & principles:
– Twelve-factor app style (externalized config via Config Server)
– Layered architecture in main service (Controller → Service → Repository)
– Microservices for separation of concerns (config, main, AI)
– Event-driven scheduling (Quartz jobs for automatic billing)
– Caching layer (Redis) for performance
– Search microservice (Elasticsearch) for text-based search
– JWT-based stateless authentication
d. Key technologies & frameworks (annotate each component)
– Frontend: React 18, Vite, TypeScript, Tailwind, React Query, Zustand
– Config Service: Spring Boot 3.3.1, Spring Cloud Config
– Main Service: Spring Boot, Spring Data JPA, Spring Security, JWT, AWS S3 SDK, Quartz, Redis, PostgreSQL, Elasticsearch
– AI Service: FastAPI, KoBERT model
– CI/CD: Jenkins, Docker, Docker Compose, NGINX, AWS EC2
Detail & granularity
• Sub-component where it matters (e.g., “Elasticsearch Indexer” inside Main Service)
• Don’t over-aggregate—show all four services (frontend, config, main, AI) separately
• Schedulers and external integrations as distinct nodes