BazzucaMedia - Social Media Management Platform

Overview

BazzucaMedia is a multi-tenant social media management platform for scheduling and publishing content across multiple social networks (X/Twitter, Instagram, Facebook, LinkedIn, TikTok, YouTube). Built with .NET 8 and PostgreSQL, it provides client management, post scheduling with calendar view, media upload via AWS S3, and automated publishing.

The project follows Clean Architecture with six layers and supports per-tenant database isolation. Publishing to X/Twitter is synchronous, while LinkedIn uses an asynchronous architecture with RabbitMQ message queues, retry with TTL, and dead-letter queues (DLQ), processed by a background Worker service using Playwright for browser automation.

---

Features

• 🏢 Multi-Tenant - Per-tenant database isolation via X-Tenant-Id header and JWT claims
• 👥 Client Management - CRUD for client accounts with soft-delete
• 📅 Post Scheduling - Calendar-based scheduling with automatic conflict resolution (30-min increments)
• 🌐 Social Network Integration - OAuth credential management for multiple networks per client
• 🐦 X/Twitter Publishing - Synchronous posting with chunked video upload (OAuth 1.0a)
• 💼 LinkedIn Publishing - Asynchronous posting via Playwright browser automation with RabbitMQ queues
• 🔄 Retry & DLQ - Automatic retry with TTL and dead-letter queue for failed LinkedIn posts
• 📦 Media Storage - File upload to AWS S3 via zTools
• 🔐 JWT Authentication - Per-tenant JWT secrets via NAuth
• 🖥️ Debug Console - CLI tool to visually debug LinkedIn publishing with visible Playwright browser

---

Technologies Used

Core

• .NET 8 - Web API, Background Worker, and Console CLI
• Entity Framework Core 9 - ORM with Lazy Loading Proxies
• PostgreSQL - Database (Npgsql provider)

Message Broker

• RabbitMQ - Async message queue for LinkedIn publishing (retry + DLQ topology)

Browser Automation

• Microsoft Playwright - LinkedIn publishing via Chromium automation

Authentication & Utilities

• NAuth - JWT authentication with IUserClient
• zTools - S3 file upload, ChatGPT, slug generation, email

External APIs

• AWS S3 - Media file storage
• X/Twitter API - OAuth 1.0a, chunked video upload
• LinkedIn - Web automation via Playwright

DevOps

• Docker / Docker Compose - Containerized deployment (API, Worker, RabbitMQ, PostgreSQL)
• GitHub Actions - Semantic versioning (GitVersion), release creation, SSH deploy

---

Project Structure

BazzucaMedia/
├── Bazzuca.API/                 # REST API (Controllers, Middlewares, Startup)
├── Bazzuca.Application/         # DI registration (Initializer.cs), Tenant services
├── Bazzuca.Domain/              # Models, Services, Factories (business logic)
├── Bazzuca.DTO/                 # Data Transfer Objects, Enums, QueueSettings
├── Bazzuca.Infra/               # DbContext, Repositories, UnitOfWork, RabbitMQ, Playwright
├── Bazzuca.Infra.Interface/     # Repository & service interfaces (zero dependencies)
├── Bazzuca.Worker/              # Background services (LinkedIn RabbitMQ consumer)
├── Bazzuca.Console/             # CLI tool for debugging LinkedIn publishing
├── docs/                        # System design diagrams
├── docker-compose.yml           # Docker local development
├── docker-compose-prod.yml      # Production deployment
├── BazzucaAPI.Dockerfile        # API container image
├── BazzucaWorker.Dockerfile     # Worker container image (Playwright + Chromium)
└── .github/workflows/           # CI/CD pipelines

Architecture Layers

API → Application → Domain → Infra.Interface
                  → Infra   → Infra.Interface
                  → DTO (no dependencies)

Layer	Project	Responsibility
API	Bazzuca.API	Controllers, TenantMiddleware, Startup/DI composition
Application	Bazzuca.Application	Initializer.cs (all DI), ITenantContext, ITenantDbContextFactory
Domain	Bazzuca.Domain	Models, Services, Factories, LinkedinService (retry/DLQ)
Infra	Bazzuca.Infra	DbContext, Repositories, RabbitAppService, LinkedinAppService
Infra.Interface	Bazzuca.Infra.Interface	Repository interfaces, IRabbitAppService, ILinkedinAppService
DTO	Bazzuca.DTO	DTOs, Enums, QueueSettings, PublishMessage
Worker	Bazzuca.Worker	LinkedinBackgroundService (RabbitMQ consumer)
Console	Bazzuca.Console	CLI for visual LinkedIn debugging

---

System Design

The following diagram illustrates the high-level architecture of BazzucaMedia:

Request Flow

Synchronous (X/Twitter): Browser → API → TenantMiddleware → NAuth → PostController → PostService → XService → X/Twitter API

Asynchronous (LinkedIn):

1. Enqueue: API → PostController publishes PublishMessage to RabbitMQ (bazzuca.linkedin.msg) → returns 202 Accepted
2. Process: Worker's LinkedinBackgroundService consumes message → extracts tenant → creates DI scope → calls LinkedinService.Process()
3. Publish: LinkedinService downloads media from S3 → calls LinkedinAppService (Playwright) → updates post status
4. On Failure: LinkedinService increments retry count → publishes to .retry queue (TTL returns to .msg) or .error queue (DLQ after max retries)

📄 Source: The editable Mermaid source is available at docs/system-design.mmd.

---

Environment Configuration

1. Copy the environment template

cp .env.example .env

2. Edit the .env file

# PostgreSQL
POSTGRES_DB=bazzuca_db
POSTGRES_USER=bazzuca_user
POSTGRES_PASSWORD=your_secure_password_here
POSTGRES_PORT=5434

# Multi-Tenant
DEFAULT_TENANT_ID=bazzuca
BAZZUCA_CONNECTION_STRING=Host=bazzuca-postgres;Port=5432;Database=bazzuca_db;Username=bazzuca_user;Password=your_secure_password_here
BAZZUCA_JWT_SECRET=your_jwt_secret_min_32_chars

# RabbitMQ
RABBITMQ_USER=guest
RABBITMQ_PASSWORD=your_rabbitmq_password_here

# API
API_HTTP_PORT=5010

The project uses three environments:

Environment	Config File	Secrets Source	Swagger
Development	appsettings.Development.json	Values inline	Yes
Docker	appsettings.Docker.json	.env file	Yes
Production	appsettings.Production.json	.env.prod file	No

---

Docker Setup

Quick Start

# Create external network (first time only)
docker network create emagine-network

# Build and start (API + Worker + RabbitMQ + PostgreSQL)
docker compose up -d --build

# Verify
docker compose ps
docker compose logs -f bazzuca-api
docker compose logs -f bazzuca-worker

Accessing the Application

Service	URL
API	http://localhost:5010
Swagger UI	http://localhost:5010/swagger
RabbitMQ Management	http://localhost:15672 (guest/guest)
PostgreSQL	localhost:5434

Docker Compose Commands

Action	Command
Start services	docker compose up -d
Start with rebuild	docker compose up -d --build
Stop services	docker compose stop
View logs	docker compose logs -f
View worker logs	docker compose logs -f bazzuca-worker
Remove containers	docker compose down
Remove with volumes (⚠️)	docker compose down -v

---

Manual Setup (Without Docker)

Prerequisites

• .NET 8 SDK
• PostgreSQL 17
• RabbitMQ 3.x (with management plugin)

Setup Steps

1. Configure the database

Create the database and run migrations:

dotnet ef database update --project Bazzuca.Infra --startup-project Bazzuca.API

2. Start RabbitMQ

# Via Docker (easiest)
docker run -d --name rabbitmq -p 5672:5672 -p 15672:15672 rabbitmq:3-management-alpine

3. Run the API

dotnet run --project Bazzuca.API

4. Run the Worker

dotnet run --project Bazzuca.Worker

5. Debug LinkedIn Publishing (optional)

# Opens visible Chromium browser for debugging
dotnet run --project Bazzuca.Console -- --postId 1 --tenantId bazzuca

---

API Documentation

With the application running, access Swagger UI at: http://localhost:5010/swagger

Authentication

All endpoints require the Authorization header with a JWT token issued by NAuth. Multi-tenant endpoints also require the X-Tenant-Id header.

Key Endpoints

Method	Endpoint	Description	Auth
GET	/client/listByUser	List user's clients	Yes
GET	/client/getById/{id}	Get client by ID	Yes
POST	/client/insert	Create client	Yes
POST	/client/update	Update client	Yes
DELETE	/client/delete/{id}	Soft-delete client	Yes
GET	/post/listByUser/{month}/{year}	Calendar month posts	Yes
GET	/post/getById/{id}	Get post by ID	Yes
POST	/post/insert	Create post	Yes
POST	/post/update	Update post	Yes
GET	/post/publish/{postId}	Publish post (sync or async)	Yes
POST	/post/search	Search posts (paginated)	Yes
GET	/socialnetwork/listByClient/{id}	List client networks	Yes
POST	/image/upload	Upload media to S3	Yes

Note: POST /post/publish/{postId} returns 200 OK for X/Twitter (synchronous) or 202 Accepted for LinkedIn (queued to RabbitMQ).

---

Database

Migrations

# Create new migration
dotnet ef migrations add <MigrationName> --project Bazzuca.Infra --startup-project Bazzuca.API

# Apply migrations
dotnet ef database update --project Bazzuca.Infra --startup-project Bazzuca.API

Backup

pg_dump -h localhost -p 5434 -U bazzuca_user -d bazzuca_db > backup.sql

Restore

psql -h localhost -p 5434 -U bazzuca_user -d bazzuca_db < backup.sql

---

Deployment

Production

Production deployment uses docker-compose-prod.yml with secrets from .env.prod. In production, PostgreSQL and RabbitMQ are managed externally on the emagine-network — only the API and Worker containers are deployed.

cp .env.prod.example .env.prod
# Edit .env.prod with production secrets (connection string, JWT, RabbitMQ credentials)

docker compose --env-file .env.prod -f docker-compose-prod.yml up --build -d

CI/CD (GitHub Actions)

Workflow	Trigger	Description
Version and Tag	Push to main	Auto-generates semantic version tag via GitVersion
Create Release	After version tag	Creates GitHub Release and release branch (minor/major only)
Deploy Production	Manual dispatch	SSH deploy to production server (API + Worker)

---

Troubleshooting

Common Issues

RabbitMQ connection refused

Check:

docker compose logs rabbitmq
docker compose ps

Common causes:

• RabbitMQ container not started or unhealthy
• Incorrect RabbitMQ:HostName in appsettings (use emagine-rabbitmq in Docker, localhost locally)

LinkedIn publishing fails

Check:

# Check worker logs
docker compose logs -f bazzuca-worker

# Check DLQ for failed messages
# Access RabbitMQ Management: http://localhost:15672 → Queues → bazzuca.linkedin.error

Common causes:

• LinkedIn credentials (user/password) incorrect in SocialNetwork entity
• LinkedIn detected automation (CAPTCHA) — use Console project with headless: false to debug
• Playwright browser data corrupted — delete playwright-data/client-{id}/ directory

Worker not consuming messages

Common causes:

• Queue topology not declared (check worker startup logs)
• Queues:LinkedIn config missing from appsettings

---

Author

Developed by Rodrigo Landim Carneiro

---

License

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