🎭 Emotube: YouTube Comment Sentiment Analyzer

An experimental web application for analyzing sentiment in YouTube comments

---

🌟 Overview

Emotube fetches YouTube video metadata and top-level comments through the YouTube Data API, classifies comments with a pretrained Hugging Face sentiment model, and stores the results in PostgreSQL. A React interface displays video details, comment filters, and sentiment charts. YouTube is the only implemented platform; replies within comment threads are not fetched.

✨ Key Features

• 🎯 Sentiment Classification - Positive, neutral, negative, and ambiguous labels with model scores
• 📊 Charts - Sentiment counts over time with date-range selection
• 🐳 Docker Compose - Frontend, backend, and database services for local development
• ⚡ Background Analysis - Comment fetching and classification through FastAPI background tasks; the frontend polls every 3 seconds
• 🔍 Comment Filters - Sentiment, author, minimum likes, text search, and sorting
• 📈 Video Metrics - View, like, and comment counts plus derived engagement statistics

---

🖥️ User Interface Showcase

🏠 Home Page & Video Analysis

Home Page

YouTube URL Input and Analysis Initiation

💬 Comment Analysis Interface

Video Information

Interactive Sentiment Analysis Charts

Chart Filtering by Time Period

Comment Sentiment Labels and Filters

✨ Key UI Features

• 🎨 Interface - React components styled with Tailwind CSS and ShadCN/UI
• 📊 Charts - Sentiment visualization with Recharts
• 🔍 Filters - Comment filtering and sorting
• ⚡ Progress Display - Periodic polling of analysis status
• 🌙 Theme Support - A theme provider with dark mode as the default

---

🏗️ Architecture

Docker Compose runs a React frontend, a FastAPI backend, and PostgreSQL. Analysis runs inside the backend process; there is no separate worker service or durable job queue.

Emotube/
├── 🐳 docker-compose.yml           # Multi-service orchestration
├── 🔧 .env.example                 # Environment configuration template
├── 🌐 frontend/                    # React TypeScript Frontend
│   ├── 🐳 Dockerfile               # Frontend container config
│   ├── 📦 package.json             # Dependencies (Bun runtime)
│   ├── ⚙️ vite.config.js           # Build configuration
│   └── 📁 src/
│       ├── 🎨 components/          # React components
│       │   ├── ui/                 # ShadCN UI components
│       │   ├── VideoDetails.tsx    # Video metadata display
│       │   ├── CommentSection.tsx  # Comment analysis interface
│       │   ├── CommentCharts.tsx   # Data visualization charts
│       │   └── AnalyzedVideoList.tsx # Video history
│       ├── 🚦 routes/              # TanStack Router pages
│       ├── 🔗 services/            # API service layer
│       ├── 🎯 hooks/               # Custom React hooks
│       └── 📝 types/               # TypeScript definitions
├── ⚡ backend/                     # FastAPI Python Backend
│   ├── 🐳 Dockerfile               # Backend container config
│   ├── 📋 requirements.txt         # Python dependencies
│   ├── 📚 overview.md              # Detailed backend documentation
│   └── 📁 app/
│       ├── 🚀 main.py              # FastAPI application entry
│       ├── 🌐 api/
│       │   ├── routes/             # API endpoints
│       │   │   ├── videos.py       # Video analysis endpoints
│       │   │   ├── comments.py     # Comment retrieval & filtering
│       │   │   └── chart_data.py   # Analytics data endpoints
│       │   └── logic/              # Business logic layer
│       ├── ⚙️ core/
│       │   ├── config.py           # Application configuration
│       │   ├── sentiment/          # Sentiment analysis engine
│       │   └── integrations/       # Social media platform integrations
│       │       └── youtube/        # YouTube API integration
│       ├── 🗄️ db/
│       │   ├── session.py          # Database session management
│       │   └── init_db.py          # Database initialization
│       ├── 📊 models/              # SQLModel database models
│       ├── 📝 schemas/             # Pydantic API schemas
│       ├── 🔧 crud/                # Database operations
│       └── 🛠️ utils/               # Utility functions
└── 🗄️ PostgreSQL Database         # Persistent data storage

---

🚀 Tech Stack

Backend

• 🐍 Python 3.11 with FastAPI - API routes and background tasks
• 🤖 Transformers & PyTorch - Pretrained sentiment classifier
• 🗄️ PostgreSQL + SQLModel - Database models and async sessions
• 📡 Google API Client - YouTube Data API integration
• ⚡ Uvicorn - ASGI server with auto-reload

Frontend

• ⚛️ React 19 with TypeScript - Frontend components
• 🦄 Bun - JavaScript runtime and package manager
• 🧭 TanStack Router - Type-safe routing with devtools
• 🔄 TanStack Query - API queries and polling
• 🎨 Tailwind CSS - Utility-first styling framework
• 🧩 ShadCN/UI - UI components
• 📊 Recharts - Responsive data visualization
• ⚡ Vite - Development server and build tool

Infrastructure

• 🐳 Docker & Docker Compose - Local service configuration
• 🔧 Environment Variables - API key, database connection, and service URLs
• 🌐 CORS-enabled API - Cross-origin resource sharing
• 📝 Auto-generated API Docs - Interactive Swagger/OpenAPI

---

🚀 Quick Start

Prerequisites

• 🐳 Docker and Docker Compose
• 🔑 YouTube Data API v3 Key (Get one here)
• Network access for dependencies, initial model downloads, and YouTube API requests; comment fetching uses your API quota.

1. Clone the Repository

git clone https://github.com/00200200/youtube-comment-sentiment-analyzer
cd youtube-comment-sentiment-analyzer

2. Environment Setup

# Copy environment template
cp .env.example .env

# Edit .env file with your configuration
nano .env  # or use your preferred editor

Required environment variables:

# YouTube API Configuration
YOUTUBE_API_KEY=your_youtube_api_key_here

# Service Configuration
HOST=0.0.0.0
BACKEND_PORT=8000
FRONTEND_PORT=3000

# Database Configuration
POSTGRES_USER=emotube_user
POSTGRES_PASSWORD=secure_password_here
POSTGRES_DB=emotube_db
POSTGRES_HOST=db
POSTGRES_PORT=5432
DATABASE_URL=postgresql+asyncpg://${POSTGRES_USER}:${POSTGRES_PASSWORD}@${POSTGRES_HOST}:${POSTGRES_PORT}/${POSTGRES_DB}

# Frontend Configuration
VITE_BACKEND_URL=http://localhost:8000

3. Launch with Docker

# Start all services
docker-compose up -d

# View logs (optional)
docker-compose logs -f

4. Access the Application

• 🌐 Frontend: http://localhost:3000
• ⚡ Backend API: http://localhost:8000
• 📚 API Docs: http://localhost:8000/docs
• 🗄️ Database: localhost:5432

The supplied Compose setup uses Vite's development server, Uvicorn auto-reload, source mounts, and published database/API ports. The API has no authentication and allows all CORS origins. Treat this as a local development setup, not a production deployment template. Initial backend startup may take time to download and load the model.

---

🎯 Usage

Analyzing YouTube Videos

1. 📱 Navigate to the frontend application
2. 🔗 Enter a YouTube video URL
3. ⚡ Watch analysis status update through periodic polling
4. 📊 Explore interactive charts and comment insights
5. 🔍 Filter comments by sentiment, author, or keywords
6. 📈 View sentiment counts over time and video engagement metrics

API Endpoints

Videos

• GET /videos - List analyzed videos (paginated)
• GET /videos?url={youtube_url} - Analyze new video or get existing analysis

Comments

• GET /comments?url={youtube_url} - Get stored comments with filtering:
  • sentiment - Filter by sentiment (positive/negative/neutral/ambiguous)
  • author - Filter by author name
  • min_likes - Minimum like count
  • phrase - Text search within comments
  • sort_by - Sort field (for example, published_at, like_count, or sentiment_label)
  • sort_order - asc/desc

Analytics

• GET /chart-data?url={youtube_url} - Get visualization data

---

🌐 Platform Scope

The only integration is backend/app/core/integrations/youtube/. The routes, URL parsing, data models, and frontend are built around YouTube videos. Supporting another platform would require implementation and testing across those parts of the application.

---

🛠️ Development

Local Development Setup

This setup requires Python 3.11 and Bun. Configure the root .env first. For processes outside Docker, use localhost in DATABASE_URL instead of the Compose hostname db. The example below uses the credentials shown above; replace them with your configured values.

# Backend development, from the repository root
docker-compose up -d db
cd backend
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate
pip install -r requirements.txt
DATABASE_URL=postgresql+asyncpg://emotube_user:secure_password_here@localhost:5432/emotube_db \
  uvicorn app.main:app --reload --env-file ../.env

# Frontend development, in a separate terminal from the repository root
cd frontend
bun install
VITE_BACKEND_URL=http://localhost:8000 bun run start

Docker Development

# Rebuild and restart services
docker-compose down
docker-compose up --build

# View service logs
docker-compose logs backend
docker-compose logs frontend
docker-compose logs db

Database Management

# Access PostgreSQL
docker-compose exec db psql -U emotube_user -d emotube_db

# Backup database
docker-compose exec db pg_dump -U emotube_user emotube_db > backup.sql

# View database logs
docker-compose logs db

---

📊 Features Deep Dive

Sentiment Analysis Engine

• 🤖 Model - cardiffnlp/twitter-xlm-roberta-base-sentiment, configured in config.py
• 🎯 Scores - Stores the model's predicted label and score; scores below 0.6 are labeled ambiguous
• 🔄 Processing - Fetches pages of top-level comments, then classifies and saves comments individually
• 📏 Input Limit - Model inputs are truncated to 512 tokens

Data Visualization

• 📊 Timeline - Area charts of comment counts by sentiment and publication date
• 🔍 Comment View - Filtered comments, model scores, and sentiment totals
• 📈 Video Details - Counts and derived engagement metrics; no word clouds or dislike-correlation analysis are implemented

Limitations

• Analysis is an on-demand background pass, not a stream of new YouTube comments. Polling refreshes stored results and status.
• Model scores are predictions, not verified measurements of a commenter's intent. The repository does not report a sentiment-accuracy evaluation on YouTube comments.
• API availability, quota, comment volume, and local inference speed affect completion time. This project does not provide a real-time latency guarantee.

---