🛒 Django + OpenSearch Product Search Backend

This project provides a lightweight, extensible backend for product search using Django and OpenSearch. It supports tag-based filtering, free-text search, and future-ready support for semantic vector search (e.g., via CLIP or custom embeddings).

---

🔧 Tech Stack

• Django (Models, Admin, API logic)
• Django REST Framework (Search endpoint)
• OpenSearch (Indexed search + filters)
• Redis (optional) for caching
• Celery (optional) for async indexing
• Docker (recommended) for local dev

---

🚀 Features

• Index products with name, description, tags, and optional tag_vector
• Search by:
  • Free-text query (across name & description)
  • Filter by one or more tags
• Built-in support for AI-generated tags (via ProductTag model and tag_vector)
• Simple REST endpoint for search

---

📦 Installation

git clone https://github.com/your-user/django-opensearch-search.git
cd django-opensearch-search

# Setup virtual environment
python -m venv venv
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

# Start OpenSearch (use Docker or local setup)
docker-compose up -d opensearch

# Run Django
python manage.py migrate
python manage.py runserver

📁 Key Files

models.py

Defines the Product, Tag, and ProductTag models with support for tag metadata and JSON vectors.

product_indexer.py

Handles indexing of Product objects to OpenSearch, including tag names and tag vectors.

views.py

Exposes a REST API endpoint (GET /search) to:

• Search by keyword
• Filter by tags
• Return matching product documents from OpenSearch

---

🔍 Example Search Request

GET /search?tags=shoes&tags=green&q=summer

Returns all products matching the term summer in name/description AND tagged with both shoes and green.

---

🧠 Future Additions

• Cosine similarity vector search (CLIP integration)
• Faceted filtering (categories, brands, price ranges)
• Real-time updates via Celery tasks
• OpenAPI/Swagger auto-docs

---

🛠️ Index Setup Tips

• Use OpenSearch's text fields for full-text queries
• Use keyword fields for tag filters
• Define custom analyzers for fuzzy matching, synonyms, or edge n-grams if needed
• Store tag_vector as dense vectors to enable semantic search in the future

---

🤝 Contributing

Pull requests and feature suggestions are welcome! Feel free to fork and build on top of this to suit your own product catalog or tagging system.

---

📄 License

MIT — Use freely, launch fast, and remember to re-index responsibly 😉