📚 Database Module Documentation

File: app/db/database.py This module provides a streamlined way to connect to and manage database sessions for your application. It supports both PostgreSQL and SQLite, with environment-based configuration for flexible deployment.

1. ⚙️ Configuration

The database connection is controlled using environment variables. If these are not set, the module uses sensible defaults.

📌 Environment Variables

💡 Example: Switch to SQLite

export DB_MODE="sqlite"

2. 🧱 Core Components

This module exposes several key components used to interface with the database.

engine

• A SQLAlchemy Engine instance.
• Manages connections, pooling, and execution context.

SessionLocal

• A factory for creating new database session objects.
• Typically accessed indirectly through the get_db dependency.

Base

• The base class all SQLAlchemy models should inherit from.
• Used to declare table mappings.

get_db()

• A FastAPI dependency function.
• Yields a new session per request and ensures it's closed afterward, even in case of errors.

3. 🚀 Usage with FastAPI

To safely interact with the database in your FastAPI routes, use the get_db() dependency.

✅ Steps to Use

1. Import get_db:

   from app.db.database import get_db

2. Add it as a dependency:

   from fastapi import Depends
   from sqlalchemy.orm import Session

3. Inject into your route handler:

   @app.get("/items/")
   def read_items(db: Session = Depends(get_db)):
       return db.query(Item).all()

This pattern ensures that every request has a dedicated, isolated database session that is properly cleaned up afterward.

Here is your fully formatted and organized documentation for the SQLAlchemy models in app/db/models.py, suitable for technical documentation or a README:

🗂️ Database Models Documentation

This document describes the SQLAlchemy models defined in app/db/models.py. These classes represent the tables used to store application data, including chat sessions, messages, and document metadata for Retrieval-Augmented Generation (RAG).

All models inherit from the Base class, imported from the database.py module.

1. 🗨️ Session Model

Represents a single conversation between a user and an AI. It serves as the container for all messages in that session.

🔑 Key Fields

🔗 Relationships

• messages: One-to-many relationship with the Message model. Deleting a session will also delete all related messages (via cascade="all, delete-orphan").

2. 💬 Message Model

Stores individual messages within a session, including user inputs and AI responses.

🔑 Key Fields

🔗 Relationships

• session: Many-to-one relationship with the Session model.

3. 📄 Document Model

Stores metadata and content of documents ingested into the system for RAG purposes.

🔑 Key Fields

🔗 Relationships

• vector_metadata: One-to-one relationship with the VectorMetadata model.

4. 🧠 VectorMetadata Model

Connects documents to their vector representations (e.g., FAISS indices) for efficient retrieval in RAG workflows.

🔑 Key Fields

🔗 Relationships

• document: Many-to-one relationship with the Document model

• session: Many-to-one relationship with the Session model