MiniDB

Overview

MiniDB is my answer to the question: "How does a database actually work under the hood?"

While most developers spend their time learning how to query databases, I wanted to learn how to build one. MiniDB isn't a wrapper around SQLite or MySQL. It is a raw, high-performance database engine written in C++17 that handles its own memory, file I/O, and binary serialization.

To make it usable for humans, I built a modern Flask (Python) web interface that communicates with the C++ engine via system subprocesses.

Key Features

Raw Engineering

• Custom Binary Storage: Data isn't saved as text; it's serialized into binary packets. This teaches the fundamentals of how real DBMS systems optimize storage.
• Dynamic Schema System: Just like a real DB, you define structure in schema.txt, and the engine adapts its binary parsing logic instantly.
• 5,200 Ops/Sec: Benchmarked to handle heavy read/write loads efficiently for a file-based system.

The Full-Stack Bridge

• Polyglot Architecture: The backend is C++ (for speed), the middleware is Python (for logic), and the frontend is HTML/JS (for interaction).
• Logical Deletes: Implemented "Soft Delete" mechanisms to preserve data integrity before purging.

System Architecture

The project demonstrates a clean separation of concerns between the Data Engine and the User Interface.

The Schema Logic

Everything starts with the schema.txt. The C++ engine reads this on boot to understand how to parse the binary blocks.

// schema.txt
id:int,name:string,age:int,class:char

The Data Flow

# How the stack communicates
1. User clicks "Save" on Web UI
2. Flask server catches request
3. Python spawns a C++ Subprocess:
   $ ./mydb.exe insert id=1 name=Herin age=21
4. C++ Engine serializes data -> writes to dbfile.db (Binary)
5. Success/Error code returned to UI