1. Overview

QueryBuilder is a modular backend system that allows loading CSV/Excel files into a database, running read-only SQL queries, inspecting tables and schemas, and exporting query results. The backend is designed to be engine-agnostic, supporting multiple SQL engines (SQLite, DuckDB planned).

Key Features

• Load CSV/Excel files into a local SQL database
• Abstracted database engine support
• Query execution with read-only validation
• Table and schema inspection
• Export query results to CSV or Excel
• Modular, extensible architecture

2. Architecture

Modules

1. Core

   • Provides constants, config, logging, and custom exceptions
   • Ensures consistent behavior and centralized logging

2. Engines

   • EngineLoader: manages connections and cleanup
   • EngineQuery: executes queries and fetches results

   Supports SQLite (currently) and planned DuckDB. We use raw database APIs instead of ORM or third-party wrappers. It handles connection pooling internally and provides a context manager for safe resource handling

3. Loader

   • Loads files into the database
   • CSV files → single table
   • Excel files → each sheet as a table
   • Tracks table names

4. Query Builder

   • High-level query execution
   • Validates queries to allow only read-only operations
   • Returns results as pandas DataFrames

5. Main (QuerySession)

   • Entry point for CLI/GUI
   • Manages engine, loader, query builder
   • Provides unified interface for queries, schema, tables, export, cleanup

3. Data Flow

1. File Upload/Load
   • User provides CSV/Excel file
   • FileLoader loads data into the selected database engine
   • Tables are tracked for inspection