Curaitor

Curaitor is a web-based application that helps you study more effectively. It takes your course materials, organizes them, and automatically generates quizzes to help you test your knowledge.

Why Curaitor?

In today's fast-paced learning environment, it's easy to get overwhelmed with the amount of information you need to process. Curaitor helps you stay on top of your studies by:

• Automating quiz generation: Spend less time creating study materials and more time learning.
• Organizing your course materials: Keep all your notes, lectures, and readings in one place.
• Visualizing your knowledge: See how different concepts connect with an interactive file graph.

Features

• Automatic Quiz Generation: Curaitor uses AI to automatically generate quizzes from your course materials.
• Course Management: Easily upload and organize your course materials.
• Interactive File Graph: Visualize the relationships between your files and concepts.
• Study Guide Generation: Generate study guides from your course materials.

Architecture

Curaitor is a monorepo with a Go backend and a Next.js frontend.

Backend (Go)

The backend is responsible for:

Configuration (internal/config)

• Loads all program configuration (e.g., Gemini API key, dump path) via environment variables.

File Operations (internal/fileops)

• Directory watcher that monitors new/updated files.
• File move and organization logic.
• Directory tree builder (two-level deep). Example:

CS101/ assignments/, homework/, exams/
CS270/ notes/, labs/
CS370/ (no subdirectories)

Gemini Integration (internal/gemini)

• prepMessage: Prepares prompts by embedding file data.
• sendMessage: Sends requests to Gemini and returns responses.
• Handles structured extraction of file metadata (FileInfo model).

Application Data (internal/data)

• Stores courses, quizzes, study guides.
• Writes updates to local JSON files (courses.json, quizzes.json, cache.json).

Caching System

• Maintains a cache.json file to avoid re-parsing files.
• Tracks CachedFile { FilePath, Content } to ensure integrity in concurrent operations.

Quiz Generation (gemini/generateQuiz.go)

• Workers listen for course signals via a channel.
• Each worker generates quiz questions from files in a course.
• Quiz data model:

type Question struct {
  Question string
  Choices  []string
  Answer   int
}
type QuizInfo struct {
  Name     string
  Code     string
  NumFiles int
  QaPairs  []Question
}

Study Guide Generation

• Generates study guides per course instead of globally.
• Uses mutexes to avoid race conditions when multiple files are processed concurrently.
• Study guide is regenerated when a new file is added to a course’s directory.

Frontend

The frontend is a Next.js application that provides a user-friendly interface for interacting with the application. It uses the following technologies:

• Next.js: A React framework for building server-rendered applications.
• Ant Design: A UI library for React.
• React Markdown: A component for rendering Markdown content.
• Vis Network: A library for creating interactive network graphs.

---

Workflow

1. File Dumping:
   • Place course materials (syllabus, notes, assignments) into the dump_files_here directory.
2. File Watching:
   • A watcher checks the directory every 10 seconds and sends new file paths to a processing channel.
3. File Processing (Concurrent):
   • Workers extract metadata with Gemini:
     • Syllabus: Creates a new course entry in courses.json.
     • Other files: Assigns to the correct course and type; validates against existing course codes.
4. File Organization:
   • Files are moved to their respective directories:
     • destpath = schoolpath + coursecode + filetype
     • destfilename = (is syllabus? same name : fileinfo.Title)
5. Caching:
   • Before parsing, file content is checked against cache.json to avoid duplicate API calls.
6. Quiz & Study Guide Generation:
   • Quizzes are generated per course on demand.
   • Study guides update whenever new course files are added.