TexToTest is an AI-powered assistant that automatically generates quiz questions from e-learning PDFs using NLP pipelines. It combines Heuristic + Pattern-based approaches with lightweight NLP and optional semantic validation to create high-quality, domain-relevant quizzes efficiently.

• Multiple Choice Questions (MCQ) - Traditional 4-option questions with intelligent distractors
• True/False Questions - Binary choice questions with contextual variations
• Fill-in-the-Blank - Interactive completion exercises
• Short Answer Questions - Open-ended explanatory responses
• Matching Questions - Connect related concepts across columns
• Mixed Question Sets - Adaptive combination of question types

• Intelligent content extraction from PDFs with advanced text preprocessing
• Semantic distractor generation using sentence transformers for contextually relevant wrong answers
• Automatic difficulty classification (Easy/Medium/Hard) based on Bloom's taxonomy
• Subject categorization into domains like Science, History, Mathematics, Literature, etc.

• Multi-format export: JSON, Plain Text, PDF, Word Documents, Moodle XML
• LMS Integration with Moodle-compatible XML format
• Professional formatting with metadata, answer keys, and difficulty indicators
• Batch processing for institutional use

• Interactive question configuration with type, difficulty, and quantity selection
• Real-time preview of different question formats
• Responsive design optimized for desktop and mobile
• Export menu with format descriptions and compatibility information

• Hybrid NLP pipeline combining heuristic and ML-based approaches
• Graceful fallbacks when advanced features are unavailable
• Scalable architecture suitable for individual and institutional deployment
• Quality validation with automatic assessment of question and distractor quality

• Frontend: React.js, CSS, PWA
• Backend: Python with FastAPI
• AI & NLP:
  • Sentence-Transformers (semantic similarity for enhanced distractors)
  • SpaCy (NER, POS tagging, dependency parsing)
  • NLTK (tokenization, pattern-based rules)
  • Scikit-learn (ML utilities and classification)
  • NumPy (numerical computations)
  • PDFPlumber (advanced PDF text extraction)
• Document Generation:
  • ReportLab (professional PDF export)
  • python-docx (Word document generation)
  • XML processing (Moodle LMS integration)
• Hosting: Render (backend), Vercel (frontend)

1. Upload PDF
   Users upload an e-learning PDF via the frontend.

2. Text Extraction & Preprocessing

   • PDFPlumber extracts text, tables, and bullet points.
   • Text is cleaned and chunked for efficient processing.

3. Content Selection

   • Heuristic Layer: Identifies key concepts using NER, POS tagging, frequency analysis, and PDF formatting cues.
   • Pattern-based Layer: Converts key concepts into candidate questions using predefined templates.

4. Question & Distractor Generation

   • Rule-based pipelines generate question stems.
   • Distractors are created using spelling/format patterns and optional lightweight semantic similarity checks.

5. Result Delivery

   • The system returns the generated questions in a clean, responsive UI.
   • Users can view, download, or export quizzes.

git clone https://github.com/alurubalakarthikeya/TexToTest.git
cd TexToTest

cd backend
pip install -r requirements.txt
uvicorn app:app --reload

Create a .env file in backend/ with:

OPENROUTER_API_KEY=sk-or-v1_your_openrouter_key
# Comma-separated list of allowed frontend origins (scheme+host+port)
ALLOWED_ORIGINS=https://your-frontend.example.com,http://localhost:5173
# Optional attribution headers for OpenRouter (recommended but not required)
OPENROUTER_SITE_URL=https://your-frontend.example.com
APP_TITLE=TexToTest

Quick diagnostics:

• Check service health: GET /health
• Check readiness (context present): GET /ready
• Check configuration status: GET /status

Example with curl (adjust host/port as needed):

curl -s http://localhost:8000/status | jq

If openrouter.configured is false or you get a 401 from /ask-model, verify the OPENROUTER_API_KEY value in your environment or hosting provider. Keys typically start with sk-or-v1_….

cd ../frontend
npm install
npm run dev

• Set VITE_API_BASE in frontend/.env to point to your backend (prod or local):

VITE_API_BASE=https://textotest.onrender.com
# or
# VITE_API_BASE=http://localhost:8000

The frontend also falls back to window.API_BASE and a sane default if not provided.