Skip to content

TESTING_GUIDE.md

Latest commit

 

History

History
171 lines (127 loc) · 4.08 KB

TESTING_GUIDE.md

File metadata and controls

171 lines (127 loc) · 4.08 KB

🧪 Testing Guide - Before Deployment

This guide will help you test both APIs before deploying to Vercel.

📋 Quick Test Process

Step 1: Test the Vercel-Compatible API (No Database Required!)

  1. Start the server:

    • Double-click start_vercel_api.bat
    • OR run: python api/vercel_index.py

  2. Open the frontend:

    • Open index.html in your browser
    • The URL should be: file:///C:/Users/rupga/OneDrive/Desktop/8/Bajaj_Project/index.html

  3. Test the features:

    • ✅ Without PDF: Click sample queries and submit (uses default policies)
    • ✅ With PDF: Upload a policy PDF, then submit queries
    • ✅ Check results: Verify APPROVED/REJECTED decisions

Step 2: Test the Original API

  1. Stop the Vercel API (Press Ctrl+C in the command window)

  2. Start the original server:

    • Double-click start_original_api.bat
    • OR run: python api/index.py

  3. Test with the same frontend:

    • Refresh the browser page
    • Test the same features

Step 3: Run Automated Tests

  1. Keep one API running (either one)

  2. Open a new terminal and run:

    python test_apis.py

  3. Follow the prompts to test each API

🔍 What to Check

✅ Vercel API Should:

  • Work WITHOUT any database
  • Process PDFs in memory
  • Handle queries with default policies
  • Return proper JSON responses
  • Show confidence scores and classifications

✅ Original API Should:

  • Have all the same features
  • Work with uploaded PDFs
  • Process complex queries

📊 Test Scenarios

Scenario 1: No PDF Upload

Query: "25F with 2 years policy, need IVF treatment"
Expected: REJECTED (Standard policy doesn't cover IVF)

Scenario 2: With PDF Upload

1. Upload a policy PDF
2. Query: "Is cardiac surgery covered?"
Expected: Decision based on PDF content

Scenario 3: Multiple Queries

- "45M requiring cardiac surgery" → Should detect cardiac procedure
- "30F pregnant, maternity benefits?" → Should detect maternity
- "Emergency treatment coverage" → Should detect emergency

🌐 API Endpoints to Test

Health Check

GET http://localhost:5000/health

Should return:

  • Status: healthy
  • Capabilities list
  • Message

Query Processing

POST http://localhost:5000/query
Body: {
  "query": "Your question here",
  "session_id": "optional_if_pdf_uploaded"
}

PDF Upload (Vercel API)

POST http://localhost:5000/upload
Body: Form-data with PDF file

🎯 Success Criteria

Your project is ready for deployment when:

  1. ✅ Health check returns "healthy"
  2. ✅ Queries work without PDF upload
  3. ✅ PDF upload processes successfully
  4. ✅ Queries return proper decisions (APPROVED/REJECTED)
  5. ✅ Confidence scores are shown
  6. ✅ Frontend displays results correctly

🚀 Ready to Deploy?

If all tests pass:

  1. Install Vercel CLI:

    npm install -g vercel

  2. Deploy:

    vercel

  3. Follow prompts and your app will be live!

🐛 Common Issues & Solutions

Issue: "Connection refused"

Solution: Make sure the API server is running

Issue: "No module named 'flask'"

Solution: Install dependencies: pip install -r requirements.txt

Issue: "PDF processing not available"

Solution: Install: pip install pdfplumber PyPDF2

Issue: Frontend not connecting to API

Solution: Check if API is running on port 5000

📝 Testing Checklist

  • Vercel API starts successfully
  • Health endpoint works
  • Query without PDF works
  • PDF upload works
  • Query with PDF works
  • Frontend displays results
  • Confidence scores shown
  • Decision classifications correct
  • Original API also works
  • Automated tests pass

💡 Tips

  1. Test locally first - Always test before deploying
  2. Check console - Browser console shows errors
  3. Use test script - python test_apis.py for automated testing
  4. Test both APIs - Ensure both work before choosing one
  5. Check logs - Server logs show processing details

Once everything works locally, you're ready to deploy to Vercel! 🎉