test_hf_spaces_session.py

#!/usr/bin/env python3
"""
Quick test to diagnose the current HF Spaces chat issue.
Run this after uploading a document to test the session state.
"""
import requests
import json

# Replace with your actual Space URL
BASE_URL = "https://hyperxd-0-cognichat.hf.space"  # Update this to your actual Space URL

def test_endpoints():
    """Test the debug and session endpoints to understand the issue."""
    
    print("=== CogniChat HF Spaces Diagnostic ===\n")
    
    # Test 1: Check debug endpoint
    print("1. Testing /debug endpoint...")
    try:
        response = requests.get(f"{BASE_URL}/debug")
        if response.status_code == 200:
            data = response.json()
            print("✓ Debug endpoint working")
            print(f"   Environment: {data.get('environment')}")
            print(f"   GROQ API Key: {'Set' if data.get('groq_api_key_set') else 'NOT SET'}")
            print(f"   Sessions count: {data.get('sessions_count')}")
            print(f"   Upload folder: {data.get('upload_folder')}")
            print(f"   Upload folder writable: {data.get('upload_folder_writable')}")
            print(f"   Flask session ID: {data.get('flask_session_id')}")
            print(f"   Session keys: {data.get('flask_session_keys')}")
        else:
            print(f"✗ Debug endpoint failed: {response.status_code}")
    except Exception as e:
        print(f"✗ Error accessing debug endpoint: {e}")
    
    print()
    
    # Test 2: Check session handling
    print("2. Testing /test-session endpoint...")
    try:
        # Create a session with cookies
        session = requests.Session()
        
        # Test session write
        response = session.post(f"{BASE_URL}/test-session")
        if response.status_code == 200:
            data = response.json()
            print("✓ Session write working")
            print(f"   Test key: {data.get('test_key')}")
            print(f"   Session keys: {data.get('session_keys')}")
        else:
            print(f"✗ Session write failed: {response.status_code}")
        
        # Test session read
        response = session.get(f"{BASE_URL}/test-session")
        if response.status_code == 200:
            data = response.json()
            print("✓ Session read working")
            print(f"   Test key persisted: {data.get('test_key')}")
            print(f"   Has session data: {data.get('has_session_data')}")
            
            if not data.get('test_key'):
                print("⚠️  WARNING: Sessions are not persisting between requests!")
                print("   This is likely the cause of the 400 chat error.")
        else:
            print(f"✗ Session read failed: {response.status_code}")
            
    except Exception as e:
        print(f"✗ Error testing sessions: {e}")
    
    print()
    
    # Test 3: Check if we can find any existing sessions
    print("3. Checking for existing RAG sessions...")
    try:
        response = requests.get(f"{BASE_URL}/debug")
        if response.status_code == 200:
            data = response.json()
            session_ids = data.get('session_ids', [])
            if session_ids:
                print(f"✓ Found {len(session_ids)} existing RAG sessions")
                print(f"   Session IDs: {session_ids[:3]}{'...' if len(session_ids) > 3 else ''}")
            else:
                print("ℹ️  No RAG sessions found (normal if no documents were uploaded)")
        
    except Exception as e:
        print(f"✗ Error checking RAG sessions: {e}")
    
    print()
    print("=== Diagnosis Complete ===")
    print()
    print("LIKELY ISSUE:")
    print("If sessions are not persisting, this is a common issue in HF Spaces")
    print("where Flask sessions don't work properly across requests.")
    print()
    print("SOLUTION:")  
    print("We need to modify the app to use a different session storage method")
    print("or pass session ID through request body instead of Flask sessions.")

if __name__ == "__main__":
    print("Before running this script:")
    print("1. Update BASE_URL with your actual HF Spaces URL")
    print("2. Make sure your Space is running")
    print("3. Optionally upload a document first")
    print()
    
    # Uncomment the next line and update the URL to run the test
    # test_endpoints()
    
    print("Update the BASE_URL variable above and uncomment the test_endpoints() call")