Skip to content

~95% on SimpleQA (e.g. Qwen3.6-27B on a 3090). Supports all local and cloud LLMs (llama.cpp, Ollama, Google, ...). 10+ search engines - arXiv, PubMed, your private documents. Everything Local & Encrypted.

9.0 KB
Raw Blame
#!/usr/bin/env python3
"""
pytest: skip

This is an end-to-end test script, not a pytest test.

End-to-end test script for verifying OpenAI API key configuration.

This script helps users verify that their OpenAI API key is properly configured
and working with Local Deep Research.

Usage:
    python tests/test_openai_api_key_e2e.py --username YOUR_USERNAME --password YOUR_PASSWORD --api-key YOUR_LDR_LLM_OPENAI_API_KEY

Or set environment variables:
    export LDR_USERNAME=your_username
    export LDR_PASSWORD=your_password
    export LDR_LLM_OPENAI_API_KEY=your_api_key
    python tests/test_openai_api_key_e2e.py
"""

import argparse
import os
import sys
from pathlib import Path
from typing import Optional

# Add project root to path
sys.path.insert(0, str(Path(__file__).parent.parent.resolve()))

from local_deep_research.database.session_context import get_user_db_session
from local_deep_research.config.llm_config import get_llm
from local_deep_research.api.research_functions import quick_summary
from local_deep_research.settings import SettingsManager
from loguru import logger
import pytest

# Skip this entire module in pytest
pytestmark = pytest.mark.skip(
    reason="End-to-end test script, not a pytest test"
)


def test_openai_api_key_e2e(
    username: str, password: str, api_key: Optional[str] = None
) -> bool:
    """
    Test OpenAI API key configuration and usage.

    Args:
        username: LDR username
        password: LDR password
        api_key: OpenAI API key (optional, can use env var)

    Returns:
        bool: True if test passed, False otherwise
    """
    # audit: PUNCHLIST reviewed 2026-05 — issue resolved by prior PR (recommendation: rename file out of test_ prefix).
    print("=" * 60)
    print("OpenAI API Key Configuration Test")
    print("=" * 60)

    try:
        # Step 1: Authenticate
        print("\n1. Authenticating with LDR...")
        with get_user_db_session(
            username=username, password=password
        ) as session:
            print("   ✓ Authentication successful")

            # Step 2: Configure settings
            print("\n2. Configuring OpenAI settings...")
            settings_manager = SettingsManager(session)

            # Set OpenAI as provider
            settings_manager.set_setting("llm.provider", "openai")
            print("   ✓ Set provider to OpenAI")

            # Set API key if provided
            if api_key:
                settings_manager.set_setting("llm.openai.api_key", api_key)
                print("   ✓ Set OpenAI API key")
            else:
                # Check if API key exists in settings or environment
                existing_key = settings_manager.get_setting(
                    "llm.openai.api_key"
                )
                env_key = os.getenv("LDR_LLM_OPENAI_API_KEY")

                if not existing_key and not env_key:
                    print("   ✗ No API key found in settings or environment")
                    print(
                        "     Please provide --api-key or set LDR_LLM_OPENAI_API_KEY environment variable"
                    )
                    return False

                if existing_key:
                    print("   ✓ Using existing API key from settings")
                else:
                    print("   ✓ Using API key from environment variable")

            # Set model
            settings_manager.set_setting("llm.model", "gpt-3.5-turbo")
            print("   ✓ Set model to gpt-3.5-turbo")

            # Get settings snapshot
            settings_snapshot = settings_manager.get_all_settings()

            # Step 3: Test LLM initialization
            print("\n3. Testing LLM initialization...")
            try:
                llm = get_llm(settings_snapshot=settings_snapshot)
                print("   ✓ LLM initialized successfully")
            except Exception as e:
                print(f"   ✗ Failed to initialize LLM: {str(e)}")
                if "api key" in str(e).lower():
                    print("     This appears to be an API key issue")
                return False

            # Step 4: Test simple LLM call
            print("\n4. Testing LLM response...")
            try:
                from langchain_core.messages.human import HumanMessage

                response = llm.invoke(
                    [
                        HumanMessage(
                            content="Say 'Hello, LDR!' in 5 words or less"
                        )
                    ]
                )
                print(f"   ✓ LLM responded: {response.content}")
            except Exception as e:
                print(f"   ✗ Failed to get LLM response: {str(e)}")
                if "401" in str(e) or "unauthorized" in str(e).lower():
                    print("     API key appears to be invalid")
                elif "429" in str(e):
                    print(
                        "     Rate limit exceeded - API key is valid but quota exhausted"
                    )
                return False

            # Step 5: Test research functionality
            print("\n5. Testing research functionality...")
            try:
                result = quick_summary(
                    query="What is the capital of France?",
                    settings_snapshot=settings_snapshot,
                    iterations=1,
                    questions_per_iteration=1,
                    search_tool="wikipedia",  # Use Wikipedia to avoid web search rate limits
                )

                print("   ✓ Research completed successfully")
                print(f"   Research ID: {result.get('research_id', 'N/A')}")
                print(
                    f"   Summary preview: {result.get('summary', '')[:100]}..."
                )

                if "paris" in result.get("summary", "").lower():
                    print("   ✓ Result contains expected content")
                else:
                    print("   ⚠ Result may not contain expected content")

            except Exception as e:
                print(f"   ✗ Failed to complete research: {str(e)}")
                return False

            # Success!
            print("\n" + "=" * 60)
            print(
                "✅ All tests passed! Your OpenAI API key is configured correctly."
            )
            print("=" * 60)

            # Show configuration summary
            print("\nCurrent Configuration:")
            print(
                f"  Provider: {settings_snapshot.get('llm.provider', {}).get('value', 'Not set')}"
            )
            print(
                f"  Model: {settings_snapshot.get('llm.model', {}).get('value', 'Not set')}"
            )
            print(f"  API Key: {'*' * 20} (hidden)")

            return True

    except Exception as e:
        print(f"\n✗ Unexpected error: {str(e)}")
        logger.exception("Test failed with exception")
        return False


def main():
    """Main entry point for the test script."""
    parser = argparse.ArgumentParser(
        description="Test OpenAI API key configuration with Local Deep Research",
        formatter_class=argparse.RawDescriptionHelpFormatter,
        epilog="""
Examples:
  # With command line arguments
  python test_openai_api_key_e2e.py --username myuser --password mypass --api-key sk-...

  # With environment variables
  export LDR_USERNAME=myuser
  export LDR_PASSWORD=mypass
  export LDR_LLM_OPENAI_API_KEY=sk-...
  python test_openai_api_key_e2e.py

Note: The LDR server must be running and you must have a user account created.
        """,
    )

    parser.add_argument(
        "--username",
        default=os.getenv("LDR_USERNAME"),
        help="LDR username (or set LDR_USERNAME env var)",
    )
    parser.add_argument(
        "--password",
        default=os.getenv("LDR_PASSWORD"),
        help="LDR password (or set LDR_PASSWORD env var)",
    )
    parser.add_argument(
        "--api-key",
        default=os.getenv("LDR_LLM_OPENAI_API_KEY"),
        help="OpenAI API key (or set LDR_LLM_OPENAI_API_KEY env var)",
    )
    parser.add_argument(
        "--verbose", action="store_true", help="Enable verbose logging"
    )

    args = parser.parse_args()

    # Validate required arguments
    if not args.username or not args.password:
        print("Error: Username and password are required")
        print(
            "Provide them via --username/--password or LDR_USERNAME/LDR_PASSWORD env vars"
        )
        parser.print_help()
        sys.exit(1)

    # Configure logging
    if not args.verbose:
        logger.remove()  # Remove default handler
        # diagnose=False: this e2e harness passes args.api_key and
        # args.password into the test function — both end up as frame locals
        # that loguru's default diagnose=True would dump into error tracebacks
        # (#4185 / #4384).
        logger.add(sys.stderr, level="ERROR", diagnose=False)

    # Run the test
    success = test_openai_api_key_e2e(
        username=args.username, password=args.password, api_key=args.api_key
    )

    # Exit with appropriate code
    sys.exit(0 if success else 1)


if __name__ == "__main__":
    main()