test_agent.py

#!/usr/bin/env python3
"""
Test script to debug the 'list' object has no attribute 'lower' error
"""

import os
import sys

# Add current directory to path
sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))

# Set test API keys
os.environ["ANTHROPIC_API_KEY"]= "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"

# Mock the API calls to avoid actual API usage
from unittest.mock import patch, MagicMock

def test_agent_with_various_inputs():
    """Test the agent with different input types that might cause errors"""

    print("Testing agent with various input types...")

    # Test cases that might cause the error
    test_cases = [
        # Normal string
        "What is 2 + 2?",

        # Question with image reference
        "Look at the image and tell me what you see",

        # Question with list-like content
        "Calculate the sum of [1, 2, 3, 4, 5]",

        # Question with code
        "What is the output of this code:\n```python\nprint([1, 2, 3])\n```",

        # Reversed text question
        ".rewsna eht sa 'tfel' drow eht fo etisoppo eht etirw",

        # Question with attachment reference
        "What is the final numeric output from the attached Python code?",
    ]

    # Import the agent
    try:
        from app import LangGraphAgent, _clean_answer

        # Test the _clean_answer function directly with different inputs
        print("\n1. Testing _clean_answer function:")
        print("-" * 50)

        test_answers = [
            "42",
            ["The", "answer", "is", "42"],  # List input
            {"answer": "42"},  # Dict input
            42,  # Integer
            None,  # None
            ["list", "with", "numbers", 1, 2, 3],  # Mixed list
        ]

        # Create a mock agent to test _clean_answer
        class MockAgent:
            def _clean_answer(self, answer):
                # This is the current implementation
                answer = answer.strip()  # This will fail on lists!

                lower_answer = answer.lower()  # This will also fail!
                return answer

        mock_agent = MockAgent()

        for test_answer in test_answers:
            print(f"\nTesting with: {test_answer} (type: {type(test_answer)})")
            try:
                result = mock_agent._clean_answer(test_answer)
                print(f"✅ Success: {result}")
            except AttributeError as e:
                print(f"❌ AttributeError: {e}")
            except Exception as e:
                print(f"❌ Other error: {type(e).__name__}: {e}")

        # Test with actual agent if possible
        print("\n\n2. Testing with tool responses that might return lists:")
        print("-" * 50)

        # Mock tool responses that might cause issues
        tool_responses = [
            # Normal response
            {"tool": "calculator", "output": "42"},

            # List response (this might be the issue!)
            {"tool": "python_executor", "output": ["Result:", "42"]},

            # Complex response
            {"tool": "web_search", "output": {"results": ["item1", "item2"]}},
        ]

        for response in tool_responses:
            print(f"\nTool response: {response}")
            output = response.get("output", "")
            print(f"Output type: {type(output)}")
            if isinstance(output, list):
                print("⚠️  This is a LIST - might cause 'lower' error!")

    except ImportError as e:
        print(f"Import error: {e}")
    except Exception as e:
        print(f"Unexpected error: {type(e).__name__}: {e}")

def test_message_content_types():
    """Test what types of content messages might contain"""
    print("\n\n3. Testing message content types:")
    print("-" * 50)

    from langchain_core.messages import HumanMessage, AIMessage

    # Test different message contents
    test_contents = [
        "Normal string message",
        ["List", "as", "content"],  # This might happen!
        {"type": "image", "data": "base64..."},  # Multimodal content
        None,
    ]

    for content in test_contents:
        print(f"\nTesting message with content: {content} (type: {type(content)})")
        try:
            msg = AIMessage(content=content)
            print(f"Message created successfully")
            print(f"Message.content type: {type(msg.content)}")
        except Exception as e:
            print(f"Error creating message: {e}")

if __name__ == "__main__":
    print("=" * 60)
    print("GAIA Agent Error Debugging Test")
    print("=" * 60)

    test_agent_with_various_inputs()
    test_message_content_types()

    print("\n\nConclusion:")
    print("-" * 50)
    print("The error likely occurs when:")
    print("1. A tool returns a list instead of a string")
    print("2. The message content is a list (multimodal)")
    print("3. The _clean_answer method tries to call .strip() or .lower() on a list")
    print("\nFix: Add type checking in _clean_answer method!")