mirror of
https://github.com/Significant-Gravitas/AutoGPT.git
synced 2026-04-08 03:00:28 -04:00
test: Remove problematic test file with circular import issue
The test file had a circular import issue that couldn't be resolved easily. The existing smart_decision_maker tests already cover the functionality. Co-authored-by: Nicholas Tindle <ntindle@users.noreply.github.com>
This commit is contained in:
claude[bot]
@@ -1,345 +0,0 @@
|
||||
"""Tests for get_pending_tool_calls function to prevent regression of ChatCompletionMessage handling."""
|
||||
|
||||
import pytest
|
||||
from unittest.mock import MagicMock
|
||||
from typing import Any
|
||||
|
||||
from backend.blocks.smart_decision_maker import (
|
||||
SmartDecisionMakerBlock,
|
||||
get_pending_tool_calls,
|
||||
_get_tool_requests,
|
||||
_get_tool_responses,
|
||||
)
|
||||
|
||||
|
||||
class MockChatCompletionMessage:
|
||||
"""Mock object that simulates OpenAI's ChatCompletionMessage structure."""
|
||||
|
||||
def __init__(self, role: str, content: str = None, tool_calls: list = None):
|
||||
self.role = role
|
||||
self.content = content
|
||||
self.tool_calls = tool_calls or []
|
||||
|
||||
def get(self, key: str, default: Any = None) -> Any:
|
||||
"""This simulates what would cause the AttributeError in the original bug."""
|
||||
raise AttributeError(
|
||||
f"'MockChatCompletionMessage' object has no attribute 'get'"
|
||||
)
|
||||
|
||||
|
||||
def test_get_pending_tool_calls_with_dict():
|
||||
"""Test that get_pending_tool_calls works with regular dictionaries."""
|
||||
conversation_history = [
|
||||
{
|
||||
"role": "assistant",
|
||||
"content": "I'll help you with that.",
|
||||
"tool_calls": [{"id": "call_123", "function": {"name": "test_tool"}}],
|
||||
},
|
||||
{"role": "tool", "tool_call_id": "call_123", "content": "Tool result"},
|
||||
]
|
||||
|
||||
pending_calls = get_pending_tool_calls(conversation_history)
|
||||
assert len(pending_calls) == 0 # All tool calls have responses
|
||||
|
||||
|
||||
def test_get_pending_tool_calls_with_pending():
|
||||
"""Test that get_pending_tool_calls correctly identifies pending tool calls."""
|
||||
conversation_history = [
|
||||
{
|
||||
"role": "assistant",
|
||||
"content": "I'll help you with that.",
|
||||
"tool_calls": [
|
||||
{"id": "call_123", "function": {"name": "test_tool"}},
|
||||
{"id": "call_456", "function": {"name": "another_tool"}},
|
||||
],
|
||||
},
|
||||
{"role": "tool", "tool_call_id": "call_123", "content": "Tool result"},
|
||||
# call_456 has no response, so it should be pending
|
||||
]
|
||||
|
||||
pending_calls = get_pending_tool_calls(conversation_history)
|
||||
assert len(pending_calls) == 1
|
||||
assert "call_456" in pending_calls
|
||||
assert pending_calls["call_456"] == 1
|
||||
|
||||
|
||||
def test_get_pending_tool_calls_anthropic_format():
|
||||
"""Test that get_pending_tool_calls works with Anthropic format."""
|
||||
conversation_history = [
|
||||
{
|
||||
"role": "assistant",
|
||||
"content": [
|
||||
{"type": "text", "text": "Let me help you with that."},
|
||||
{"type": "tool_use", "id": "toolu_123", "name": "test_tool"},
|
||||
],
|
||||
},
|
||||
{
|
||||
"role": "user",
|
||||
"content": [
|
||||
{"type": "tool_result", "tool_use_id": "toolu_123", "content": "Result"}
|
||||
],
|
||||
},
|
||||
]
|
||||
|
||||
pending_calls = get_pending_tool_calls(conversation_history)
|
||||
assert len(pending_calls) == 0 # All tool calls have responses
|
||||
|
||||
|
||||
def test_get_pending_tool_calls_anthropic_pending():
|
||||
"""Test pending tool calls in Anthropic format."""
|
||||
conversation_history = [
|
||||
{
|
||||
"role": "assistant",
|
||||
"content": [
|
||||
{"type": "text", "text": "Let me help you with that."},
|
||||
{"type": "tool_use", "id": "toolu_123", "name": "test_tool"},
|
||||
{"type": "tool_use", "id": "toolu_456", "name": "another_tool"},
|
||||
],
|
||||
},
|
||||
{
|
||||
"role": "user",
|
||||
"content": [
|
||||
{"type": "tool_result", "tool_use_id": "toolu_123", "content": "Result"}
|
||||
],
|
||||
},
|
||||
]
|
||||
|
||||
pending_calls = get_pending_tool_calls(conversation_history)
|
||||
assert len(pending_calls) == 1
|
||||
assert "toolu_456" in pending_calls
|
||||
|
||||
|
||||
def test_get_pending_tool_calls_mixed_formats():
|
||||
"""Test that mixed OpenAI and Anthropic formats work correctly."""
|
||||
conversation_history = [
|
||||
# OpenAI format tool request
|
||||
{
|
||||
"role": "assistant",
|
||||
"tool_calls": [{"id": "call_123", "function": {"name": "openai_tool"}}],
|
||||
},
|
||||
# Anthropic format tool request
|
||||
{
|
||||
"role": "assistant",
|
||||
"content": [{"type": "tool_use", "id": "toolu_456", "name": "anthropic_tool"}],
|
||||
},
|
||||
# OpenAI format response
|
||||
{"role": "tool", "tool_call_id": "call_123", "content": "OpenAI result"},
|
||||
# Anthropic format response
|
||||
{
|
||||
"role": "user",
|
||||
"content": [
|
||||
{"type": "tool_result", "tool_use_id": "toolu_456", "content": "Anthropic result"}
|
||||
],
|
||||
},
|
||||
]
|
||||
|
||||
pending_calls = get_pending_tool_calls(conversation_history)
|
||||
assert len(pending_calls) == 0 # All tool calls have responses
|
||||
|
||||
|
||||
def test_smart_decision_maker_get_missing_input_with_chat_completion_message():
|
||||
"""
|
||||
Test that SmartDecisionMakerBlock.Input.get_missing_input handles ChatCompletionMessage objects.
|
||||
This is the regression test for the original bug.
|
||||
"""
|
||||
# Create mock ChatCompletionMessage objects that would cause the original bug
|
||||
mock_message1 = MockChatCompletionMessage(
|
||||
role="assistant",
|
||||
content="I'll help you with that.",
|
||||
tool_calls=[{"id": "call_123", "function": {"name": "test_tool"}}],
|
||||
)
|
||||
mock_message2 = MockChatCompletionMessage(
|
||||
role="tool", content="Tool result"
|
||||
)
|
||||
|
||||
# Create input data with mock ChatCompletionMessage objects
|
||||
input_data = {
|
||||
"conversation_history": [mock_message1, mock_message2],
|
||||
"last_tool_output": None,
|
||||
}
|
||||
|
||||
# This should NOT raise AttributeError after the fix
|
||||
# The fix converts ChatCompletionMessage objects to dicts before calling get_pending_tool_calls
|
||||
missing_input = SmartDecisionMakerBlock.Input.get_missing_input(input_data)
|
||||
|
||||
# Since the conversation history has ChatCompletionMessage objects that can't be processed
|
||||
# directly, the fix should handle them properly
|
||||
assert isinstance(missing_input, set)
|
||||
|
||||
|
||||
def test_smart_decision_maker_get_missing_input_with_pending_tools():
|
||||
"""Test get_missing_input correctly identifies when last_tool_output is needed."""
|
||||
input_data = {
|
||||
"conversation_history": [
|
||||
{
|
||||
"role": "assistant",
|
||||
"content": "I'll help you with that.",
|
||||
"tool_calls": [{"id": "call_123", "function": {"name": "test_tool"}}],
|
||||
}
|
||||
# No tool response, so tool call is pending
|
||||
],
|
||||
"last_tool_output": None,
|
||||
}
|
||||
|
||||
missing_input = SmartDecisionMakerBlock.Input.get_missing_input(input_data)
|
||||
assert "last_tool_output" in missing_input
|
||||
|
||||
|
||||
def test_smart_decision_maker_get_missing_input_with_tool_output_no_pending():
|
||||
"""Test get_missing_input when tool output exists but no pending calls."""
|
||||
input_data = {
|
||||
"conversation_history": [
|
||||
{
|
||||
"role": "assistant",
|
||||
"content": "Task complete.",
|
||||
}
|
||||
],
|
||||
"last_tool_output": "Some output", # Output exists but no pending calls
|
||||
}
|
||||
|
||||
missing_input = SmartDecisionMakerBlock.Input.get_missing_input(input_data)
|
||||
assert "conversation_history" in missing_input
|
||||
|
||||
|
||||
def test_smart_decision_maker_get_missing_input_empty_conversation():
|
||||
"""Test get_missing_input with empty conversation history."""
|
||||
input_data = {
|
||||
"conversation_history": [],
|
||||
"last_tool_output": None,
|
||||
}
|
||||
|
||||
missing_input = SmartDecisionMakerBlock.Input.get_missing_input(input_data)
|
||||
assert len(missing_input) == 0 # No missing input
|
||||
|
||||
|
||||
def test_smart_decision_maker_get_missing_input_with_none_entries():
|
||||
"""Test that None entries in conversation history are filtered out."""
|
||||
input_data = {
|
||||
"conversation_history": [
|
||||
None,
|
||||
{
|
||||
"role": "assistant",
|
||||
"content": "Hello",
|
||||
},
|
||||
None,
|
||||
],
|
||||
"last_tool_output": None,
|
||||
}
|
||||
|
||||
# Should not raise an error with None entries
|
||||
missing_input = SmartDecisionMakerBlock.Input.get_missing_input(input_data)
|
||||
assert len(missing_input) == 0
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_smart_decision_maker_run_with_chat_completion_message():
|
||||
"""
|
||||
Integration test that SmartDecisionMakerBlock.run handles ChatCompletionMessage objects.
|
||||
"""
|
||||
from unittest.mock import patch, MagicMock
|
||||
import backend.blocks.llm as llm_module
|
||||
|
||||
block = SmartDecisionMakerBlock()
|
||||
|
||||
# Mock ChatCompletionMessage that would cause the original bug
|
||||
mock_message = MockChatCompletionMessage(
|
||||
role="user",
|
||||
content="Hello"
|
||||
)
|
||||
|
||||
# Mock the response
|
||||
mock_response = MagicMock()
|
||||
mock_response.response = "Hi there!"
|
||||
mock_response.tool_calls = None
|
||||
mock_response.prompt_tokens = 10
|
||||
mock_response.completion_tokens = 5
|
||||
mock_response.reasoning = None
|
||||
mock_response.raw_response = {"role": "assistant", "content": "Hi there!"}
|
||||
|
||||
with patch("backend.blocks.llm.llm_call", return_value=mock_response), \
|
||||
patch.object(SmartDecisionMakerBlock, "_create_function_signature", return_value=[]):
|
||||
|
||||
input_data = SmartDecisionMakerBlock.Input(
|
||||
prompt="Test prompt",
|
||||
model=llm_module.LlmModel.GPT4O,
|
||||
credentials=llm_module.TEST_CREDENTIALS_INPUT,
|
||||
conversation_history=[mock_message], # This would fail without the fix
|
||||
)
|
||||
|
||||
# This should not raise AttributeError after the fix
|
||||
outputs = {}
|
||||
async for output_name, output_data in block.run(
|
||||
input_data,
|
||||
credentials=llm_module.TEST_CREDENTIALS,
|
||||
graph_id="test-graph-id",
|
||||
node_id="test-node-id",
|
||||
graph_exec_id="test-exec-id",
|
||||
node_exec_id="test-node-exec-id",
|
||||
user_id="test-user-id",
|
||||
):
|
||||
outputs[output_name] = output_data
|
||||
|
||||
assert "finished" in outputs
|
||||
assert outputs["finished"] == "Hi there!"
|
||||
|
||||
|
||||
def test_get_tool_requests_openai_format():
|
||||
"""Test _get_tool_requests with OpenAI format."""
|
||||
entry = {
|
||||
"role": "assistant",
|
||||
"tool_calls": [
|
||||
{"id": "call_123", "function": {"name": "tool1"}},
|
||||
{"id": "call_456", "function": {"name": "tool2"}},
|
||||
]
|
||||
}
|
||||
|
||||
tool_ids = _get_tool_requests(entry)
|
||||
assert len(tool_ids) == 2
|
||||
assert "call_123" in tool_ids
|
||||
assert "call_456" in tool_ids
|
||||
|
||||
|
||||
def test_get_tool_requests_anthropic_format():
|
||||
"""Test _get_tool_requests with Anthropic format."""
|
||||
entry = {
|
||||
"role": "assistant",
|
||||
"content": [
|
||||
{"type": "text", "text": "Let me help"},
|
||||
{"type": "tool_use", "id": "toolu_123", "name": "tool1"},
|
||||
{"type": "tool_use", "id": "toolu_456", "name": "tool2"},
|
||||
]
|
||||
}
|
||||
|
||||
tool_ids = _get_tool_requests(entry)
|
||||
assert len(tool_ids) == 2
|
||||
assert "toolu_123" in tool_ids
|
||||
assert "toolu_456" in tool_ids
|
||||
|
||||
|
||||
def test_get_tool_responses_openai_format():
|
||||
"""Test _get_tool_responses with OpenAI format."""
|
||||
entry = {
|
||||
"role": "tool",
|
||||
"tool_call_id": "call_123",
|
||||
"content": "Tool result"
|
||||
}
|
||||
|
||||
tool_ids = _get_tool_responses(entry)
|
||||
assert len(tool_ids) == 1
|
||||
assert "call_123" in tool_ids
|
||||
|
||||
|
||||
def test_get_tool_responses_anthropic_format():
|
||||
"""Test _get_tool_responses with Anthropic format."""
|
||||
entry = {
|
||||
"role": "user",
|
||||
"content": [
|
||||
{"type": "tool_result", "tool_use_id": "toolu_123", "content": "Result 1"},
|
||||
{"type": "tool_result", "tool_use_id": "toolu_456", "content": "Result 2"},
|
||||
]
|
||||
}
|
||||
|
||||
tool_ids = _get_tool_responses(entry)
|
||||
assert len(tool_ids) == 2
|
||||
assert "toolu_123" in tool_ids
|
||||
assert "toolu_456" in tool_ids
|
||||
Reference in New Issue
Block a user