"""
Unit tests for boring.llm.openai_compat module.
测试原则：
1. 测试决策结果：给定 prompt，应该返回什么响应
2. 只 mock 边界：requests（外部 HTTP API）
3. 测试名称即规格：清楚说明输入和期望输出
"""
from unittest.mock import MagicMock, patch
import pytest
from boring.llm.openai_compat import OpenAICompatProvider
@pytest.fixture
def temp_project(tmp_path):
project = tmp_path / "project"
project.mkdir()
return project
class TestOpenAICompatProvider:
"""Tests for OpenAICompatProvider class."""
def test_init(self, temp_project):
"""Test initialization."""
provider = OpenAICompatProvider(
model_name="test-model",
base_url="http://localhost:1234/v1",
log_dir=temp_project / "logs",
)
assert provider.model_name == "test-model"
assert provider.base_url == "http://localhost:1234/v1"
assert provider.api_key == "lm-studio"
def test_model_name_property(self, temp_project):
"""Test model_name property."""
provider = OpenAICompatProvider("test-model", log_dir=temp_project / "logs")
assert provider.model_name == "test-model"
def test_provider_name_property(self, temp_project):
"""Test provider_name property."""
provider = OpenAICompatProvider("test-model", log_dir=temp_project / "logs")
assert provider.provider_name == "openai_compat"
def test_base_url_property(self, temp_project):
"""Test base_url property."""
provider = OpenAICompatProvider(
"test-model", base_url="http://localhost:5678/v1", log_dir=temp_project / "logs"
)
assert provider.base_url == "http://localhost:5678/v1"
def test_当服务器可达时_应返回is_available为True(self, temp_project):
"""规格：requests.get() 返回 status_code=200 → is_available 应为 True"""
with patch("boring.llm.openai_compat.requests.get") as mock_get:
# Mock 外部 HTTP API（边界）
mock_response = MagicMock()
mock_response.status_code = 200
mock_get.return_value = mock_response
provider = OpenAICompatProvider("test-model", log_dir=temp_project / "logs")
# 测试结果：应该标记为可用
assert provider.is_available is True
def test_当服务器不可达时_应返回is_available为False(self, temp_project):
"""规格：requests.get() 抛出异常 → is_available 应为 False"""
with patch(
"boring.llm.openai_compat.requests.get", side_effect=Exception("Connection error")
):
provider = OpenAICompatProvider("test-model", log_dir=temp_project / "logs")
# 测试结果：应该标记为不可用
assert provider.is_available is False
def test_当API返回成功时_生成应返回响应文本(self, temp_project):
"""规格：requests.post() 返回 status_code=200 → generate() 应返回成功和响应文本"""
with (
patch("boring.llm.openai_compat.requests.post") as mock_post,
patch("boring.llm.openai_compat.log_status"),
):
# Mock 外部 HTTP API（边界）
mock_response = MagicMock()
mock_response.status_code = 200
mock_response.json.return_value = {
"choices": [{"message": {"content": "Generated text"}}]
}
mock_post.return_value = mock_response
provider = OpenAICompatProvider("test-model", log_dir=temp_project / "logs")
result, success = provider.generate("Test prompt")
# 测试结果：应该返回成功和响应
assert success is True
assert "Generated text" in result
def test_当提供context时_生成应包含context在请求中(self, temp_project):
"""规格：generate(prompt, context="...") → 请求应包含 context"""
with (
patch("boring.llm.openai_compat.requests.post") as mock_post,
patch("boring.llm.openai_compat.log_status"),
):
mock_response = MagicMock()
mock_response.status_code = 200
mock_response.json.return_value = {"choices": [{"message": {"content": "Response"}}]}
mock_post.return_value = mock_response
provider = OpenAICompatProvider("test-model", log_dir=temp_project / "logs")
result, success = provider.generate("Prompt", context="Context")
# 测试结果：应该成功，且 context 应包含在请求中
assert success is True
call_args = mock_post.call_args
assert call_args is not None
def test_当API返回错误时_生成应返回失败(self, temp_project):
"""规格：requests.post() 返回 status_code!=200 → generate() 应返回失败"""
with (
patch("boring.llm.openai_compat.requests.post") as mock_post,
patch("boring.llm.openai_compat.log_status"),
):
# Mock API 错误（边界错误）
mock_response = MagicMock()
mock_response.status_code = 400
mock_response.text = "Bad Request"
mock_post.return_value = mock_response
provider = OpenAICompatProvider("test-model", log_dir=temp_project / "logs")
result, success = provider.generate("Test prompt")
# 测试结果：应该返回失败
assert success is False
assert "Error" in result
def test_当网络异常时_生成应返回失败(self, temp_project):
"""规格：requests.post() 抛出异常 → generate() 应返回失败"""
with (
patch("boring.llm.openai_compat.requests.post", side_effect=Exception("Network error")),
patch("boring.llm.openai_compat.log_status"),
):
provider = OpenAICompatProvider("test-model", log_dir=temp_project / "logs")
result, success = provider.generate("Test prompt")
# 测试结果：应该优雅处理错误
assert success is False
assert "Network error" in result or "Exception" in result
def test_generate_with_tools(self, temp_project):
"""Test generate_with_tools."""
with (
patch("boring.llm.openai_compat.requests.post") as mock_post,
patch("boring.llm.openai_compat.log_status"),
):
mock_response = MagicMock()
mock_response.status_code = 200
mock_response.json.return_value = {
"choices": [{"message": {"content": "Response", "tool_calls": []}}]
}
mock_post.return_value = mock_response
provider = OpenAICompatProvider("test-model", log_dir=temp_project / "logs")
result = provider.generate_with_tools("Test prompt")
assert result.success is True
assert "Response" in result.text
