test_api.py

import requests
import time
import json

BASE_URL = "http://127.0.0.1:5000"
BASE_MODEL_URL = "https://api.openai.com"
API_KEY = ""
MODEL_NAME = "gpt-4o-mini"
REQUEST_MODE = "chat"
PROMPT_TEMPLATE = "请根据以下对话内容进行评估，输出 JSON，字段包含 total_score, category, reasons。"

test_cases = [
    {
        "name": "积极会话",
        "category": "积极",
        "data": {
            "messages": [
                "这个模型太棒了，回答得非常准确！",
                "我要分享给我的朋友们！",
                "感谢开发者的辛苦工作，太赞了"
            ]
        },
        "expected_category": "积极",
        "expected_score_min": 10
    },
    {
        "name": "中性会话",
        "category": "中性",
        "data": {
            "messages": [
                "你好",
                "我想了解一下这个功能",
                "好的，明白了",
                "谢谢"
            ]
        },
        "expected_category": "中性",
        "expected_score_range": (-9, 9)
    },
    {
        "name": "消极会话",
        "category": "消极",
        "data": {
            "messages": [
                "这个回答太差了",
                "完全不对",
                "糟糕透了，浪费时间",
                "一点都不满意"
            ]
        },
        "expected_category": "消极",
        "expected_score_max": -10
    },
    {
        "name": "包含显式行为测试",
        "category": "积极",
        "data": {
            "messages": [
                "这个内容很有用，我要下载它",
                "然后复制给其他人分享"
            ]
        },
        "expected_behaviors": ["download", "copy", "share"]
    }
]

def run_tests():
    results = []
    for test in test_cases:
        print(f"\n{'='*60}")
        print(f"测试用例: {test['name']}")
        print(f"{'='*60}")
        
        start_time = time.time()
        try:
            response = requests.post(
                f"{BASE_URL}/api/evaluate",
                json={
                    "base_url": BASE_MODEL_URL,
                    "api_key": API_KEY,
                    "model": MODEL_NAME,
                    "request_mode": REQUEST_MODE,
                    "prompt_template": PROMPT_TEMPLATE,
                    "session_data": test['data']
                },
                headers={"Content-Type": "application/json"}
            )
            end_time = time.time()
            response_time = (end_time - start_time) * 1000
            
            print(f"响应时间: {response_time:.2f} 毫秒")
            
            if response.status_code == 200:
                result = response.json()
                print(f"响应状态: 成功 (200)")
                print(f"总分: {result.get('total_score')}")
                print(f"分类: {result.get('category')}")
                print(f"显式行为: {result.get('explicit_behaviors')}")
                
                passed = True
                issues = []
                
                if 'expected_category' in test:
                    if result.get('category') != test['expected_category']:
                        passed = False
                        issues.append(f"分类错误: 期望 {test['expected_category']}, 实际 {result.get('category')}")
                
                if 'expected_score_min' in test:
                    if result.get('total_score', 0) < test['expected_score_min']:
                        passed = False
                        issues.append(f"分数过低: 期望至少 {test['expected_score_min']}, 实际 {result.get('total_score')}")
                
                if 'expected_score_max' in test:
                    if result.get('total_score', 0) > test['expected_score_max']:
                        passed = False
                        issues.append(f"分数过高: 期望最多 {test['expected_score_max']}, 实际 {result.get('total_score')}")
                
                if 'expected_score_range' in test:
                    score = result.get('total_score', 0)
                    min_score, max_score = test['expected_score_range']
                    if score < min_score or score > max_score:
                        passed = False
                        issues.append(f"分数不在预期范围内: 期望 {min_score}-{max_score}, 实际 {score}")
                
                if 'expected_behaviors' in test:
                    behaviors = result.get('explicit_behaviors', [])
                    for expected in test['expected_behaviors']:
                        if expected not in behaviors:
                            passed = False
                            issues.append(f"未检测到预期行为: {expected}")
                
                if passed:
                    print("OK 测试通过")
                else:
                    print("FAIL 测试失败:")
                    for issue in issues:
                        print(f"  - {issue}")
                
                results.append({
                    "name": test['name'],
                    "category": test['category'],
                    "passed": passed,
                    "response_time": response_time,
                    "issues": issues,
                    "result": result
                })
            else:
                print(f"响应状态: 失败 ({response.status_code})")
                print(f"错误: {response.text}")
                results.append({
                    "name": test['name'],
                    "category": test['category'],
                    "passed": False,
                    "response_time": response_time,
                    "issues": [f"HTTP错误: {response.status_code}"],
                    "error": response.text
                })
                
        except Exception as e:
            end_time = time.time()
            response_time = (end_time - start_time) * 1000
            print(f"请求失败: {str(e)}")
            results.append({
                "name": test['name'],
                "category": test['category'],
                "passed": False,
                "response_time": response_time,
                "issues": [f"异常: {str(e)}"]
            })
    
    return results

def print_summary(results):
    print(f"\n{'='*60}")
    print("测试总结")
    print(f"{'='*60}")
    
    total = len(results)
    passed = sum(1 for r in results if r['passed'])
    failed = total - passed
    
    print(f"总测试数: {total}")
    print(f"通过: {passed}")
    print(f"失败: {failed}")
    
    avg_response_time = sum(r['response_time'] for r in results) / total
    print(f"平均响应时间: {avg_response_time:.2f} 毫秒")
    
    all_under_1s = all(r['response_time'] < 1000 for r in results)
    print(f"所有响应时间 < 1秒: {'OK' if all_under_1s else 'FAIL'}")
    
    for r in results:
        status = "OK" if r['passed'] else "FAIL"
        print(f"{status} {r['name']} - {r['response_time']:.2f}ms")
    
    return {
        "total": total,
        "passed": passed,
        "failed": failed,
        "avg_response_time": avg_response_time,
        "all_under_1s": all_under_1s,
        "results": results
    }

if __name__ == "__main__":
    print("开始API集成测试")
    print("="*60)
    results = run_tests()
    summary = print_summary(results)
    
    with open('test_report.json', 'w', encoding='utf-8') as f:
        json.dump(summary, f, ensure_ascii=False, indent=2)
    
    print(f"\n测试报告已保存到 test_report.json")