import os import logging from pathlib import Path from typing import Optional import yaml from agent.llm_client import LLMClient, LLMConfig from agent.git_ops import GitOps, GitRepoInfo from agent.code_analyzer import CodeAnalyzer, FileAnalysis from agent.domain_extractor import DomainExtractor, DomainModel from agent.prompt_builder import PromptBuilder from agent.test_generator import TestGenerator logger = logging.getLogger(__name__) class TestAgent: def __init__(self, config_path: str = "config.yaml"): self.config = self._load_config(config_path) self._setup_logging() self.llm_config = LLMConfig( provider=self.config.get("agent", {}).get("provider", "deepseek"), model=self.config.get("agent", {}).get("model", "deepseek-chat"), base_url=self.config.get("agent", {}).get("base_url", "https://api.deepseek.com"), temperature=self.config.get("agent", {}).get("temperature", 0.3), max_tokens=self.config.get("agent", {}).get("max_tokens", 8192), ) self.work_dir = Path(self.config.get("git", {}).get("work_dir", "./work")) self.output_dir = Path(self.config.get("test_generation", {}).get("output_dir", "tests")) self.llm_client: Optional[LLMClient] = None self.git_ops = GitOps(self.work_dir) self.prompt_builder: Optional[PromptBuilder] = None self._api_key: str = "" def _load_config(self, config_path: str) -> dict: path = Path(config_path) if path.exists(): with open(path, "r", encoding="utf-8") as f: return yaml.safe_load(f) return {} def _setup_logging(self): level = getattr(logging, self.config.get("logging", {}).get("level", "INFO").upper(), logging.INFO) fmt = self.config.get("logging", {}).get("format", "%(asctime)s [%(levelname)s] %(name)s: %(message)s") logging.basicConfig(level=level, format=fmt) def set_api_key(self, api_key: str): self._api_key = api_key self.llm_config.api_key = api_key self.llm_client = LLMClient(self.llm_config) base_dir = Path(__file__).parent.parent skills_dir = base_dir / "skills" soul_path = base_dir / "SOUL.md" self.prompt_builder = PromptBuilder(skills_dir, soul_path) def run( self, repo_url: str, branch: str = "main", test_types: list[str] = None, output_dir: str = None, ) -> dict: if not self._api_key: raise ValueError("请先调用 set_api_key(api_key) 设置 API Key") if test_types is None: test_types = ["smoke", "unit", "integration"] if output_dir: self.output_dir = Path(output_dir) logger.info(f"=== TestAgent 开始工作 ===") logger.info(f"仓库: {repo_url}") logger.info(f"分支: {branch}") logger.info(f"测试类型: {test_types}") logger.info(f"[1/5] 拉取代码...") repo_info = self.git_ops.clone_or_pull(repo_url, branch) logger.info(f" 语言: {repo_info.language}") logger.info(f" 框架: {repo_info.framework}") logger.info(f" 源文件: {len(repo_info.files)} 个") logger.info(f"[2/5] 分析代码结构...") analyzer = CodeAnalyzer(repo_info.local_path) analyses = analyzer.analyze_all(repo_info.files) logger.info(f" 分析完成: {len(analyses)} 个文件") class_count = sum(len(a.classes) for a in analyses) func_count = sum(len(a.functions) for a in analyses) logger.info(f" 类: {class_count}, 函数: {func_count}") logger.info(f"[3/5] 提取 DDD 领域模型...") extractor = DomainExtractor(repo_info.local_path) domain_model = extractor.extract_domains(analyses) logger.info(f" 限界上下文: {len(domain_model.bounded_contexts)} 个") for ctx in domain_model.bounded_contexts: logger.info(f" - {ctx.name}: {len(ctx.files)} 个文件") logger.info(f"[4/5] 读取源代码...") source_codes = {} for analysis in analyses: source = self.git_ops.get_file_content(repo_info.local_path, analysis.path) source_codes[analysis.path] = source logger.info(f"[5/5] 调用 LLM 生成测试用例...") generator = TestGenerator(self.llm_client, self.prompt_builder) all_tests = generator.generate_tests_for_all( analyses, source_codes, domain_model, test_types ) test_output_dir = self.output_dir written_files = generator.write_tests(all_tests, test_output_dir, domain_model) logger.info(f" 生成测试文件: {len(written_files)} 个") report = self._build_report(repo_info, analyses, domain_model, written_files, test_types) return report def run_diff( self, repo_url: str, branch: str = "main", against: str = "HEAD~1", test_types: list[str] = None, ) -> dict: if not self._api_key: raise ValueError("请先调用 set_api_key(api_key) 设置 API Key") if test_types is None: test_types = ["regression", "unit"] logger.info(f"=== TestAgent 精确回归模式 ===") logger.info(f"仓库: {repo_url}") logger.info(f"对比: {against}") repo_info = self.git_ops.clone_or_pull(repo_url, branch) diff_files = self.git_ops.get_diff(repo_info.local_path, against) if not diff_files: logger.info("没有检测到代码变更,跳过") return {"changed_files": [], "generated_tests": []} logger.info(f"变更文件: {len(diff_files)} 个") for f in diff_files: logger.info(f" - {f}") analyzer = CodeAnalyzer(repo_info.local_path) analyses = analyzer.analyze_all(diff_files) extractor = DomainExtractor(repo_info.local_path) domain_model = extractor.extract_domains(analyses) source_codes = {} for analysis in analyses: source = self.git_ops.get_file_content(repo_info.local_path, analysis.path) source_codes[analysis.path] = source generator = TestGenerator(self.llm_client, self.prompt_builder) all_tests = generator.generate_tests_for_all( analyses, source_codes, domain_model, test_types ) written_files = generator.write_tests(all_tests, self.output_dir, domain_model) return { "changed_files": diff_files, "generated_tests": [str(f) for f in written_files], "domain_contexts": [ctx.name for ctx in domain_model.bounded_contexts], } def _build_report( self, repo_info: GitRepoInfo, analyses: list[FileAnalysis], domain_model: DomainModel, written_files: list[Path], test_types: list[str], ) -> dict: return { "repo_url": repo_info.url, "branch": repo_info.branch, "language": repo_info.language, "framework": repo_info.framework, "files_analyzed": len(analyses), "classes_found": sum(len(a.classes) for a in analyses), "functions_found": sum(len(a.functions) for a in analyses), "bounded_contexts": len(domain_model.bounded_contexts), "context_names": [ctx.name for ctx in domain_model.bounded_contexts], "test_types": test_types, "test_files_generated": len(written_files), "output_dir": str(self.output_dir), "test_files": [str(f) for f in written_files], }