jarvis-ai-assistant 0.1.72__tar.gz → 0.1.74__tar.gz

{jarvis_ai_assistant-0.1.72/src/jarvis_ai_assistant.egg-info → jarvis_ai_assistant-0.1.74}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: jarvis-ai-assistant
-Version: 0.1.72
+Version: 0.1.74
 Summary: Jarvis: An AI assistant that uses tools to interact with the system
 Home-page: https://github.com/skyfireitdiy/Jarvis
 Author: skyfire

{jarvis_ai_assistant-0.1.72 → jarvis_ai_assistant-0.1.74}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "jarvis-ai-assistant"
-version = "0.1.72"
+version = "0.1.74"
 description = "Jarvis: An AI assistant that uses tools to interact with the system"
 readme = "README.md"
 authors = [{ name = "Your Name", email = "your.email@example.com" }]

{jarvis_ai_assistant-0.1.72 → jarvis_ai_assistant-0.1.74}/setup.py

@@ -2,7 +2,7 @@ from setuptools import setup, find_packages
 
 setup(
     name="jarvis-ai-assistant",
-    version="0.1.72",
+    version="0.1.74",
     author="skyfire",
     author_email="skyfireitdiy@hotmail.com",
     description="An AI assistant that uses various tools to interact with the system",

{jarvis_ai_assistant-0.1.72 → jarvis_ai_assistant-0.1.74}/src/jarvis/__init__.py

@@ -1,3 +1,3 @@
 """Jarvis AI Assistant"""
 
-__version__ = "0.1.72" 
+__version__ = "0.1.74" 

{jarvis_ai_assistant-0.1.72 → jarvis_ai_assistant-0.1.74}/src/jarvis/agent.py

@@ -277,6 +277,7 @@ arguments:
 6. 处理问题的每个步骤不是必须有的，可按情况省略
 7. 在执行一些可能对系统或者用户代码库造成破坏的工具时，请先询问用户
 8. 在多次迭代却没有任何进展时，可请求用户指导
+9. 如果返回的yaml字符串中包含冒号，请将整个字符串用引号包裹，避免yaml解析错误
 
 -------------------------------------------------------------
 

{jarvis_ai_assistant-0.1.72 → jarvis_ai_assistant-0.1.74}/src/jarvis/jarvis_codebase/main.py

@@ -310,27 +310,103 @@ class CodeBase:
         self.build_index()
         self.save_cache()
 
-    def generate_codebase(self):
-        """生成代码库索引"""
-        files_deleted = self.clean_cache()  # 清理过期缓存
-        processed_files = []
+
+    def generate_codebase(self, force: bool = False):
+        """生成代码库索引
+        Args:
+            force: 是否强制重建索引，不询问用户
+        """
+        # 更新 git 文件列表
+        self.git_file_list = self.get_git_file_list()
+        
+        # 检查文件变化
+        changes_detected = False
+        new_files = []
+        modified_files = []
+        deleted_files = []
         
-        # 使用线程池处理文件
-        with ThreadPoolExecutor(max_workers=self.thread_count) as executor:
-            futures = [executor.submit(self.process_file, file) for file in self.git_file_list]
-            for future in concurrent.futures.as_completed(futures):
-                result = future.result()
-                if result:
-                    processed_files.append(result)
-                    PrettyOutput.print(f"索引文件: {result}", output_type=OutputType.INFO)
-
-        if files_deleted or processed_files:
+        # 检查删除的文件
+        files_to_delete = []
+        for file_path in list(self.vector_cache.keys()):
+            if file_path not in self.git_file_list:
+                deleted_files.append(file_path)
+                files_to_delete.append(file_path)
+                changes_detected = True
+        
+        # 检查新增和修改的文件
+        for file_path in self.git_file_list:
+            if not os.path.exists(file_path) or not self.is_text_file(file_path):
+                continue
+            
+            try:
+                current_md5 = hashlib.md5(open(file_path, "rb").read()).hexdigest()
+                
+                if file_path not in self.vector_cache:
+                    new_files.append(file_path)
+                    changes_detected = True
+                elif self.vector_cache[file_path].get("md5") != current_md5:
+                    modified_files.append(file_path)
+                    changes_detected = True
+            except Exception as e:
+                PrettyOutput.print(f"检查文件失败 {file_path}: {str(e)}", 
+                                 output_type=OutputType.ERROR)
+                continue
+        
+        # 如果检测到变化，显示变化并询问用户
+        if changes_detected:
+            PrettyOutput.print("\n检测到以下变化:", output_type=OutputType.WARNING)
+            if new_files:
+                PrettyOutput.print("\n新增文件:", output_type=OutputType.INFO)
+                for f in new_files:
+                    PrettyOutput.print(f"  {f}", output_type=OutputType.INFO)
+            if modified_files:
+                PrettyOutput.print("\n修改的文件:", output_type=OutputType.INFO)
+                for f in modified_files:
+                    PrettyOutput.print(f"  {f}", output_type=OutputType.INFO)
+            if deleted_files:
+                PrettyOutput.print("\n删除的文件:", output_type=OutputType.INFO)
+                for f in deleted_files:
+                    PrettyOutput.print(f"  {f}", output_type=OutputType.INFO)
+            
+
+            # 如果force为True，直接继续
+            if not force:
+                # 询问用户是否继续
+                while True:
+                    response = input("\n是否重建索引？[y/N] ").lower().strip()
+                    if response in ['y', 'yes']:
+                        break
+                    elif response in ['', 'n', 'no']:
+                        PrettyOutput.print("取消重建索引", output_type=OutputType.INFO)
+                        return
+                    else:
+                        PrettyOutput.print("请输入 y 或 n", output_type=OutputType.WARNING)
+            
+            # 清理已删除的文件
+            for file_path in files_to_delete:
+                del self.vector_cache[file_path]
+            if files_to_delete:
+                PrettyOutput.print(f"清理了 {len(files_to_delete)} 个文件的缓存", 
+                                 output_type=OutputType.INFO)
+            
+            # 处理新文件和修改的文件
+            processed_files = []
+            files_to_process = new_files + modified_files
+            
+            # 使用线程池处理文件
+            with ThreadPoolExecutor(max_workers=self.thread_count) as executor:
+                futures = [executor.submit(self.process_file, file) for file in files_to_process]
+                for future in concurrent.futures.as_completed(futures):
+                    result = future.result()
+                    if result:
+                        processed_files.append(result)
+                        PrettyOutput.print(f"索引文件: {result}", output_type=OutputType.INFO)
+
             PrettyOutput.print("重新生成向量数据库", output_type=OutputType.INFO)
             self.gen_vector_db_from_cache()
+            PrettyOutput.print(f"成功为 {len(processed_files)} 个文件生成索引", output_type=OutputType.INFO)
         else:
-            PrettyOutput.print("没有新的文件变更，跳过向量数据库生成", output_type=OutputType.INFO)
-            
-        PrettyOutput.print(f"成功为 {len(processed_files)} 个文件生成索引", output_type=OutputType.INFO)
+            PrettyOutput.print("没有检测到文件变更，无需重建索引", output_type=OutputType.INFO)
 
     def rerank_results(self, query: str, initial_results: List[Tuple[str, float, str]]) -> List[Tuple[str, float, str]]:
         """使用大模型对搜索结果重新排序"""
@@ -377,10 +453,12 @@ class CodeBase:
                 try:
                     file_path, score_str = line.split(':', 1)
                     file_path = file_path.strip()
-                    score = float(score_str.strip())
-                    # 找到对应的原始描述
-                    desc = next((desc for p, _, desc in initial_results if p == file_path), "")
-                    scored_results.append((file_path, score/100.0, desc))
+                    score = float(score_str.strip()) / 100.0  # 转换为0-1范围
+                    # 只保留相关度大于等于0.7的结果
+                    if score >= 0.7:
+                        # 找到对应的原始描述
+                        desc = next((desc for p, _, desc in initial_results if p == file_path), "")
+                        scored_results.append((file_path, score, desc))
                 except:
                     continue
             
@@ -413,8 +491,6 @@ class CodeBase:
         PrettyOutput.print(f"查询:  {query}", output_type=OutputType.INFO)
         
         # 为每个查询获取相似文件
-        all_results = {}  # 文件路径 -> (总分数, 出现次数, 描述)
-        
         q_vector = self.get_embedding(query)
         q_vector = q_vector.reshape(1, -1)
             
@@ -429,12 +505,18 @@ class CodeBase:
                 continue
                 
             similarity = 1.0 / (1.0 + float(distance))
-            PrettyOutput.print(f"  {self.file_paths[i]} : 距离 {distance:.3f}, 相似度 {similarity:.3f}", 
-                                output_type=OutputType.INFO)
+            # 只保留相似度大于等于0.5的结果
+            if similarity >= 0.5:
+                PrettyOutput.print(f"  {self.file_paths[i]} : 距离 {distance:.3f}, 相似度 {similarity:.3f}", 
+                                 output_type=OutputType.INFO)
                 
-            file_path = self.file_paths[i]
-            data = self.vector_cache[file_path]
-            initial_results.append((file_path, similarity, data["description"]))
+                file_path = self.file_paths[i]
+                data = self.vector_cache[file_path]
+                initial_results.append((file_path, similarity, data["description"]))
+
+        if not initial_results:
+            PrettyOutput.print("没有找到相似度大于0.5的文件", output_type=OutputType.WARNING)
+            return []
 
         # 使用大模型重新排序
         PrettyOutput.print("使用大模型重新排序...", output_type=OutputType.INFO)
@@ -482,22 +564,56 @@ class CodeBase:
         finally:
             model.delete_chat()
 
+    def is_index_generated(self) -> bool:
+        """检查索引是否已经生成"""
+        # 检查缓存文件是否存在
+        if not os.path.exists(self.cache_path):
+            return False
+        
+        # 检查缓存是否有效
+        try:
+            with open(self.cache_path, 'rb') as f:
+                cache_data = pickle.load(f)
+                if not cache_data.get("vectors") or not cache_data.get("file_paths"):
+                    return False
+        except Exception:
+            return False
+        
+        # 检查索引是否已构建
+        if not hasattr(self, 'index') or self.index is None:
+            return False
+        
+        # 检查向量缓存和文件路径列表是否非空
+        if not self.vector_cache or not self.file_paths:
+            return False
+        
+        return True
+
+
 
 def main():
     parser = argparse.ArgumentParser(description='Codebase management and search tool')
     parser.add_argument('--search', type=str, help='Search query to find similar code files')
     parser.add_argument('--top-k', type=int, default=20, help='Number of results to return (default: 20)')
     parser.add_argument('--ask', type=str, help='Ask a question about the codebase')
+    parser.add_argument('--generate', action='store_true', help='Generate codebase index')
     args = parser.parse_args()
     
     current_dir = find_git_root()
     codebase = CodeBase(current_dir)
 
-    try:
-        codebase.generate_codebase()
-        PrettyOutput.print("\nCodebase generation completed", output_type=OutputType.SUCCESS)
-    except Exception as e:
-        PrettyOutput.print(f"Error during codebase generation: {str(e)}", output_type=OutputType.ERROR)
+    # 如果没有生成索引，且不是生成命令，提示用户先生成索引
+    if not codebase.is_index_generated() and not args.generate:
+        PrettyOutput.print("索引尚未生成，请先运行 --generate 生成索引", output_type=OutputType.WARNING)
+        return
+
+
+    if args.generate:
+        try:
+            codebase.generate_codebase(force=True)
+            PrettyOutput.print("\nCodebase generation completed", output_type=OutputType.SUCCESS)
+        except Exception as e:
+            PrettyOutput.print(f"Error during codebase generation: {str(e)}", output_type=OutputType.ERROR)
     
     if args.search:
         results = codebase.search_similar(args.search, args.top_k)

{jarvis_ai_assistant-0.1.72 → jarvis_ai_assistant-0.1.74}/src/jarvis/jarvis_coder/main.py

@@ -278,12 +278,12 @@ class JarvisCoder:
         
         return True, ""
 
-    def _save_edit_record(self, feature: str, patches: List[str]) -> None:
+    def _save_edit_record(self, commit_message: str, git_diff: str) -> None:
         """保存代码修改记录
         
         Args:
-            feature: 需求描述
-            patches: 补丁列表
+            commit_message: 提交信息
+            git_diff: git diff --cached的输出
         """
             
         # 获取下一个序号
@@ -296,8 +296,8 @@ class JarvisCoder:
         # 创建记录文件
         record = {
             "timestamp": time.strftime("%Y-%m-%d %H:%M:%S"),
-            "feature": feature,
-            "patches": patches
+            "commit_message": commit_message,
+            "git_diff": git_diff
         }
         
         record_path = os.path.join(self.record_dir, f"{next_num:04d}.yaml")
@@ -308,14 +308,21 @@ class JarvisCoder:
 
 
 
+
     def _prepare_execution(self) -> None:
         """准备执行环境"""
         self.main_model = self._new_model()
         self._codebase.generate_codebase()
 
+
     def _load_related_files(self, feature: str) -> List[Dict]:
         """加载相关文件内容"""
         ret = []
+        # 确保索引数据库已生成
+        if not self._codebase.is_index_generated():
+            PrettyOutput.print("检测到索引数据库未生成，正在生成...", OutputType.WARNING)
+            self._codebase.generate_codebase()
+            
         related_files = self._codebase.search_similar(feature)
         for file, score, _ in related_files:
             PrettyOutput.print(f"相关文件: {file} 相关度: {score:.3f}", OutputType.SUCCESS)
@@ -339,7 +346,7 @@ class JarvisCoder:
             if success:
                 user_confirm = input("是否确认修改？(y/n)")
                 if user_confirm.lower() == "y":
-                    self._finalize_changes(feature, patches)
+                    self._finalize_changes(feature)
                     return {
                         "success": True,
                         "stdout": f"已完成功能开发{feature}",
@@ -369,12 +376,74 @@ class JarvisCoder:
 """
                 patches = self._remake_patch(retry_prompt)
 
-    def _finalize_changes(self, feature: str, patches: List[str]) -> None:
+
+
+
+    def _generate_commit_message(self, git_diff: str, feature: str) -> str:
+        """根据git diff和功能描述生成commit信息
+        
+        Args:
+            git_diff: git diff --cached的输出
+            feature: 用户的功能描述
+            
+        Returns:
+            str: 生成的commit信息
+        """
+        
+        # 生成提示词
+        prompt = f"""你是一个经验丰富的程序员，请根据以下代码变更和功能描述生成简洁明了的commit信息：
+
+功能描述：
+{feature}
+
+代码变更：
+"""
+        # 添加git diff内容
+        prompt += f"Git Diff:\n{git_diff}\n\n"
+            
+        prompt += """
+请遵循以下规则：
+1. 使用英文编写
+2. 采用常规的commit message格式：<type>(<scope>): <subject>
+3. 保持简洁，不超过50个字符
+4. 准确描述代码变更的主要内容
+5. 优先考虑功能描述和git diff中的变更内容
+"""
+        
+        # 使用normal模型生成commit信息
+        model = PlatformRegistry().get_global_platform_registry().create_platform(self.platform)
+        model.set_model_name(self.model)
+        model.set_suppress_output(True)
+        success, response = self._call_model_with_retry(model, prompt)
+        if not success:
+            return "Update code changes"
+            
+        # 清理响应内容
+        return response.strip().split("\n")[0]
+
+    def _finalize_changes(self, feature: str) -> None:
         """完成修改并提交"""
         PrettyOutput.print("修改确认成功，提交修改", OutputType.INFO)
-        os.system(f"git add .")
-        os.system(f"git commit -m '{feature}'")
-        self._save_edit_record(feature, patches)
+
+        # 只添加已经在 git 控制下的修改文件
+        os.system("git add -u")
+        
+        # 然后获取 git diff
+        git_diff = os.popen("git diff --cached").read()
+        
+        # 自动生成commit信息，传入feature
+        commit_message = self._generate_commit_message(git_diff, feature)
+        
+        # 显示并确认commit信息
+        PrettyOutput.print(f"自动生成的commit信息: {commit_message}", OutputType.INFO)
+        user_confirm = input("是否使用该commit信息？(y/n) [y]: ") or "y"
+        
+        if user_confirm.lower() != "y":
+            commit_message = input("请输入新的commit信息: ")
+        
+        # 不需要再次 git add，因为已经添加过了
+        os.system(f"git commit -m '{commit_message}'")
+        self._save_edit_record(commit_message, git_diff)
 
     def _revert_changes(self) -> None:
         """回退所有修改"""

{jarvis_ai_assistant-0.1.72 → jarvis_ai_assistant-0.1.74}/src/jarvis/tools/codebase_qa.py

@@ -45,8 +45,12 @@ class CodebaseQATool:
                     "stderr": "错误：当前目录不在Git仓库中",
                     "error": "NotInGitRepository"
                 }
+
             os.chdir(root_dir)
             codebase = CodeBase(root_dir)
+            # 生成索引
+
+            codebase.generate_codebase(force=True)
             # 执行问答
             response = codebase.ask_codebase(question, top_k)
             os.chdir(current_dir)

{jarvis_ai_assistant-0.1.72 → jarvis_ai_assistant-0.1.74/src/jarvis_ai_assistant.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: jarvis-ai-assistant
-Version: 0.1.72
+Version: 0.1.74
 Summary: Jarvis: An AI assistant that uses tools to interact with the system
 Home-page: https://github.com/skyfireitdiy/Jarvis
 Author: skyfire