jarvis-ai-assistant 0.1.124__py3-none-any.whl → 0.1.126__py3-none-any.whl

jarvis/jarvis_codebase/main.py

@@ -7,14 +7,18 @@ from typing import List, Tuple, Optional, Dict
 from jarvis.jarvis_platform.registry import PlatformRegistry
 import concurrent.futures
 from concurrent.futures import ThreadPoolExecutor
-from jarvis.jarvis_utils import OutputType, PrettyOutput, find_git_root, get_context_token_count, get_embedding, get_file_md5, get_max_token_count, get_thread_count, load_embedding_model, user_confirm
-from jarvis.jarvis_utils import init_env
 import argparse
 import pickle
 import lzma  # 添加 lzma 导入
 from tqdm import tqdm
 import re
 
+from jarvis.jarvis_utils.config import get_max_token_count, get_thread_count
+from jarvis.jarvis_utils.embedding import get_embedding, load_embedding_model, get_context_token_count
+from jarvis.jarvis_utils.git_utils import find_git_root
+from jarvis.jarvis_utils.output import OutputType, PrettyOutput
+from jarvis.jarvis_utils.utils import  get_file_md5, init_env, user_confirm
+
 class CodeBase:
     def __init__(self, root_dir: str):
         init_env()
@@ -72,13 +76,13 @@ class CodeBase:
             model.set_suppress_output(True)
         else:
             PrettyOutput.print(f"为 {file_path} 生成描述 ...", output_type=OutputType.PROGRESS)
-        prompt = f"""Please analyze the following code file and generate a detailed description. The description should include:
-1. Overall file functionality description
-2. description for each global variable, function, type definition, class, method, and other code elements
+        prompt = f"""请分析以下代码文件并生成详细描述。描述应包含：
+1. 文件整体功能描述
+2. 对每个全局变量、函数、类型定义、类、方法和其他代码元素的描述
 
-Please use concise and professional language, emphasizing technical functionality to facilitate subsequent code retrieval.
-File path: {file_path}
-Code content:
+请使用简洁专业的语言，强调技术功能，以便于后续代码检索。
+文件路径: {file_path}
+代码内容:
 {content}
 """
         response = model.chat_until_success(prompt)
@@ -316,7 +320,7 @@ Content: {content}
             ids = []
             self.file_paths = []  # Reset the file path list
             
-            for i, (file_path, data) in enumerate(self.vector_cache.items()):
+            for i, ( file_path, data) in enumerate(self.vector_cache.items()):
                 if "vector" not in data:
                     PrettyOutput.print(f"无效的缓存数据 {file_path}: 缺少向量", 
                                      output_type=OutputType.WARNING)
@@ -591,24 +595,24 @@ Content: {content}
             
     def _process_batch(self, query: str, files_info: List[str]) -> List[Dict[str, str]]:
         """Process a batch of files"""
-        prompt = f"""As a code analysis expert, please help identify the most relevant files for the given query using chain-of-thought reasoning.
+        prompt = f"""作为一名代码分析专家，请使用链式思维推理帮助识别与给定查询最相关的文件。
 
-Query: {query}
+查询: {query}
 
-Available files:
+可用文件:
 {''.join(files_info)}
 
-Think through this step by step:
-1. First, analyze the query to identify key requirements and technical concepts
-2. For each file:
-   - Examine its path and content
-   - Assess how it relates to the query's requirements
-   - Consider both direct and indirect relationships
-   - Rate its relevance (high/medium/low)
-3. Select only files with clear relevance to the query
-4. Order files by relevance, with most relevant first
-
-Please output your selection in YAML format:
+请按以下步骤思考：
+1. 首先，分析查询以识别关键需求和技术概念
+2. 对于每个文件：
+   - 检查其路径和内容
+   - 评估其与查询需求的关系
+   - 考虑直接和间接关系
+   - 评估其相关性（高/中/低）
+3. 仅选择与查询明确相关的文件
+4. 按相关性排序，最相关的文件在前
+
+请以YAML格式输出您的选择：
 <FILES>
 - file: path/to/most/relevant.py
   reason: xxxxxxxxxx
@@ -616,12 +620,12 @@ Please output your selection in YAML format:
   reason: yyyyyyyyyy
 </FILES>
 
-Important:
-- Only include files that are truly relevant
-- Exclude files with weak or unclear connections
-- Focus on implementation rather than test files
-- Consider both file paths and content
-- Only output the file paths, no other text
+重要提示：
+- 仅包含真正相关的文件
+- 排除连接不明确或较弱的文件
+- 重点关注实现文件而非测试文件
+- 同时考虑文件路径和内容
+- 仅输出文件路径，不要包含其他文本
 """
 
         # Use a large model to evaluate
@@ -653,29 +657,28 @@ Important:
         """
         model = PlatformRegistry.get_global_platform_registry().get_normal_platform()
         model.set_suppress_output(True)
-        prompt = f"""Please generate 10 different expressions optimized for vector search based on the following query. Each expression should:
-
-1. Focus on key technical concepts and terminology
-2. Use clear and specific language
-3. Include important contextual terms
-4. Avoid general or ambiguous words
-5. Maintain semantic similarity with original query
-6. Be suitable for embedding-based search
-
-Original query: 
+        prompt = f"""请基于以下查询生成10个针对向量搜索优化的不同表达。每个表达应满足：
+1. 聚焦关键技术概念和术语
+2. 使用清晰明确的语言
+3. 包含重要的上下文术语
+4. 避免使用通用或模糊的词语
+5. 保持与原始查询的语义相似性
+6. 适合基于嵌入的搜索
+
+原始查询: 
 {query}
 
-Example transformations:
-Query: "How to handle user login?"
-Output format:
+示例转换：
+查询: "如何处理用户登录？"
+输出格式:
 <QUESTION>
-- user authentication implementation and flow
-- login system architecture and components
-- credential validation and session management
+- 用户认证的实现与流程
+- 登录系统架构与组件
+- 凭证验证与会话管理
 - ...
 </QUESTION>
 
-Please provide 10 search-optimized expressions in the specified format.
+请以指定格式提供10个搜索优化的表达。
 """
         response = model.chat_until_success(prompt)
         
@@ -790,52 +793,52 @@ Please provide 10 search-optimized expressions in the specified format.
             return [], ""
         
         prompt = f"""
-# 🤖 Role Definition
-You are a code analysis expert who provides comprehensive and accurate answers about codebases.
-
-# 🎯 Core Responsibilities
-- Analyze code files thoroughly
-- Explain technical concepts clearly
-- Provide relevant code examples
-- Identify missing information
-- Answer in user's language
-
-# 📋 Response Requirements
-## Content Quality
-- Focus on implementation details
-- Be technically precise
-- Include relevant code snippets
-- Indicate any missing information
-- Use professional terminology
-
-## Response Format
-- question: [Restate the question]
+# 🤖 角色定义
+您是一位代码分析专家，能够提供关于代码库的全面且准确的回答。
+
+# 🎯 核心职责
+- 深入分析代码文件
+- 清晰解释技术概念
+- 提供相关代码示例
+- 识别缺失的信息
+- 使用用户的语言进行回答
+
+# 📋 回答要求
+## 内容质量
+- 关注实现细节
+- 保持技术准确性
+- 包含相关代码片段
+- 指出任何缺失的信息
+- 使用专业术语
+
+## 回答格式
+- question: [重述问题]
   answer: |
-    [Detailed technical answer with:
-    - Implementation details
-    - Code examples (if relevant)
-    - Missing information (if any)
-    - Related technical concepts]
+    [详细的技术回答，包含：
+    - 实现细节
+    - 代码示例（如果相关）
+    - 缺失的信息（如果有）
+    - 相关技术概念]
 
-- question: [Follow-up question if needed]
+- question: [如果需要，提出后续问题]
   answer: |
-    [Additional technical details]
+    [额外的技术细节]
 
-# 🔍 Analysis Context
-Question: {query}
+# 🔍 分析上下文
+问题: {query}
 
-Relevant Code Files (by relevance):
+相关代码文件（按相关性排序）:
 """
 
-        # Add context with length control
-        available_count = self.max_token_count - get_context_token_count(prompt) - 1000  # Reserve space for answer
+        # 添加上下文，控制长度
+        available_count = self.max_token_count - get_context_token_count(prompt) - 1000  # 为回答预留空间
         current_count = 0
         
         for path in files_from_codebase:
             try:
                 content = open(path["file"], "r", encoding="utf-8").read()
                 file_content = f"""
-## File: {path["file"]}
+## 文件: {path["file"]}
 ```
 {content}
 ```
@@ -857,13 +860,13 @@ Relevant Code Files (by relevance):
                 continue
 
         prompt += """
-# ❗ Important Rules
-1. Always base answers on provided code
-2. Use technical precision
-3. Include code examples when relevant
-4. Indicate any missing information
-5. Maintain professional language
-6. Answer in user's language
+# ❗ 重要规则
+1. 始终基于提供的代码进行回答
+2. 保持技术准确性
+3. 在相关时包含代码示例
+4. 指出任何缺失的信息
+5. 保持专业语言
+6. 使用用户的语言进行回答
 """
 
         model = PlatformRegistry.get_global_platform_registry().get_thinking_platform()