ishangsf/workflows
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

refactor: Update rewrite.py with Chinese prompts and symbol analysis

Browse Source 
- Translate prompts to Chinese for better localization
- Add symbol analysis functionality to detect missing definitions
- Implement context code gathering for better refactoring results
- Add helper functions for project root detection and symbol definition lookup
This commit is contained in:
bobo.yang 2025-03-07 12:00:49 +08:00
parent f4a4dbc529
commit 021efac1d9
1 changed files with 387 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

399
merico/refactor/rewrite.py
View File

@ -2,9 +2,12 @@ import os
import re
import sys
from devchat.llm import chat
from typing import List, Dict, Any, Optional, Tuple
from devchat.llm import chat, chat_json
from lib.ide_service import IDEService
from lib.chatmark import Step, Form, TextEditor
def get_selected_code():
@ -37,21 +40,33 @@ def get_selected_code():
REWRITE_PROMPT = prompt = """
Your task is:
你是一个代码重构专家，你的任务是根据用户的需求重写代码。你需要根据用户的需求，重写代码，并保证代码的语法正确性和逻辑正确性。
你的重构目标：
{question}
Following the task requirements, modify only the selected portion of the code. \
Please ensure that the revised code segment maintains the same indentation as the \
selected code to seamlessly integrate with the existing code structure and maintain \
correct syntax. Just refactor the selected code. Keep all other information as it is. \
Here is the relevant context \
information for your reference:
1. selected code info: {selected_text}
待重构的代码：
{selected_text}
项目中其他相关上下文代码：
{context_code}
围绕重构目标对代码进行重构，不要进行重构目标之外的其他代码优化修改。
输出重构后的代码，代码需要用markdown代码块包裹,代码需要与原重构代码保持同样的缩进格式，并且代码块的语言类型需要与原重构代码保持一致，例如：
```python
...
```
以及
```java
...
```
只输出最终重构后代码，不要输出其他任何内容。
"""
@chat(prompt=REWRITE_PROMPT, stream_out=True)
# pylint: disable=unused-argument
def ai_rewrite(question, selected_text):
def ai_rewrite(question, selected_text, context_code):
"""
call ai to rewrite selected code
"""
@ -113,13 +128,373 @@ def replace_selected(new_code):
file.write(modified_text)
# 定义用于分析代码中缺少定义的符号的提示
SYMBOL_ANALYSIS_PROMPT = """
角色：你是一个资深的程序工程师，擅长代码分析与代码重构。
重构目标：
{task}
当前用户选择代码片段：
{code}
你的任务：
针对当前已知代码，找出所有可能缺少定义的符号（变量、函数、类等）。这些符号在代码中被使用，但可能没有在当前上下文中定义。
输出要求，输出markdown代码块形式的JSON块，形式为：
```json
[
{{
"symbol": "<identifier>",
"line": "<identifier所在行完整代码，包含缩进空格>",
"value": "<影响重要度0-1之间>",
"reason": "<影响描述>"
}},
......
]
```
示例：
针对以下代码片段：
```python
def fun1():
fun2("hello")
fun3(
"hello"
)
```
进行重构，希望fun1执行打印输出hello DevChat。
此时由于缺少fun2,fun3的函数定义，所以并不清楚fun1的完整执行逻辑。所以输出：
```json
[
{{
"symbol": "fun2",
"line": " fun2(\"hello\")",
"value": 0.8,
"reason": "不能确定fun2中具体代码逻辑，是否会打印输出参数'hello'，如果是，那么重构只需要修改fun1中参数信息"
}},
{{
"symbol": "fun3",
"line": " fun3(",
"value": 0.8,
"reason": "不能确定fun3中具体代码逻辑，是否会打印输出参数'hello'，如果是，那么重构只需要修改fun1中参数信息"
}}
]
```
请确保返回的JSON格式正确，只包含实际缺少定义的符号。只输出最终结果JSON，不要输出其他解释描述。
"""
# 定义用于生成符号使用建议的提示
SYMBOL_USAGE_PROMPT = """
基于以下符号的定义，生成正确使用这些符号的建议。
符号定义:
{symbol_definitions}
原始代码片段:
```
{original_code}
```
请提供以下内容：
1. 对每个符号的正确使用方法的解释
2. 在原始代码中可能存在的符号使用错误
3. 修复建议
4. 示例代码，展示如何正确使用这些符号
请以JSON格式返回，包含以下字段：
1. "suggestions": 包含所有符号使用建议的数组，每个建议包含：
- "symbol": 符号名称
- "explanation": 符号的正确使用方法解释
- "errors": 在原始代码中可能存在的使用错误
- "fix": 修复建议
- "example": 示例代码
请确保返回的JSON格式正确。
"""
@chat_json(prompt=SYMBOL_ANALYSIS_PROMPT)
def analyze_missing_symbols(code: str, task: str) -> Dict[str, List[Dict[str, Any]]]:
"""
使用大模型分析代码片段中缺少定义的符号
Args:
code: 需要分析的代码片段
Returns:
包含缺少定义的符号列表的字典
"""
pass
@chat_json(prompt=SYMBOL_USAGE_PROMPT)
def generate_symbol_usage_suggestions(symbol_definitions: str, original_code: str) -> Dict[str, List[Dict[str, Any]]]:
"""
基于符号定义生成符号使用建议
Args:
symbol_definitions: 符号定义的文本表示
original_code: 原始代码片段
Returns:
包含符号使用建议的字典
"""
pass
def get_symbol_definition(abspath: str, line: int, character: int, symbol_name: str, symbol_type: str, project_root_path: str) -> List[Tuple]:
"""
获取符号定义的代码
Args:
abspath: 文件的绝对路径
line: 符号所在行
character: 符号所在列
symbol_type: 符号类型
Returns:
符号定义的代码，如果找不到则返回None
"""
ide_service = IDEService()
locations = []
has_visited = set()
# 根据符号类型选择合适的查找方法
locations1 = ide_service.find_type_def_locations(abspath, line, character)
locations2 = ide_service.find_def_locations(abspath, line, character)
locations3 = ide_service.find_type_def_locations(abspath, line, character + len(symbol_name) - 1)
locations4 = ide_service.find_def_locations(abspath, line, character + len(symbol_name) - 1)
for location in locations1 + locations2 + locations3 + locations4:
if not location.abspath.startswith(project_root_path):
continue
key = (location.abspath, location.range.start.line, location.range.end.line)
key_str = f"{location.abspath}:{location.range.start.line}:{location.range.end.line}"
if key_str not in has_visited:
has_visited.add(key_str)
locations.append(key)
return locations
def format_symbol_results(symbols: List[Dict[str, Any]], definitions: Dict[str, str]) -> str:
"""
格式化符号分析结果为Markdown格式
Args:
symbols: 符号列表
definitions: 符号定义字典
Returns:
格式化后的Markdown文本
"""
result = "## 符号分析结果\n\n"
if not symbols:
return result + "没有找到缺少定义的符号。"
result += f"找到 {len(symbols)} 个可能缺少定义的符号：\n\n"
for i, symbol in enumerate(symbols):
result += f"### {i+1}. {symbol['name']} ({symbol['type']})\n\n"
result += f"- 位置: 第{symbol['line'] + 1}行，第{symbol['character'] + 1}列\n"
result += f"- 原因: {symbol['reason']}\n\n"
if symbol['name'] in definitions and definitions[symbol['name']]:
result += "#### 找到的定义:\n\n"
result += f"{definitions[symbol['name']]}\n\n"
else:
result += "#### 未找到定义\n\n"
result += "无法在当前项目中找到此符号的定义。可能是外部库、内置函数或拼写错误。\n\n"
return result
def format_usage_suggestions(suggestions: List[Dict[str, Any]]) -> str:
"""
格式化符号使用建议为Markdown格式
Args:
suggestions: 符号使用建议列表
Returns:
格式化后的Markdown文本
"""
if not suggestions:
return ""
result = "## 符号使用建议\n\n"
for i, suggestion in enumerate(suggestions):
result += f"### {i+1}. {suggestion['symbol']}\n\n"
result += f"**正确使用方法**:\n{suggestion['explanation']}\n\n"
if suggestion.get('errors'):
result += f"**可能存在的错误**:\n{suggestion['errors']}\n\n"
if suggestion.get('fix'):
result += f"**修复建议**:\n{suggestion['fix']}\n\n"
if suggestion.get('example'):
result += "**示例代码**:\n```python\n" + suggestion['example'] + "\n```\n\n"
return result
def find_project_root(file_path: str) -> str:
"""
根据文件路径查找项目根目录
通过向上遍历目录，查找 .git 或 .svn 目录来确定项目根目录
Args:
file_path: 文件的绝对路径
Returns:
项目根目录的绝对路径，如果找不到则返回原始文件所在目录
"""
if not os.path.isabs(file_path):
file_path = os.path.abspath(file_path)
current_dir = os.path.dirname(file_path)
# 向上遍历目录，直到找到包含 .git 或 .svn 的目录，或者到达根目录
while current_dir and current_dir != '/':
# 检查当前目录是否包含 .git 或 .svn
if os.path.exists(os.path.join(current_dir, '.git')) or \
os.path.exists(os.path.join(current_dir, '.svn')):
return current_dir
# 向上移动一级目录
parent_dir = os.path.dirname(current_dir)
if parent_dir == current_dir: # 防止在Windows根目录下无限循环
break
current_dir = parent_dir
# 如果没有找到版本控制目录，返回文件所在目录
return os.path.dirname(file_path)
def main():
ide_service = IDEService()
question = sys.argv[1]
rafact_task = sys.argv[1]
# prepare code
selected_text = get_selected_code()
# 步骤1: 获取用户选中的代码片段
with Step("获取选中的代码片段..."):
selected_code = ide_service.get_selected_range()
if not selected_code or not selected_code.text.strip():
print("请先选择一段代码片段再执行此命令。")
return
# print(selected_code)
selected_text = selected_code.text
project_root_path = find_project_root(selected_code.abspath)
print(f"项目根目录: {project_root_path}\n\n")
# 步骤2: 分析代码片段中缺少定义的符号
with Step("分析代码中缺少定义的符号..."):
try:
analysis_result = analyze_missing_symbols(code=selected_code.text, task=rafact_task)
missing_symbols = analysis_result # 直接获取返回的列表
if not missing_symbols:
print("没有找到缺少定义的符号。")
return
ide_service.ide_logging("info", f"找到 {len(missing_symbols)} 个可能缺少定义的符号")
except Exception as e:
ide_service.ide_logging("error", f"分析符号时出错: {str(e)}")
print(f"分析代码时出错: {str(e)}")
return
current_filepath = selected_code.abspath
base_line = selected_code.range.start.line
# range = "abspath='/Users/boboyang/.chat/scripts/merico/symbol_resolver/command.py' range=line=248 character=0 - line=248 character=24 text=' print(selected_code)'"
# 步骤3: 将分析结果转换为可处理的结构
with Step("处理符号信息..."):
symbols = []
code_lines = selected_code.text.splitlines()
for symbol_info in missing_symbols:
symbol_name = symbol_info["symbol"]
symbol_line_text = symbol_info["line"]
# 在代码中查找匹配的行
line_index = -1
for i, line in enumerate(code_lines):
if line == symbol_line_text:
line_index = i
break
if line_index == -1:
ide_service.ide_logging("warning", f"找不到符号 {symbol_name} 所在行")
continue
# 在行中查找符号位置
char_index = symbol_line_text.find(symbol_name)
if char_index == -1:
ide_service.ide_logging("warning", f"在行中找不到符号 {symbol_name}")
continue
# 构建符号信息
symbol = {
"name": symbol_name,
"line": base_line + line_index,
"character": char_index,
"type": "unknown", # 默认类型
"reason": symbol_info.get("reason", "未知原因")
}
symbols.append(symbol)
# 步骤3: 查找符号的实际定义
with Step("查找符号的实际定义..."):
symbol_definitions = {}
for symbol in symbols:
symbol_name = symbol["name"]
symbol_line = symbol.get("line", 0)
symbol_char = symbol.get("character", 0)
symbol_type = symbol.get("type", "unknown")
definitions = get_symbol_definition(
selected_code.abspath,
symbol_line,
symbol_char,
symbol_name,
symbol_type,
project_root_path
)
symbol_definitions[symbol_name] = definitions
# 计算每个文件被引用次数
files_ref_counts = {}
# 当前选中代码文件，默认计算100
files_ref_counts[selected_code.abspath] = 100
for symbol in symbol_definitions:
for definition in symbol_definitions[symbol]:
if definition[0] not in files_ref_counts:
files_ref_counts[definition[0]] = 0
files_ref_counts[definition[0]] += 1
context_code = ""
for filepath, ref_count in files_ref_counts.items():
if ref_count > 0:
with open(filepath, "r") as f:
context_code += f"文件名：{filepath}\n\n"
context_code += "文件内容：\n"
context_code += f.read()
context_code += "\n\n"
# rewrite
response = ai_rewrite(question=question, selected_text=selected_text)
response = ai_rewrite(question=question, selected_text=selected_text, context_code=context_code)
if not response:
sys.exit(1)