code-refactor（代码重构）

Code Refactor

Overview

Perform systematic code refactoring operations across files using pattern-based search and replace. This skill focuses on bulk transformations that maintain code functionality while improving structure, naming, or updating APIs.

When to Use This Skill

Activate this skill when users request:

• Rename variables/functions: "Rename getUserData to fetchUserData everywhere"
• Replace deprecated patterns: "Replace all var declarations with let or const"
• Update API calls: "Update all calls to the old authentication API"
• Refactor patterns: "Convert all callbacks to async/await"
• Standardize code: "Make all import statements use absolute paths"

Activation phrases:

• "rename [identifier] to [new_name]"
• "replace all [pattern] with [replacement]"
• "refactor to use [new_pattern]"
• "update all calls to [function/API]"
• "convert [old_pattern] to [new_pattern]"
• "standardize [aspect] across the codebase"

Core Capabilities

1. Find All Occurrences

Before refactoring, locate all instances of the pattern to understand scope and impact.

Search for exact matches:

Grep(pattern="getUserData", output_mode="files_with_matches")

Search with context to verify usage:

Grep(pattern="getUserData", output_mode="content", -n=true, -B=2, -A=2)

Case-insensitive search:

Grep(pattern="getuserdata", -i=true, output_mode="files_with_matches")

Regex patterns:

Grep(pattern="get\\w+Data", output_mode="content")

2. Replace All Instances

Use the Edit tool with replace_all=true for bulk replacements:

Simple replacement:

Edit(
  file_path="src/api.js",
  old_string="getUserData",
  new_string="fetchUserData",
  replace_all=true
)

Multi-line replacement:

Edit(
  file_path="src/auth.js",
  old_string="function authenticate(user) {\\n  return user.valid;\\n}",
  new_string="async function authenticate(user) {\\n  return await validateUser(user);\\n}",
  replace_all=true
)

3. Targeted Replacement

For single occurrences or when context matters, use Edit without replace_all:

Edit(
  file_path="src/config.js",
  old_string="const API_URL = 'http://old-api.com'",
  new_string="const API_URL = 'https://new-api.com'"
)

Workflow Examples

Example 1: Rename Function Across Codebase

User: "Rename getUserData to fetchUserData everywhere"

Workflow:

1. Find all occurrences to understand scope:

   Grep(pattern="getUserData", output_mode="files_with_matches")

1. Preview changes by viewing context:

   Grep(pattern="getUserData", output_mode="content", -n=true, -B=1, -A=1)

1. Inform user of scope: "Found 15 occurrences in 5 files"

1. Replace in each file:

   Edit(file_path="src/api.js", old_string="getUserData", new_string="fetchUserData", replace_all=true)
   Edit(file_path="src/utils.js", old_string="getUserData", new_string="fetchUserData", replace_all=true)
   # ... for each file

1. Verify changes: Re-run Grep to confirm all instances were replaced

1. Suggest testing: Recommend running tests to ensure refactoring didn't break functionality

Example 2: Replace Deprecated Pattern

User: "Replace all var declarations with let or const"

Workflow:

1. Find all var declarations:

   Grep(pattern="\\bvar\\s+\\w+", output_mode="content", -n=true)

1. Analyze each occurrence to determine whether to use let or const:

- Check if variable is reassigned → use let - Check if variable is constant → use const

1. Replace systematically:

   Edit(file_path="src/index.js", old_string="var count = 0", new_string="let count = 0", replace_all=false)
   Edit(file_path="src/index.js", old_string="var MAX_SIZE = 100", new_string="const MAX_SIZE = 100", replace_all=false)

1. Use linter if available to verify syntax:

   npm run lint

Example 3: Update API Calls

User: "Update all authentication API calls to use the new endpoint"

Workflow:

1. Find all API calls:

   Grep(pattern="/api/auth/login", output_mode="content", -n=true)

1. Identify variation patterns:

- Look for fetch calls, axios calls, or other HTTP methods - Check for different HTTP methods (GET, POST, etc.)

1. Replace with updated endpoint:

   Edit(
     file_path="src/auth.js",
     old_string="fetch('/api/auth/login', {",
     new_string="fetch('/api/v2/authentication/login', {",
     replace_all=true
   )

1. Update any response handling if API contract changed

1. Recommend integration tests

Example 4: Convert Callbacks to Async/Await

User: "Refactor this file to use async/await instead of callbacks"

Workflow:

1. Analyze callback patterns:

   Grep(pattern="function\\s*\\([^)]*\\)\\s*{", output_mode="content", -n=true)

1. Identify callback functions (typically with callback, cb, or done parameters)

1. Transform each function:

   Edit(
     file_path="src/data.js",
     old_string="function loadData(callback) {\\n  db.query('SELECT *', callback);\\n}",
     new_string="async function loadData() {\\n  return await db.query('SELECT *');\\n}"
   )

1. Update call sites to use await:

   Edit(
     file_path="src/app.js",
     old_string="loadData((err, data) => {\\n  if (err) throw err;\\n  process(data);\\n});",
     new_string="const data = await loadData();\\nprocess(data);"
   )

1. Add try-catch for error handling where callbacks had error parameters

Example 5: Standardize Import Paths

User: "Convert all relative imports to absolute imports"

Workflow:

1. Find all relative imports:

   Grep(pattern="from ['\"]\\.\\.?/", output_mode="content", -n=true)

1. Calculate absolute path for each import based on project structure

1. Replace each import:

   Edit(
     file_path="src/components/Button.jsx",
     old_string="import { theme } from '../../utils/theme'",
     new_string="import { theme } from '@/utils/theme'",
     replace_all=true
   )

1. Verify module resolution works with new paths

Best Practices

Planning Refactoring

Before executing refactoring:

1. Understand scope: Use Grep to find all affected locations
2. Assess impact: Review context of each occurrence
3. Inform user: Report how many files/instances will be changed
4. Consider edge cases: Look for string literals, comments, documentation

Safe Refactoring Process

The refactoring workflow should be:

1. Search → Find all instances
2. Analyze → Verify each match is appropriate to change
3. Inform → Tell user the scope
4. Execute → Make the changes
5. Verify → Confirm changes were applied correctly
6. Test → Suggest running tests

Handling Special Cases

String literals and comments:

• Ask user if they want to update strings/comments containing the identifier
• Usually rename in code only, not in user-facing strings

Exported APIs:

• Warn if renaming exported functions (breaking change for consumers)
• Suggest deprecation warnings or maintaining aliases

Case sensitivity:

• Be explicit about case-sensitive vs case-insensitive replacements
• Use -i flag in Grep for case-insensitive when appropriate

Verification

After refactoring:

1. Re-run Grep to verify all instances were updated
2. Check syntax: Use linter or language-specific checker
3. Read affected files: Spot-check critical files
4. Recommend testing: Suggest running test suite

Error Handling

Common Issues and Solutions

Issue: "Replacement created syntax errors"

• Solution: Ensure old_string and new_string maintain proper syntax
• Prevention: Include sufficient context in old_string to ensure valid replacement points

Issue: "Not all instances were replaced"

• Solution: Check for variations in whitespace, quotes, or formatting
• Alternative: Use regex patterns in Grep to find variations, then handle each

Issue: "Replaced instances in comments/strings unintentionally"

• Solution: Be more specific with old_string context
• Consider: Use language-aware tools if available (parsers, AST tools)

Issue: "Replace broke tests"

• Solution: Review test files separately, update test expectations
• Prevention: Preview test file changes before applying

Tool Usage Reference

Edit Tool Parameters

Required:

• file_path: File to modify
• old_string: Exact string to find
• new_string: Replacement string

Optional:

• replace_all: Boolean (default: false)

- true: Replace all occurrences in file - false: Replace only first occurrence (or fail if multiple matches)

Important notes:

• old_string must match EXACTLY (including whitespace, quotes)
• If multiple matches exist without replace_all=true, Edit will fail
• Use replace_all=true for bulk refactoring

Grep for Refactoring

Useful Grep options for refactoring:

• -n=true: Show line numbers (helps locate changes)
• -B=N, -A=N: Show context (verify match is correct)
• -i=true: Case-insensitive (find variations)
• output_mode="content": See actual code
• output_mode="count": Count occurrences per file
• type: Filter by file type (e.g., type="py")

Integration with Other Tools

Working with Claude's Native Tools

• Grep: Find patterns across codebase
• Read: Analyze files before refactoring
• Edit: Execute replacements (with replace_all)
• Bash: Run linters, formatters, tests after refactoring
• Glob: Find files by pattern for targeted refactoring

Working with Other Skills

• test-fixing: Fix tests broken by refactoring
• code-transfer: Move refactored code to better locations
• feature-planning: Plan large-scale refactoring efforts

Language-Specific Patterns

JavaScript/TypeScript

Common refactoring patterns:

• var → let/const
• Callbacks → Promises/async-await
• require() → import statements
• CommonJS → ES modules
• Class components → Function components (React)

Python

Common refactoring patterns:

• Old string formatting → f-strings
• % formatting → .format() or f-strings
• Dict access → .get() with defaults
• Type hints additions

General

Universal refactoring patterns:

• Function/variable renaming
• API endpoint updates
• Library version migrations
• Naming convention standardization
• Import path updates