phase-8-review（第 8 阶段审查）

Phase 8: Architecture/Convention Review

Overall codebase quality verification

Purpose

Review the entire codebase before deployment. Identify architecture consistency, convention compliance, and potential issues.

What to Do in This Phase

1. Architecture Review: Review structural consistency
2. Convention Review: Verify rule compliance
3. Code Quality Review: Duplication, complexity, potential bugs
4. Refactoring: Fix discovered issues

Deliverables

docs/03-analysis/
├── architecture-review.md      # Architecture review
├── convention-review.md        # Convention review
└── refactoring-plan.md         # Refactoring plan

PDCA Application

• Plan: Define review scope/criteria
• Design: Design checklist
• Do: Execute code review
• Check: Analyze issues
• Act: Refactor and proceed to Phase 9

Level-wise Application

Full Phase Verification Matrix

Cross-Phase Consistency Verification

Phase 8 verifies that all Phase outputs and rules are consistently applied.

┌─────────────────────────────────────────────────────────────────┐
│                    Cross-Phase Dependency Flow                    │
├─────────────────────────────────────────────────────────────────┤
│                                                                  │
│   Phase 1 (Schema/Terminology)                                   │
│       ↓ Glossary, entity definitions                             │
│   Phase 2 (Coding Convention)                                    │
│       ↓ Naming rules, environment variable conventions           │
│   Phase 3 (Mockup)                                               │
│       ↓ Component structure, Props design                        │
│   Phase 4 (API)                                                  │
│       ↓ RESTful principles, response format, error codes         │
│   Phase 5 (Design System)                                        │
│       ↓ Design tokens, component variants                        │
│   Phase 6 (UI Implementation)                                    │
│       ↓ API client, type sharing, error handling                 │
│   Phase 7 (SEO/Security)                                         │
│       ↓ Security rules, metadata                                 │
│   Phase 8 (Review) ← Current stage: Full verification            │
│       ↓                                                          │
│   Phase 9 (Deployment)                                           │
│                                                                  │
└─────────────────────────────────────────────────────────────────┘

Phase-specific Verification Checklist

Phase 1 → Verify: Terminology/Schema Consistency

□ Are glossary.md terms consistently used in code?
  - Business terms → Code naming matching
  - Global standard terms → API response field names matching
□ Do entity definitions match actual types?
□ Do relationship definitions match actual implementation?

Phase 2 → Verify: Convention Compliance

□ Naming rule compliance (PascalCase, camelCase, UPPER_SNAKE_CASE)
□ Folder structure rule compliance
□ Environment variable naming rule compliance (NEXT_PUBLIC_*, DB_*, API_*, etc.)
□ .env.example template completion
□ Environment variable validation logic (lib/env.ts) exists

Phase 4 → Verify: API Consistency

□ RESTful principle compliance
  - Resource-based URLs (nouns, plural)
  - Correct HTTP method usage
  - Status code consistency
□ Response format consistency
  - Success: { data, meta? }
  - Error: { error: { code, message, details? } }
  - Pagination: { data, pagination }
□ Error code standardization (matches ERROR_CODES constant)

Phase 5 → Verify: Design System Consistency

□ Are design tokens defined? (CSS Variables / ThemeData)
□ Do components use tokens? (no hardcoded colors)
□ Are component variants consistent?
□ Dark mode support (if defined)

Phase 6 → Verify: UI-API Integration Consistency

□ API client layer structure compliance
  - Components → hooks → services → apiClient
  - No direct fetch calls
□ Type consistency
  - Phase 4 API spec types = Phase 6 client types
□ Error handling consistency
  - Global error handler usage
  - Error code-specific handling logic
□ State management pattern consistency

Phase 7 → Verify: Security/SEO Application

□ Authentication/authorization middleware applied
□ Input validation (server-side)
□ XSS, CSRF defense
□ No sensitive info exposed to client
□ SEO meta tags applied

Phase 9 → Verify: Deployment Readiness

□ Environment variable Secrets registered (based on Phase 2 list)
□ Environment separation (dev/staging/prod)
□ Build successful
□ Environment variable validation script passes

Clean Architecture Verification

Layer Separation Principles

┌─────────────────────────────────────────────────────────────┐
│                      Presentation Layer                      │
│         (pages, components, hooks - UI concerns)             │
├─────────────────────────────────────────────────────────────┤
│                      Application Layer                       │
│         (services, use-cases - business logic)               │
├─────────────────────────────────────────────────────────────┤
│                       Domain Layer                           │
│         (entities, types - core domain models)               │
├─────────────────────────────────────────────────────────────┤
│                    Infrastructure Layer                      │
│         (api client, db, external services)                  │
└─────────────────────────────────────────────────────────────┘

Dependency direction: Outside → Inside (Presentation → Domain)
Inner layers must not know about outer layers

Layer-specific Verification Checklist

Presentation Layer (UI)

□ Is there business logic in components?
□ Are there direct API calls? (should go through hooks)
□ Is state management properly separated?
□ Do components have single responsibility?

Application Layer (Services)

□ Are domain logic and infrastructure logic separated?
□ Are external dependencies abstracted?
□ Is the structure testable?
□ Are use cases clearly defined?

Domain Layer (Types/Entities)

□ Are there no external library dependencies?
□ Does it contain only pure business rules?
□ Do types match Phase 1 schema?

Infrastructure Layer (API Client)

□ Are external service calls abstracted?
□ Is error handling consistent?
□ Is configuration managed via environment variables?

Architecture Review Checklist

Structure

• Does folder structure match conventions
• Is separation of concerns well done
• Is dependency direction correct (outside → inside)

Patterns

• Are consistent patterns used
• Is there unnecessary abstraction
• Is proper encapsulation done

Convention Review Checklist

Naming

• Does it follow Phase 2 defined rules
• Are meaningful names used
• Is there consistency

Code Style

• Unified indentation, quotes, etc.
• Is file length appropriate
• Is function length appropriate

Code Quality Checklist

• Is there duplicate code
• Are there highly complex functions
• Is error handling appropriate
• Is type safety ensured

AI-Assisted Review

Request code review from Claude:

"Review this project's code.
- Does it follow CONVENTIONS.md rules
- Is there architecture consistency
- Are there potential bugs or improvements"

Template

See templates/pipeline/phase-8-review.template.md

Next Phase

Phase 9: Deployment → After review completion, deploy to production

6. In-Depth Code Quality Review

6.1 Duplicate Code Detection

Detection Methods

# 1. Search for similar function names
grep -r "function.*format" src/
grep -r "function.*calculate" src/
grep -r "function.*get.*By" src/

# 2. Search for similar patterns
grep -rn "reduce.*sum" src/
grep -rn "filter.*map" src/
grep -rn "useState.*useEffect" src/

Handling by Duplication Type

Duplicate Code Checklist

□ Is the same logic in 2+ places?
□ Are there multiple functions with similar names?
□ Is the same data transformation repeated?
□ Are similar UI patterns repeated?
□ Is the same API call pattern repeated?
□ Is similar error handling repeated?

6.2 Reusability Assessment

Assessment Criteria

Reusability Checklist

Check for each function/component:
□ Is it tied to a specific domain?
□ Does it depend on external state?
□ Are parameters generic?
□ Is the return value predictable?
□ Are there side effects?

6.3 Extensibility Assessment

Extensibility Check

When new requirements come:
□ Can it be added without modifying existing code?
□ Can behavior be changed by configuration only?
□ Is adding new types/cases easy?
□ Can it be extended without adding conditionals?

Extensibility Anti-patterns

// ❌ Requires modification for each extension
function process(type: string) {
  if (type === 'a') { /* ... */ }
  else if (type === 'b') { /* ... */ }
  // Add else if for each new type...
}

// ❌ Hardcoded list
const ALLOWED_TYPES = ['a', 'b', 'c']

// ❌ Enumerated switch statements
switch (action.type) {
  case 'ADD': // ...
  case 'REMOVE': // ...
  // Add case for each new action...
}

Good Extensibility Patterns

// ✅ Registry pattern
const handlers: Record<string, Handler> = {}
function register(type: string, handler: Handler) {
  handlers[type] = handler
}
function process(type: string, data: unknown) {
  return handlers[type]?.(data)
}

// ✅ Configuration-based
const CONFIG = {
  types: ['a', 'b', 'c'],
  handlers: { ... }
}

// ✅ Plugin structure
interface Plugin { execute(data): Result }
const plugins: Plugin[] = []

6.4 Object-Oriented Principles Check

SOLID Principles Checklist

S - Single Responsibility (SRP)

□ Does the class/function change for only one reason?
□ Does the name clearly explain the role?
□ Is "and" in the name? → Needs separation

O - Open/Closed (OCP)

□ Is it open for extension? (new features can be added)
□ Is it closed for modification? (no existing code changes needed)
□ Are interfaces/abstractions used?

L - Liskov Substitution (LSP)

□ Can subtypes replace parent types?
□ Do overridden methods keep the contract?

I - Interface Segregation (ISP)

□ Is the interface too large?
□ Must unused methods be implemented?
□ Can the interface be split smaller?

D - Dependency Inversion (DIP)

□ Does it depend on abstractions instead of concrete classes?
□ Are dependencies injected? (DI)
□ Is the structure testable?

6.5 Refactoring Priority

Urgent (Required before deployment):
1. Duplication that can cause bugs
2. Security vulnerabilities
3. Performance bottlenecks

High (As soon as possible):
4. Same logic duplicated in 3+ places
5. Files over 200 lines
6. Nesting deeper than 5 levels

Medium (Next sprint):
7. Structure lacking extensibility
8. Naming inconsistencies
9. Structure difficult to test

Low (Backlog):
10. Style inconsistencies
11. Excessive comments
12. Unused code

7. AI Code Review Request Template

Please review the code from these perspectives:

1. Duplicate Code
   - Are there similar functions/components?
   - Is there common logic that can be extracted?

2. Reusability
   - Can it be used generically?
   - Is it tied to a specific domain?

3. Extensibility
   - Can it flexibly respond to new requirements?
   - Are there hardcoded parts?

4. SOLID Principles
   - Does it follow single responsibility?
   - Is it open for extension and closed for modification?

5. Convention Compliance
   - Does it follow CONVENTIONS.md rules?
   - Is naming consistent?

Please identify parts that need refactoring and their priority.

8. Gap Analysis (Design vs Implementation)

Gap Analysis Report Template

# Gap Analysis Report

## Analysis Target
- Design document: docs/02-design/{feature}.design.md
- Implementation path: src/features/{feature}/

## Results by Category

### API Endpoints
| Design | Implementation | Status |
|--------|----------------|--------|
| POST /api/users | POST /api/users | ✅ Match |
| GET /api/users/:id | - | ❌ Not implemented |
| - | DELETE /api/users/:id | ⚠️ Missing from design |

### Data Model
| Design Entity | Implementation | Status |
|---------------|----------------|--------|
| User | types/user.ts | ✅ Match |
| UserRole | - | ❌ Not implemented |

### Match Rate
- Total items: 10
- Matches: 7
- Not implemented: 2
- Missing from design: 1
- **Match Rate: 70%**

Gap Types and Actions

When to Run Gap Analysis

• After completing feature implementation
• Before deployment
• When design document is updated
• During code review