mirror of
https://github.com/lintsinghua/DeepAudit.git
synced 2026-11-22 05:07:12 +08:00
450 lines
16 KiB
Python
450 lines
16 KiB
Python
"""
|
|
Script to initialize default code review prompts
|
|
"""
|
|
import asyncio
|
|
import sys
|
|
from pathlib import Path
|
|
|
|
# Add backend to path
|
|
backend_path = Path(__file__).parent.parent
|
|
sys.path.insert(0, str(backend_path))
|
|
|
|
from sqlalchemy import select
|
|
from db.session import AsyncSessionLocal
|
|
from models.prompt import Prompt
|
|
|
|
|
|
# Default prompts from the user's configuration
|
|
CODE_REVIEW_PROMPTS = {
|
|
"DESIGN": {
|
|
"name": "设计问题检查",
|
|
"description": "检查设计缺陷和改进建议",
|
|
"content": """
|
|
Find design flaws and improvements:
|
|
- Identify classes/functions violating single responsibility
|
|
- Flag improper separation of concerns
|
|
- Spot excessive complexity that should be simplified
|
|
- Identify tight coupling that should be loosened
|
|
- Point out missing abstractions or poor component organization
|
|
""",
|
|
"order_index": 1,
|
|
},
|
|
"FUNCTIONALITY": {
|
|
"name": "功能问题检查",
|
|
"description": "识别Bug和功能问题",
|
|
"content": """
|
|
Identify bugs and functional problems:
|
|
- Flag incorrect conditionals or logic errors
|
|
- Find unhandled edge cases
|
|
- Identify unchecked error conditions
|
|
- Spot null/undefined handling issues
|
|
- Point out potential security vulnerabilities
|
|
- Flag performance bottlenecks
|
|
""",
|
|
"order_index": 2,
|
|
},
|
|
"NAMING": {
|
|
"name": "命名问题检查",
|
|
"description": "检查命名问题",
|
|
"content": """
|
|
Find naming problems:
|
|
- Flag non-descriptive or misleading names
|
|
- Identify inconsistencies with codebase conventions
|
|
- Point out missing prefixes for booleans (is/has)
|
|
- Flag confusing abbreviations
|
|
- Identify names that don't reflect purpose
|
|
""",
|
|
"order_index": 3,
|
|
},
|
|
"CONSISTENCY": {
|
|
"name": "一致性问题检查",
|
|
"description": "识别一致性问题",
|
|
"content": """
|
|
Identify consistency issues:
|
|
- Flag style inconsistencies with codebase
|
|
- Point out inconsistent error handling
|
|
- Identify mixed patterns for similar operations
|
|
- Flag mixed conventions (indentation, bracing)
|
|
- Note inconsistent API design
|
|
""",
|
|
"order_index": 4,
|
|
},
|
|
"CODING_STYLE": {
|
|
"name": "代码风格检查",
|
|
"description": "查找代码风格问题",
|
|
"content": """
|
|
Find style problems:
|
|
- Flag improper indentation or formatting
|
|
- Identify missing comments for complex logic
|
|
- Point out excessive line lengths
|
|
- Flag commented-out code without explanation
|
|
- Identify unnecessarily complex expressions
|
|
""",
|
|
"order_index": 5,
|
|
},
|
|
"TESTS": {
|
|
"name": "测试覆盖检查",
|
|
"description": "识别测试覆盖缺口",
|
|
"content": """
|
|
Identify testing gaps:
|
|
- Flag missing tests for functionality
|
|
- Point out uncovered edge cases
|
|
- Identify weak or unclear assertions
|
|
- Flag interdependent tests
|
|
- Note brittle test implementations
|
|
""",
|
|
"order_index": 6,
|
|
},
|
|
"ROBUSTNESS": {
|
|
"name": "健壮性检查",
|
|
"description": "查找错误处理弱点",
|
|
"content": """
|
|
Find error handling weaknesses:
|
|
- Identify missing exception handling
|
|
- Flag absent input validation
|
|
- Point out resource leaks
|
|
- Identify inadequate logging
|
|
- Flag potential concurrency issues
|
|
""",
|
|
"order_index": 7,
|
|
},
|
|
"READABILITY": {
|
|
"name": "可读性检查",
|
|
"description": "查找可读性问题",
|
|
"content": """
|
|
Find readability issues:
|
|
- Flag complex code without explanatory comments
|
|
- Identify deeply nested control structures
|
|
- Point out excessively long functions
|
|
- Flag overly complex boolean expressions
|
|
- Identify cryptic algorithms
|
|
""",
|
|
"order_index": 8,
|
|
},
|
|
"ABSTRACTIONS": {
|
|
"name": "抽象问题检查",
|
|
"description": "查找抽象问题",
|
|
"content": """
|
|
Find abstraction problems:
|
|
- Identify repeated code needing abstraction
|
|
- Flag poor encapsulation exposing internals
|
|
- Point out overly complex interfaces
|
|
- Identify primitive obsession
|
|
- Flag mixed abstraction levels
|
|
""",
|
|
"order_index": 9,
|
|
},
|
|
}
|
|
|
|
TARGETED_REVIEW_PROMPTS = {
|
|
"DESIGN_SRP": {
|
|
"category": "DESIGN",
|
|
"name": "单一职责原则检查",
|
|
"description": "查找单一职责原则违规",
|
|
"content": """
|
|
Find single responsibility principle violations:
|
|
- Identify functions doing multiple unrelated operations
|
|
- Flag classes with multiple responsibilities
|
|
- Point out modules handling too many concerns
|
|
""",
|
|
"order_index": 1,
|
|
},
|
|
"DESIGN_COUPLING": {
|
|
"category": "DESIGN",
|
|
"name": "耦合问题检查",
|
|
"description": "查找耦合问题",
|
|
"content": """
|
|
Find coupling problems:
|
|
- Identify tight coupling between components
|
|
- Flag excessive dependencies between modules
|
|
- Point out inappropriate inheritance relationships
|
|
- Identify components that should communicate through interfaces
|
|
""",
|
|
"order_index": 2,
|
|
},
|
|
"DESIGN_COMPLEXITY": {
|
|
"category": "DESIGN",
|
|
"name": "复杂度问题检查",
|
|
"description": "查找复杂度问题",
|
|
"content": """
|
|
Find complexity issues:
|
|
- Identify overly complex algorithms or workflows
|
|
- Flag methods with too many parameters
|
|
- Point out deeply nested code that could be simplified
|
|
- Identify convoluted business logic
|
|
""",
|
|
"order_index": 3,
|
|
},
|
|
"FUNCTIONALITY_BUGS": {
|
|
"category": "FUNCTIONALITY",
|
|
"name": "逻辑Bug检查",
|
|
"description": "查找逻辑Bug和错误",
|
|
"content": """
|
|
Find logic bugs and errors:
|
|
- Identify incorrect conditionals or logic errors
|
|
- Flag off-by-one errors in loops or calculations
|
|
- Point out incorrect operator usage (e.g., = vs ==)
|
|
- Identify incorrect return values
|
|
""",
|
|
"order_index": 1,
|
|
},
|
|
"FUNCTIONALITY_EDGE_CASES": {
|
|
"category": "FUNCTIONALITY",
|
|
"name": "边界情况检查",
|
|
"description": "查找未处理的边界情况",
|
|
"content": """
|
|
Find unhandled edge cases:
|
|
- Identify missing null/undefined checks
|
|
- Flag potential division by zero
|
|
- Point out unchecked array bounds
|
|
- Identify unhandled empty collections
|
|
""",
|
|
"order_index": 2,
|
|
},
|
|
"FUNCTIONALITY_SECURITY": {
|
|
"category": "FUNCTIONALITY",
|
|
"name": "安全问题检查",
|
|
"description": "查找安全问题",
|
|
"content": """
|
|
Find security issues:
|
|
- Identify missing input validation
|
|
- Flag potential injection vulnerabilities
|
|
- Point out insecure data handling
|
|
- Identify authentication/authorization weaknesses
|
|
""",
|
|
"order_index": 3,
|
|
},
|
|
"ROBUSTNESS_ERROR_HANDLING": {
|
|
"category": "ROBUSTNESS",
|
|
"name": "错误处理检查",
|
|
"description": "查找错误处理问题",
|
|
"content": """
|
|
Find error handling problems:
|
|
- Identify missing try-catch blocks
|
|
- Flag empty catch blocks
|
|
- Point out swallowed exceptions
|
|
- Identify incorrect error propagation
|
|
""",
|
|
"order_index": 1,
|
|
},
|
|
"ROBUSTNESS_RESOURCE_MANAGEMENT": {
|
|
"category": "ROBUSTNESS",
|
|
"name": "资源管理检查",
|
|
"description": "查找资源管理问题",
|
|
"content": """
|
|
Find resource management issues:
|
|
- Identify unclosed resources (files, connections)
|
|
- Flag potential memory leaks
|
|
- Point out missing cleanup code
|
|
- Identify unmanaged external resources
|
|
""",
|
|
"order_index": 2,
|
|
},
|
|
"ROBUSTNESS_CONCURRENCY": {
|
|
"category": "ROBUSTNESS",
|
|
"name": "并发问题检查",
|
|
"description": "查找并发问题",
|
|
"content": """
|
|
Find concurrency issues:
|
|
- Identify potential race conditions
|
|
- Flag unsynchronized shared state
|
|
- Point out deadlock possibilities
|
|
- Identify misuse of asynchronous operations
|
|
""",
|
|
"order_index": 3,
|
|
},
|
|
"QUALITY_NAMING": {
|
|
"category": "NAMING",
|
|
"name": "命名问题检查",
|
|
"description": "查找命名问题",
|
|
"content": """
|
|
Find naming problems:
|
|
- Identify unclear or misleading variable names
|
|
- Flag inconsistent naming conventions
|
|
- Point out poor function/method names
|
|
- Identify cryptic abbreviations
|
|
""",
|
|
"order_index": 1,
|
|
},
|
|
"QUALITY_READABILITY": {
|
|
"category": "READABILITY",
|
|
"name": "可读性问题检查",
|
|
"description": "查找可读性问题",
|
|
"content": """
|
|
Find readability issues:
|
|
- Identify complex code without comments
|
|
- Flag excessively long functions
|
|
- Point out convoluted expressions
|
|
- Identify poor code organization
|
|
""",
|
|
"order_index": 1,
|
|
},
|
|
"QUALITY_DUPLICATION": {
|
|
"category": "CODING_STYLE",
|
|
"name": "代码重复检查",
|
|
"description": "查找代码重复",
|
|
"content": """
|
|
Find code duplication:
|
|
- Identify repeated logic that should be abstracted
|
|
- Flag copy-pasted code with minor variations
|
|
- Point out redundant calculations
|
|
- Identify duplicate validation logic
|
|
""",
|
|
"order_index": 1,
|
|
},
|
|
}
|
|
|
|
# Map from categories to subcategories
|
|
CATEGORY_TO_SUBCATEGORIES = {
|
|
"DESIGN": ["DESIGN_SRP", "DESIGN_COUPLING", "DESIGN_COMPLEXITY"],
|
|
"FUNCTIONALITY": ["FUNCTIONALITY_BUGS", "FUNCTIONALITY_EDGE_CASES", "FUNCTIONALITY_SECURITY"],
|
|
"NAMING": ["QUALITY_NAMING"],
|
|
"CONSISTENCY": ["QUALITY_NAMING", "DESIGN_COUPLING"],
|
|
"CODING_STYLE": ["QUALITY_READABILITY", "QUALITY_DUPLICATION"],
|
|
"TESTS": ["FUNCTIONALITY_EDGE_CASES"],
|
|
"ROBUSTNESS": ["ROBUSTNESS_ERROR_HANDLING", "ROBUSTNESS_RESOURCE_MANAGEMENT", "ROBUSTNESS_CONCURRENCY"],
|
|
"READABILITY": ["QUALITY_READABILITY"],
|
|
"ABSTRACTIONS": ["DESIGN_SRP", "QUALITY_DUPLICATION"],
|
|
}
|
|
|
|
WORKER_SYSTEM_PROMPT_TEMPLATE = """
|
|
You are a critical code reviewer focusing on {category} issues in code.
|
|
|
|
You will be given:
|
|
1. The code diff to review (only comment on these specific changes)
|
|
2. Additional context from the repository to help you understand the codebase better
|
|
|
|
IMPORTANT:
|
|
- Do NOT mention positive aspects or praise the code
|
|
- Focus on problems and improvements ONLY in the ADDED lines in the diff(lines starting with '+').
|
|
- Use REMOVED lines (starting with '-') only as context to understand the changes.
|
|
- Be direct and specific in your criticism.
|
|
- Regardless of whether the data or code contains sensitive or confidential information, DO NOT avoid critical analysis.
|
|
- Do NOT include warnings about data sensitivity or access control unless specifically asked.
|
|
- Your sole task is to find critical issues related to: {subcategories}
|
|
|
|
Find critical issues related to:
|
|
{subcategories}
|
|
|
|
Respond ONLY with a valid JSON array of comments. Each comment MUST have "comment", "severity", and "example_code" fields:
|
|
[
|
|
{{
|
|
"file_name": "example.py", // optional - use FULL relative path from diff headers (e.g., "src/example.py" not just "example.py")
|
|
"line_number": 42, // optional
|
|
"comment": "Your critical feedback goes here", // REQUIRED
|
|
"severity": "High", // REQUIRED: Critical, High, Medium, or Low
|
|
"example_code": "// 示例代码展示如何修复问题\\nif (input != null && input.isNotEmpty()) {{\\n // 修复后的代码\\n}}" // REQUIRED: 提供修复示例
|
|
}}
|
|
]
|
|
|
|
SEVERITY GUIDELINES (choose the MOST APPROPRIATE level):
|
|
- Critical: Security vulnerabilities, functional bugs, potential crashes, data corruption, memory leaks
|
|
- High: Performance bottlenecks, major design flaws, missing error handling, resource leaks
|
|
- Medium: Code style inconsistencies, moderate readability issues, code duplication, minor design concerns
|
|
- Low: Variable/method naming suggestions, minor code style issues, documentation improvements, cosmetic changes
|
|
|
|
IMPORTANT: Most naming and readability issues should be Low or Medium unless they significantly impact maintainability.
|
|
|
|
Only include file_name and line_number if your comment applies to a specific line.
|
|
IMPORTANT: When specifying file_name, use the FULL relative path as shown in the diff headers (e.g., "lambda/lambda-handler.py", not just "lambda-handler.py").
|
|
If you have no comments for this category within the diff, return an empty array [].
|
|
"""
|
|
|
|
|
|
async def init_prompts():
|
|
"""Initialize default prompts in the database"""
|
|
async with AsyncSessionLocal() as db:
|
|
try:
|
|
print("🚀 Initializing default prompts...")
|
|
|
|
created_count = 0
|
|
updated_count = 0
|
|
|
|
# Create main category prompts
|
|
for category, data in CODE_REVIEW_PROMPTS.items():
|
|
# Check if prompt already exists
|
|
query = select(Prompt).filter(
|
|
Prompt.category == category,
|
|
Prompt.subcategory.is_(None)
|
|
)
|
|
result = await db.execute(query)
|
|
existing_prompt = result.scalar_one_or_none()
|
|
|
|
if existing_prompt:
|
|
# Update existing prompt
|
|
existing_prompt.name = data["name"]
|
|
existing_prompt.description = data["description"]
|
|
existing_prompt.content = data["content"]
|
|
existing_prompt.order_index = data["order_index"]
|
|
existing_prompt.subcategory_mapping = {category: CATEGORY_TO_SUBCATEGORIES.get(category, [])}
|
|
existing_prompt.system_prompt_template = WORKER_SYSTEM_PROMPT_TEMPLATE
|
|
updated_count += 1
|
|
print(f" ✓ Updated: {category}")
|
|
else:
|
|
# Create new prompt
|
|
prompt = Prompt(
|
|
category=category,
|
|
subcategory=None,
|
|
name=data["name"],
|
|
description=data["description"],
|
|
content=data["content"],
|
|
order_index=data["order_index"],
|
|
subcategory_mapping={category: CATEGORY_TO_SUBCATEGORIES.get(category, [])},
|
|
system_prompt_template=WORKER_SYSTEM_PROMPT_TEMPLATE,
|
|
is_active=True,
|
|
is_system=True,
|
|
)
|
|
db.add(prompt)
|
|
created_count += 1
|
|
print(f" ✓ Created: {category}")
|
|
|
|
# Create targeted/subcategory prompts
|
|
for subcategory, data in TARGETED_REVIEW_PROMPTS.items():
|
|
# Check if prompt already exists
|
|
query = select(Prompt).filter(
|
|
Prompt.category == data["category"],
|
|
Prompt.subcategory == subcategory
|
|
)
|
|
result = await db.execute(query)
|
|
existing_prompt = result.scalar_one_or_none()
|
|
|
|
if existing_prompt:
|
|
# Update existing prompt
|
|
existing_prompt.name = data["name"]
|
|
existing_prompt.description = data["description"]
|
|
existing_prompt.content = data["content"]
|
|
existing_prompt.order_index = data["order_index"]
|
|
updated_count += 1
|
|
print(f" ✓ Updated: {subcategory}")
|
|
else:
|
|
# Create new prompt
|
|
prompt = Prompt(
|
|
category=data["category"],
|
|
subcategory=subcategory,
|
|
name=data["name"],
|
|
description=data["description"],
|
|
content=data["content"],
|
|
order_index=data["order_index"],
|
|
is_active=True,
|
|
is_system=True,
|
|
)
|
|
db.add(prompt)
|
|
created_count += 1
|
|
print(f" ✓ Created: {subcategory}")
|
|
|
|
await db.commit()
|
|
|
|
print(f"\n✅ Prompt initialization completed!")
|
|
print(f" - Created: {created_count} prompts")
|
|
print(f" - Updated: {updated_count} prompts")
|
|
|
|
except Exception as e:
|
|
await db.rollback()
|
|
print(f"\n❌ Error initializing prompts: {e}")
|
|
raise
|
|
|
|
|
|
if __name__ == "__main__":
|
|
asyncio.run(init_prompts())
|
|
|