222 lines
7.1 KiB
Python
222 lines
7.1 KiB
Python
#!/usr/bin/env python3
|
||
# -*- coding: utf-8 -*-
|
||
"""
|
||
向量记忆系统 - 自动备份脚本
|
||
支持：
|
||
1. 导出到本地 JSON/Markdown
|
||
2. 同步到飞书云盘
|
||
3. 定时自动备份
|
||
"""
|
||
|
||
import os
|
||
import sys
|
||
import json
|
||
import sqlite3
|
||
import shutil
|
||
from datetime import datetime
|
||
from pathlib import Path
|
||
|
||
sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
|
||
|
||
from vector_memory import VectorMemorySystem
|
||
|
||
|
||
class MemoryBackup:
|
||
"""记忆备份管理器"""
|
||
|
||
def __init__(self, backup_dir: str = "./backups"):
|
||
self.backup_dir = Path(backup_dir)
|
||
self.backup_dir.mkdir(exist_ok=True)
|
||
|
||
def export_json(self, vm: VectorMemorySystem, filepath: str = None) -> str:
|
||
"""导出为 JSON 格式"""
|
||
if not filepath:
|
||
filepath = self.backup_dir / f"memory_backup_{datetime.now().strftime('%Y%m%d_%H%M%S')}.json"
|
||
|
||
# 从 SQLite 读取全部记忆
|
||
conn = sqlite3.connect(vm.db_path)
|
||
conn.row_factory = sqlite3.Row
|
||
cursor = conn.execute("SELECT * FROM memories ORDER BY created_at DESC")
|
||
|
||
memories = []
|
||
for row in cursor.fetchall():
|
||
memories.append({
|
||
'id': row['id'],
|
||
'content': row['content'],
|
||
'metadata': json.loads(row['metadata']) if row['metadata'] else {},
|
||
'importance': row['importance'],
|
||
'tier': row['tier'],
|
||
'created_at': row['created_at'],
|
||
'updated_at': row['updated_at']
|
||
})
|
||
|
||
conn.close()
|
||
|
||
# 写入文件
|
||
backup_data = {
|
||
'version': '1.0',
|
||
'export_time': datetime.now().isoformat(),
|
||
'total_count': len(memories),
|
||
'memories': memories
|
||
}
|
||
|
||
with open(filepath, 'w', encoding='utf-8') as f:
|
||
json.dump(backup_data, f, ensure_ascii=False, indent=2)
|
||
|
||
print(f"✅ JSON 备份已保存: {filepath}")
|
||
return str(filepath)
|
||
|
||
def export_markdown(self, vm: VectorMemorySystem, filepath: str = None) -> str:
|
||
"""导出为 Markdown 格式（可读性好）"""
|
||
if not filepath:
|
||
filepath = self.backup_dir / f"memory_backup_{datetime.now().strftime('%Y%m%d_%H%M%S')}.md"
|
||
|
||
# 从 SQLite 读取
|
||
conn = sqlite3.connect(vm.db_path)
|
||
conn.row_factory = sqlite3.Row
|
||
cursor = conn.execute("SELECT * FROM memories ORDER BY created_at DESC")
|
||
|
||
rows = cursor.fetchall()
|
||
conn.close()
|
||
|
||
# 生成 Markdown
|
||
md_lines = [
|
||
"# 🧠 OpenClaw 向量记忆备份",
|
||
f"",
|
||
f"> 导出时间: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}",
|
||
f"> 总数量: {len(rows)} 条",
|
||
f"",
|
||
f"---",
|
||
f""
|
||
]
|
||
|
||
for i, row in enumerate(rows, 1):
|
||
metadata = json.loads(row['metadata']) if row['metadata'] else {}
|
||
tags = metadata.get('tags', [])
|
||
|
||
md_lines.append(f"## {i}. [{'⭐' * row['importance']}] {row['content'][:60]}...")
|
||
md_lines.append(f"")
|
||
md_lines.append(f"- **ID**: `{row['id']}`")
|
||
md_lines.append(f"- **重要性**: {row['importance']}/5")
|
||
md_lines.append(f"- **层级**: {row['tier']}")
|
||
md_lines.append(f"- **标签**: {', '.join(tags) if tags else '无'}")
|
||
md_lines.append(f"- **创建时间**: {row['created_at']}")
|
||
md_lines.append(f"- **内容**: {row['content']}")
|
||
md_lines.append(f"")
|
||
md_lines.append(f"---")
|
||
md_lines.append(f"")
|
||
|
||
with open(filepath, 'w', encoding='utf-8') as f:
|
||
f.write('\n'.join(md_lines))
|
||
|
||
print(f"✅ Markdown 备份已保存: {filepath}")
|
||
return str(filepath)
|
||
|
||
def backup_all(self, vm: VectorMemorySystem) -> dict:
|
||
"""执行完整备份"""
|
||
timestamp = datetime.now().strftime('%Y%m%d_%H%M%S')
|
||
|
||
results = {
|
||
'json': None,
|
||
'markdown': None,
|
||
'vector': None,
|
||
'timestamp': timestamp
|
||
}
|
||
|
||
# 1. JSON 备份
|
||
results['json'] = self.export_json(vm)
|
||
|
||
# 2. Markdown 备份
|
||
results['markdown'] = self.export_markdown(vm)
|
||
|
||
# 3. 向量数据库备份（复制整个目录）
|
||
vector_backup = self.backup_dir / f"vector_{timestamp}"
|
||
shutil.copytree(
|
||
os.path.dirname(vm.db_path),
|
||
vector_backup,
|
||
dirs_exist_ok=True
|
||
)
|
||
results['vector'] = str(vector_backup)
|
||
|
||
# 生成备份清单
|
||
manifest = {
|
||
'backup_time': timestamp,
|
||
'files': results,
|
||
'total_memories': vm.count()
|
||
}
|
||
|
||
manifest_file = self.backup_dir / f"manifest_{timestamp}.json"
|
||
with open(manifest_file, 'w', encoding='utf-8') as f:
|
||
json.dump(manifest, f, indent=2)
|
||
|
||
print(f"\n📋 备份清单: {manifest_file}")
|
||
|
||
return results
|
||
|
||
def restore_from_json(self, json_file: str, vm: VectorMemorySystem = None):
|
||
"""从 JSON 恢复记忆"""
|
||
with open(json_file, 'r', encoding='utf-8') as f:
|
||
data = json.load(f)
|
||
|
||
if not vm:
|
||
api_key = os.getenv("SILICONFLOW_API_KEY")
|
||
vm = VectorMemorySystem(api_key=api_key)
|
||
|
||
print(f"📥 开始恢复 {len(data['memories'])} 条记忆...")
|
||
|
||
for mem in data['memories']:
|
||
try:
|
||
vm.add_memory(
|
||
content=mem['content'],
|
||
metadata=mem['metadata'],
|
||
importance=mem['importance']
|
||
)
|
||
except Exception as e:
|
||
print(f" ⚠️ 跳过重复: {mem['id']}")
|
||
|
||
print(f"✅ 恢复完成！")
|
||
|
||
|
||
def main():
|
||
import argparse
|
||
|
||
parser = argparse.ArgumentParser(description="向量记忆备份工具")
|
||
subparsers = parser.add_subparsers(dest="command")
|
||
|
||
# 导出命令
|
||
subparsers.add_parser("export-json", help="导出为 JSON")
|
||
subparsers.add_parser("export-markdown", help="导出为 Markdown")
|
||
subparsers.add_parser("backup", help="完整备份（JSON + Markdown + 向量库）")
|
||
|
||
# 恢复命令
|
||
restore_parser = subparsers.add_parser("restore", help="从 JSON 恢复")
|
||
restore_parser.add_argument("file", help="备份文件路径")
|
||
|
||
args = parser.parse_args()
|
||
|
||
# 获取 API Key
|
||
api_key = os.getenv("SILICONFLOW_API_KEY")
|
||
if not api_key:
|
||
print("❌ 请设置 SILICONFLOW_API_KEY 环境变量")
|
||
sys.exit(1)
|
||
|
||
# 初始化
|
||
vm = VectorMemorySystem(api_key=api_key)
|
||
backup = MemoryBackup()
|
||
|
||
# 执行命令
|
||
if args.command == "export-json":
|
||
backup.export_json(vm)
|
||
elif args.command == "export-markdown":
|
||
backup.export_markdown(vm)
|
||
elif args.command == "backup":
|
||
backup.backup_all(vm)
|
||
elif args.command == "restore":
|
||
backup.restore_from_json(args.file, vm)
|
||
else:
|
||
parser.print_help()
|
||
|
||
|
||
if __name__ == "__main__":
|
||
main()
|