Spaces:
Sleeping
Sleeping
Update scripts/backup_to_dataset.py
Browse files- scripts/backup_to_dataset.py +97 -22
scripts/backup_to_dataset.py
CHANGED
|
@@ -2,15 +2,49 @@
|
|
| 2 |
import os
|
| 3 |
import sys
|
| 4 |
import datetime
|
|
|
|
|
|
|
|
|
|
| 5 |
from huggingface_hub import HfApi, login
|
| 6 |
|
| 7 |
# ===== 配置 =====
|
| 8 |
-
DB_PATH = "/app/server/data/freeapi.db"
|
| 9 |
-
DATASET_REPO = "lydgs/freellm-backup"
|
| 10 |
-
BACKUP_PREFIX = "freeapi_backup"
|
| 11 |
-
|
|
|
|
| 12 |
# ================
|
| 13 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
def backup_database():
|
| 15 |
if not os.path.exists(DB_PATH):
|
| 16 |
print(f"❌ 数据库文件不存在: {DB_PATH}")
|
|
@@ -23,11 +57,12 @@ def backup_database():
|
|
| 23 |
|
| 24 |
timestamp = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
|
| 25 |
backup_name = f"{BACKUP_PREFIX}_{timestamp}.db"
|
|
|
|
| 26 |
|
| 27 |
login(token=token)
|
| 28 |
api = HfApi()
|
| 29 |
|
| 30 |
-
# 上传
|
| 31 |
try:
|
| 32 |
api.upload_file(
|
| 33 |
path_or_fileobj=DB_PATH,
|
|
@@ -35,29 +70,69 @@ def backup_database():
|
|
| 35 |
repo_id=DATASET_REPO,
|
| 36 |
repo_type="dataset"
|
| 37 |
)
|
| 38 |
-
print(f"✅ 备份成功: {backup_name}")
|
| 39 |
except Exception as e:
|
| 40 |
-
print(f"❌ 上传失败: {e}")
|
| 41 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 42 |
|
| 43 |
-
# 清理旧备份(
|
| 44 |
try:
|
| 45 |
files = api.list_repo_files(repo_id=DATASET_REPO, repo_type="dataset")
|
| 46 |
now = datetime.datetime.now()
|
| 47 |
deleted = 0
|
| 48 |
for f in files:
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
|
| 58 |
-
|
| 59 |
-
|
| 60 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 61 |
if deleted:
|
| 62 |
print(f"✅ 清理完成,共删除 {deleted} 个旧备份")
|
| 63 |
else:
|
|
@@ -68,7 +143,7 @@ def backup_database():
|
|
| 68 |
return True
|
| 69 |
|
| 70 |
if __name__ == "__main__":
|
| 71 |
-
print(f"[{datetime.datetime.now()}] 开始备份...")
|
| 72 |
if backup_database():
|
| 73 |
print("备份任务完成")
|
| 74 |
sys.exit(0)
|
|
|
|
| 2 |
import os
|
| 3 |
import sys
|
| 4 |
import datetime
|
| 5 |
+
import json
|
| 6 |
+
import sqlite3
|
| 7 |
+
import tarfile
|
| 8 |
from huggingface_hub import HfApi, login
|
| 9 |
|
| 10 |
# ===== 配置 =====
|
| 11 |
+
DB_PATH = "/app/server/data/freeapi.db" # 数据库路径
|
| 12 |
+
DATASET_REPO = "lydgs/freellm-backup" # 数据集名称
|
| 13 |
+
BACKUP_PREFIX = "freeapi_backup" # 二进制备份前缀
|
| 14 |
+
CONFIG_PREFIX = "config_export" # 配置导出前缀
|
| 15 |
+
RETENTION_DAYS = 30
|
| 16 |
# ================
|
| 17 |
|
| 18 |
+
def export_db_to_json(db_path, temp_dir):
|
| 19 |
+
"""将数据库中所有表导出为JSON文件,保存在temp_dir下,返回生成的文件列表"""
|
| 20 |
+
if not os.path.exists(db_path):
|
| 21 |
+
print(f"⚠️ 数据库文件不存在: {db_path}")
|
| 22 |
+
return []
|
| 23 |
+
|
| 24 |
+
conn = sqlite3.connect(db_path)
|
| 25 |
+
cursor = conn.cursor()
|
| 26 |
+
|
| 27 |
+
# 获取所有表名
|
| 28 |
+
cursor.execute("SELECT name FROM sqlite_master WHERE type='table';")
|
| 29 |
+
tables = cursor.fetchall()
|
| 30 |
+
|
| 31 |
+
exported_files = []
|
| 32 |
+
for (table_name,) in tables:
|
| 33 |
+
cursor.execute(f"SELECT * FROM {table_name}")
|
| 34 |
+
rows = cursor.fetchall()
|
| 35 |
+
# 获取列名
|
| 36 |
+
col_names = [desc[0] for desc in cursor.description]
|
| 37 |
+
# 转换为字典列表
|
| 38 |
+
data = [dict(zip(col_names, row)) for row in rows]
|
| 39 |
+
json_path = os.path.join(temp_dir, f"{table_name}.json")
|
| 40 |
+
with open(json_path, 'w', encoding='utf-8') as f:
|
| 41 |
+
json.dump(data, f, indent=2, ensure_ascii=False)
|
| 42 |
+
exported_files.append(json_path)
|
| 43 |
+
print(f"📄 导出表 {table_name} 到 {json_path} ({len(data)} 行)")
|
| 44 |
+
|
| 45 |
+
conn.close()
|
| 46 |
+
return exported_files
|
| 47 |
+
|
| 48 |
def backup_database():
|
| 49 |
if not os.path.exists(DB_PATH):
|
| 50 |
print(f"❌ 数据库文件不存在: {DB_PATH}")
|
|
|
|
| 57 |
|
| 58 |
timestamp = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
|
| 59 |
backup_name = f"{BACKUP_PREFIX}_{timestamp}.db"
|
| 60 |
+
config_tar_name = f"{CONFIG_PREFIX}_{timestamp}.tar.gz"
|
| 61 |
|
| 62 |
login(token=token)
|
| 63 |
api = HfApi()
|
| 64 |
|
| 65 |
+
# 1. 上传二进制备份
|
| 66 |
try:
|
| 67 |
api.upload_file(
|
| 68 |
path_or_fileobj=DB_PATH,
|
|
|
|
| 70 |
repo_id=DATASET_REPO,
|
| 71 |
repo_type="dataset"
|
| 72 |
)
|
| 73 |
+
print(f"✅ 二进制备份成功: {backup_name}")
|
| 74 |
except Exception as e:
|
| 75 |
+
print(f"❌ 二进制上传失败: {e}")
|
| 76 |
+
# 继续尝试导出配置,不中断
|
| 77 |
+
|
| 78 |
+
# 2. 导出 JSON 配置并打包上传
|
| 79 |
+
try:
|
| 80 |
+
# 创建临时目录
|
| 81 |
+
temp_export_dir = f"/tmp/config_export_{timestamp}"
|
| 82 |
+
os.makedirs(temp_export_dir, exist_ok=True)
|
| 83 |
+
|
| 84 |
+
export_db_to_json(DB_PATH, temp_export_dir)
|
| 85 |
+
|
| 86 |
+
# 打包成 tar.gz
|
| 87 |
+
tar_path = f"/tmp/{config_tar_name}"
|
| 88 |
+
with tarfile.open(tar_path, "w:gz") as tar:
|
| 89 |
+
for file in os.listdir(temp_export_dir):
|
| 90 |
+
full_path = os.path.join(temp_export_dir, file)
|
| 91 |
+
tar.add(full_path, arcname=file)
|
| 92 |
+
|
| 93 |
+
# 上传压缩包
|
| 94 |
+
api.upload_file(
|
| 95 |
+
path_or_fileobj=tar_path,
|
| 96 |
+
path_in_repo=config_tar_name,
|
| 97 |
+
repo_id=DATASET_REPO,
|
| 98 |
+
repo_type="dataset"
|
| 99 |
+
)
|
| 100 |
+
print(f"✅ 配置导出包上传成功: {config_tar_name}")
|
| 101 |
+
|
| 102 |
+
# 清理临时文件
|
| 103 |
+
import shutil
|
| 104 |
+
shutil.rmtree(temp_export_dir)
|
| 105 |
+
os.remove(tar_path)
|
| 106 |
+
except Exception as e:
|
| 107 |
+
print(f"❌ 配置导出失败: {e}")
|
| 108 |
|
| 109 |
+
# 3. 清理旧备份(二进制和配置包)
|
| 110 |
try:
|
| 111 |
files = api.list_repo_files(repo_id=DATASET_REPO, repo_type="dataset")
|
| 112 |
now = datetime.datetime.now()
|
| 113 |
deleted = 0
|
| 114 |
for f in files:
|
| 115 |
+
# 删除超过 RETENTION_DAYS 天的二进制备份和配置包
|
| 116 |
+
if f.startswith(BACKUP_PREFIX) or f.startswith(CONFIG_PREFIX):
|
| 117 |
+
# 提取时间戳
|
| 118 |
+
parts = f.replace(".db", "").replace(".tar.gz", "").split("_")
|
| 119 |
+
if len(parts) >= 3:
|
| 120 |
+
ts_str = f"{parts[2]}_{parts[3]}" if len(parts) > 3 else parts[2]
|
| 121 |
+
# 文件名格式: freeapi_backup_20260602_061721.db 或 config_export_20260602_061721.tar.gz
|
| 122 |
+
# 简化:取最后两个部分(日期_时间)
|
| 123 |
+
try:
|
| 124 |
+
# 尝试解析时间戳
|
| 125 |
+
if f.startswith(BACKUP_PREFIX):
|
| 126 |
+
ts_part = f.replace(BACKUP_PREFIX + "_", "").replace(".db", "")
|
| 127 |
+
else:
|
| 128 |
+
ts_part = f.replace(CONFIG_PREFIX + "_", "").replace(".tar.gz", "")
|
| 129 |
+
file_time = datetime.datetime.strptime(ts_part, "%Y%m%d_%H%M%S")
|
| 130 |
+
if (now - file_time).days > RETENTION_DAYS:
|
| 131 |
+
api.delete_file(path_in_repo=f, repo_id=DATASET_REPO, repo_type="dataset")
|
| 132 |
+
print(f"🗑️ 已删除旧备份: {f}")
|
| 133 |
+
deleted += 1
|
| 134 |
+
except Exception as parse_err:
|
| 135 |
+
print(f"⚠️ 跳过无法解析的文件: {f} ({parse_err})")
|
| 136 |
if deleted:
|
| 137 |
print(f"✅ 清理完成,共删除 {deleted} 个旧备份")
|
| 138 |
else:
|
|
|
|
| 143 |
return True
|
| 144 |
|
| 145 |
if __name__ == "__main__":
|
| 146 |
+
print(f"[{datetime.datetime.now()}] 开始备份及配置导出...")
|
| 147 |
if backup_database():
|
| 148 |
print("备份任务完成")
|
| 149 |
sys.exit(0)
|