Spaces:
Sleeping
Sleeping
Update scripts/restore_from_dataset.py
Browse files- scripts/restore_from_dataset.py +100 -29
scripts/restore_from_dataset.py
CHANGED
|
@@ -2,60 +2,101 @@
|
|
| 2 |
import os
|
| 3 |
import sys
|
| 4 |
import shutil
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
import datetime
|
| 6 |
from huggingface_hub import HfApi, login, hf_hub_download
|
| 7 |
|
| 8 |
DB_TARGET = "/app/server/data/freeapi.db"
|
| 9 |
DATASET_REPO = "lydgs/freellm-backup"
|
|
|
|
| 10 |
BACKUP_PREFIX = "freeapi_backup"
|
| 11 |
|
| 12 |
-
def
|
| 13 |
-
|
| 14 |
-
|
| 15 |
return False
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
token = os.getenv("HF_TOKEN")
|
| 17 |
if not token:
|
| 18 |
-
print("❌ 环境变量 HF_TOKEN 未设置")
|
| 19 |
return False
|
| 20 |
-
timestamp = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
|
| 21 |
-
backup_name = f"{backup_prefix}_{timestamp}.db"
|
| 22 |
login(token=token)
|
| 23 |
api = HfApi()
|
| 24 |
try:
|
| 25 |
-
api.
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
print(f"
|
| 32 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
except Exception as e:
|
| 34 |
-
print(f"
|
| 35 |
return False
|
| 36 |
|
| 37 |
-
def
|
|
|
|
| 38 |
token = os.getenv("HF_TOKEN")
|
| 39 |
if not token:
|
| 40 |
-
|
| 41 |
-
return
|
| 42 |
login(token=token)
|
| 43 |
api = HfApi()
|
| 44 |
try:
|
| 45 |
files = api.list_repo_files(repo_id=DATASET_REPO, repo_type="dataset")
|
| 46 |
db_files = [f for f in files if f.endswith(".db") and f.startswith(BACKUP_PREFIX)]
|
| 47 |
if not db_files:
|
| 48 |
-
|
| 49 |
-
# 如果没有备份,但本地数据库存在,则自动创建一份
|
| 50 |
-
if os.path.exists(DB_TARGET):
|
| 51 |
-
print("🔄 检测到本地数据库存在,将创建初始备份...")
|
| 52 |
-
upload_backup(DB_TARGET, DATASET_REPO, BACKUP_PREFIX)
|
| 53 |
-
else:
|
| 54 |
-
print("ℹ️ 本地数据库也不存在,跳过恢复")
|
| 55 |
-
return
|
| 56 |
db_files.sort()
|
| 57 |
latest = db_files[-1]
|
| 58 |
-
print(f"🔄
|
| 59 |
downloaded = hf_hub_download(
|
| 60 |
repo_id=DATASET_REPO,
|
| 61 |
filename=latest,
|
|
@@ -64,10 +105,40 @@ def restore_latest_backup():
|
|
| 64 |
)
|
| 65 |
os.makedirs(os.path.dirname(DB_TARGET), exist_ok=True)
|
| 66 |
shutil.copy(downloaded, DB_TARGET)
|
| 67 |
-
print(f"✅ 数据库恢复成功: {DB_TARGET}")
|
| 68 |
os.remove(downloaded)
|
|
|
|
| 69 |
except Exception as e:
|
| 70 |
-
print(f"❌
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 71 |
|
| 72 |
if __name__ == "__main__":
|
| 73 |
restore_latest_backup()
|
|
|
|
| 2 |
import os
|
| 3 |
import sys
|
| 4 |
import shutil
|
| 5 |
+
import json
|
| 6 |
+
import sqlite3
|
| 7 |
+
import tarfile
|
| 8 |
+
import tempfile
|
| 9 |
+
import subprocess
|
| 10 |
import datetime
|
| 11 |
from huggingface_hub import HfApi, login, hf_hub_download
|
| 12 |
|
| 13 |
DB_TARGET = "/app/server/data/freeapi.db"
|
| 14 |
DATASET_REPO = "lydgs/freellm-backup"
|
| 15 |
+
CONFIG_PREFIX = "config_export"
|
| 16 |
BACKUP_PREFIX = "freeapi_backup"
|
| 17 |
|
| 18 |
+
def import_json_to_db(json_dir, db_path):
|
| 19 |
+
"""从 JSON 文件目录重建数据库"""
|
| 20 |
+
if not os.path.exists(json_dir):
|
| 21 |
return False
|
| 22 |
+
os.makedirs(os.path.dirname(db_path), exist_ok=True)
|
| 23 |
+
if os.path.exists(db_path):
|
| 24 |
+
os.remove(db_path)
|
| 25 |
+
conn = sqlite3.connect(db_path)
|
| 26 |
+
cursor = conn.cursor()
|
| 27 |
+
for json_file in os.listdir(json_dir):
|
| 28 |
+
if not json_file.endswith('.json'):
|
| 29 |
+
continue
|
| 30 |
+
table_name = json_file[:-5]
|
| 31 |
+
with open(os.path.join(json_dir, json_file), 'r', encoding='utf-8') as f:
|
| 32 |
+
data = json.load(f)
|
| 33 |
+
if not data:
|
| 34 |
+
continue
|
| 35 |
+
# 获取列名
|
| 36 |
+
columns = list(data[0].keys())
|
| 37 |
+
# 创建表
|
| 38 |
+
cursor.execute(f"DROP TABLE IF EXISTS {table_name}")
|
| 39 |
+
cursor.execute(f"CREATE TABLE {table_name} ({', '.join(columns)})")
|
| 40 |
+
# 插入数据
|
| 41 |
+
placeholders = ', '.join('?' for _ in columns)
|
| 42 |
+
for row in data:
|
| 43 |
+
values = [row[col] for col in columns]
|
| 44 |
+
cursor.execute(f"INSERT INTO {table_name} VALUES ({placeholders})", values)
|
| 45 |
+
conn.commit()
|
| 46 |
+
conn.close()
|
| 47 |
+
return True
|
| 48 |
+
|
| 49 |
+
def restore_from_json():
|
| 50 |
+
"""优先尝试从 JSON 包恢复,成功返回 True"""
|
| 51 |
token = os.getenv("HF_TOKEN")
|
| 52 |
if not token:
|
|
|
|
| 53 |
return False
|
|
|
|
|
|
|
| 54 |
login(token=token)
|
| 55 |
api = HfApi()
|
| 56 |
try:
|
| 57 |
+
files = api.list_repo_files(repo_id=DATASET_REPO, repo_type="dataset")
|
| 58 |
+
config_files = [f for f in files if f.startswith(CONFIG_PREFIX) and f.endswith('.tar.gz')]
|
| 59 |
+
if not config_files:
|
| 60 |
+
return False
|
| 61 |
+
config_files.sort(reverse=True)
|
| 62 |
+
latest_config = config_files[0]
|
| 63 |
+
print(f"🔄 发现 JSON 配置包: {latest_config},重建数据库...")
|
| 64 |
+
with tempfile.TemporaryDirectory() as tmpdir:
|
| 65 |
+
downloaded = hf_hub_download(
|
| 66 |
+
repo_id=DATASET_REPO,
|
| 67 |
+
filename=latest_config,
|
| 68 |
+
repo_type="dataset",
|
| 69 |
+
local_dir=tmpdir
|
| 70 |
+
)
|
| 71 |
+
extract_dir = os.path.join(tmpdir, "extract")
|
| 72 |
+
os.makedirs(extract_dir, exist_ok=True)
|
| 73 |
+
with tarfile.open(downloaded, "r:gz") as tar:
|
| 74 |
+
tar.extractall(extract_dir)
|
| 75 |
+
if import_json_to_db(extract_dir, DB_TARGET):
|
| 76 |
+
print(f"✅ 数据库从 JSON 配置重建成功: {DB_TARGET}")
|
| 77 |
+
return True
|
| 78 |
+
else:
|
| 79 |
+
print("⚠️ JSON 导入失败")
|
| 80 |
+
return False
|
| 81 |
except Exception as e:
|
| 82 |
+
print(f"⚠️ JSON 恢复出错: {e}")
|
| 83 |
return False
|
| 84 |
|
| 85 |
+
def restore_from_binary():
|
| 86 |
+
"""回退:从二进制 .db 文件恢复"""
|
| 87 |
token = os.getenv("HF_TOKEN")
|
| 88 |
if not token:
|
| 89 |
+
return False
|
|
|
|
| 90 |
login(token=token)
|
| 91 |
api = HfApi()
|
| 92 |
try:
|
| 93 |
files = api.list_repo_files(repo_id=DATASET_REPO, repo_type="dataset")
|
| 94 |
db_files = [f for f in files if f.endswith(".db") and f.startswith(BACKUP_PREFIX)]
|
| 95 |
if not db_files:
|
| 96 |
+
return False
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 97 |
db_files.sort()
|
| 98 |
latest = db_files[-1]
|
| 99 |
+
print(f"🔄 使用二进制备份恢复: {latest}")
|
| 100 |
downloaded = hf_hub_download(
|
| 101 |
repo_id=DATASET_REPO,
|
| 102 |
filename=latest,
|
|
|
|
| 105 |
)
|
| 106 |
os.makedirs(os.path.dirname(DB_TARGET), exist_ok=True)
|
| 107 |
shutil.copy(downloaded, DB_TARGET)
|
| 108 |
+
print(f"✅ 数据库从二进制备份恢复成功: {DB_TARGET}")
|
| 109 |
os.remove(downloaded)
|
| 110 |
+
return True
|
| 111 |
except Exception as e:
|
| 112 |
+
print(f"❌ 二进制恢复失败: {e}")
|
| 113 |
+
return False
|
| 114 |
+
|
| 115 |
+
def create_full_backup():
|
| 116 |
+
"""调用备份脚本,生成完整的 .db 和 JSON 包"""
|
| 117 |
+
backup_script = "/app/scripts/backup_to_dataset.py"
|
| 118 |
+
if not os.path.exists(backup_script):
|
| 119 |
+
print("⚠️ 备份脚本不存在,无法创建初始备份")
|
| 120 |
+
return False
|
| 121 |
+
try:
|
| 122 |
+
subprocess.run(["python3", backup_script], check=True)
|
| 123 |
+
print("✅ 已执行完整备份(包括 JSON 包)")
|
| 124 |
+
return True
|
| 125 |
+
except subprocess.CalledProcessError as e:
|
| 126 |
+
print(f"❌ 执行备份脚本失败: {e}")
|
| 127 |
+
return False
|
| 128 |
+
|
| 129 |
+
def restore_latest_backup():
|
| 130 |
+
# 1. 优先尝试 JSON 恢复
|
| 131 |
+
if restore_from_json():
|
| 132 |
+
return
|
| 133 |
+
# 2. 回退到二进制恢复
|
| 134 |
+
if restore_from_binary():
|
| 135 |
+
return
|
| 136 |
+
# 3. 完全无备份,但本地数据库存在 -> 创建完整备份
|
| 137 |
+
if os.path.exists(DB_TARGET):
|
| 138 |
+
print("🔄 无任何备份,检测到本地数据库,将创建完整备份(包括 JSON 包)...")
|
| 139 |
+
create_full_backup()
|
| 140 |
+
else:
|
| 141 |
+
print("ℹ️ 本地数据库也不存在,跳过恢复")
|
| 142 |
|
| 143 |
if __name__ == "__main__":
|
| 144 |
restore_latest_backup()
|