Spaces:
Paused
Paused
more fixes
Browse files- backend/scripts/restore.py +61 -42
backend/scripts/restore.py
CHANGED
|
@@ -8,21 +8,28 @@ from pathlib import Path
|
|
| 8 |
|
| 9 |
from huggingface_hub import HfApi, hf_hub_download
|
| 10 |
|
| 11 |
-
#
|
| 12 |
-
# 1
|
| 13 |
-
#
|
| 14 |
-
|
| 15 |
-
|
|
|
|
|
|
|
|
|
|
| 16 |
|
| 17 |
-
#
|
| 18 |
-
|
|
|
|
|
|
|
|
|
|
| 19 |
DB_GPG_PATH = os.path.join(RESTORE_BACKUP_DIR, "webui.db.gpg")
|
| 20 |
-
|
| 21 |
|
| 22 |
-
# Paths
|
| 23 |
REPO_TIMESTAMP_FILE = "db_backup/last_backup_time.txt"
|
| 24 |
REPO_DB_GPG_FILE = "db_backup/webui.db.gpg"
|
| 25 |
|
|
|
|
| 26 |
def check_requirements():
|
| 27 |
"""
|
| 28 |
Verify that GPG is installed and available in the system.
|
|
@@ -31,9 +38,10 @@ def check_requirements():
|
|
| 31 |
subprocess.run(["gpg", "--version"], check=True, capture_output=True)
|
| 32 |
return True
|
| 33 |
except (subprocess.CalledProcessError, FileNotFoundError):
|
| 34 |
-
print("Error: gpg is not installed")
|
| 35 |
return False
|
| 36 |
|
|
|
|
| 37 |
def validate_secrets():
|
| 38 |
"""
|
| 39 |
Ensure all required environment variables are set: BACKUP_PASSPHRASE, HF_TOKEN, SPACE_ID.
|
|
@@ -46,21 +54,27 @@ def validate_secrets():
|
|
| 46 |
return False
|
| 47 |
return True
|
| 48 |
|
|
|
|
| 49 |
def ensure_directories():
|
| 50 |
"""
|
| 51 |
-
Create necessary directories for database and backup files (
|
|
|
|
| 52 |
"""
|
| 53 |
try:
|
| 54 |
-
|
| 55 |
-
|
|
|
|
|
|
|
|
|
|
| 56 |
return True
|
| 57 |
except Exception as e:
|
| 58 |
print(f"Error creating directories: {e}")
|
| 59 |
return False
|
| 60 |
|
|
|
|
| 61 |
def get_latest_backup_info(repo_id, hf_token):
|
| 62 |
"""
|
| 63 |
-
Check if a backup exists in the HF Space. If so, return True
|
| 64 |
"""
|
| 65 |
api = HfApi()
|
| 66 |
try:
|
|
@@ -69,12 +83,12 @@ def get_latest_backup_info(repo_id, hf_token):
|
|
| 69 |
# Check if the encrypted DB is present
|
| 70 |
backup_exists = (REPO_DB_GPG_FILE in files)
|
| 71 |
if not backup_exists:
|
| 72 |
-
print("No backup file found in the repository")
|
| 73 |
return False, None
|
| 74 |
|
| 75 |
# Attempt to fetch the timestamp file
|
| 76 |
-
|
| 77 |
-
|
| 78 |
timestamp_file = hf_hub_download(
|
| 79 |
repo_id=repo_id,
|
| 80 |
repo_type="space",
|
|
@@ -82,22 +96,24 @@ def get_latest_backup_info(repo_id, hf_token):
|
|
| 82 |
token=hf_token
|
| 83 |
)
|
| 84 |
with open(timestamp_file, "r", encoding="utf-8") as f:
|
| 85 |
-
|
|
|
|
| 86 |
print(f"Found backup from: {timestamp} UTC")
|
| 87 |
return True, timestamp
|
| 88 |
-
|
| 89 |
-
print("
|
| 90 |
return True, None
|
| 91 |
-
|
| 92 |
-
print(
|
| 93 |
return True, None
|
| 94 |
except Exception as e:
|
| 95 |
print(f"Error checking repository: {e}")
|
| 96 |
return False, None
|
| 97 |
|
|
|
|
| 98 |
def download_backup(repo_id, hf_token):
|
| 99 |
"""
|
| 100 |
-
Download the encrypted database backup from Hugging Face
|
| 101 |
"""
|
| 102 |
try:
|
| 103 |
print("Downloading encrypted database backup...")
|
|
@@ -107,19 +123,19 @@ def download_backup(repo_id, hf_token):
|
|
| 107 |
filename=REPO_DB_GPG_FILE,
|
| 108 |
token=hf_token
|
| 109 |
)
|
| 110 |
-
|
| 111 |
# Move the downloaded file to DB_GPG_PATH
|
| 112 |
-
os.makedirs(os.path.dirname(DB_GPG_PATH), exist_ok=True)
|
| 113 |
os.replace(temp_file, DB_GPG_PATH)
|
| 114 |
-
print("Backup downloaded successfully")
|
| 115 |
return True
|
| 116 |
except Exception as e:
|
| 117 |
print(f"Error downloading backup: {e}")
|
| 118 |
return False
|
| 119 |
|
|
|
|
| 120 |
def decrypt_database(passphrase):
|
| 121 |
"""
|
| 122 |
-
Decrypt the database
|
|
|
|
| 123 |
"""
|
| 124 |
if not os.path.exists(DB_GPG_PATH):
|
| 125 |
print("No encrypted backup found locally. Proceeding with fresh DB.")
|
|
@@ -150,12 +166,13 @@ def decrypt_database(passphrase):
|
|
| 150 |
print(f"GPG error output: {e.stderr.decode(errors='ignore')}")
|
| 151 |
return False
|
| 152 |
|
|
|
|
| 153 |
def verify_database():
|
| 154 |
"""
|
| 155 |
-
Verify the integrity of the restored
|
| 156 |
"""
|
| 157 |
if not os.path.exists(DB_FILE_PATH):
|
| 158 |
-
# If there's no DB yet,
|
| 159 |
return True
|
| 160 |
|
| 161 |
try:
|
|
@@ -168,15 +185,15 @@ def verify_database():
|
|
| 168 |
tables = cursor.fetchall()
|
| 169 |
|
| 170 |
if result.lower() == "ok" and len(tables) > 0:
|
| 171 |
-
print("Database integrity verified successfully")
|
| 172 |
-
print(f"Found {len(tables)} tables in database")
|
| 173 |
return True
|
| 174 |
else:
|
| 175 |
-
print("Database integrity check failed")
|
| 176 |
if result.lower() != "ok":
|
| 177 |
print(f"Integrity check result: {result}")
|
| 178 |
if len(tables) == 0:
|
| 179 |
-
print("No tables found in database")
|
| 180 |
return False
|
| 181 |
except sqlite3.Error as e:
|
| 182 |
print(f"Database verification failed: {e}")
|
|
@@ -185,13 +202,14 @@ def verify_database():
|
|
| 185 |
print(f"Unexpected error during database verification: {e}")
|
| 186 |
return False
|
| 187 |
|
|
|
|
| 188 |
def restore_db():
|
| 189 |
"""
|
| 190 |
Main restore function:
|
| 191 |
-
|
| 192 |
-
|
| 193 |
-
|
| 194 |
-
|
| 195 |
"""
|
| 196 |
if not check_requirements() or not validate_secrets():
|
| 197 |
return False
|
|
@@ -206,25 +224,26 @@ def restore_db():
|
|
| 206 |
backup_exists, timestamp = get_latest_backup_info(space_id, hf_token)
|
| 207 |
if backup_exists:
|
| 208 |
if not download_backup(space_id, hf_token):
|
| 209 |
-
print("Failed to download backup")
|
| 210 |
return False
|
| 211 |
|
| 212 |
if not decrypt_database(passphrase):
|
| 213 |
-
print("Failed to decrypt database")
|
| 214 |
return False
|
| 215 |
|
| 216 |
if not verify_database():
|
| 217 |
-
print("Database integrity verification failed")
|
| 218 |
-
# Remove the corrupted DB so we don
|
| 219 |
if os.path.exists(DB_FILE_PATH):
|
| 220 |
os.unlink(DB_FILE_PATH)
|
| 221 |
return False
|
| 222 |
else:
|
| 223 |
-
print("No backup found - starting with an empty/fresh database")
|
| 224 |
|
| 225 |
print("Database restore completed successfully!")
|
| 226 |
return True
|
| 227 |
|
|
|
|
| 228 |
if __name__ == "__main__":
|
| 229 |
success = restore_db()
|
| 230 |
sys.exit(0 if success else 1)
|
|
|
|
| 8 |
|
| 9 |
from huggingface_hub import HfApi, hf_hub_download
|
| 10 |
|
| 11 |
+
###############################################################################
|
| 12 |
+
# 1) Determine the *final* database file location.
|
| 13 |
+
# - We read DATA_DIR from env, defaulting to "/app/backend/data".
|
| 14 |
+
# - The final DB (decrypted) should be "/app/backend/data/webui.db" so
|
| 15 |
+
# the app will use the restored DB on startup.
|
| 16 |
+
###############################################################################
|
| 17 |
+
DATA_DIR = os.environ.get("DATA_DIR", "/app/backend/data")
|
| 18 |
+
DB_FILE_PATH = os.path.join(DATA_DIR, "webui.db")
|
| 19 |
|
| 20 |
+
###############################################################################
|
| 21 |
+
# 2) Where we put the encrypted backup and timestamp locally.
|
| 22 |
+
# - A writable directory on Hugging Face is "/tmp/open_webui/db_backup".
|
| 23 |
+
###############################################################################
|
| 24 |
+
RESTORE_BACKUP_DIR = "/tmp/open_webui/db_backup"
|
| 25 |
DB_GPG_PATH = os.path.join(RESTORE_BACKUP_DIR, "webui.db.gpg")
|
| 26 |
+
TIMESTAMP_FILE_PATH = os.path.join(RESTORE_BACKUP_DIR, "last_backup_time.txt")
|
| 27 |
|
| 28 |
+
# Paths in the Hugging Face repo
|
| 29 |
REPO_TIMESTAMP_FILE = "db_backup/last_backup_time.txt"
|
| 30 |
REPO_DB_GPG_FILE = "db_backup/webui.db.gpg"
|
| 31 |
|
| 32 |
+
|
| 33 |
def check_requirements():
|
| 34 |
"""
|
| 35 |
Verify that GPG is installed and available in the system.
|
|
|
|
| 38 |
subprocess.run(["gpg", "--version"], check=True, capture_output=True)
|
| 39 |
return True
|
| 40 |
except (subprocess.CalledProcessError, FileNotFoundError):
|
| 41 |
+
print("Error: gpg is not installed or not in PATH.")
|
| 42 |
return False
|
| 43 |
|
| 44 |
+
|
| 45 |
def validate_secrets():
|
| 46 |
"""
|
| 47 |
Ensure all required environment variables are set: BACKUP_PASSPHRASE, HF_TOKEN, SPACE_ID.
|
|
|
|
| 54 |
return False
|
| 55 |
return True
|
| 56 |
|
| 57 |
+
|
| 58 |
def ensure_directories():
|
| 59 |
"""
|
| 60 |
+
Create necessary directories for database (DATA_DIR) and backup files (RESTORE_BACKUP_DIR).
|
| 61 |
+
We must ensure the final DB directory is writable so the decrypted DB can be placed there.
|
| 62 |
"""
|
| 63 |
try:
|
| 64 |
+
# Create the final DB folder (if not already existing)
|
| 65 |
+
os.makedirs(DATA_DIR, mode=0o755, exist_ok=True)
|
| 66 |
+
|
| 67 |
+
# Create the backup folder in /tmp
|
| 68 |
+
os.makedirs(RESTORE_BACKUP_DIR, mode=0o755, exist_ok=True)
|
| 69 |
return True
|
| 70 |
except Exception as e:
|
| 71 |
print(f"Error creating directories: {e}")
|
| 72 |
return False
|
| 73 |
|
| 74 |
+
|
| 75 |
def get_latest_backup_info(repo_id, hf_token):
|
| 76 |
"""
|
| 77 |
+
Check if a backup (webui.db.gpg) exists in the HF Space. If so, return True + timestamp (or None).
|
| 78 |
"""
|
| 79 |
api = HfApi()
|
| 80 |
try:
|
|
|
|
| 83 |
# Check if the encrypted DB is present
|
| 84 |
backup_exists = (REPO_DB_GPG_FILE in files)
|
| 85 |
if not backup_exists:
|
| 86 |
+
print("No backup file found in the repository.")
|
| 87 |
return False, None
|
| 88 |
|
| 89 |
# Attempt to fetch the timestamp file
|
| 90 |
+
if REPO_TIMESTAMP_FILE in files:
|
| 91 |
+
try:
|
| 92 |
timestamp_file = hf_hub_download(
|
| 93 |
repo_id=repo_id,
|
| 94 |
repo_type="space",
|
|
|
|
| 96 |
token=hf_token
|
| 97 |
)
|
| 98 |
with open(timestamp_file, "r", encoding="utf-8") as f:
|
| 99 |
+
timestamp_str = f.read().strip()
|
| 100 |
+
timestamp = datetime.datetime.fromisoformat(timestamp_str)
|
| 101 |
print(f"Found backup from: {timestamp} UTC")
|
| 102 |
return True, timestamp
|
| 103 |
+
except Exception as e:
|
| 104 |
+
print(f"Could not read timestamp (possibly first run): {e}")
|
| 105 |
return True, None
|
| 106 |
+
else:
|
| 107 |
+
print("No timestamp file found, but backup file exists.")
|
| 108 |
return True, None
|
| 109 |
except Exception as e:
|
| 110 |
print(f"Error checking repository: {e}")
|
| 111 |
return False, None
|
| 112 |
|
| 113 |
+
|
| 114 |
def download_backup(repo_id, hf_token):
|
| 115 |
"""
|
| 116 |
+
Download the encrypted database backup from Hugging Face to /tmp/open_webui/db_backup.
|
| 117 |
"""
|
| 118 |
try:
|
| 119 |
print("Downloading encrypted database backup...")
|
|
|
|
| 123 |
filename=REPO_DB_GPG_FILE,
|
| 124 |
token=hf_token
|
| 125 |
)
|
|
|
|
| 126 |
# Move the downloaded file to DB_GPG_PATH
|
|
|
|
| 127 |
os.replace(temp_file, DB_GPG_PATH)
|
| 128 |
+
print("Backup downloaded successfully.")
|
| 129 |
return True
|
| 130 |
except Exception as e:
|
| 131 |
print(f"Error downloading backup: {e}")
|
| 132 |
return False
|
| 133 |
|
| 134 |
+
|
| 135 |
def decrypt_database(passphrase):
|
| 136 |
"""
|
| 137 |
+
Decrypt the database into DATA_DIR/webui.db so the app can use it.
|
| 138 |
+
The encrypted file is at /tmp/open_webui/db_backup/webui.db.gpg.
|
| 139 |
"""
|
| 140 |
if not os.path.exists(DB_GPG_PATH):
|
| 141 |
print("No encrypted backup found locally. Proceeding with fresh DB.")
|
|
|
|
| 166 |
print(f"GPG error output: {e.stderr.decode(errors='ignore')}")
|
| 167 |
return False
|
| 168 |
|
| 169 |
+
|
| 170 |
def verify_database():
|
| 171 |
"""
|
| 172 |
+
Verify the integrity of the restored DB at DATA_DIR/webui.db using SQLite's built-in checks.
|
| 173 |
"""
|
| 174 |
if not os.path.exists(DB_FILE_PATH):
|
| 175 |
+
# If there's no DB yet, might be fresh start, not an immediate error
|
| 176 |
return True
|
| 177 |
|
| 178 |
try:
|
|
|
|
| 185 |
tables = cursor.fetchall()
|
| 186 |
|
| 187 |
if result.lower() == "ok" and len(tables) > 0:
|
| 188 |
+
print("Database integrity verified successfully.")
|
| 189 |
+
print(f"Found {len(tables)} tables in database.")
|
| 190 |
return True
|
| 191 |
else:
|
| 192 |
+
print("Database integrity check failed.")
|
| 193 |
if result.lower() != "ok":
|
| 194 |
print(f"Integrity check result: {result}")
|
| 195 |
if len(tables) == 0:
|
| 196 |
+
print("No tables found in database.")
|
| 197 |
return False
|
| 198 |
except sqlite3.Error as e:
|
| 199 |
print(f"Database verification failed: {e}")
|
|
|
|
| 202 |
print(f"Unexpected error during database verification: {e}")
|
| 203 |
return False
|
| 204 |
|
| 205 |
+
|
| 206 |
def restore_db():
|
| 207 |
"""
|
| 208 |
Main restore function:
|
| 209 |
+
1. Check GPG + environment
|
| 210 |
+
2. Create directories
|
| 211 |
+
3. Check if remote backup exists, download + decrypt
|
| 212 |
+
4. Verify DB
|
| 213 |
"""
|
| 214 |
if not check_requirements() or not validate_secrets():
|
| 215 |
return False
|
|
|
|
| 224 |
backup_exists, timestamp = get_latest_backup_info(space_id, hf_token)
|
| 225 |
if backup_exists:
|
| 226 |
if not download_backup(space_id, hf_token):
|
| 227 |
+
print("Failed to download backup.")
|
| 228 |
return False
|
| 229 |
|
| 230 |
if not decrypt_database(passphrase):
|
| 231 |
+
print("Failed to decrypt database.")
|
| 232 |
return False
|
| 233 |
|
| 234 |
if not verify_database():
|
| 235 |
+
print("Database integrity verification failed.")
|
| 236 |
+
# Remove the corrupted DB so we don’t keep a broken file
|
| 237 |
if os.path.exists(DB_FILE_PATH):
|
| 238 |
os.unlink(DB_FILE_PATH)
|
| 239 |
return False
|
| 240 |
else:
|
| 241 |
+
print("No backup found - starting with an empty/fresh database.")
|
| 242 |
|
| 243 |
print("Database restore completed successfully!")
|
| 244 |
return True
|
| 245 |
|
| 246 |
+
|
| 247 |
if __name__ == "__main__":
|
| 248 |
success = restore_db()
|
| 249 |
sys.exit(0 if success else 1)
|