erebion/libervia-backend
Watch
1
0
Fork
You've already forked libervia-backend
 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
libervia-backend/libervia/cli/cmd_backup.py
Goffi f22af42ece cli (backup): add backup commands: 
Add the `li backup` commands with `export`, `import` and `verify`:

- data types selection (`-i`/`-e`) and profile selection (`-p`), `--no-db` on export, `--use-db` on
  import
- GPG symmetric encryption of the archive (`--encrypt`)
- progression is shown, with ETA
- the commands can run without a backend connection (import from DB MUST be run without
  backend)
- cli (base): `BRIDGELESS_COMMANDS` holds the name of the few commands which can be run
  without backend.

rel 480
2026-09-15 11:48:11 +02:00

740 lines
28 KiB
Python
Raw Blame History

#!/usr/bin/env python3
# Libervia CLI
# Copyright (C) 2009-2026 Jérôme Poisson (goffi@goffi.org)
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU Affero General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU Affero General Public License for more details.
# You should have received a copy of the GNU Affero General Public License
# along with this program. If not, see <http://www.gnu.org/licenses/>.
from collections.abc import Callable, Iterable, Iterator
from contextlib import contextmanager
import os
from pathlib import Path
import time
from rich.progress import Progress
from libervia.backend.core.constants import Const as CBackend
from libervia.backend.core.i18n import _
from libervia.backend.tools import backup
from libervia.backend.tools.common import utils
from libervia.backend.tools.common.gpg_provider import get_provider
from libervia.backend.tools.common.gpg_provider.exceptions import (
DecryptionFailed,
GPGProviderError,
)
from . import base
from .constants import Const as C
__commands__ = ["Backup"]
PARAM_FILE_NAME = "param"
# errors the GPG provider can raise (they have no common base class)
GPG_ERRORS = (GPGProviderError, DecryptionFailed)
# progress bar labels of the global (not per profile) export handlers
GLOBAL_EXPORT_TASKS = {
"db": "snapshotting database",
"files": "copying files",
"gpg": "copying GPG homes",
}
# progress bar labels of the global (not per profile) import handlers
GLOBAL_IMPORT_TASKS = {
"db": "restoring database",
"files": "copying files",
"gpg": "merging GPG homes",
}
# fixed width of the entry name in the verify progress bar, so the bar does
# not shift horizontally as names of different lengths are displayed
VERIFY_NAME_WIDTH = 30
# max number of profiles listed by verify in non-verbose mode, beyond that a
# simple count is shown
MAX_PROFILES_SHOWN = 3
def _get_types(parser, args) -> list[str]:
"""Compute the list of types to handle.
Types are generated from --include/--exclude:
- no option = all types
- --include is a whitelist (if given, only these types are handled)
- --no-db is a named exclusion shortcut (export only: the database
is not imported by default on import).
"""
included = set(args.include)
excluded = set(args.exclude)
if getattr(args, "no_db", False):
excluded.add("db")
unknown = (included | excluded) - set(backup.DATA_TYPES)
if unknown:
parser.error(f"unknown type: {', '.join(sorted(unknown))}")
both = included & excluded
if both:
parser.error("type both included and excluded: " + ", ".join(sorted(both)))
types = [t for t in backup.DATA_TYPES if t not in excluded]
if included:
types = [t for t in types if t in included]
return types
def _print_types(command, types: list[str]) -> None:
"""Print the selected types, so the user can check before it starts"""
command.print(_("types included: {types}").format(types=", ".join(types)))
excluded = [t for t in backup.DATA_TYPES if t not in types]
if excluded:
command.print(_("types excluded: {types}").format(types=", ".join(excluded)))
def _print_counted(command, label: str, messages: list[str], error: bool = False) -> None:
"""Print messages, counting repetitions (e.g. ``2x imported: ...``)"""
counts: dict[str, int] = {}
for msg in messages:
counts[msg] = counts.get(msg, 0) + 1
for msg, count in counts.items():
prefix = f"{count}x " if count > 1 else ""
(command.disp if error else command.print)(f"{label}{prefix}{msg}")
def _split_handlers(
types: list[str],
) -> tuple[list[backup.ProfileHandler], list[backup.GlobalHandler]]:
"""Split the handlers to run, by kind (per profile / global)"""
handlers = backup.Backup.get_handlers(types)
return (
[h for h in handlers if isinstance(h, backup.ProfileHandler)],
[h for h in handlers if isinstance(h, backup.GlobalHandler)],
)
def _selected_profiles(
parser, requested: list[str], available: Iterable[str], not_found: str
) -> set[str]:
"""Check the requested profiles exist, and return the names to handle
@param parser: argparse parser (used to report the error)
@param requested: profile names requested with -p
@param available: names of the existing profiles
@param not_found: error message, with a ``{names}`` placeholder
"""
unknown = set(requested) - set(available)
if unknown:
parser.error(not_found.format(names=", ".join(sorted(unknown))))
return set(requested)
def _fixed_width(name: str, width: int) -> str:
"""Truncate a name to a fixed width, left-justified
Used for the progress bar descriptions, so that the line does not shift
while names of different lengths are displayed
"""
if len(name) > width:
name = name[: width - 1] + "…"
return f"{name:<{width}}"
@contextmanager
def _byte_task(progress: Progress, label: str) -> Iterator[Callable[[int, int], None]]:
"""Progress task for byte-based operations (building/extracting an archive)
Yields a ``callable(done, total)`` progress callback
"""
task = progress.add_task(_(label), total=0)
try:
yield lambda done, total: progress.update(task, total=total, completed=done)
finally:
progress.remove_task(task)
def _run_global_handlers(
progress: Progress,
handlers: list[backup.GlobalHandler],
labels: dict[str, str],
verb: str,
run: Callable[[backup.GlobalHandler], None],
) -> None:
"""Run the global handlers, each with its own progress task
@param labels: progress bar label by handler name (a default is derived
from ``verb`` for the other handlers)
@param run: called for each handler
"""
for handler in handlers:
label = labels.get(handler.name, f"{verb} {handler.name} data")
task = progress.add_task(_(label), total=None)
run(handler)
progress.remove_task(task)
def _local_paths(host) -> tuple[Path, Path, Path]:
"""Paths of the local data
@param host: LiberviaCli instance
@return: (local_dir, db_path, param_file)
"""
local_dir = Path(host.settings.common.local_dir)
return local_dir, local_dir / CBackend.SAVEFILE_DATABASE, local_dir / PARAM_FILE_NAME
def _gpg_symmetric(path: Path, passphrase: str) -> Path:
"""Encrypt a file with GPG symmetric encryption, replacing it
@param path: file to encrypt
@param passphrase: passphrase to use
@return: path of the encrypted file
@raise backup.BackupError: if the encryption fails
"""
provider = get_provider()
encrypted = path.with_name(path.name + ".gpg")
try:
# TODO: encryption is currently done in memory, need to move to streaming
# encryption (same for decryption in _gpg_decrypt)
data = provider.encrypt_symmetrically(path.read_bytes(), passphrase)
except GPG_ERRORS as e:
encrypted.unlink(missing_ok=True)
raise backup.BackupError(f"GPG encryption failed: {e}") from e
encrypted.write_bytes(data)
path.unlink()
os.chmod(encrypted, 0o600)
return encrypted
def _gpg_decrypt(in_path: Path, out_path: Path, passphrase: str) -> None:
"""Decrypt a GPG symmetric encrypted file
@param in_path: file to decrypt (name must end with .gpg)
@param out_path: path where the decrypted file is written
@param passphrase: passphrase to use
@raise backup.BackupError: if the decryption fails
"""
provider = get_provider()
try:
data = provider.decrypt_symmetrically(in_path.read_bytes(), passphrase)
except DecryptionFailed as e:
out_path.unlink(missing_ok=True)
raise backup.BackupError("GPG decryption failed, check the passphrase") from e
except GPGProviderError as e:
out_path.unlink(missing_ok=True)
raise backup.BackupError(f"GPG decryption failed: {e}") from e
out_path.write_bytes(data)
os.chmod(out_path, 0o600)
async def _resolve_archive(host, args, data: backup.Staging) -> Path:
"""Check a backup archive exists and decrypt it to the staging if needed
@param host: LiberviaCli instance
@param args: parsed arguments (``path`` and ``passphrase`` attributes)
@param data: staging used to hold the decrypted archive (owner-only dir, wiped with
it: no plaintext is kept next to the .gpg file)
@return: path of the (decrypted) archive
"""
path = Path(args.path)
if not path.is_file():
host.disp(_("no such file: {path}").format(path=path), error=True)
host.quit(C.EXIT_ERROR)
if path.name.endswith(".gpg"):
passphrase = args.passphrase
if passphrase is None:
passphrase = await host.a_pwd_input(_("please enter the GPG passphrase: "))
out = data.path("archive.tar.xz")
with Progress(console=host.console) as progress:
progress.add_task(_("decrypting archive"), total=None)
_gpg_decrypt(path, out, passphrase)
path = out
return path
class Export(base.CommandBase):
def __init__(self, host):
super(Export, self).__init__(
host,
"export",
use_profile=False,
help=_("export Libervia data to a backup archive"),
)
def add_parser_options(self):
self.parser.add_argument(
"-o",
"--output-dir",
default="backup",
help=_("directory where the archive is written (default: backup)"),
)
self.parser.add_argument(
"-p",
"--profile",
action="append",
default=[],
help=_("profile to export (default: all profiles)"),
)
self.parser.add_argument(
"-i",
"--include",
action="append",
default=[],
help=_(
"type to include (repeatable; if given, only these types are "
"exported, default: all)"
),
)
self.parser.add_argument(
"-e",
"--exclude",
action="append",
default=[],
help=_("type to exclude (repeatable)"),
)
self.parser.add_argument(
"--no-db",
action="store_true",
help=_(
"don't export the non-portable full database backup "
"(it is exported by default, as it is the only part covering "
"all the data; without it, data such as message history is "
"not backed up)"
),
)
self.parser.add_argument(
"--encrypt",
nargs="?",
const=True,
default=None,
metavar="PASSPHRASE",
help=_(
"encrypt the archive with GPG symmetric encryption "
"(passphrase asked if not given)"
),
)
async def start(self):
types = _get_types(self.parser, self.args)
local_dir, db_path, param_file = _local_paths(self.host)
if not db_path.is_file():
self.disp(_("no database found at {path}").format(path=db_path), error=True)
self.host.quit(C.EXIT_ERROR)
_print_types(self, types)
passphrase = self.args.encrypt
if passphrase is True:
passphrase = await self.host.a_pwd_input(_("please enter a GPG passphrase: "))
ctx = None
staging = backup.Staging()
start_time = time.monotonic()
try:
ctx = await backup.Backup.open_context(
backup.sqlite_db_config(local_dir), param_file, local_dir, readonly=True
)
profiles = await backup.Backup.get_profiles(ctx)
if self.args.profile:
selected = _selected_profiles(
self.parser,
self.args.profile,
(p.name for p in profiles),
"unknown profile: {names}",
)
profiles = [p for p in profiles if p.name in selected]
# fixed width for the profile name in the progress bar, so that the line
# doesn't shift with names of different lengths
name_width = min(max((len(p.name) for p in profiles), default=0), 20)
prof_handlers, glob_handlers = _split_handlers(types)
with Progress(console=self.console) as progress:
task = progress.add_task(_("exporting"), total=len(profiles))
global_report = backup.Report()
profiles_meta: dict[str, backup.ProfileMeta] = {}
reports: list[backup.Report] = []
for profile in profiles:
reports.append(
await backup.Backup.export_profile(
ctx, staging, profile, prof_handlers
)
)
profiles_meta[profile.name] = backup.ProfileMeta(
jid=profile.jid, types=types
)
progress.advance(task)
progress.update(
task,
description=_("exporting: {name}").format(
name=_fixed_width(profile.name, name_width)
),
)
progress.remove_task(task)
profile_names = [p.name for p in profiles]
_run_global_handlers(
progress,
glob_handlers,
GLOBAL_EXPORT_TASKS,
"exporting",
lambda h: h.export(
local_dir, staging, global_report, profiles=profile_names
),
)
reports.append(global_report)
manifest = backup.Backup.build_manifest(
backup.Backup.get_app_version(),
await backup.Backup.get_alembic_rev(ctx),
profiles_meta,
staging,
)
with _byte_task(progress, "building archive") as update:
path = backup.Backup.make_archive(
staging,
manifest,
Path(self.args.output_dir),
progress_cb=update,
)
if passphrase:
task = progress.add_task(_("encrypting"), total=None)
path = _gpg_symmetric(path, passphrase)
progress.remove_task(task)
self.print(
_("backup written to {path} in {elapsed}s").format(
path=path, elapsed=f"{time.monotonic() - start_time:.1f}"
)
)
_print_counted(
self,
" - ",
[msg for report in reports for msg in report.done + report.skipped],
)
finally:
staging.close()
if ctx is not None:
await ctx.close()
self.host.quit(0)
class Import(base.CommandBase):
def __init__(self, host):
super(Import, self).__init__(
host,
"import",
use_profile=False,
help=_("import Libervia data from a backup archive"),
)
def add_parser_options(self):
self.parser.add_argument(
"path", help=_("path to the backup archive (optionally GPG encrypted)")
)
self.parser.add_argument(
"-p",
"--profile",
action="append",
default=[],
help=_("profile to import (default: all profiles found in the archive)"),
)
self.parser.add_argument(
"-i",
"--include",
action="append",
default=[],
help=_(
"type to include (repeatable; if given, only these types are "
"imported, default: all)"
),
)
self.parser.add_argument(
"-e",
"--exclude",
action="append",
default=[],
help=_("type to exclude (repeatable)"),
)
self.parser.add_argument(
"--use-db",
action="store_true",
help=_(
"restore the full database backup (it replaces the current "
"database), importing only the data not stored in it "
"(gpg keys, files)"
),
)
self.parser.add_argument(
"--passphrase",
default=None,
help=_("GPG passphrase, if the archive is encrypted (asked if needed)"),
)
self.parser.add_argument(
"--force",
action="store_true",
help=_(
"do not ask for confirmation of dangerous operations "
"(database restore, diverged GPG keyring)"
),
)
async def start(self):
types = _get_types(self.parser, self.args)
if self.args.use_db:
if "db" in set(self.args.exclude):
self.parser.error("type both included and excluded: db")
# a database restore replaces the database: the data stored in it
# comes from the restore, so import only the data not stored in
# the database on top of it
wanted = {t for t in types if t not in backup.DB_BACKED_TYPES}
wanted.add("db")
types = [t for t in backup.DATA_TYPES if t in wanted]
elif "db" not in self.args.include:
# a database restore is destructive (the database is replaced):
# it only happens when explicitly requested (--use-db or -i db)
types = [t for t in types if t != "db"]
if "db" in types:
# a database restore replaces the database file: it must
# be done while no backend has it open
if self.host.bridge is not None or utils.is_process_running(
Path(self.host.settings.common.pid_dir) / f"{CBackend.APP_NAME_FILE}.pid"
):
self.disp(
_(
"the backend is running: a database restore must "
"be done while it is stopped, please stop it and re-run"
),
error=True,
)
self.host.quit(C.EXIT_ERROR)
_print_types(self, types)
# a database restore is destructive: it is confirmed right away (before the
# long archive extraction, and outside of the progress bar: the prompt and
# the bar don't play well together on the console)
db_confirmed = self.args.force
if "db" in types and not db_confirmed:
await self.host.confirm_or_quit(
_(
"restoring the database backup replaces the current "
"database, are you sure?"
)
)
db_confirmed = True
data = backup.Staging()
report = backup.Report()
manifest: backup.Manifest | None = None
try:
path = await _resolve_archive(self.host, self.args, data)
with Progress(console=self.console) as progress:
with _byte_task(progress, "extracting archive") as update:
manifest = backup.Backup.read_archive(path, data, progress_cb=update)
# read_archive succeeded (the command has quit otherwise)
assert manifest is not None
self.print(
_("backup of {app_version} from {date}").format(
app_version=manifest.app_version, date=manifest.date
)
)
profiles = manifest.profiles
if self.args.profile:
selected = _selected_profiles(
self.parser,
self.args.profile,
profiles,
"profile not in archive: {names}",
)
profiles = {n: d for n, d in profiles.items() if n in selected}
local_dir, db_path, param_file = _local_paths(self.host)
prof_handlers, glob_handlers = _split_handlers(types)
if any(h.uses_db for h in prof_handlers):
# only the handlers that read/write the database need a
# database context (the gpg and files imports don't)
if not db_path.is_file():
self.disp(
_(
"no database found at {path}, please install "
"Libervia first"
).format(path=db_path),
error=True,
)
self.host.quit(C.EXIT_ERROR)
ctx = await backup.Backup.open_context(
backup.sqlite_db_config(local_dir), param_file, local_dir
)
try:
name_width = min(max((len(n) for n in profiles), default=0), 20)
task = progress.add_task(_("importing"), total=len(profiles))
for name in sorted(profiles):
await backup.Backup.import_profile(
ctx,
data,
name,
prof_handlers,
report,
force=self.args.force,
)
progress.advance(task)
progress.update(
task,
description=_("importing: {name}").format(
name=_fixed_width(name, name_width)
),
)
progress.remove_task(task)
finally:
await ctx.close()
profile_names = sorted(profiles)
_run_global_handlers(
progress,
glob_handlers,
GLOBAL_IMPORT_TASKS,
"importing",
# the db restore was already authorized (--force or confirmed
# above), --force is passed to the other handlers only
lambda h: h.import_(
local_dir,
data,
report,
profiles=profile_names,
force=self.args.force or (db_confirmed and h.name == "db"),
),
)
except backup.BackupError as e:
self.disp(str(e), error=True)
self.host.quit(C.EXIT_ERROR)
finally:
data.close()
_print_counted(self, "imported: ", report.done)
_print_counted(self, "skipped: ", report.skipped, error=True)
self.host.quit(0)
class Verify(base.CommandBase):
def __init__(self, host):
super(Verify, self).__init__(
host,
"verify",
use_profile=False,
use_verbose=True,
help=_("check a backup archive integrity and display its content"),
)
def add_parser_options(self):
self.parser.add_argument(
"path", help=_("path to the backup archive (optionally GPG encrypted)")
)
self.parser.add_argument(
"--passphrase",
default=None,
help=_("GPG passphrase, if the archive is encrypted (asked if needed)"),
)
async def start(self):
data = backup.Staging()
manifest: backup.Manifest | None = None
results: dict[str, dict] = {}
verbose = self.args.verbose > 0
try:
path = await _resolve_archive(self.host, self.args, data)
try:
self.disp(_("starting verification..."))
with Progress(console=self.console) as progress:
task = progress.add_task(_("verifying"), total=None)
def progress_cb(name: str, done: int, total: int) -> None:
progress.update(
task,
completed=done,
total=total,
description=_fixed_width(name, VERIFY_NAME_WIDTH),
)
def entry_cb(name: str, info: dict) -> None:
# a failed check stays on its own line, ok lines are
# verbose only
if info["checksum"] != "ok" or info["pie"]:
status = []
if info["checksum"] != "ok":
status.append(_("checksum mismatch (archive tampered?)"))
if info["pie"]:
status.append(info["pie"])
progress.console.print(
f"[red]{name}: {', '.join(status)}[/red]"
)
elif verbose:
progress.console.print(
_(" {name} ({size} bytes): ok").format(
name=name, size=info["size"]
)
)
manifest, results = backup.Backup.verify_archive(
path, progress_cb=progress_cb, entry_cb=entry_cb
)
except backup.BackupError as e:
self.disp(str(e), error=True)
self.host.quit(C.EXIT_ERROR)
finally:
data.close()
# verify_archive succeeded (the command has quit otherwise)
assert manifest is not None
self.print(
_("format: {fmt} v{version}").format(
fmt=manifest.format, version=manifest.format_version
)
)
self.print(
_("app: {app} {version}").format(
app=manifest.app, version=manifest.app_version
)
)
self.print(_("alembic: {alembic}").format(alembic=manifest.alembic))
self.print(_("date: {date}").format(date=manifest.date))
profiles = sorted(manifest.profiles.items())
if verbose or len(profiles) <= MAX_PROFILES_SHOWN:
for name, prof in profiles:
self.print(
_("profile {name}: {jid} ({types})").format(
name=name, jid=prof.jid or "-", types=", ".join(prof.types)
)
)
else:
self.print(_("({count} profiles)").format(count=len(profiles)))
if not results:
# only the manifest is in the archive (the selected types did not
# produce any data), there is nothing to check
self.console.print(
_("[yellow]archive contains no data entries (only the manifest)[/yellow]")
)
self.host.quit(0)
succeeded = sum(
1 for i in results.values() if i["checksum"] == "ok" and not i["pie"]
)
if succeeded == len(results):
self.console.print(
_(
"[green]{done}/{total} checks successful, archive is sane[/green]"
).format(done=succeeded, total=len(results))
)
self.host.quit(0)
self.console.print(
_(
"[red]{done}/{total} checks successful, archive integrity FAILED[/red]"
).format(done=succeeded, total=len(results))
)
self.host.quit(C.EXIT_ERROR)
class Backup(base.CommandBase):
"""Backup/export/import of Libervia data"""
subcommands = (Export, Import, Verify)
def __init__(self, host):
super(Backup, self).__init__(
host,
"backup",
use_profile=False,
help=_("backup/restore Libervia data (XEP-0227 based format)"),
)
Reference in a new issue View git blame Copy permalink