archiver.py 23 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506
  1. import argparse
  2. from binascii import hexlify
  3. from datetime import datetime
  4. from operator import attrgetter
  5. import os
  6. import stat
  7. import sys
  8. from attic import __version__
  9. from attic.archive import Archive
  10. from attic.repository import Repository
  11. from attic.cache import Cache
  12. from attic.key import key_creator
  13. from attic.helpers import location_validator, format_time, \
  14. format_file_mode, IncludePattern, ExcludePattern, exclude_path, adjust_patterns, to_localtime, \
  15. get_cache_dir, get_keys_dir, format_timedelta, prune_split, Manifest, Location, remove_surrogates
  16. from attic.remote import RepositoryServer, RemoteRepository, ConnectionClosed
  17. class Archiver:
  18. def __init__(self):
  19. self.exit_code = 0
  20. def open_repository(self, location, create=False):
  21. if location.proto == 'ssh':
  22. repository = RemoteRepository(location, create=create)
  23. else:
  24. repository = Repository(location.path, create=create)
  25. repository._location = location
  26. return repository
  27. def print_error(self, msg, *args):
  28. msg = args and msg % args or msg
  29. self.exit_code = 1
  30. print('attic: ' + msg, file=sys.stderr)
  31. def print_verbose(self, msg, *args, **kw):
  32. if self.verbose:
  33. msg = args and msg % args or msg
  34. if kw.get('newline', True):
  35. print(msg)
  36. else:
  37. print(msg, end=' ')
  38. def do_serve(self, args):
  39. return RepositoryServer().serve()
  40. def do_init(self, args):
  41. """Initialize an empty repository
  42. """
  43. print('Initializing repository at "%s"' % args.repository.orig)
  44. repository = self.open_repository(args.repository, create=True)
  45. key = key_creator(repository, args)
  46. manifest = Manifest()
  47. manifest.repository = repository
  48. manifest.key = key
  49. manifest.write()
  50. repository.commit()
  51. return self.exit_code
  52. def do_change_passphrase(self, args):
  53. """Change repository key file passphrase
  54. """
  55. repository = self.open_repository(args.repository)
  56. manifest, key = Manifest.load(repository)
  57. key.change_passphrase()
  58. return self.exit_code
  59. def do_create(self, args):
  60. """Create new archive
  61. """
  62. t0 = datetime.now()
  63. repository = self.open_repository(args.archive)
  64. manifest, key = Manifest.load(repository)
  65. cache = Cache(repository, key, manifest)
  66. archive = Archive(repository, key, manifest, args.archive.archive, cache=cache,
  67. create=True, checkpoint_interval=args.checkpoint_interval,
  68. numeric_owner=args.numeric_owner)
  69. # Add Attic cache dir to inode_skip list
  70. skip_inodes = set()
  71. try:
  72. st = os.stat(get_cache_dir())
  73. skip_inodes.add((st.st_ino, st.st_dev))
  74. except IOError:
  75. pass
  76. # Add local repository dir to inode_skip list
  77. if not args.archive.host:
  78. try:
  79. st = os.stat(args.archive.path)
  80. skip_inodes.add((st.st_ino, st.st_dev))
  81. except IOError:
  82. pass
  83. for path in args.paths:
  84. path = os.path.normpath(path)
  85. if args.dontcross:
  86. try:
  87. restrict_dev = os.lstat(path).st_dev
  88. except OSError as e:
  89. self.print_error('%s: %s', path, e)
  90. continue
  91. else:
  92. restrict_dev = None
  93. self._process(archive, cache, args.excludes, skip_inodes, path, restrict_dev)
  94. archive.save()
  95. if args.stats:
  96. t = datetime.now()
  97. diff = t - t0
  98. print('-' * 40)
  99. print('Archive name: %s' % args.archive.archive)
  100. print('Archive fingerprint: %s' % hexlify(archive.id).decode('ascii'))
  101. print('Start time: %s' % t0.strftime('%c'))
  102. print('End time: %s' % t.strftime('%c'))
  103. print('Duration: %s' % format_timedelta(diff))
  104. archive.stats.print_()
  105. print('-' * 40)
  106. return self.exit_code
  107. def _process(self, archive, cache, excludes, skip_inodes, path, restrict_dev):
  108. if exclude_path(path, excludes):
  109. return
  110. try:
  111. st = os.lstat(path)
  112. except OSError as e:
  113. self.print_error('%s: %s', path, e)
  114. return
  115. if (st.st_ino, st.st_dev) in skip_inodes:
  116. return
  117. # Entering a new filesystem?
  118. if restrict_dev and st.st_dev != restrict_dev:
  119. return
  120. # Ignore unix sockets
  121. if stat.S_ISSOCK(st.st_mode):
  122. return
  123. self.print_verbose(remove_surrogates(path))
  124. if stat.S_ISREG(st.st_mode):
  125. try:
  126. archive.process_file(path, st, cache)
  127. except IOError as e:
  128. self.print_error('%s: %s', path, e)
  129. elif stat.S_ISDIR(st.st_mode):
  130. archive.process_item(path, st)
  131. try:
  132. entries = os.listdir(path)
  133. except OSError as e:
  134. self.print_error('%s: %s', path, e)
  135. else:
  136. for filename in sorted(entries):
  137. self._process(archive, cache, excludes, skip_inodes,
  138. os.path.join(path, filename), restrict_dev)
  139. elif stat.S_ISLNK(st.st_mode):
  140. archive.process_symlink(path, st)
  141. elif stat.S_ISFIFO(st.st_mode):
  142. archive.process_item(path, st)
  143. elif stat.S_ISCHR(st.st_mode) or stat.S_ISBLK(st.st_mode):
  144. archive.process_dev(path, st)
  145. else:
  146. self.print_error('Unknown file type: %s', path)
  147. def do_extract(self, args):
  148. """Extract archive contents
  149. """
  150. repository = self.open_repository(args.archive)
  151. manifest, key = Manifest.load(repository)
  152. archive = Archive(repository, key, manifest, args.archive.archive,
  153. numeric_owner=args.numeric_owner)
  154. patterns = adjust_patterns(args.paths, args.excludes)
  155. dirs = []
  156. for item, peek in archive.iter_items(lambda item: not exclude_path(item[b'path'], patterns)):
  157. while dirs and not item[b'path'].startswith(dirs[-1][b'path']):
  158. archive.extract_item(dirs.pop(-1))
  159. self.print_verbose(remove_surrogates(item[b'path']))
  160. try:
  161. if stat.S_ISDIR(item[b'mode']):
  162. dirs.append(item)
  163. archive.extract_item(item, restore_attrs=False)
  164. else:
  165. archive.extract_item(item, peek=peek)
  166. except IOError as e:
  167. self.print_error('%s: %s', remove_surrogates(item[b'path']), e)
  168. while dirs:
  169. archive.extract_item(dirs.pop(-1))
  170. return self.exit_code
  171. def do_delete(self, args):
  172. """Delete archive
  173. """
  174. repository = self.open_repository(args.archive)
  175. manifest, key = Manifest.load(repository)
  176. cache = Cache(repository, key, manifest)
  177. archive = Archive(repository, key, manifest, args.archive.archive, cache=cache)
  178. archive.delete(cache)
  179. return self.exit_code
  180. def do_mount(self, args):
  181. """Mount archive as a FUSE fileystem
  182. """
  183. try:
  184. from attic.fuse import AtticOperations
  185. except ImportError:
  186. self.print_error('the "llfuse" module is required to use this feature')
  187. return self.exit_code
  188. if not os.path.isdir(args.mountpoint) or not os.access(args.mountpoint, os.R_OK | os.W_OK | os.X_OK):
  189. self.print_error('%s: Mountpoint must be a writable directory' % args.mountpoint)
  190. return self.exit_code
  191. repository = self.open_repository(args.archive)
  192. manifest, key = Manifest.load(repository)
  193. self.print_verbose("Loading archive metadata...", newline=False)
  194. archive = Archive(repository, key, manifest, args.archive.archive)
  195. self.print_verbose('done')
  196. operations = AtticOperations(key, repository, archive)
  197. self.print_verbose("Mounting filesystem")
  198. try:
  199. operations.mount(args.mountpoint, args.options, args.foreground)
  200. except RuntimeError:
  201. # Relevant error message already printed to stderr by fuse
  202. self.exit_code = 1
  203. return self.exit_code
  204. def do_list(self, args):
  205. """List archive or repository contents
  206. """
  207. repository = self.open_repository(args.src)
  208. manifest, key = Manifest.load(repository)
  209. if args.src.archive:
  210. tmap = {1: 'p', 2: 'c', 4: 'd', 6: 'b', 0o10: '-', 0o12: 'l', 0o14: 's'}
  211. archive = Archive(repository, key, manifest, args.src.archive)
  212. for item, _ in archive.iter_items():
  213. type = tmap.get(item[b'mode'] // 4096, '?')
  214. mode = format_file_mode(item[b'mode'])
  215. size = 0
  216. if type == '-':
  217. try:
  218. size = sum(size for _, size, _ in item[b'chunks'])
  219. except KeyError:
  220. pass
  221. mtime = format_time(datetime.fromtimestamp(item[b'mtime'] / 10**9))
  222. if b'source' in item:
  223. if type == 'l':
  224. extra = ' -> %s' % item[b'source']
  225. else:
  226. type = 'h'
  227. extra = ' link to %s' % item[b'source']
  228. else:
  229. extra = ''
  230. print('%s%s %-6s %-6s %8d %s %s%s' % (type, mode, item[b'user'] or item[b'uid'],
  231. item[b'group'] or item[b'gid'], size, mtime,
  232. remove_surrogates(item[b'path']), extra))
  233. else:
  234. for archive in sorted(Archive.list_archives(repository, key, manifest), key=attrgetter('ts')):
  235. print('%-20s %s' % (archive.metadata[b'name'], to_localtime(archive.ts).strftime('%c')))
  236. return self.exit_code
  237. def do_verify(self, args):
  238. """Verify archive consistency
  239. """
  240. repository = self.open_repository(args.archive)
  241. manifest, key = Manifest.load(repository)
  242. archive = Archive(repository, key, manifest, args.archive.archive)
  243. patterns = adjust_patterns(args.paths, args.excludes)
  244. def start_cb(item):
  245. self.print_verbose('%s ...', remove_surrogates(item[b'path']), newline=False)
  246. def result_cb(item, success):
  247. if success:
  248. self.print_verbose('OK')
  249. else:
  250. self.print_verbose('ERROR')
  251. self.print_error('%s: verification failed' % remove_surrogates(item[b'path']))
  252. for item, peek in archive.iter_items(lambda item: not exclude_path(item[b'path'], patterns)):
  253. if stat.S_ISREG(item[b'mode']) and b'chunks' in item:
  254. archive.verify_file(item, start_cb, result_cb, peek=peek)
  255. return self.exit_code
  256. def do_info(self, args):
  257. """Show archive details such as disk space used
  258. """
  259. repository = self.open_repository(args.archive)
  260. manifest, key = Manifest.load(repository)
  261. cache = Cache(repository, key, manifest)
  262. archive = Archive(repository, key, manifest, args.archive.archive, cache=cache)
  263. stats = archive.calc_stats(cache)
  264. print('Name:', archive.name)
  265. print('Fingerprint: %s' % hexlify(archive.id).decode('ascii'))
  266. print('Hostname:', archive.metadata[b'hostname'])
  267. print('Username:', archive.metadata[b'username'])
  268. print('Time: %s' % to_localtime(archive.ts).strftime('%c'))
  269. print('Command line:', remove_surrogates(' '.join(archive.metadata[b'cmdline'])))
  270. stats.print_()
  271. return self.exit_code
  272. def do_prune(self, args):
  273. """Prune repository archives according to specified rules
  274. """
  275. repository = self.open_repository(args.repository)
  276. manifest, key = Manifest.load(repository)
  277. cache = Cache(repository, key, manifest)
  278. archives = list(sorted(Archive.list_archives(repository, key, manifest, cache),
  279. key=attrgetter('ts'), reverse=True))
  280. if args.hourly + args.daily + args.weekly + args.monthly + args.yearly == 0:
  281. self.print_error('At least one of the "hourly", "daily", "weekly", "monthly" or "yearly" '
  282. 'settings must be specified')
  283. return 1
  284. if args.prefix:
  285. archives = [archive for archive in archives if archive.name.startswith(args.prefix)]
  286. keep = []
  287. if args.hourly:
  288. keep += prune_split(archives, '%Y-%m-%d %H', args.hourly)
  289. if args.daily:
  290. keep += prune_split(archives, '%Y-%m-%d', args.daily, keep)
  291. if args.weekly:
  292. keep += prune_split(archives, '%G-%V', args.weekly, keep)
  293. if args.monthly:
  294. keep += prune_split(archives, '%Y-%m', args.monthly, keep)
  295. if args.yearly:
  296. keep += prune_split(archives, '%Y', args.yearly, keep)
  297. keep.sort(key=attrgetter('ts'), reverse=True)
  298. to_delete = [a for a in archives if a not in keep]
  299. for archive in keep:
  300. self.print_verbose('Keeping archive "%s"' % archive.name)
  301. for archive in to_delete:
  302. self.print_verbose('Pruning archive "%s"', archive.name)
  303. archive.delete(cache)
  304. return self.exit_code
  305. def run(self, args=None):
  306. keys_dir = get_keys_dir()
  307. if not os.path.exists(keys_dir):
  308. os.makedirs(keys_dir)
  309. os.chmod(keys_dir, stat.S_IRWXU)
  310. cache_dir = get_cache_dir()
  311. if not os.path.exists(cache_dir):
  312. os.makedirs(cache_dir)
  313. os.chmod(cache_dir, stat.S_IRWXU)
  314. common_parser = argparse.ArgumentParser(add_help=False)
  315. common_parser.add_argument('-v', '--verbose', dest='verbose', action='store_true',
  316. default=False,
  317. help='verbose output')
  318. parser = argparse.ArgumentParser(description='Attic %s - Deduplicated Backups' % __version__)
  319. subparsers = parser.add_subparsers(title='Available commands')
  320. subparser = subparsers.add_parser('serve', parents=[common_parser])
  321. subparser.set_defaults(func=self.do_serve)
  322. subparser = subparsers.add_parser('init', parents=[common_parser],
  323. description=self.do_init.__doc__)
  324. subparser.set_defaults(func=self.do_init)
  325. subparser.add_argument('repository', metavar='REPOSITORY',
  326. type=location_validator(archive=False),
  327. help='repository to create')
  328. subparser.add_argument('--key-file', dest='keyfile',
  329. action='store_true', default=False,
  330. help='enable key file based encryption')
  331. subparser.add_argument('--passphrase', dest='passphrase',
  332. action='store_true', default=False,
  333. help='enable passphrase based encryption')
  334. subparser = subparsers.add_parser('change-passphrase', parents=[common_parser],
  335. description=self.do_change_passphrase.__doc__)
  336. subparser.set_defaults(func=self.do_change_passphrase)
  337. subparser.add_argument('repository', metavar='REPOSITORY',
  338. type=location_validator(archive=False))
  339. subparser = subparsers.add_parser('create', parents=[common_parser],
  340. description=self.do_create.__doc__)
  341. subparser.set_defaults(func=self.do_create)
  342. subparser.add_argument('-s', '--stats', dest='stats',
  343. action='store_true', default=False,
  344. help='print statistics for the created archive')
  345. subparser.add_argument('-e', '--exclude', dest='excludes',
  346. type=ExcludePattern, action='append',
  347. metavar="PATTERN", help='exclude paths matching PATTERN')
  348. subparser.add_argument('-c', '--checkpoint-interval', dest='checkpoint_interval',
  349. type=int, default=300, metavar='SECONDS',
  350. help='write checkpoint every SECONDS seconds (Default: 300)')
  351. subparser.add_argument('--do-not-cross-mountpoints', dest='dontcross',
  352. action='store_true', default=False,
  353. help='do not cross mount points')
  354. subparser.add_argument('--numeric-owner', dest='numeric_owner',
  355. action='store_true', default=False,
  356. help='only store numeric user and group identifiers')
  357. subparser.add_argument('archive', metavar='ARCHIVE',
  358. type=location_validator(archive=True),
  359. help='archive to create')
  360. subparser.add_argument('paths', metavar='PATH', nargs='+', type=str,
  361. help='paths to archive')
  362. subparser = subparsers.add_parser('extract', parents=[common_parser],
  363. description=self.do_extract.__doc__)
  364. subparser.set_defaults(func=self.do_extract)
  365. subparser.add_argument('-e', '--exclude', dest='excludes',
  366. type=ExcludePattern, action='append',
  367. metavar="PATTERN", help='exclude paths matching PATTERN')
  368. subparser.add_argument('--numeric-owner', dest='numeric_owner',
  369. action='store_true', default=False,
  370. help='only obey numeric user and group identifiers')
  371. subparser.add_argument('archive', metavar='ARCHIVE',
  372. type=location_validator(archive=True),
  373. help='archive to extract')
  374. subparser.add_argument('paths', metavar='PATH', nargs='*', type=str,
  375. help='paths to extract')
  376. subparser = subparsers.add_parser('delete', parents=[common_parser],
  377. description=self.do_delete.__doc__)
  378. subparser.set_defaults(func=self.do_delete)
  379. subparser.add_argument('archive', metavar='ARCHIVE',
  380. type=location_validator(archive=True),
  381. help='archive to delete')
  382. subparser = subparsers.add_parser('list', parents=[common_parser],
  383. description=self.do_list.__doc__)
  384. subparser.set_defaults(func=self.do_list)
  385. subparser.add_argument('src', metavar='REPOSITORY_OR_ARCHIVE', type=location_validator(),
  386. help='repository/archive to list contents of')
  387. subparser = subparsers.add_parser('mount', parents=[common_parser],
  388. description=self.do_mount.__doc__)
  389. subparser.set_defaults(func=self.do_mount)
  390. subparser.add_argument('archive', metavar='ARCHIVE', type=location_validator(archive=True),
  391. help='archive to mount')
  392. subparser.add_argument('mountpoint', metavar='MOUNTPOINT', type=str,
  393. help='where to mount filesystem')
  394. subparser.add_argument('-f', '--foreground', dest='foreground',
  395. action='store_true', default=False,
  396. help='stay in foreground, do not daemonize')
  397. subparser.add_argument('-o', dest='options', type=str,
  398. help='Extra mount options')
  399. subparser = subparsers.add_parser('verify', parents=[common_parser],
  400. description=self.do_verify.__doc__)
  401. subparser.set_defaults(func=self.do_verify)
  402. subparser.add_argument('-e', '--exclude', dest='excludes',
  403. type=ExcludePattern, action='append',
  404. metavar="PATTERN", help='exclude paths matching PATTERN')
  405. subparser.add_argument('archive', metavar='ARCHIVE',
  406. type=location_validator(archive=True),
  407. help='archive to verity integrity of')
  408. subparser.add_argument('paths', metavar='PATH', nargs='*', type=str,
  409. help='paths to verify')
  410. subparser = subparsers.add_parser('info', parents=[common_parser],
  411. description=self.do_info.__doc__)
  412. subparser.set_defaults(func=self.do_info)
  413. subparser.add_argument('archive', metavar='ARCHIVE',
  414. type=location_validator(archive=True),
  415. help='archive to display information about')
  416. subparser = subparsers.add_parser('prune', parents=[common_parser],
  417. description=self.do_prune.__doc__)
  418. subparser.set_defaults(func=self.do_prune)
  419. subparser.add_argument('-H', '--hourly', dest='hourly', type=int, default=0,
  420. help='number of hourly archives to keep')
  421. subparser.add_argument('-d', '--daily', dest='daily', type=int, default=0,
  422. help='number of daily archives to keep')
  423. subparser.add_argument('-w', '--weekly', dest='weekly', type=int, default=0,
  424. help='number of daily archives to keep')
  425. subparser.add_argument('-m', '--monthly', dest='monthly', type=int, default=0,
  426. help='number of monthly archives to keep')
  427. subparser.add_argument('-y', '--yearly', dest='yearly', type=int, default=0,
  428. help='number of yearly archives to keep')
  429. subparser.add_argument('-p', '--prefix', dest='prefix', type=str,
  430. help='only consider archive names starting with this prefix')
  431. subparser.add_argument('repository', metavar='REPOSITORY',
  432. type=location_validator(archive=False),
  433. help='repository to prune')
  434. args = parser.parse_args(args or ['-h'])
  435. self.verbose = args.verbose
  436. return args.func(args)
  437. def main():
  438. archiver = Archiver()
  439. try:
  440. exit_code = archiver.run(sys.argv[1:])
  441. except Repository.DoesNotExist:
  442. archiver.print_error('Error: Repository not found')
  443. exit_code = 1
  444. except Repository.AlreadyExists:
  445. archiver.print_error('Error: Repository already exists')
  446. exit_code = 1
  447. except Archive.AlreadyExists as e:
  448. archiver.print_error('Error: Archive "%s" already exists', e)
  449. exit_code = 1
  450. except Archive.DoesNotExist as e:
  451. archiver.print_error('Error: Archive "%s" does not exist', e)
  452. exit_code = 1
  453. except ConnectionClosed:
  454. archiver.print_error('Connection closed by remote host')
  455. exit_code = 1
  456. except KeyboardInterrupt:
  457. archiver.print_error('Error: Keyboard interrupt')
  458. exit_code = 1
  459. else:
  460. if exit_code:
  461. archiver.print_error('Exiting with failure status due to previous errors')
  462. sys.exit(exit_code)
  463. if __name__ == '__main__':
  464. main()