archiver.py 17 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396
  1. import argparse
  2. from datetime import datetime
  3. from operator import attrgetter
  4. import os
  5. import stat
  6. import sys
  7. from .archive import Archive
  8. from .store import Store
  9. from .cache import Cache
  10. from .key import Key
  11. from .helpers import location_validator, format_time, \
  12. format_file_mode, IncludePattern, ExcludePattern, exclude_path, to_localtime, \
  13. get_cache_dir, format_timedelta, purge_split, Manifest
  14. from .remote import StoreServer, RemoteStore
  15. class Archiver(object):
  16. def __init__(self):
  17. self.exit_code = 0
  18. def open_store(self, location, create=False):
  19. if location.proto == 'ssh':
  20. return RemoteStore(location, create=create)
  21. else:
  22. return Store(location.path, create=create)
  23. def print_error(self, msg, *args):
  24. msg = args and msg % args or msg
  25. self.exit_code = 1
  26. print >> sys.stderr, msg
  27. def print_verbose(self, msg, *args, **kw):
  28. if self.verbose:
  29. msg = args and msg % args or msg
  30. if kw.get('newline', True):
  31. print msg
  32. else:
  33. print msg,
  34. def do_serve(self, args):
  35. return StoreServer().serve()
  36. def do_init(self, args):
  37. store = self.open_store(args.store, create=True)
  38. Key.create(store, args.store.to_key_filename(),
  39. password=args.password)
  40. key = Key(store)
  41. manifest = Manifest(store, key, dont_load=True)
  42. manifest.write()
  43. store.commit()
  44. return self.exit_code
  45. def do_create(self, args):
  46. t0 = datetime.now()
  47. store = self.open_store(args.archive)
  48. key = Key(store)
  49. manifest = Manifest(store, key)
  50. cache = Cache(store, key, manifest)
  51. archive = Archive(store, key, manifest, args.archive.archive, cache=cache,
  52. create=True, checkpoint_interval=args.checkpoint_interval)
  53. # Add darc cache dir to inode_skip list
  54. skip_inodes = set()
  55. try:
  56. st = os.stat(get_cache_dir())
  57. skip_inodes.add((st.st_ino, st.st_dev))
  58. except IOError:
  59. pass
  60. # Add local store dir to inode_skip list
  61. if not args.archive.host:
  62. try:
  63. st = os.stat(args.archive.path)
  64. skip_inodes.add((st.st_ino, st.st_dev))
  65. except IOError:
  66. pass
  67. for path in args.paths:
  68. self._process(archive, cache, args.patterns, skip_inodes, path)
  69. archive.save()
  70. if args.stats:
  71. t = datetime.now()
  72. diff = t - t0
  73. print '-' * 40
  74. print 'Archive name: %s' % args.archive.archive
  75. print 'Archive fingerprint: %s' % archive.id.encode('hex')
  76. print 'Start time: %s' % t0.strftime('%c')
  77. print 'End time: %s' % t.strftime('%c')
  78. print 'Duration: %s' % format_timedelta(diff)
  79. archive.stats.print_()
  80. print '-' * 40
  81. return self.exit_code
  82. def _process(self, archive, cache, patterns, skip_inodes, path):
  83. if exclude_path(path, patterns):
  84. return
  85. try:
  86. st = os.lstat(path)
  87. except OSError, e:
  88. self.print_error('%s: %s', path, e)
  89. return
  90. if (st.st_ino, st.st_dev) in skip_inodes:
  91. return
  92. # Ignore unix sockets
  93. if stat.S_ISSOCK(st.st_mode):
  94. return
  95. self.print_verbose(path)
  96. if stat.S_ISDIR(st.st_mode):
  97. archive.process_dir(path, st)
  98. try:
  99. entries = os.listdir(path)
  100. except OSError, e:
  101. self.print_error('%s: %s', path, e)
  102. else:
  103. for filename in sorted(entries):
  104. self._process(archive, cache, patterns, skip_inodes,
  105. os.path.join(path, filename))
  106. elif stat.S_ISLNK(st.st_mode):
  107. archive.process_symlink(path, st)
  108. elif stat.S_ISFIFO(st.st_mode):
  109. archive.process_fifo(path, st)
  110. elif stat.S_ISREG(st.st_mode):
  111. try:
  112. archive.process_file(path, st, cache)
  113. except IOError, e:
  114. self.print_error('%s: %s', path, e)
  115. else:
  116. self.print_error('Unknown file type: %s', path)
  117. def do_extract(self, args):
  118. def start_cb(item):
  119. self.print_verbose(item['path'])
  120. def extract_cb(item):
  121. if exclude_path(item['path'], args.patterns):
  122. return
  123. if stat.S_ISDIR(item['mode']):
  124. dirs.append(item)
  125. archive.extract_item(item, args.dest, start_cb, restore_attrs=False)
  126. else:
  127. archive.extract_item(item, args.dest, start_cb)
  128. if dirs and not item['path'].startswith(dirs[-1]['path']):
  129. def cb(_, __, item):
  130. # Extract directories twice to make sure mtime is correctly restored
  131. archive.extract_item(item, args.dest)
  132. store.add_callback(cb, dirs.pop(-1))
  133. store = self.open_store(args.archive)
  134. key = Key(store)
  135. manifest = Manifest(store, key)
  136. archive = Archive(store, key, manifest, args.archive.archive)
  137. dirs = []
  138. archive.iter_items(extract_cb)
  139. store.flush_rpc()
  140. while dirs:
  141. archive.extract_item(dirs.pop(-1), args.dest)
  142. return self.exit_code
  143. def do_delete(self, args):
  144. store = self.open_store(args.archive)
  145. key = Key(store)
  146. manifest = Manifest(store, key)
  147. cache = Cache(store, key, manifest)
  148. archive = Archive(store, key, manifest, args.archive.archive, cache=cache)
  149. archive.delete(cache)
  150. return self.exit_code
  151. def do_list(self, args):
  152. def callback(item):
  153. type = tmap.get(item['mode'] / 4096, '?')
  154. mode = format_file_mode(item['mode'])
  155. size = 0
  156. if type == '-':
  157. try:
  158. size = sum(size for _, size, _ in item['chunks'])
  159. except KeyError:
  160. pass
  161. mtime = format_time(datetime.fromtimestamp(item['mtime']))
  162. if 'source' in item:
  163. if type == 'l':
  164. extra = ' -> %s' % item['source']
  165. else:
  166. type = 'h'
  167. extra = ' link to %s' % item['source']
  168. else:
  169. extra = ''
  170. print '%s%s %-6s %-6s %8d %s %s%s' % (type, mode, item['user'],
  171. item['group'], size, mtime,
  172. item['path'], extra)
  173. store = self.open_store(args.src)
  174. key = Key(store)
  175. manifest = Manifest(store, key)
  176. if args.src.archive:
  177. tmap = {1: 'p', 2: 'c', 4: 'd', 6: 'b', 010: '-', 012: 'l', 014: 's'}
  178. archive = Archive(store, key, manifest, args.src.archive)
  179. archive.iter_items(callback)
  180. store.flush_rpc()
  181. else:
  182. for archive in sorted(Archive.list_archives(store, key, manifest), key=attrgetter('ts')):
  183. print '%-20s %s' % (archive.metadata['name'], to_localtime(archive.ts).strftime('%c'))
  184. return self.exit_code
  185. def do_verify(self, args):
  186. store = self.open_store(args.archive)
  187. key = Key(store)
  188. manifest = Manifest(store, key)
  189. archive = Archive(store, key, manifest, args.archive.archive)
  190. def start_cb(item):
  191. self.print_verbose('%s ...', item['path'], newline=False)
  192. def result_cb(item, success):
  193. if success:
  194. self.print_verbose('OK')
  195. else:
  196. self.print_verbose('ERROR')
  197. self.print_error('%s: verification failed' % item['path'])
  198. def callback(item):
  199. if exclude_path(item['path'], args.patterns):
  200. return
  201. if stat.S_ISREG(item['mode']) and 'chunks' in item:
  202. archive.verify_file(item, start_cb, result_cb)
  203. archive.iter_items(callback)
  204. store.flush_rpc()
  205. return self.exit_code
  206. def do_info(self, args):
  207. store = self.open_store(args.archive)
  208. key = Key(store)
  209. manifest = Manifest(store, key)
  210. cache = Cache(store, key, manifest)
  211. archive = Archive(store, key, manifest, args.archive.archive, cache=cache)
  212. stats = archive.calc_stats(cache)
  213. print 'Name:', archive.name
  214. print 'Fingerprint: %s' % archive.id.encode('hex')
  215. print 'Hostname:', archive.metadata['hostname']
  216. print 'Username:', archive.metadata['username']
  217. print 'Time:', to_localtime(archive.ts).strftime('%c')
  218. print 'Command line:', ' '.join(archive.metadata['cmdline'])
  219. stats.print_()
  220. return self.exit_code
  221. def do_purge(self, args):
  222. store = self.open_store(args.store)
  223. key = Key(store)
  224. manifest = Manifest(store, key)
  225. cache = Cache(store, key, manifest)
  226. archives = list(sorted(Archive.list_archives(store, key, manifest, cache),
  227. key=attrgetter('ts'), reverse=True))
  228. if args.hourly + args.daily + args.weekly + args.monthly + args.yearly == 0:
  229. self.print_error('At least one of the "hourly", "daily", "weekly", "monthly" or "yearly" '
  230. 'settings must be specified')
  231. return 1
  232. if args.prefix:
  233. archives = [archive for archive in archives if archive.name.startswith(args.prefix)]
  234. keep = []
  235. if args.hourly:
  236. keep += purge_split(archives, '%Y-%m-%d %H', args.hourly)
  237. if args.daily:
  238. keep += purge_split(archives, '%Y-%m-%d', args.daily, keep)
  239. if args.weekly:
  240. keep += purge_split(archives, '%Y-%V', args.weekly, keep)
  241. if args.monthly:
  242. keep += purge_split(archives, '%Y-%m', args.monthly, keep)
  243. if args.yearly:
  244. keep += purge_split(archives, '%Y', args.yearly, keep)
  245. keep.sort(key=attrgetter('ts'), reverse=True)
  246. to_delete = [a for a in archives if a not in keep]
  247. for archive in keep:
  248. self.print_verbose('Keeping archive "%s"' % archive.name)
  249. for archive in to_delete:
  250. if args.really:
  251. self.print_verbose('Purging archive "%s"', archive.name)
  252. archive.delete(cache)
  253. else:
  254. print ('Archive "%s" marked for deletion. '
  255. 'Use the "--really" option to actually delete it'
  256. % archive.metadata['name'])
  257. return self.exit_code
  258. def run(self, args=None):
  259. dot_path = os.path.join(os.path.expanduser('~'), '.darc')
  260. if not os.path.exists(dot_path):
  261. os.mkdir(dot_path)
  262. os.mkdir(os.path.join(dot_path, 'keys'))
  263. os.mkdir(os.path.join(dot_path, 'cache'))
  264. common_parser = argparse.ArgumentParser(add_help=False)
  265. common_parser.add_argument('-v', '--verbose', dest='verbose', action='store_true',
  266. default=False,
  267. help='Verbose output')
  268. parser = argparse.ArgumentParser(description='DARC - Deduplicating Archiver')
  269. subparsers = parser.add_subparsers(title='Available subcommands')
  270. subparser = subparsers.add_parser('serve', parents=[common_parser])
  271. subparser.set_defaults(func=self.do_serve)
  272. subparser = subparsers.add_parser('init', parents=[common_parser])
  273. subparser.set_defaults(func=self.do_init)
  274. subparser.add_argument('-p', '--password', dest='password',
  275. help='Protect store key with password (Default: prompt)')
  276. subparser.add_argument('store',
  277. type=location_validator(archive=False),
  278. help='Store to create')
  279. subparser = subparsers.add_parser('create', parents=[common_parser])
  280. subparser.set_defaults(func=self.do_create)
  281. subparser.add_argument('-s', '--stats', dest='stats',
  282. action='store_true', default=False,
  283. help='Print statistics for the created archive')
  284. subparser.add_argument('-i', '--include', dest='patterns',
  285. type=IncludePattern, action='append',
  286. help='Include condition')
  287. subparser.add_argument('-e', '--exclude', dest='patterns',
  288. type=ExcludePattern, action='append',
  289. help='Include condition')
  290. subparser.add_argument('-c', '--checkpoint-interval', dest='checkpoint_interval',
  291. type=int, default=300, metavar='SECONDS',
  292. help='Write checkpointe ever SECONDS seconds (Default: 300)')
  293. subparser.add_argument('archive', metavar='ARCHIVE',
  294. type=location_validator(archive=True),
  295. help='Archive to create')
  296. subparser.add_argument('paths', metavar='PATH', nargs='+', type=str,
  297. help='Paths to add to archive')
  298. subparser = subparsers.add_parser('extract', parents=[common_parser])
  299. subparser.set_defaults(func=self.do_extract)
  300. subparser.add_argument('-i', '--include', dest='patterns',
  301. type=IncludePattern, action='append',
  302. help='Include condition')
  303. subparser.add_argument('-e', '--exclude', dest='patterns',
  304. type=ExcludePattern, action='append',
  305. help='Include condition')
  306. subparser.add_argument('archive', metavar='ARCHIVE',
  307. type=location_validator(archive=True),
  308. help='Archive to create')
  309. subparser.add_argument('dest', metavar='DEST', type=str, nargs='?',
  310. help='Where to extract files')
  311. subparser = subparsers.add_parser('delete', parents=[common_parser])
  312. subparser.set_defaults(func=self.do_delete)
  313. subparser.add_argument('archive', metavar='ARCHIVE',
  314. type=location_validator(archive=True),
  315. help='Archive to delete')
  316. subparser = subparsers.add_parser('list', parents=[common_parser])
  317. subparser.set_defaults(func=self.do_list)
  318. subparser.add_argument('src', metavar='SRC', type=location_validator(),
  319. help='Store/Archive to list contents of')
  320. subparser = subparsers.add_parser('verify', parents=[common_parser])
  321. subparser.set_defaults(func=self.do_verify)
  322. subparser.add_argument('-i', '--include', dest='patterns',
  323. type=IncludePattern, action='append',
  324. help='Include condition')
  325. subparser.add_argument('-e', '--exclude', dest='patterns',
  326. type=ExcludePattern, action='append',
  327. help='Include condition')
  328. subparser.add_argument('archive', metavar='ARCHIVE',
  329. type=location_validator(archive=True),
  330. help='Archive to verity integrity of')
  331. subparser = subparsers.add_parser('info', parents=[common_parser])
  332. subparser.set_defaults(func=self.do_info)
  333. subparser.add_argument('archive', metavar='ARCHIVE',
  334. type=location_validator(archive=True),
  335. help='Archive to display information about')
  336. subparser = subparsers.add_parser('purge', parents=[common_parser])
  337. subparser.set_defaults(func=self.do_purge)
  338. subparser.add_argument('-H', '--hourly', dest='hourly', type=int, default=0,
  339. help='Number of hourly archives to keep')
  340. subparser.add_argument('-d', '--daily', dest='daily', type=int, default=0,
  341. help='Number of daily archives to keep')
  342. subparser.add_argument('-w', '--weekly', dest='weekly', type=int, default=0,
  343. help='Number of daily archives to keep')
  344. subparser.add_argument('-m', '--monthly', dest='monthly', type=int, default=0,
  345. help='Number of monthly archives to keep')
  346. subparser.add_argument('-y', '--yearly', dest='yearly', type=int, default=0,
  347. help='Number of yearly archives to keep')
  348. subparser.add_argument('-p', '--prefix', dest='prefix', type=str,
  349. help='Only consider archive names starting with this prefix')
  350. subparser.add_argument('-r', '--really', dest='really',
  351. action='store_true', default=False,
  352. help='Actually delete archives')
  353. subparser.add_argument('store', metavar='STORE',
  354. type=location_validator(archive=False),
  355. help='Store to purge')
  356. args = parser.parse_args(args)
  357. self.verbose = args.verbose
  358. return args.func(args)
  359. def main():
  360. archiver = Archiver()
  361. sys.exit(archiver.run())
  362. if __name__ == '__main__':
  363. main()