remote.py 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321
  1. import errno
  2. import fcntl
  3. import msgpack
  4. import os
  5. import select
  6. import shutil
  7. from subprocess import Popen, PIPE
  8. import sys
  9. import tempfile
  10. from .hashindex import NSIndex
  11. from .helpers import Error, IntegrityError
  12. from .repository import Repository
  13. BUFSIZE = 10 * 1024 * 1024
  14. class ConnectionClosed(Error):
  15. """Connection closed by remote host"""
  16. class PathNotAllowed(Error):
  17. """Repository path not allowed"""
  18. class RepositoryServer(object):
  19. def __init__(self, restrict_to_paths):
  20. self.repository = None
  21. self.restrict_to_paths = restrict_to_paths
  22. def serve(self):
  23. # Make stdin non-blocking
  24. fl = fcntl.fcntl(sys.stdin.fileno(), fcntl.F_GETFL)
  25. fcntl.fcntl(sys.stdin.fileno(), fcntl.F_SETFL, fl | os.O_NONBLOCK)
  26. # Make stdout blocking
  27. fl = fcntl.fcntl(sys.stdout.fileno(), fcntl.F_GETFL)
  28. fcntl.fcntl(sys.stdout.fileno(), fcntl.F_SETFL, fl & ~os.O_NONBLOCK)
  29. unpacker = msgpack.Unpacker(use_list=False)
  30. while True:
  31. r, w, es = select.select([sys.stdin], [], [], 10)
  32. if r:
  33. data = os.read(sys.stdin.fileno(), BUFSIZE)
  34. if not data:
  35. return
  36. unpacker.feed(data)
  37. for unpacked in unpacker:
  38. if not (isinstance(unpacked, tuple) and len(unpacked) == 4):
  39. raise Exception("Unexpected RPC data format.")
  40. type, msgid, method, args = unpacked
  41. method = method.decode('ascii')
  42. try:
  43. try:
  44. f = getattr(self, method)
  45. except AttributeError:
  46. f = getattr(self.repository, method)
  47. res = f(*args)
  48. except Exception as e:
  49. sys.stdout.buffer.write(msgpack.packb((1, msgid, e.__class__.__name__, e.args)))
  50. else:
  51. sys.stdout.buffer.write(msgpack.packb((1, msgid, None, res)))
  52. sys.stdout.flush()
  53. if es:
  54. return
  55. def negotiate(self, versions):
  56. return 1
  57. def open(self, path, create=False):
  58. path = os.fsdecode(path)
  59. if path.startswith('/~'):
  60. path = path[1:]
  61. path = os.path.realpath(os.path.expanduser(path))
  62. if self.restrict_to_paths:
  63. for restrict_to_path in self.restrict_to_paths:
  64. if path.startswith(os.path.realpath(restrict_to_path)):
  65. break
  66. else:
  67. raise PathNotAllowed(path)
  68. self.repository = Repository(path, create)
  69. return self.repository.id
  70. class RemoteRepository(object):
  71. extra_test_args = []
  72. class RPCError(Exception):
  73. def __init__(self, name):
  74. self.name = name
  75. def __init__(self, location, create=False):
  76. self.location = location
  77. self.preload_ids = []
  78. self.msgid = 0
  79. self.to_send = b''
  80. self.cache = {}
  81. self.ignore_responses = set()
  82. self.responses = {}
  83. self.unpacker = msgpack.Unpacker(use_list=False)
  84. self.p = None
  85. if location.host == '__testsuite__':
  86. args = [sys.executable, '-m', 'attic.archiver', 'serve'] + self.extra_test_args
  87. else:
  88. args = ['ssh']
  89. if location.port:
  90. args += ['-p', str(location.port)]
  91. if location.user:
  92. args.append('%s@%s' % (location.user, location.host))
  93. else:
  94. args.append('%s' % location.host)
  95. args += ['attic', 'serve']
  96. self.p = Popen(args, bufsize=0, stdin=PIPE, stdout=PIPE)
  97. self.stdin_fd = self.p.stdin.fileno()
  98. self.stdout_fd = self.p.stdout.fileno()
  99. fcntl.fcntl(self.stdin_fd, fcntl.F_SETFL, fcntl.fcntl(self.stdin_fd, fcntl.F_GETFL) | os.O_NONBLOCK)
  100. fcntl.fcntl(self.stdout_fd, fcntl.F_SETFL, fcntl.fcntl(self.stdout_fd, fcntl.F_GETFL) | os.O_NONBLOCK)
  101. self.r_fds = [self.stdout_fd]
  102. self.x_fds = [self.stdin_fd, self.stdout_fd]
  103. version = self.call('negotiate', 1)
  104. if version != 1:
  105. raise Exception('Server insisted on using unsupported protocol version %d' % version)
  106. self.id = self.call('open', location.path, create)
  107. def __del__(self):
  108. self.close()
  109. def call(self, cmd, *args, **kw):
  110. for resp in self.call_many(cmd, [args], **kw):
  111. return resp
  112. def call_many(self, cmd, calls, wait=True, is_preloaded=False):
  113. if not calls:
  114. return
  115. def fetch_from_cache(args):
  116. msgid = self.cache[args].pop(0)
  117. if not self.cache[args]:
  118. del self.cache[args]
  119. return msgid
  120. calls = list(calls)
  121. waiting_for = []
  122. w_fds = [self.stdin_fd]
  123. while wait or calls:
  124. while waiting_for:
  125. try:
  126. error, res = self.responses.pop(waiting_for[0])
  127. waiting_for.pop(0)
  128. if error:
  129. if error == b'DoesNotExist':
  130. raise Repository.DoesNotExist(self.location.orig)
  131. elif error == b'AlreadyExists':
  132. raise Repository.AlreadyExists(self.location.orig)
  133. elif error == b'CheckNeeded':
  134. raise Repository.CheckNeeded(self.location.orig)
  135. elif error == b'IntegrityError':
  136. raise IntegrityError(res)
  137. elif error == b'PathNotAllowed':
  138. raise PathNotAllowed(*res)
  139. if error == b'ObjectNotFound':
  140. raise Repository.ObjectNotFound(res[0], self.location.orig)
  141. raise self.RPCError(error)
  142. else:
  143. yield res
  144. if not waiting_for and not calls:
  145. return
  146. except KeyError:
  147. break
  148. r, w, x = select.select(self.r_fds, w_fds, self.x_fds, 1)
  149. if x:
  150. raise Exception('FD exception occured')
  151. if r:
  152. data = os.read(self.stdout_fd, BUFSIZE)
  153. if not data:
  154. raise ConnectionClosed()
  155. self.unpacker.feed(data)
  156. for unpacked in self.unpacker:
  157. if not (isinstance(unpacked, tuple) and len(unpacked) == 4):
  158. raise Exception("Unexpected RPC data format.")
  159. type, msgid, error, res = unpacked
  160. if msgid in self.ignore_responses:
  161. self.ignore_responses.remove(msgid)
  162. else:
  163. self.responses[msgid] = error, res
  164. if w:
  165. while not self.to_send and (calls or self.preload_ids) and len(waiting_for) < 100:
  166. if calls:
  167. if is_preloaded:
  168. if calls[0] in self.cache:
  169. waiting_for.append(fetch_from_cache(calls.pop(0)))
  170. else:
  171. args = calls.pop(0)
  172. if cmd == 'get' and args in self.cache:
  173. waiting_for.append(fetch_from_cache(args))
  174. else:
  175. self.msgid += 1
  176. waiting_for.append(self.msgid)
  177. self.to_send = msgpack.packb((1, self.msgid, cmd, args))
  178. if not self.to_send and self.preload_ids:
  179. args = (self.preload_ids.pop(0),)
  180. self.msgid += 1
  181. self.cache.setdefault(args, []).append(self.msgid)
  182. self.to_send = msgpack.packb((1, self.msgid, cmd, args))
  183. if self.to_send:
  184. try:
  185. self.to_send = self.to_send[os.write(self.stdin_fd, self.to_send):]
  186. except OSError as e:
  187. # io.write might raise EAGAIN even though select indicates
  188. # that the fd should be writable
  189. if e.errno != errno.EAGAIN:
  190. raise
  191. if not self.to_send and not (calls or self.preload_ids):
  192. w_fds = []
  193. self.ignore_responses |= set(waiting_for)
  194. def check(self, repair=False):
  195. return self.call('check', repair)
  196. def commit(self, *args):
  197. return self.call('commit')
  198. def rollback(self, *args):
  199. return self.call('rollback')
  200. def __len__(self):
  201. return self.call('__len__')
  202. def list(self, limit=None, marker=None):
  203. return self.call('list', limit, marker)
  204. def get(self, id_):
  205. for resp in self.get_many([id_]):
  206. return resp
  207. def get_many(self, ids, is_preloaded=False):
  208. for resp in self.call_many('get', [(id_,) for id_ in ids], is_preloaded=is_preloaded):
  209. yield resp
  210. def put(self, id_, data, wait=True):
  211. return self.call('put', id_, data, wait=wait)
  212. def delete(self, id_, wait=True):
  213. return self.call('delete', id_, wait=wait)
  214. def close(self):
  215. if self.p:
  216. self.p.stdin.close()
  217. self.p.stdout.close()
  218. self.p.wait()
  219. self.p = None
  220. def preload(self, ids):
  221. self.preload_ids += ids
  222. class RepositoryCache:
  223. """A caching Repository wrapper
  224. Caches Repository GET operations using a temporary file
  225. """
  226. def __init__(self, repository):
  227. self.tmppath = None
  228. self.index = None
  229. self.data_fd = None
  230. self.repository = repository
  231. self.entries = {}
  232. self.initialize()
  233. def __del__(self):
  234. self.cleanup()
  235. def initialize(self):
  236. self.tmppath = tempfile.mkdtemp()
  237. self.index = NSIndex()
  238. self.data_fd = open(os.path.join(self.tmppath, 'data'), 'a+b')
  239. def cleanup(self):
  240. del self.index
  241. if self.data_fd:
  242. self.data_fd.close()
  243. if self.tmppath:
  244. shutil.rmtree(self.tmppath)
  245. def load_object(self, offset, size):
  246. self.data_fd.seek(offset)
  247. data = self.data_fd.read(size)
  248. assert len(data) == size
  249. return data
  250. def store_object(self, key, data):
  251. self.data_fd.seek(0, os.SEEK_END)
  252. self.data_fd.write(data)
  253. offset = self.data_fd.tell()
  254. self.index[key] = offset - len(data), len(data)
  255. def get(self, key):
  256. return next(self.get_many([key]))
  257. def get_many(self, keys):
  258. unknown_keys = [key for key in keys if not key in self.index]
  259. repository_iterator = zip(unknown_keys, self.repository.get_many(unknown_keys))
  260. for key in keys:
  261. try:
  262. yield self.load_object(*self.index[key])
  263. except KeyError:
  264. for key_, data in repository_iterator:
  265. if key_ == key:
  266. self.store_object(key, data)
  267. yield data
  268. break
  269. # Consume any pending requests
  270. for _ in repository_iterator:
  271. pass
  272. def cache_if_remote(repository):
  273. if isinstance(repository, RemoteRepository):
  274. return RepositoryCache(repository)
  275. return repository