lib/bup/client.py

   1 import re, struct, errno, time, zlib
   2 from bup import git, ssh
   3 from bup.helpers import *
   4
   5 bwlimit = None
   6
   7
   8 class ClientError(Exception):
   9     pass
  10
  11
  12 def _raw_write_bwlimit(f, buf, bwcount, bwtime):
  13     if not bwlimit:
  14         f.write(buf)
  15         return (len(buf), time.time())
  16     else:
  17         # We want to write in reasonably large blocks, but not so large that
  18         # they're likely to overflow a router's queue.  So our bwlimit timing
  19         # has to be pretty granular.  Also, if it takes too long from one
  20         # transmit to the next, we can't just make up for lost time to bring
  21         # the average back up to bwlimit - that will risk overflowing the
  22         # outbound queue, which defeats the purpose.  So if we fall behind
  23         # by more than one block delay, we shouldn't ever try to catch up.
  24         for i in xrange(0,len(buf),4096):
  25             now = time.time()
  26             next = max(now, bwtime + 1.0*bwcount/bwlimit)
  27             time.sleep(next-now)
  28             sub = buf[i:i+4096]
  29             f.write(sub)
  30             bwcount = len(sub)  # might be less than 4096
  31             bwtime = next
  32         return (bwcount, bwtime)
  33
  34
  35 def parse_remote(remote):
  36     protocol = r'([a-z]+)://'
  37     host = r'(?P<sb>\[)?((?(sb)[0-9a-f:]+|[^:/]+))(?(sb)\])'
  38     port = r'(?::(\d+))?'
  39     path = r'(/.*)?'
  40     url_match = re.match(
  41             '%s(?:%s%s)?%s' % (protocol, host, port, path), remote, re.I)
  42     if url_match:
  43         assert(url_match.group(1) in ('ssh', 'bup', 'file'))
  44         return url_match.group(1,3,4,5)
  45     else:
  46         rs = remote.split(':', 1)
  47         if len(rs) == 1 or rs[0] in ('', '-'):
  48             return 'file', None, None, rs[-1]
  49         else:
  50             return 'ssh', rs[0], None, rs[1]
  51
  52
  53 class Client:
  54     def __init__(self, remote, create=False):
  55         self._busy = self.conn = None
  56         self.sock = self.p = self.pout = self.pin = None
  57         is_reverse = os.environ.get('BUP_SERVER_REVERSE')
  58         if is_reverse:
  59             assert(not remote)
  60             remote = '%s:' % is_reverse
  61         (self.protocol, self.host, self.port, self.dir) = parse_remote(remote)
  62         self.cachedir = git.repo('index-cache/%s'
  63                                  % re.sub(r'[^@\w]', '_',
  64                                           "%s:%s" % (self.host, self.dir)))
  65         if is_reverse:
  66             self.pout = os.fdopen(3, 'rb')
  67             self.pin = os.fdopen(4, 'wb')
  68         else:
  69             if self.protocol in ('ssh', 'file'):
  70                 try:
  71                     # FIXME: ssh and file shouldn't use the same module
  72                     self.p = ssh.connect(self.host, self.port, 'server')
  73                     self.pout = self.p.stdout
  74                     self.pin = self.p.stdin
  75                 except OSError, e:
  76                     raise ClientError, 'connect: %s' % e, sys.exc_info()[2]
  77             elif self.protocol == 'bup':
  78                 self.sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
  79                 self.sock.connect((self.host, self.port or 1982))
  80                 self.pout = self.sock.makefile('rb')
  81                 self.pin = self.sock.makefile('wb')
  82         self.conn = Conn(self.pout, self.pin)
  83         if self.dir:
  84             self.dir = re.sub(r'[\r\n]', ' ', self.dir)
  85             if create:
  86                 self.conn.write('init-dir %s\n' % self.dir)
  87             else:
  88                 self.conn.write('set-dir %s\n' % self.dir)
  89             self.check_ok()
  90         self.sync_indexes()
  91
  92     def __del__(self):
  93         try:
  94             self.close()
  95         except IOError, e:
  96             if e.errno == errno.EPIPE:
  97                 pass
  98             else:
  99                 raise
 100
 101     def close(self):
 102         if self.conn and not self._busy:
 103             self.conn.write('quit\n')
 104         if self.pin and self.pout:
 105             self.pin.close()
 106             while self.pout.read(65536):
 107                 pass
 108             self.pout.close()
 109         if self.sock:
 110             self.sock.close()
 111         if self.p:
 112             self.p.wait()
 113             rv = self.p.wait()
 114             if rv:
 115                 raise ClientError('server tunnel returned exit code %d' % rv)
 116         self.conn = None
 117         self.sock = self.p = self.pin = self.pout = None
 118
 119     def check_ok(self):
 120         if self.p:
 121             rv = self.p.poll()
 122             if rv != None:
 123                 raise ClientError('server exited unexpectedly with code %r'
 124                                   % rv)
 125         try:
 126             return self.conn.check_ok()
 127         except Exception, e:
 128             raise ClientError, e, sys.exc_info()[2]
 129
 130     def check_busy(self):
 131         if self._busy:
 132             raise ClientError('already busy with command %r' % self._busy)
 133
 134     def ensure_busy(self):
 135         if not self._busy:
 136             raise ClientError('expected to be busy, but not busy?!')
 137
 138     def _not_busy(self):
 139         self._busy = None
 140
 141     def sync_indexes(self):
 142         self.check_busy()
 143         conn = self.conn
 144         mkdirp(self.cachedir)
 145         # All cached idxs are extra until proven otherwise
 146         extra = set()
 147         for f in os.listdir(self.cachedir):
 148             debug1('%s\n' % f)
 149             if f.endswith('.idx'):
 150                 extra.add(f)
 151         needed = set()
 152         conn.write('list-indexes\n')
 153         for line in linereader(conn):
 154             if not line:
 155                 break
 156             assert(line.find('/') < 0)
 157             parts = line.split(' ')
 158             idx = parts[0]
 159             if len(parts) == 2 and parts[1] == 'load' and idx not in extra:
 160                 # If the server requests that we load an idx and we don't
 161                 # already have a copy of it, it is needed
 162                 needed.add(idx)
 163             # Any idx that the server has heard of is proven not extra
 164             extra.discard(idx)
 165
 166         self.check_ok()
 167         debug1('client: removing extra indexes: %s\n' % extra)
 168         for idx in extra:
 169             os.unlink(os.path.join(self.cachedir, idx))
 170         debug1('client: server requested load of: %s\n' % needed)
 171         for idx in needed:
 172             self.sync_index(idx)
 173         git.auto_midx(self.cachedir)
 174
 175
 176     def sync_index(self, name):
 177         #debug1('requesting %r\n' % name)
 178         self.check_busy()
 179         mkdirp(self.cachedir)
 180         self.conn.write('send-index %s\n' % name)
 181         n = struct.unpack('!I', self.conn.read(4))[0]
 182         assert(n)
 183         fn = os.path.join(self.cachedir, name)
 184         f = open(fn + '.tmp', 'w')
 185         count = 0
 186         progress('Receiving index from server: %d/%d\r' % (count, n))
 187         for b in chunkyreader(self.conn, n):
 188             f.write(b)
 189             count += len(b)
 190             progress('Receiving index from server: %d/%d\r' % (count, n))
 191         progress('Receiving index from server: %d/%d, done.\n' % (count, n))
 192         self.check_ok()
 193         f.close()
 194         os.rename(fn + '.tmp', fn)
 195
 196     def _make_objcache(self):
 197         return git.PackIdxList(self.cachedir)
 198
 199     def _suggest_packs(self):
 200         ob = self._busy
 201         if ob:
 202             assert(ob == 'receive-objects-v2')
 203             self.conn.write('\xff\xff\xff\xff')  # suspend receive-objects-v2
 204         suggested = []
 205         for line in linereader(self.conn):
 206             if not line:
 207                 break
 208             debug2('%s\n' % line)
 209             if line.startswith('index '):
 210                 idx = line[6:]
 211                 debug1('client: received index suggestion: %s\n' % idx)
 212                 suggested.append(idx)
 213             else:
 214                 assert(line.endswith('.idx'))
 215                 debug1('client: completed writing pack, idx: %s\n' % line)
 216                 suggested.append(line)
 217         self.check_ok()
 218         if ob:
 219             self._busy = None
 220         for idx in suggested:
 221             self.sync_index(idx)
 222         git.auto_midx(self.cachedir)
 223         if ob:
 224             self._busy = ob
 225             self.conn.write('%s\n' % ob)
 226         return idx
 227
 228     def new_packwriter(self):
 229         self.check_busy()
 230         def _set_busy():
 231             self._busy = 'receive-objects-v2'
 232             self.conn.write('receive-objects-v2\n')
 233         return PackWriter_Remote(self.conn,
 234                                  objcache_maker = self._make_objcache,
 235                                  suggest_packs = self._suggest_packs,
 236                                  onopen = _set_busy,
 237                                  onclose = self._not_busy,
 238                                  ensure_busy = self.ensure_busy)
 239
 240     def read_ref(self, refname):
 241         self.check_busy()
 242         self.conn.write('read-ref %s\n' % refname)
 243         r = self.conn.readline().strip()
 244         self.check_ok()
 245         if r:
 246             assert(len(r) == 40)   # hexified sha
 247             return r.decode('hex')
 248         else:
 249             return None   # nonexistent ref
 250
 251     def update_ref(self, refname, newval, oldval):
 252         self.check_busy()
 253         self.conn.write('update-ref %s\n%s\n%s\n'
 254                         % (refname, newval.encode('hex'),
 255                            (oldval or '').encode('hex')))
 256         self.check_ok()
 257
 258     def cat(self, id):
 259         self.check_busy()
 260         self._busy = 'cat'
 261         self.conn.write('cat %s\n' % re.sub(r'[\n\r]', '_', id))
 262         while 1:
 263             sz = struct.unpack('!I', self.conn.read(4))[0]
 264             if not sz: break
 265             yield self.conn.read(sz)
 266         e = self.check_ok()
 267         self._not_busy()
 268         if e:
 269             raise KeyError(str(e))
 270
 271
 272 class PackWriter_Remote(git.PackWriter):
 273     def __init__(self, conn, objcache_maker, suggest_packs,
 274                  onopen, onclose,
 275                  ensure_busy):
 276         git.PackWriter.__init__(self, objcache_maker)
 277         self.file = conn
 278         self.filename = 'remote socket'
 279         self.suggest_packs = suggest_packs
 280         self.onopen = onopen
 281         self.onclose = onclose
 282         self.ensure_busy = ensure_busy
 283         self._packopen = False
 284         self._bwcount = 0
 285         self._bwtime = time.time()
 286
 287     def _open(self):
 288         if not self._packopen:
 289             self.onopen()
 290             self._packopen = True
 291
 292     def _end(self):
 293         if self._packopen and self.file:
 294             self.file.write('\0\0\0\0')
 295             self._packopen = False
 296             self.onclose() # Unbusy
 297             self.objcache = None
 298             return self.suggest_packs() # Returns last idx received
 299
 300     def close(self):
 301         id = self._end()
 302         self.file = None
 303         return id
 304
 305     def abort(self):
 306         raise GitError("don't know how to abort remote pack writing")
 307
 308     def _raw_write(self, datalist, sha):
 309         assert(self.file)
 310         if not self._packopen:
 311             self._open()
 312         self.ensure_busy()
 313         data = ''.join(datalist)
 314         assert(data)
 315         assert(sha)
 316         crc = zlib.crc32(data) & 0xffffffff
 317         outbuf = ''.join((struct.pack('!I', len(data) + 20 + 4),
 318                           sha,
 319                           struct.pack('!I', crc),
 320                           data))
 321         (self._bwcount, self._bwtime) = \
 322             _raw_write_bwlimit(self.file, outbuf, self._bwcount, self._bwtime)
 323         self.outbytes += len(data) - 20 - 4 # Don't count sha1+crc
 324         self.count += 1
 325
 326         if self.file.has_input():
 327             self.suggest_packs()
 328             self.objcache.refresh()
 329
 330         return sha, crc