]> arthur.barton.de Git - bup.git/blob - lib/bup/client.py
hashsplit.py: remove PackWriter-specific knowledge.
[bup.git] / 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             self.conn = Conn(self.pout, self.pin)
69         else:
70             if self.protocol in ('ssh', 'file'):
71                 try:
72                     # FIXME: ssh and file shouldn't use the same module
73                     self.p = ssh.connect(self.host, self.port, 'server')
74                     self.pout = self.p.stdout
75                     self.pin = self.p.stdin
76                     self.conn = Conn(self.pout, self.pin)
77                 except OSError, e:
78                     raise ClientError, 'connect: %s' % e, sys.exc_info()[2]
79             elif self.protocol == 'bup':
80                 self.sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
81                 self.sock.connect((self.host, atoi(self.port) or 1982))
82                 self.sockw = self.sock.makefile('wb')
83                 self.conn = DemuxConn(self.sock.fileno(), self.sockw)
84         if self.dir:
85             self.dir = re.sub(r'[\r\n]', ' ', self.dir)
86             if create:
87                 self.conn.write('init-dir %s\n' % self.dir)
88             else:
89                 self.conn.write('set-dir %s\n' % self.dir)
90             self.check_ok()
91         self.sync_indexes()
92
93     def __del__(self):
94         try:
95             self.close()
96         except IOError, e:
97             if e.errno == errno.EPIPE:
98                 pass
99             else:
100                 raise
101
102     def close(self):
103         if self.conn and not self._busy:
104             self.conn.write('quit\n')
105         if self.pin:
106             self.pin.close()
107         if self.sock and self.sockw:
108             self.sockw.close()
109             self.sock.shutdown(socket.SHUT_WR)
110         if self.conn:
111             self.conn.close()
112         if self.pout:
113             self.pout.close()
114         if self.sock:
115             self.sock.close()
116         if self.p:
117             self.p.wait()
118             rv = self.p.wait()
119             if rv:
120                 raise ClientError('server tunnel returned exit code %d' % rv)
121         self.conn = None
122         self.sock = self.p = self.pin = self.pout = None
123
124     def check_ok(self):
125         if self.p:
126             rv = self.p.poll()
127             if rv != None:
128                 raise ClientError('server exited unexpectedly with code %r'
129                                   % rv)
130         try:
131             return self.conn.check_ok()
132         except Exception, e:
133             raise ClientError, e, sys.exc_info()[2]
134
135     def check_busy(self):
136         if self._busy:
137             raise ClientError('already busy with command %r' % self._busy)
138         
139     def ensure_busy(self):
140         if not self._busy:
141             raise ClientError('expected to be busy, but not busy?!')
142         
143     def _not_busy(self):
144         self._busy = None
145
146     def sync_indexes(self):
147         self.check_busy()
148         conn = self.conn
149         mkdirp(self.cachedir)
150         # All cached idxs are extra until proven otherwise
151         extra = set()
152         for f in os.listdir(self.cachedir):
153             debug1('%s\n' % f)
154             if f.endswith('.idx'):
155                 extra.add(f)
156         needed = set()
157         conn.write('list-indexes\n')
158         for line in linereader(conn):
159             if not line:
160                 break
161             assert(line.find('/') < 0)
162             parts = line.split(' ')
163             idx = parts[0]
164             if len(parts) == 2 and parts[1] == 'load' and idx not in extra:
165                 # If the server requests that we load an idx and we don't
166                 # already have a copy of it, it is needed
167                 needed.add(idx)
168             # Any idx that the server has heard of is proven not extra
169             extra.discard(idx)
170
171         self.check_ok()
172         debug1('client: removing extra indexes: %s\n' % extra)
173         for idx in extra:
174             os.unlink(os.path.join(self.cachedir, idx))
175         debug1('client: server requested load of: %s\n' % needed)
176         for idx in needed:
177             self.sync_index(idx)
178         git.auto_midx(self.cachedir)
179
180     def sync_index(self, name):
181         #debug1('requesting %r\n' % name)
182         self.check_busy()
183         mkdirp(self.cachedir)
184         fn = os.path.join(self.cachedir, name)
185         if os.path.exists(fn):
186             msg = "won't request existing .idx, try `bup bloom --check %s`" % fn
187             raise ClientError(msg)
188         self.conn.write('send-index %s\n' % name)
189         n = struct.unpack('!I', self.conn.read(4))[0]
190         assert(n)
191         f = open(fn + '.tmp', 'w')
192         count = 0
193         progress('Receiving index from server: %d/%d\r' % (count, n))
194         for b in chunkyreader(self.conn, n):
195             f.write(b)
196             count += len(b)
197             qprogress('Receiving index from server: %d/%d\r' % (count, n))
198         progress('Receiving index from server: %d/%d, done.\n' % (count, n))
199         self.check_ok()
200         f.close()
201         os.rename(fn + '.tmp', fn)
202
203     def _make_objcache(self):
204         return git.PackIdxList(self.cachedir)
205
206     def _suggest_packs(self):
207         ob = self._busy
208         if ob:
209             assert(ob == 'receive-objects-v2')
210             self.conn.write('\xff\xff\xff\xff')  # suspend receive-objects-v2
211         suggested = []
212         for line in linereader(self.conn):
213             if not line:
214                 break
215             debug2('%s\n' % line)
216             if line.startswith('index '):
217                 idx = line[6:]
218                 debug1('client: received index suggestion: %s\n'
219                        % git.shorten_hash(idx))
220                 suggested.append(idx)
221             else:
222                 assert(line.endswith('.idx'))
223                 debug1('client: completed writing pack, idx: %s\n'
224                        % git.shorten_hash(line))
225                 suggested.append(line)
226         self.check_ok()
227         if ob:
228             self._busy = None
229         idx = None
230         for idx in suggested:
231             self.sync_index(idx)
232         git.auto_midx(self.cachedir)
233         if ob:
234             self._busy = ob
235             self.conn.write('%s\n' % ob)
236         return idx
237
238     def new_packwriter(self):
239         self.check_busy()
240         def _set_busy():
241             self._busy = 'receive-objects-v2'
242             self.conn.write('receive-objects-v2\n')
243         return PackWriter_Remote(self.conn,
244                                  objcache_maker = self._make_objcache,
245                                  suggest_packs = self._suggest_packs,
246                                  onopen = _set_busy,
247                                  onclose = self._not_busy,
248                                  ensure_busy = self.ensure_busy)
249
250     def read_ref(self, refname):
251         self.check_busy()
252         self.conn.write('read-ref %s\n' % refname)
253         r = self.conn.readline().strip()
254         self.check_ok()
255         if r:
256             assert(len(r) == 40)   # hexified sha
257             return r.decode('hex')
258         else:
259             return None   # nonexistent ref
260
261     def update_ref(self, refname, newval, oldval):
262         self.check_busy()
263         self.conn.write('update-ref %s\n%s\n%s\n' 
264                         % (refname, newval.encode('hex'),
265                            (oldval or '').encode('hex')))
266         self.check_ok()
267
268     def cat(self, id):
269         self.check_busy()
270         self._busy = 'cat'
271         self.conn.write('cat %s\n' % re.sub(r'[\n\r]', '_', id))
272         while 1:
273             sz = struct.unpack('!I', self.conn.read(4))[0]
274             if not sz: break
275             yield self.conn.read(sz)
276         e = self.check_ok()
277         self._not_busy()
278         if e:
279             raise KeyError(str(e))
280
281
282 class PackWriter_Remote(git.PackWriter):
283     def __init__(self, conn, objcache_maker, suggest_packs,
284                  onopen, onclose,
285                  ensure_busy):
286         git.PackWriter.__init__(self, objcache_maker)
287         self.file = conn
288         self.filename = 'remote socket'
289         self.suggest_packs = suggest_packs
290         self.onopen = onopen
291         self.onclose = onclose
292         self.ensure_busy = ensure_busy
293         self._packopen = False
294         self._bwcount = 0
295         self._bwtime = time.time()
296
297     def _open(self):
298         if not self._packopen:
299             self.onopen()
300             self._packopen = True
301
302     def _end(self):
303         if self._packopen and self.file:
304             self.file.write('\0\0\0\0')
305             self._packopen = False
306             self.onclose() # Unbusy
307             self.objcache = None
308             return self.suggest_packs() # Returns last idx received
309
310     def close(self):
311         id = self._end()
312         self.file = None
313         return id
314
315     def abort(self):
316         raise ClientError("don't know how to abort remote pack writing")
317
318     def _raw_write(self, datalist, sha):
319         assert(self.file)
320         if not self._packopen:
321             self._open()
322         self.ensure_busy()
323         data = ''.join(datalist)
324         assert(data)
325         assert(sha)
326         crc = zlib.crc32(data) & 0xffffffff
327         outbuf = ''.join((struct.pack('!I', len(data) + 20 + 4),
328                           sha,
329                           struct.pack('!I', crc),
330                           data))
331         try:
332             (self._bwcount, self._bwtime) = _raw_write_bwlimit(
333                     self.file, outbuf, self._bwcount, self._bwtime)
334         except IOError, e:
335             raise ClientError, e, sys.exc_info()[2]
336         self.outbytes += len(data)
337         self.count += 1
338
339         if self.file.has_input():
340             self.suggest_packs()
341             self.objcache.refresh()
342
343         return sha, crc