dulwich/client.py

   1 # client.py -- Implementation of the server side git protocols
   2 # Copyright (C) 2008-2009 Jelmer Vernooij <jelmer@samba.org>
   3 # Copyright (C) 2008 John Carr
   4 #
   5 # This program is free software; you can redistribute it and/or
   6 # modify it under the terms of the GNU General Public License
   7 # as published by the Free Software Foundation; either version 2
   8 # or (at your option) a later version of the License.
   9 #
  10 # This program is distributed in the hope that it will be useful,
  11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13 # GNU General Public License for more details.
  14 #
  15 # You should have received a copy of the GNU General Public License
  16 # along with this program; if not, write to the Free Software
  17 # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston,
  18 # MA  02110-1301, USA.
  19
  20 """Client side support for the Git protocol."""
  21
  22 __docformat__ = 'restructuredText'
  23
  24 import os
  25 import select
  26 import socket
  27 import subprocess
  28
  29 from dulwich.errors import (
  30     ChecksumMismatch,
  31     SendPackError,
  32     UpdateRefsError,
  33     )
  34 from dulwich.protocol import (
  35     Protocol,
  36     TCP_GIT_PORT,
  37     ZERO_SHA,
  38     extract_capabilities,
  39     )
  40 from dulwich.pack import (
  41     write_pack_data,
  42     )
  43
  44
  45 def _fileno_can_read(fileno):
  46     """Check if a file descriptor is readable."""
  47     return len(select.select([fileno], [], [], 0)[0]) > 0
  48
  49 COMMON_CAPABILITIES = ["ofs-delta"]
  50 FETCH_CAPABILITIES = ["multi_ack", "side-band-64k"] + COMMON_CAPABILITIES
  51 SEND_CAPABILITIES = ['report-status'] + COMMON_CAPABILITIES
  52
  53 # TODO(durin42): this doesn't correctly degrade if the server doesn't
  54 # support some capabilities. This should work properly with servers
  55 # that don't support side-band-64k and multi_ack.
  56 class GitClient(object):
  57     """Git smart server client.
  58
  59     """
  60
  61     def __init__(self, can_read, read, write, thin_packs=True,
  62         report_activity=None):
  63         """Create a new GitClient instance.
  64
  65         :param can_read: Function that returns True if there is data available
  66             to be read.
  67         :param read: Callback for reading data, takes number of bytes to read
  68         :param write: Callback for writing data
  69         :param thin_packs: Whether or not thin packs should be retrieved
  70         :param report_activity: Optional callback for reporting transport
  71             activity.
  72         """
  73         self.proto = Protocol(read, write, report_activity)
  74         self._can_read = can_read
  75         self._fetch_capabilities = list(FETCH_CAPABILITIES)
  76         self._send_capabilities = list(SEND_CAPABILITIES)
  77         if thin_packs:
  78             self._fetch_capabilities.append("thin-pack")
  79
  80     def read_refs(self):
  81         server_capabilities = None
  82         refs = {}
  83         # Receive refs from server
  84         for pkt in self.proto.read_pkt_seq():
  85             (sha, ref) = pkt.rstrip("\n").split(" ", 1)
  86             if server_capabilities is None:
  87                 (ref, server_capabilities) = extract_capabilities(ref)
  88             refs[ref] = sha
  89         return refs, server_capabilities
  90
  91     def _parse_status_report(self):
  92         unpack = self.proto.read_pkt_line().strip()
  93         if unpack != 'unpack ok':
  94             st = True
  95             # flush remaining error data
  96             while st is not None:
  97                 st = self.proto.read_pkt_line()
  98             raise SendPackError(unpack)
  99         statuses = []
 100         errs = False
 101         ref_status = self.proto.read_pkt_line()
 102         while ref_status:
 103             ref_status = ref_status.strip()
 104             statuses.append(ref_status)
 105             if not ref_status.startswith('ok '):
 106                 errs = True
 107             ref_status = self.proto.read_pkt_line()
 108
 109         if errs:
 110             ref_status = {}
 111             ok = set()
 112             for status in statuses:
 113                 if ' ' not in status:
 114                     # malformed response, move on to the next one
 115                     continue
 116                 status, ref = status.split(' ', 1)
 117
 118                 if status == 'ng':
 119                     if ' ' in ref:
 120                         ref, status = ref.split(' ', 1)
 121                 else:
 122                     ok.add(ref)
 123                 ref_status[ref] = status
 124             raise UpdateRefsError('%s failed to update' %
 125                                   ', '.join([ref for ref in ref_status
 126                                              if ref not in ok]),
 127                                   ref_status=ref_status)
 128
 129
 130     # TODO(durin42): add side-band-64k capability support here and advertise it
 131     def send_pack(self, path, determine_wants, generate_pack_contents):
 132         """Upload a pack to a remote repository.
 133
 134         :param path: Repository path
 135         :param generate_pack_contents: Function that can return the shas of the
 136             objects to upload.
 137
 138         :raises SendPackError: if server rejects the pack data
 139         :raises UpdateRefsError: if the server supports report-status
 140                                  and rejects ref updates
 141         """
 142         old_refs, server_capabilities = self.read_refs()
 143         if 'report-status' not in server_capabilities:
 144             self._send_capabilities.remove('report-status')
 145         new_refs = determine_wants(old_refs)
 146         if not new_refs:
 147             self.proto.write_pkt_line(None)
 148             return {}
 149         want = []
 150         have = [x for x in old_refs.values() if not x == ZERO_SHA]
 151         sent_capabilities = False
 152         for refname in set(new_refs.keys() + old_refs.keys()):
 153             old_sha1 = old_refs.get(refname, ZERO_SHA)
 154             new_sha1 = new_refs.get(refname, ZERO_SHA)
 155             if old_sha1 != new_sha1:
 156                 if sent_capabilities:
 157                     self.proto.write_pkt_line("%s %s %s" % (old_sha1, new_sha1,
 158                                                             refname))
 159                 else:
 160                     self.proto.write_pkt_line(
 161                       "%s %s %s\0%s" % (old_sha1, new_sha1, refname,
 162                                         ' '.join(self._send_capabilities)))
 163                     sent_capabilities = True
 164             if new_sha1 not in have and new_sha1 != ZERO_SHA:
 165                 want.append(new_sha1)
 166         self.proto.write_pkt_line(None)
 167         if not want:
 168             return new_refs
 169         objects = generate_pack_contents(have, want)
 170         (entries, sha) = write_pack_data(self.proto.write_file(), objects,
 171                                          len(objects))
 172
 173         if 'report-status' in self._send_capabilities:
 174             self._parse_status_report()
 175         # wait for EOF before returning
 176         data = self.proto.read()
 177         if data:
 178             raise SendPackError('Unexpected response %r' % data)
 179         return new_refs
 180
 181     def fetch(self, path, target, determine_wants=None, progress=None):
 182         """Fetch into a target repository.
 183
 184         :param path: Path to fetch from
 185         :param target: Target repository to fetch into
 186         :param determine_wants: Optional function to determine what refs
 187             to fetch
 188         :param progress: Optional progress function
 189         :return: remote refs
 190         """
 191         if determine_wants is None:
 192             determine_wants = target.object_store.determine_wants_all
 193         f, commit = target.object_store.add_pack()
 194         try:
 195             return self.fetch_pack(path, determine_wants,
 196                 target.get_graph_walker(), f.write, progress)
 197         finally:
 198             commit()
 199
 200     def fetch_pack(self, path, determine_wants, graph_walker, pack_data,
 201                    progress):
 202         """Retrieve a pack from a git smart server.
 203
 204         :param determine_wants: Callback that returns list of commits to fetch
 205         :param graph_walker: Object with next() and ack().
 206         :param pack_data: Callback called for each bit of data in the pack
 207         :param progress: Callback for progress reports (strings)
 208         """
 209         (refs, server_capabilities) = self.read_refs()
 210         wants = determine_wants(refs)
 211         if not wants:
 212             self.proto.write_pkt_line(None)
 213             return refs
 214         assert isinstance(wants, list) and type(wants[0]) == str
 215         self.proto.write_pkt_line("want %s %s\n" % (
 216             wants[0], ' '.join(self._fetch_capabilities)))
 217         for want in wants[1:]:
 218             self.proto.write_pkt_line("want %s\n" % want)
 219         self.proto.write_pkt_line(None)
 220         have = graph_walker.next()
 221         while have:
 222             self.proto.write_pkt_line("have %s\n" % have)
 223             if self._can_read():
 224                 pkt = self.proto.read_pkt_line()
 225                 parts = pkt.rstrip("\n").split(" ")
 226                 if parts[0] == "ACK":
 227                     graph_walker.ack(parts[1])
 228                     assert parts[2] == "continue"
 229             have = graph_walker.next()
 230         self.proto.write_pkt_line("done\n")
 231         pkt = self.proto.read_pkt_line()
 232         while pkt:
 233             parts = pkt.rstrip("\n").split(" ")
 234             if parts[0] == "ACK":
 235                 graph_walker.ack(pkt.split(" ")[1])
 236             if len(parts) < 3 or parts[2] != "continue":
 237                 break
 238             pkt = self.proto.read_pkt_line()
 239         # TODO(durin42): this is broken if the server didn't support the
 240         # side-band-64k capability.
 241         for pkt in self.proto.read_pkt_seq():
 242             channel = ord(pkt[0])
 243             pkt = pkt[1:]
 244             if channel == 1:
 245                 pack_data(pkt)
 246             elif channel == 2:
 247                 if progress is not None:
 248                     progress(pkt)
 249             else:
 250                 raise AssertionError("Invalid sideband channel %d" % channel)
 251         return refs
 252
 253
 254 class TCPGitClient(GitClient):
 255     """A Git Client that works over TCP directly (i.e. git://)."""
 256
 257     def __init__(self, host, port=None, *args, **kwargs):
 258         self._socket = socket.socket(type=socket.SOCK_STREAM)
 259         if port is None:
 260             port = TCP_GIT_PORT
 261         self._socket.connect((host, port))
 262         self.rfile = self._socket.makefile('rb', -1)
 263         self.wfile = self._socket.makefile('wb', 0)
 264         self.host = host
 265         super(TCPGitClient, self).__init__(lambda: _fileno_can_read(self._socket.fileno()), self.rfile.read, self.wfile.write, *args, **kwargs)
 266
 267     def send_pack(self, path, changed_refs, generate_pack_contents):
 268         """Send a pack to a remote host.
 269
 270         :param path: Path of the repository on the remote host
 271         """
 272         self.proto.send_cmd("git-receive-pack", path, "host=%s" % self.host)
 273         return super(TCPGitClient, self).send_pack(path, changed_refs, generate_pack_contents)
 274
 275     def fetch_pack(self, path, determine_wants, graph_walker, pack_data, progress):
 276         """Fetch a pack from the remote host.
 277
 278         :param path: Path of the reposiutory on the remote host
 279         :param determine_wants: Callback that receives available refs dict and
 280             should return list of sha's to fetch.
 281         :param graph_walker: GraphWalker instance used to find missing shas
 282         :param pack_data: Callback for writing pack data
 283         :param progress: Callback for writing progress
 284         """
 285         self.proto.send_cmd("git-upload-pack", path, "host=%s" % self.host)
 286         return super(TCPGitClient, self).fetch_pack(path, determine_wants,
 287             graph_walker, pack_data, progress)
 288
 289
 290 class SubprocessGitClient(GitClient):
 291     """Git client that talks to a server using a subprocess."""
 292
 293     def __init__(self, *args, **kwargs):
 294         self.proc = None
 295         self._args = args
 296         self._kwargs = kwargs
 297
 298     def _connect(self, service, *args, **kwargs):
 299         argv = [service] + list(args)
 300         self.proc = subprocess.Popen(argv, bufsize=0,
 301                                 stdin=subprocess.PIPE,
 302                                 stdout=subprocess.PIPE)
 303         def read_fn(size):
 304             return self.proc.stdout.read(size)
 305         def write_fn(data):
 306             self.proc.stdin.write(data)
 307             self.proc.stdin.flush()
 308         return GitClient(lambda: _fileno_can_read(self.proc.stdout.fileno()), read_fn, write_fn, *args, **kwargs)
 309
 310     def send_pack(self, path, changed_refs, generate_pack_contents):
 311         """Upload a pack to the server.
 312
 313         :param path: Path to the git repository on the server
 314         :param changed_refs: Dictionary with new values for the refs
 315         :param generate_pack_contents: Function that returns an iterator over
 316             objects to send
 317         """
 318         client = self._connect("git-receive-pack", path)
 319         return client.send_pack(path, changed_refs, generate_pack_contents)
 320
 321     def fetch_pack(self, path, determine_wants, graph_walker, pack_data,
 322         progress):
 323         """Retrieve a pack from the server
 324
 325         :param path: Path to the git repository on the server
 326         :param determine_wants: Function that receives existing refs
 327             on the server and returns a list of desired shas
 328         :param graph_walker: GraphWalker instance
 329         :param pack_data: Function that can write pack data
 330         :param progress: Function that can write progress texts
 331         """
 332         client = self._connect("git-upload-pack", path)
 333         return client.fetch_pack(path, determine_wants, graph_walker, pack_data,
 334                                  progress)
 335
 336
 337 class SSHSubprocess(object):
 338     """A socket-like object that talks to an ssh subprocess via pipes."""
 339
 340     def __init__(self, proc):
 341         self.proc = proc
 342         self.read = self.recv = proc.stdout.read
 343         self.write = self.send = proc.stdin.write
 344
 345     def close(self):
 346         self.proc.stdin.close()
 347         self.proc.stdout.close()
 348         self.proc.wait()
 349
 350
 351 class SSHVendor(object):
 352
 353     def connect_ssh(self, host, command, username=None, port=None):
 354         #FIXME: This has no way to deal with passwords..
 355         args = ['ssh', '-x']
 356         if port is not None:
 357             args.extend(['-p', str(port)])
 358         if username is not None:
 359             host = "%s@%s" % (username, host)
 360         args.append(host)
 361         proc = subprocess.Popen(args + command,
 362                                 stdin=subprocess.PIPE,
 363                                 stdout=subprocess.PIPE)
 364         return SSHSubprocess(proc)
 365
 366 # Can be overridden by users
 367 get_ssh_vendor = SSHVendor
 368
 369
 370 class SSHGitClient(GitClient):
 371
 372     def __init__(self, host, port=None, username=None, *args, **kwargs):
 373         self.host = host
 374         self.port = port
 375         self.username = username
 376         self._args = args
 377         self._kwargs = kwargs
 378
 379     def send_pack(self, path, determine_wants, generate_pack_contents):
 380         remote = get_ssh_vendor().connect_ssh(
 381             self.host, ["git-receive-pack '%s'" % path],
 382             port=self.port, username=self.username)
 383         client = GitClient(lambda: _fileno_can_read(remote.proc.stdout.fileno()), remote.recv, remote.send, *self._args, **self._kwargs)
 384         return client.send_pack(path, determine_wants, generate_pack_contents)
 385
 386     def fetch_pack(self, path, determine_wants, graph_walker, pack_data,
 387         progress):
 388         remote = get_ssh_vendor().connect_ssh(self.host, ["git-upload-pack '%s'" % path], port=self.port, username=self.username)
 389         client = GitClient(lambda: _fileno_can_read(remote.proc.stdout.fileno()), remote.recv, remote.send, *self._args, **self._kwargs)
 390         return client.fetch_pack(path, determine_wants, graph_walker, pack_data,
 391                                  progress)
 392
 393
 394 def get_transport_and_path(uri):
 395     """Obtain a git client from a URI or path.
 396
 397     :param uri: URI or path
 398     :return: Tuple with client instance and relative path.
 399     """
 400     from dulwich.client import TCPGitClient, SSHGitClient, SubprocessGitClient
 401     for handler, transport in (("git://", TCPGitClient), ("git+ssh://", SSHGitClient)):
 402         if uri.startswith(handler):
 403             host, path = uri[len(handler):].split("/", 1)
 404             return transport(host), "/"+path
 405     # if its not git or git+ssh, try a local url..
 406     return SubprocessGitClient(), uri