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     )
  32 from dulwich.protocol import (
  33     Protocol,
  34     TCP_GIT_PORT,
  35     extract_capabilities,
  36     )
  37 from dulwich.pack import (
  38     write_pack_data,
  39     )
  40
  41
  42 def _fileno_can_read(fileno):
  43     """Check if a file descriptor is readable."""
  44     return len(select.select([fileno], [], [], 0)[0]) > 0
  45
  46
  47 CAPABILITIES = ["multi_ack", "side-band-64k", "ofs-delta"]
  48
  49
  50 class GitClient(object):
  51     """Git smart server client.
  52
  53     """
  54
  55     def __init__(self, can_read, read, write, thin_packs=True,
  56         report_activity=None):
  57         """Create a new GitClient instance.
  58
  59         :param can_read: Function that returns True if there is data available
  60             to be read.
  61         :param read: Callback for reading data, takes number of bytes to read
  62         :param write: Callback for writing data
  63         :param thin_packs: Whether or not thin packs should be retrieved
  64         :param report_activity: Optional callback for reporting transport
  65             activity.
  66         """
  67         self.proto = Protocol(read, write, report_activity)
  68         self._can_read = can_read
  69         self._capabilities = list(CAPABILITIES)
  70         if thin_packs:
  71             self._capabilities.append("thin-pack")
  72
  73     def capabilities(self):
  74         return " ".join(self._capabilities)
  75
  76     def read_refs(self):
  77         server_capabilities = None
  78         refs = {}
  79         # Receive refs from server
  80         for pkt in self.proto.read_pkt_seq():
  81             (sha, ref) = pkt.rstrip("\n").split(" ", 1)
  82             if server_capabilities is None:
  83                 (ref, server_capabilities) = extract_capabilities(ref)
  84             refs[ref] = sha
  85         return refs, server_capabilities
  86
  87     def send_pack(self, path, determine_wants, generate_pack_contents):
  88         """Upload a pack to a remote repository.
  89
  90         :param path: Repository path
  91         :param generate_pack_contents: Function that can return the shas of the
  92             objects to upload.
  93         """
  94         old_refs, server_capabilities = self.read_refs()
  95         new_refs = determine_wants(old_refs)
  96         if not new_refs:
  97             self.proto.write_pkt_line(None)
  98             return {}
  99         want = []
 100         have = [x for x in old_refs.values() if not x == "0" * 40]
 101         sent_capabilities = False
 102         for refname in set(new_refs.keys() + old_refs.keys()):
 103             old_sha1 = old_refs.get(refname, "0" * 40)
 104             new_sha1 = new_refs.get(refname, "0" * 40)
 105             if old_sha1 != new_sha1:
 106                 if sent_capabilities:
 107                     self.proto.write_pkt_line("%s %s %s" % (old_sha1, new_sha1, refname))
 108                 else:
 109                     self.proto.write_pkt_line("%s %s %s\0%s" % (old_sha1, new_sha1, refname, self.capabilities()))
 110                     sent_capabilities = True
 111             if not new_sha1 in (have, "0" * 40):
 112                 want.append(new_sha1)
 113         self.proto.write_pkt_line(None)
 114         if not want:
 115             return new_refs
 116         objects = generate_pack_contents(have, want)
 117         (entries, sha) = write_pack_data(self.proto.write_file(), objects,
 118                                          len(objects))
 119
 120         # read the final confirmation sha
 121         client_sha = self.proto.read(20)
 122         if not client_sha in (None, "", sha):
 123             raise ChecksumMismatch(sha, client_sha)
 124
 125         return new_refs
 126
 127     def fetch_pack(self, path, determine_wants, graph_walker, pack_data,
 128                    progress):
 129         """Retrieve a pack from a git smart server.
 130
 131         :param determine_wants: Callback that returns list of commits to fetch
 132         :param graph_walker: Object with next() and ack().
 133         :param pack_data: Callback called for each bit of data in the pack
 134         :param progress: Callback for progress reports (strings)
 135         """
 136         (refs, server_capabilities) = self.read_refs()
 137         wants = determine_wants(refs)
 138         if not wants:
 139             self.proto.write_pkt_line(None)
 140             return refs
 141         assert isinstance(wants, list) and type(wants[0]) == str
 142         self.proto.write_pkt_line("want %s %s\n" % (wants[0], self.capabilities()))
 143         for want in wants[1:]:
 144             self.proto.write_pkt_line("want %s\n" % want)
 145         self.proto.write_pkt_line(None)
 146         have = graph_walker.next()
 147         while have:
 148             self.proto.write_pkt_line("have %s\n" % have)
 149             if self._can_read():
 150                 pkt = self.proto.read_pkt_line()
 151                 parts = pkt.rstrip("\n").split(" ")
 152                 if parts[0] == "ACK":
 153                     graph_walker.ack(parts[1])
 154                     assert parts[2] == "continue"
 155             have = graph_walker.next()
 156         self.proto.write_pkt_line("done\n")
 157         pkt = self.proto.read_pkt_line()
 158         while pkt:
 159             parts = pkt.rstrip("\n").split(" ")
 160             if parts[0] == "ACK":
 161                 graph_walker.ack(pkt.split(" ")[1])
 162             if len(parts) < 3 or parts[2] != "continue":
 163                 break
 164             pkt = self.proto.read_pkt_line()
 165         for pkt in self.proto.read_pkt_seq():
 166             channel = ord(pkt[0])
 167             pkt = pkt[1:]
 168             if channel == 1:
 169                 pack_data(pkt)
 170             elif channel == 2:
 171                 progress(pkt)
 172             else:
 173                 raise AssertionError("Invalid sideband channel %d" % channel)
 174         return refs
 175
 176
 177 class TCPGitClient(GitClient):
 178     """A Git Client that works over TCP directly (i.e. git://)."""
 179
 180     def __init__(self, host, port=None, *args, **kwargs):
 181         self._socket = socket.socket(type=socket.SOCK_STREAM)
 182         if port is None:
 183             port = TCP_GIT_PORT
 184         self._socket.connect((host, port))
 185         self.rfile = self._socket.makefile('rb', -1)
 186         self.wfile = self._socket.makefile('wb', 0)
 187         self.host = host
 188         super(TCPGitClient, self).__init__(lambda: _fileno_can_read(self._socket.fileno()), self.rfile.read, self.wfile.write, *args, **kwargs)
 189
 190     def send_pack(self, path, changed_refs, generate_pack_contents):
 191         """Send a pack to a remote host.
 192
 193         :param path: Path of the repository on the remote host
 194         """
 195         self.proto.send_cmd("git-receive-pack", path, "host=%s" % self.host)
 196         return super(TCPGitClient, self).send_pack(path, changed_refs, generate_pack_contents)
 197
 198     def fetch_pack(self, path, determine_wants, graph_walker, pack_data, progress):
 199         """Fetch a pack from the remote host.
 200
 201         :param path: Path of the reposiutory on the remote host
 202         :param determine_wants: Callback that receives available refs dict and
 203             should return list of sha's to fetch.
 204         :param graph_walker: GraphWalker instance used to find missing shas
 205         :param pack_data: Callback for writing pack data
 206         :param progress: Callback for writing progress
 207         """
 208         self.proto.send_cmd("git-upload-pack", path, "host=%s" % self.host)
 209         return super(TCPGitClient, self).fetch_pack(path, determine_wants,
 210             graph_walker, pack_data, progress)
 211
 212
 213 class SubprocessGitClient(GitClient):
 214     """Git client that talks to a server using a subprocess."""
 215
 216     def __init__(self, *args, **kwargs):
 217         self.proc = None
 218         self._args = args
 219         self._kwargs = kwargs
 220
 221     def _connect(self, service, *args, **kwargs):
 222         argv = [service] + list(args)
 223         self.proc = subprocess.Popen(argv, bufsize=0,
 224                                 stdin=subprocess.PIPE,
 225                                 stdout=subprocess.PIPE)
 226         def read_fn(size):
 227             return self.proc.stdout.read(size)
 228         def write_fn(data):
 229             self.proc.stdin.write(data)
 230             self.proc.stdin.flush()
 231         return GitClient(lambda: _fileno_can_read(self.proc.stdout.fileno()), read_fn, write_fn, *args, **kwargs)
 232
 233     def send_pack(self, path, changed_refs, generate_pack_contents):
 234         """Upload a pack to the server.
 235
 236         :param path: Path to the git repository on the server
 237         :param changed_refs: Dictionary with new values for the refs
 238         :param generate_pack_contents: Function that returns an iterator over
 239             objects to send
 240         """
 241         client = self._connect("git-receive-pack", path)
 242         return client.send_pack(path, changed_refs, generate_pack_contents)
 243
 244     def fetch_pack(self, path, determine_wants, graph_walker, pack_data,
 245         progress):
 246         """Retrieve a pack from the server
 247
 248         :param path: Path to the git repository on the server
 249         :param determine_wants: Function that receives existing refs
 250             on the server and returns a list of desired shas
 251         :param graph_walker: GraphWalker instance
 252         :param pack_data: Function that can write pack data
 253         :param progress: Function that can write progress texts
 254         """
 255         client = self._connect("git-upload-pack", path)
 256         return client.fetch_pack(path, determine_wants, graph_walker, pack_data,
 257                                  progress)
 258
 259
 260 class SSHSubprocess(object):
 261     """A socket-like object that talks to an ssh subprocess via pipes."""
 262
 263     def __init__(self, proc):
 264         self.proc = proc
 265
 266     def send(self, data):
 267         return os.write(self.proc.stdin.fileno(), data)
 268
 269     def recv(self, count):
 270         return self.proc.stdout.read(count)
 271
 272     def close(self):
 273         self.proc.stdin.close()
 274         self.proc.stdout.close()
 275         self.proc.wait()
 276
 277
 278 class SSHVendor(object):
 279
 280     def connect_ssh(self, host, command, username=None, port=None):
 281         #FIXME: This has no way to deal with passwords..
 282         args = ['ssh', '-x']
 283         if port is not None:
 284             args.extend(['-p', str(port)])
 285         if username is not None:
 286             host = "%s@%s" % (username, host)
 287         args.append(host)
 288         proc = subprocess.Popen(args + command,
 289                                 stdin=subprocess.PIPE,
 290                                 stdout=subprocess.PIPE)
 291         return SSHSubprocess(proc)
 292
 293 # Can be overridden by users
 294 get_ssh_vendor = SSHVendor
 295
 296
 297 class SSHGitClient(GitClient):
 298
 299     def __init__(self, host, port=None, username=None, *args, **kwargs):
 300         self.host = host
 301         self.port = port
 302         self.username = username
 303         self._args = args
 304         self._kwargs = kwargs
 305
 306     def send_pack(self, path, determine_wants, generate_pack_contents):
 307         remote = get_ssh_vendor().connect_ssh(self.host, ["git-receive-pack '%s'" % path], port=self.port, username=self.username)
 308         client = GitClient(lambda: _fileno_can_read(remote.proc.stdout.fileno()), remote.recv, remote.send, *self._args, **self._kwargs)
 309         return client.send_pack(path, determine_wants, generate_pack_contents)
 310
 311     def fetch_pack(self, path, determine_wants, graph_walker, pack_data,
 312         progress):
 313         remote = get_ssh_vendor().connect_ssh(self.host, ["git-upload-pack '%s'" % path], port=self.port, username=self.username)
 314         client = GitClient(lambda: _fileno_can_read(remote.proc.stdout.fileno()), remote.recv, remote.send, *self._args, **self._kwargs)
 315         return client.fetch_pack(path, determine_wants, graph_walker, pack_data,
 316                                  progress)
 317