15
17
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
18
## XXX: This is pretty slow on high-latency connections because it
19
## doesn't keep the HTTP connection alive. If you have a smart local
20
## proxy it may be much better. Eventually I want to switch to
21
## urlgrabber which should use HTTP much more efficiently.
24
import urllib2, gzip, zlib
20
"""Proxy object for access to remote branches.
22
At the moment remote branches are only for HTTP and only for read
26
28
from cStringIO import StringIO
28
from errors import BzrError
29
from revision import Revision
30
from inventory import Inventory
32
# h = HTTPConnection('localhost:8000')
33
# h = HTTPConnection('bazaar-ng.org')
35
# velocitynet.com.au transparently proxies connections and thereby
36
# breaks keep-alive -- sucks!
39
import urlgrabber.keepalive
40
urlgrabber.keepalive.DEBUG = 2
44
prefix = 'http://localhost:8000'
45
# prefix = 'http://bazaar-ng.org/bzr/main/'
47
def get_url(path, compressed=False):
33
from bzrlib.errors import BzrError, BzrCheckError
34
from bzrlib.branch import Branch, BZR_BRANCH_FORMAT
35
from bzrlib.trace import mutter
36
from bzrlib.xml import serializer_v4
39
ENABLE_URLGRABBER = False
41
from bzrlib.errors import BzrError, NoSuchRevision
43
class GetFailed(BzrError):
44
def __init__(self, url, status):
45
BzrError.__init__(self, "Get %s failed with status %s" % (url, status))
50
import util.urlgrabber
51
import util.urlgrabber.keepalive
52
util.urlgrabber.keepalive.DEBUG = 0
53
def get_url(path, compressed=False):
58
mutter("grab url %s" % url)
59
url_f = util.urlgrabber.urlopen(url, keepalive=1, close_connection=0)
60
if url_f.status != 200:
61
raise GetFailed(url, url_f.status)
65
return gzip.GzipFile(fileobj=StringIO(url_f.read()))
66
except urllib2.URLError, e:
67
raise BzrError("remote fetch failed: %r: %s" % (url, e))
69
def get_url(url, compressed=False):
52
url_f = urlgrabber.urlopen(url, keepalive=1, close_connection=0)
73
mutter("get_url %s" % url)
74
url_f = urllib2.urlopen(url)
76
return gzip.GzipFile(fileobj=StringIO(url_f.read()))
56
return gzip.GzipFile(fileobj=StringIO(url_f.read()))
57
except urllib2.URLError, e:
58
raise BzrError("remote fetch failed: %r: %s" % (url, e))
65
history = get_url('/.bzr/revision-history').readlines()
66
num_revs = len(history)
67
for i, rev_id in enumerate(history):
68
rev_id = rev_id.rstrip()
69
print 'read revision %d/%d' % (i, num_revs)
71
# python gzip needs a seekable file (!!) but the HTTP response
72
# isn't, so we need to buffer it
74
rev_f = get_url('/.bzr/revision-store/%s' % rev_id,
77
rev = Revision.read_xml(rev_f)
79
inv_id = rev.inventory_id
80
if inv_id not in got_invs:
81
print 'get inventory %s' % inv_id
82
inv_f = get_url('/.bzr/inventory-store/%s' % inv_id,
84
inv = Inventory.read_xml(inv_f)
85
print '%4d inventory entries' % len(inv)
87
for path, ie in inv.iter_entries():
91
if text_id in got_texts:
93
print ' fetch %s text {%s}' % (path, text_id)
94
text_f = get_url('/.bzr/text-store/%s' % text_id,
96
got_texts.add(text_id)
82
def _find_remote_root(url):
83
"""Return the prefix URL that corresponds to the branch root."""
87
fmt_url = url + '/.bzr/branch-format'
92
fmt = fmt.rstrip('\r\n')
93
if fmt != BZR_BRANCH_FORMAT.rstrip('\r\n'):
94
raise BzrError("sorry, branch format %r not supported at url %s"
98
except urllib2.URLError:
101
scheme, host, path = list(urlparse.urlparse(url))[:3]
102
# discard params, query, fragment
104
# strip off one component of the path component
105
idx = path.rfind('/')
106
if idx == -1 or path == '/':
107
raise BzrError('no branch root found for URL %s'
108
' or enclosing directories'
111
url = urlparse.urlunparse((scheme, host, path, '', '', ''))
115
class RemoteBranch(Branch):
116
def __init__(self, baseurl, find_root=True):
117
"""Create new proxy for a remote branch."""
119
self.base = _find_remote_root(baseurl)
124
self.inventory_store = RemoteStore(baseurl + '/.bzr/inventory-store/')
125
self.text_store = RemoteStore(baseurl + '/.bzr/text-store/')
126
self.revision_store = RemoteStore(baseurl + '/.bzr/revision-store/')
129
b = getattr(self, 'baseurl', 'undefined')
130
return '%s(%r)' % (self.__class__.__name__, b)
134
def controlfile(self, filename, mode):
135
if mode not in ('rb', 'rt', 'r'):
136
raise BzrError("file mode %r not supported for remote branches" % mode)
137
return get_url(self.base + '/.bzr/' + filename, False)
141
# no locking for remote branches yet
144
def lock_write(self):
145
from errors import LockError
146
raise LockError("write lock not supported for remote branch %s"
153
def relpath(self, path):
154
if not path.startswith(self.base):
155
raise BzrError('path %r is not under base URL %r'
158
return path[pl:].lstrip('/')
161
def get_revision(self, revision_id):
163
revf = self.revision_store[revision_id]
165
raise NoSuchRevision(self, revision_id)
166
r = serializer_v4.read_revision(revf)
167
if r.revision_id != revision_id:
168
raise BzrCheckError('revision stored as {%s} actually contains {%s}'
169
% (revision_id, r.revision_id))
173
class RemoteStore(object):
174
def __init__(self, baseurl):
175
self._baseurl = baseurl
178
def _path(self, name):
180
raise ValueError('invalid store id', name)
181
return self._baseurl + '/' + name
183
def __getitem__(self, fileid):
184
p = self._path(fileid)
186
return get_url(p, compressed=True)
188
raise KeyError(fileid)