14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
18
import urllib, urllib2
20
from StringIO import StringIO
22
import bzrlib # for the version
23
from bzrlib.errors import (TransportNotPossible, NoSuchFile, BzrError,
24
TransportError, ConnectionError)
25
from bzrlib.trace import mutter
26
from bzrlib.transport import register_urlparse_netloc_protocol
27
from bzrlib.transport.http import (HttpTransportBase, HttpServer,
28
extract_auth, response)
30
register_urlparse_netloc_protocol('http+urllib')
33
class Request(urllib2.Request):
34
"""Request object for urllib2 that allows the method to be overridden."""
39
if self.method is not None:
42
return urllib2.Request.get_method(self)
45
class HttpTransport_urllib(HttpTransportBase):
17
from cStringIO import StringIO
26
from bzrlib.transport import http
27
# TODO: handle_response should be integrated into the http/__init__.py
28
from bzrlib.transport.http.response import handle_response
29
from bzrlib.transport.http._urllib2_wrappers import (
35
class HttpTransport_urllib(http.HttpTransportBase):
46
36
"""Python urllib transport for http and https."""
48
# TODO: Implement pipelined versions of all of the *_multi() functions.
50
def __init__(self, base, from_transport=None):
51
"""Set the base path where files will be stored."""
52
super(HttpTransport_urllib, self).__init__(base)
53
# HttpTransport_urllib doesn't maintain any per-transport state yet
54
# so nothing to do with from_transport
56
def _get(self, relpath, ranges, tail_amount=0):
59
path = self._real_abspath(relpath)
60
resp = self._get_url_impl(path, method='GET', ranges=ranges,
61
tail_amount=tail_amount)
62
return resp.code, response.handle_response(path,
63
resp.code, resp.headers, resp)
64
except urllib2.HTTPError, e:
65
mutter('url error code: %s for has url: %r', e.code, path)
67
raise NoSuchFile(path, extra=e)
69
except (BzrError, IOError), e:
70
if getattr(e, 'errno', None) is not None:
71
mutter('io error: %s %s for has url: %r',
72
e.errno, errno.errorcode.get(e.errno), path)
73
if e.errno == errno.ENOENT:
74
raise NoSuchFile(path, extra=e)
75
raise ConnectionError(msg = "Error retrieving %s: %s"
76
% (self.abspath(relpath), str(e)),
79
def _get_url_impl(self, url, method, ranges, tail_amount=0):
80
"""Actually pass get request into urllib
82
:returns: urllib Response object
84
manager = urllib2.HTTPPasswordMgrWithDefaultRealm()
85
url = extract_auth(url, manager)
86
auth_handler = urllib2.HTTPBasicAuthHandler(manager)
87
opener = urllib2.build_opener(auth_handler)
88
request = Request(url)
89
request.method = method
90
request.add_header('Pragma', 'no-cache')
91
request.add_header('Cache-control', 'max-age=0')
92
request.add_header('User-Agent',
93
'bzr/%s (urllib)' % (bzrlib.__version__,))
94
if ranges or tail_amount:
95
bytes = 'bytes=' + self.range_header(ranges, tail_amount)
96
request.add_header('Range', bytes)
97
response = opener.open(request)
100
def should_cache(self):
101
"""Return True if the data pulled across should be cached locally.
38
# In order to debug we have to issue our traces in sync with
39
# httplib, which use print :(
42
_opener_class = Opener
44
def __init__(self, base, _from_transport=None):
45
super(HttpTransport_urllib, self).__init__(
46
base, _from_transport=_from_transport)
47
if _from_transport is not None:
48
self._opener = _from_transport._opener
50
self._opener = self._opener_class()
52
def _perform(self, request):
53
"""Send the request to the server and handles common errors.
55
:returns: urllib2 Response object
57
connection = self._get_connection()
58
if connection is not None:
59
# Give back shared info
60
request.connection = connection
61
(auth, proxy_auth) = self._get_credentials()
62
# Clean the httplib.HTTPConnection pipeline in case the previous
63
# request couldn't do it
64
connection.cleanup_pipe()
66
# First request, initialize credentials.
67
# scheme and realm will be set by the _urllib2_wrappers.AuthHandler
68
auth = self._create_auth()
69
# Proxy initialization will be done by the first proxied request
71
# Ensure authentication info is provided
73
request.proxy_auth = proxy_auth
75
if self._debuglevel > 0:
76
print 'perform: %s base: %s, url: %s' % (request.method, self.base,
77
request.get_full_url())
78
response = self._opener.open(request)
79
if self._get_connection() is not request.connection:
80
# First connection or reconnection
81
self._set_connection(request.connection,
82
(request.auth, request.proxy_auth))
84
# http may change the credentials while keeping the
86
self._update_credentials((request.auth, request.proxy_auth))
89
if request.follow_redirections is False \
90
and code in (301, 302, 303, 307):
91
raise errors.RedirectRequested(request.get_full_url(),
92
request.redirected_to,
93
is_permanent=(code == 301),
94
qual_proto=self._scheme)
96
if request.redirected_to is not None:
97
trace.mutter('redirected from: %s to: %s' % (request.get_full_url(),
98
request.redirected_to))
102
def _get(self, relpath, offsets, tail_amount=0):
103
"""See HttpTransport._get"""
105
abspath = self._remote_path(relpath)
107
accepted_errors = [200, 404]
108
if offsets or tail_amount:
109
range_header = self._attempted_range_header(offsets, tail_amount)
110
if range_header is not None:
111
accepted_errors.append(206)
112
accepted_errors.append(400)
113
accepted_errors.append(416)
114
bytes = 'bytes=' + range_header
115
headers = {'Range': bytes}
117
request = Request('GET', abspath, None, headers,
118
accepted_errors=accepted_errors)
119
response = self._perform(request)
122
if code == 404: # not found
123
raise errors.NoSuchFile(abspath)
124
elif code in (400, 416):
125
# We don't know which, but one of the ranges we specified was
127
raise errors.InvalidHttpRange(abspath, range_header,
128
'Server return code %d' % code)
130
data = handle_response(abspath, code, response.info(), response)
133
def _post(self, body_bytes):
134
abspath = self._remote_path('.bzr/smart')
135
# We include 403 in accepted_errors so that send_http_smart_request can
136
# handle a 403. Otherwise a 403 causes an unhandled TransportError.
137
response = self._perform(Request('POST', abspath, body_bytes,
138
accepted_errors=[200, 403]))
140
data = handle_response(abspath, code, response.info(), response)
143
def _head(self, relpath):
144
"""Request the HEAD of a file.
146
Performs the request and leaves callers handle the results.
148
abspath = self._remote_path(relpath)
149
request = Request('HEAD', abspath,
150
accepted_errors=[200, 404])
151
response = self._perform(request)
105
155
def has(self, relpath):
106
156
"""Does the target location exist?
108
abspath = self._real_abspath(relpath)
110
f = self._get_url_impl(abspath, 'HEAD', [])
111
# Without the read and then close()
112
# we tend to have busy sockets.
158
response = self._head(relpath)
161
if code == 200: # "ok",
116
except urllib2.HTTPError, e:
117
mutter('url error code: %s, for has url: %r', e.code, abspath)
121
except urllib2.URLError, e:
122
mutter('url error: %s, for has url: %r', e.reason, abspath)
125
mutter('io error: %s %s for has url: %r',
126
e.errno, errno.errorcode.get(e.errno), abspath)
127
if e.errno == errno.ENOENT:
129
raise TransportError(orig_error=e)
131
def copy_to(self, relpaths, other, mode=None, pb=None):
132
"""Copy a set of entries from self into another Transport.
134
:param relpaths: A list/generator of entries to be copied.
136
TODO: if other is LocalTransport, is it possible to
137
do better than put(get())?
139
# At this point HttpTransport_urllib might be able to check and see if
140
# the remote location is the same, and rather than download, and
141
# then upload, it could just issue a remote copy_this command.
142
if isinstance(other, HttpTransport_urllib):
143
raise TransportNotPossible('http cannot be the target of copy_to()')
145
return super(HttpTransport_urllib, self).copy_to(relpaths, other, mode=mode, pb=pb)
147
def move(self, rel_from, rel_to):
148
"""Move the item at rel_from to the location at rel_to"""
149
raise TransportNotPossible('http does not support move()')
151
def delete(self, relpath):
152
"""Delete the item at relpath"""
153
raise TransportNotPossible('http does not support delete()')
156
class HttpServer_urllib(HttpServer):
157
"""Subclass of HttpServer that gives http+urllib urls.
159
This is for use in testing: connections to this server will always go
160
through urllib where possible.
163
# urls returned by this server should require the urllib client impl
164
_url_protocol = 'http+urllib'
167
167
def get_test_permutations():
168
168
"""Return the permutations to be used in testing."""
169
from bzrlib.tests.http_server import HttpServer_urllib
169
170
return [(HttpTransport_urllib, HttpServer_urllib),
added
removed
bzrlib/transport/http/_urllib.py
