# Copyright (C) 2005-2010 Canonical Ltd
#
# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
"""Base implementation of Transport over http.
There are separate implementation modules for each http client implementation.
"""
from cStringIO import StringIO
import mimetools
import re
import urlparse
import urllib
import sys
import weakref
from bzrlib import (
debug,
errors,
ui,
urlutils,
)
from bzrlib.smart import medium
from bzrlib.symbol_versioning import (
deprecated_method,
)
from bzrlib.trace import mutter
from bzrlib.transport import (
ConnectedTransport,
_CoalescedOffset,
get_transport,
Transport,
)
# TODO: This is not used anymore by HttpTransport_urllib
# (extracting the auth info and prompting the user for a password
# have been split), only the tests still use it. It should be
# deleted and the tests rewritten ASAP to stay in sync.
def extract_auth(url, password_manager):
"""Extract auth parameters from am HTTP/HTTPS url and add them to the given
password manager. Return the url, minus those auth parameters (which
confuse urllib2).
"""
if not re.match(r'^(https?)(\+\w+)?://', url):
raise ValueError(
'invalid absolute url %r' % (url,))
scheme, netloc, path, query, fragment = urlparse.urlsplit(url)
if '@' in netloc:
auth, netloc = netloc.split('@', 1)
if ':' in auth:
username, password = auth.split(':', 1)
else:
username, password = auth, None
if ':' in netloc:
host = netloc.split(':', 1)[0]
else:
host = netloc
username = urllib.unquote(username)
if password is not None:
password = urllib.unquote(password)
else:
password = ui.ui_factory.get_password(
prompt='HTTP %(user)s@%(host)s password',
user=username, host=host)
password_manager.add_password(None, host, username, password)
url = urlparse.urlunsplit((scheme, netloc, path, query, fragment))
return url
class HttpTransportBase(ConnectedTransport):
"""Base class for http implementations.
Does URL parsing, etc, but not any network IO.
The protocol can be given as e.g. http+urllib://host/ to use a particular
implementation.
"""
# _unqualified_scheme: "http" or "https"
# _scheme: may have "+pycurl", etc
def __init__(self, base, _impl_name, _from_transport=None):
"""Set the base path where files will be stored."""
proto_match = re.match(r'^(https?)(\+\w+)?://', base)
if not proto_match:
raise AssertionError("not a http url: %r" % base)
self._unqualified_scheme = proto_match.group(1)
self._impl_name = _impl_name
super(HttpTransportBase, self).__init__(base,
_from_transport=_from_transport)
self._medium = None
# range hint is handled dynamically throughout the life
# of the transport object. We start by trying multi-range
# requests and if the server returns bogus results, we
# retry with single range requests and, finally, we
# forget about range if the server really can't
# understand. Once acquired, this piece of info is
# propagated to clones.
if _from_transport is not None:
self._range_hint = _from_transport._range_hint
else:
self._range_hint = 'multi'
def has(self, relpath):
raise NotImplementedError("has() is abstract on %r" % self)
def get(self, relpath):
"""Get the file at the given relative path.
:param relpath: The relative path to the file
"""
code, response_file = self._get(relpath, None)
# FIXME: some callers want an iterable... One step forward, three steps
# backwards :-/ And not only an iterable, but an iterable that can be
# seeked backwards, so we will never be able to do that. One such
# known client is bzrlib.bundle.serializer.v4.get_bundle_reader. At the
# time of this writing it's even the only known client -- vila20071203
return StringIO(response_file.read())
def _get(self, relpath, ranges, tail_amount=0):
"""Get a file, or part of a file.
:param relpath: Path relative to transport base URL
:param ranges: None to get the whole file;
or a list of _CoalescedOffset to fetch parts of a file.
:param tail_amount: The amount to get from the end of the file.
:returns: (http_code, result_file)
"""
raise NotImplementedError(self._get)
def _remote_path(self, relpath):
"""See ConnectedTransport._remote_path.
user and passwords are not embedded in the path provided to the server.
"""
relative = urlutils.unescape(relpath).encode('utf-8')
path = self._combine_paths(self._path, relative)
return self._unsplit_url(self._unqualified_scheme,
None, None, self._host, self._port, path)
def _create_auth(self):
"""Returns a dict containing the credentials provided at build time."""
auth = dict(host=self._host, port=self._port,
user=self._user, password=self._password,
protocol=self._unqualified_scheme,
path=self._path)
return auth
def get_smart_medium(self):
"""See Transport.get_smart_medium."""
if self._medium is None:
# Since medium holds some state (smart server probing at least), we
# need to keep it around. Note that this is needed because medium
# has the same 'base' attribute as the transport so it can't be
# shared between transports having different bases.
self._medium = SmartClientHTTPMedium(self)
return self._medium
def _degrade_range_hint(self, relpath, ranges, exc_info):
if self._range_hint == 'multi':
self._range_hint = 'single'
mutter('Retry "%s" with single range request' % relpath)
elif self._range_hint == 'single':
self._range_hint = None
mutter('Retry "%s" without ranges' % relpath)
else:
# We tried all the tricks, but nothing worked. We re-raise the
# original exception; the 'mutter' calls above will indicate that
# further tries were unsuccessful
raise exc_info[0], exc_info[1], exc_info[2]
# _coalesce_offsets is a helper for readv, it try to combine ranges without
# degrading readv performances. _bytes_to_read_before_seek is the value
# used for the limit parameter and has been tuned for other transports. For
# HTTP, the name is inappropriate but the parameter is still useful and
# helps reduce the number of chunks in the response. The overhead for a
# chunk (headers, length, footer around the data itself is variable but
# around 50 bytes. We use 128 to reduce the range specifiers that appear in
# the header, some servers (notably Apache) enforce a maximum length for a
# header and issue a '400: Bad request' error when too much ranges are
# specified.
_bytes_to_read_before_seek = 128
# No limit on the offset number that get combined into one, we are trying
# to avoid downloading the whole file.
_max_readv_combine = 0
# By default Apache has a limit of ~400 ranges before replying with a 400
# Bad Request. So we go underneath that amount to be safe.
_max_get_ranges = 200
# We impose no limit on the range size. But see _pycurl.py for a different
# use.
_get_max_size = 0
def _readv(self, relpath, offsets):
"""Get parts of the file at the given relative path.
:param offsets: A list of (offset, size) tuples.
:param return: A list or generator of (offset, data) tuples
"""
# offsets may be a generator, we will iterate it several times, so
# build a list
offsets = list(offsets)
try_again = True
retried_offset = None
while try_again:
try_again = False
# Coalesce the offsets to minimize the GET requests issued
sorted_offsets = sorted(offsets)
coalesced = self._coalesce_offsets(
sorted_offsets, limit=self._max_readv_combine,
fudge_factor=self._bytes_to_read_before_seek,
max_size=self._get_max_size)
# Turn it into a list, we will iterate it several times
coalesced = list(coalesced)
if 'http' in debug.debug_flags:
mutter('http readv of %s offsets => %s collapsed %s',
relpath, len(offsets), len(coalesced))
# Cache the data read, but only until it's been used
data_map = {}
# We will iterate on the data received from the GET requests and
# serve the corresponding offsets respecting the initial order. We
# need an offset iterator for that.
iter_offsets = iter(offsets)
cur_offset_and_size = iter_offsets.next()
try:
for cur_coal, rfile in self._coalesce_readv(relpath, coalesced):
# Split the received chunk
for offset, size in cur_coal.ranges:
start = cur_coal.start + offset
rfile.seek(start, 0)
data = rfile.read(size)
data_len = len(data)
if data_len != size:
raise errors.ShortReadvError(relpath, start, size,
actual=data_len)
if (start, size) == cur_offset_and_size:
# The offset requested are sorted as the coalesced
# ones, no need to cache. Win !
yield cur_offset_and_size[0], data
cur_offset_and_size = iter_offsets.next()
else:
# Different sorting. We need to cache.
data_map[(start, size)] = data
# Yield everything we can
while cur_offset_and_size in data_map:
# Clean the cached data since we use it
# XXX: will break if offsets contains duplicates --
# vila20071129
this_data = data_map.pop(cur_offset_and_size)
yield cur_offset_and_size[0], this_data
cur_offset_and_size = iter_offsets.next()
except (errors.ShortReadvError, errors.InvalidRange,
errors.InvalidHttpRange), e:
mutter('Exception %r: %s during http._readv',e, e)
if (not isinstance(e, errors.ShortReadvError)
or retried_offset == cur_offset_and_size):
# We don't degrade the range hint for ShortReadvError since
# they do not indicate a problem with the server ability to
# handle ranges. Except when we fail to get back a required
# offset twice in a row. In that case, falling back to
# single range or whole file should help or end up in a
# fatal exception.
self._degrade_range_hint(relpath, coalesced, sys.exc_info())
# Some offsets may have been already processed, so we retry
# only the unsuccessful ones.
offsets = [cur_offset_and_size] + [o for o in iter_offsets]
retried_offset = cur_offset_and_size
try_again = True
def _coalesce_readv(self, relpath, coalesced):
"""Issue several GET requests to satisfy the coalesced offsets"""
def get_and_yield(relpath, coalesced):
if coalesced:
# Note that the _get below may raise
# errors.InvalidHttpRange. It's the caller's responsibility to
# decide how to retry since it may provide different coalesced
# offsets.
code, rfile = self._get(relpath, coalesced)
for coal in coalesced:
yield coal, rfile
if self._range_hint is None:
# Download whole file
for c, rfile in get_and_yield(relpath, coalesced):
yield c, rfile
else:
total = len(coalesced)
if self._range_hint == 'multi':
max_ranges = self._max_get_ranges
elif self._range_hint == 'single':
max_ranges = total
else:
raise AssertionError("Unknown _range_hint %r"
% (self._range_hint,))
# TODO: Some web servers may ignore the range requests and return
# the whole file, we may want to detect that and avoid further
# requests.
# Hint: test_readv_multiple_get_requests will fail once we do that
cumul = 0
ranges = []
for coal in coalesced:
if ((self._get_max_size > 0
and cumul + coal.length > self._get_max_size)
or len(ranges) >= max_ranges):
# Get that much and yield
for c, rfile in get_and_yield(relpath, ranges):
yield c, rfile
# Restart with the current offset
ranges = [coal]
cumul = coal.length
else:
ranges.append(coal)
cumul += coal.length
# Get the rest and yield
for c, rfile in get_and_yield(relpath, ranges):
yield c, rfile
def recommended_page_size(self):
"""See Transport.recommended_page_size().
For HTTP we suggest a large page size to reduce the overhead
introduced by latency.
"""
return 64 * 1024
def _post(self, body_bytes):
"""POST body_bytes to .bzr/smart on this transport.
:returns: (response code, response body file-like object).
"""
# TODO: Requiring all the body_bytes to be available at the beginning of
# the POST may require large client buffers. It would be nice to have
# an interface that allows streaming via POST when possible (and
# degrades to a local buffer when not).
raise NotImplementedError(self._post)
def put_file(self, relpath, f, mode=None):
"""Copy the file-like object into the location.
:param relpath: Location to put the contents, relative to base.
:param f: File-like object.
"""
raise errors.TransportNotPossible('http PUT not supported')
def mkdir(self, relpath, mode=None):
"""Create a directory at the given path."""
raise errors.TransportNotPossible('http does not support mkdir()')
def rmdir(self, relpath):
"""See Transport.rmdir."""
raise errors.TransportNotPossible('http does not support rmdir()')
def append_file(self, relpath, f, mode=None):
"""Append the text in the file-like object into the final
location.
"""
raise errors.TransportNotPossible('http does not support append()')
def copy(self, rel_from, rel_to):
"""Copy the item at rel_from to the location at rel_to"""
raise errors.TransportNotPossible('http does not support copy()')
def copy_to(self, relpaths, other, mode=None, pb=None):
"""Copy a set of entries from self into another Transport.
:param relpaths: A list/generator of entries to be copied.
TODO: if other is LocalTransport, is it possible to
do better than put(get())?
"""
# At this point HttpTransport might be able to check and see if
# the remote location is the same, and rather than download, and
# then upload, it could just issue a remote copy_this command.
if isinstance(other, HttpTransportBase):
raise errors.TransportNotPossible(
'http cannot be the target of copy_to()')
else:
return super(HttpTransportBase, self).\
copy_to(relpaths, other, mode=mode, pb=pb)
def move(self, rel_from, rel_to):
"""Move the item at rel_from to the location at rel_to"""
raise errors.TransportNotPossible('http does not support move()')
def delete(self, relpath):
"""Delete the item at relpath"""
raise errors.TransportNotPossible('http does not support delete()')
def external_url(self):
"""See bzrlib.transport.Transport.external_url."""
# HTTP URL's are externally usable as long as they don't mention their
# implementation qualifier
return self._unsplit_url(self._unqualified_scheme,
self._user, self._password,
self._host, self._port,
self._path)
def is_readonly(self):
"""See Transport.is_readonly."""
return True
def listable(self):
"""See Transport.listable."""
return False
def stat(self, relpath):
"""Return the stat information for a file.
"""
raise errors.TransportNotPossible('http does not support stat()')
def lock_read(self, relpath):
"""Lock the given file for shared (read) access.
:return: A lock object, which should be passed to Transport.unlock()
"""
# The old RemoteBranch ignore lock for reading, so we will
# continue that tradition and return a bogus lock object.
class BogusLock(object):
def __init__(self, path):
self.path = path
def unlock(self):
pass
return BogusLock(relpath)
def lock_write(self, relpath):
"""Lock the given file for exclusive (write) access.
WARNING: many transports do not support this, so trying avoid using it
:return: A lock object, which should be passed to Transport.unlock()
"""
raise errors.TransportNotPossible('http does not support lock_write()')
def _attempted_range_header(self, offsets, tail_amount):
"""Prepare a HTTP Range header at a level the server should accept.
:return: the range header representing offsets/tail_amount or None if
no header can be built.
"""
if self._range_hint == 'multi':
# Generate the header describing all offsets
return self._range_header(offsets, tail_amount)
elif self._range_hint == 'single':
# Combine all the requested ranges into a single
# encompassing one
if len(offsets) > 0:
if tail_amount not in (0, None):
# Nothing we can do here to combine ranges with tail_amount
# in a single range, just returns None. The whole file
# should be downloaded.
return None
else:
start = offsets[0].start
last = offsets[-1]
end = last.start + last.length - 1
whole = self._coalesce_offsets([(start, end - start + 1)],
limit=0, fudge_factor=0)
return self._range_header(list(whole), 0)
else:
# Only tail_amount, requested, leave range_header
# do its work
return self._range_header(offsets, tail_amount)
else:
return None
@staticmethod
def _range_header(ranges, tail_amount):
"""Turn a list of bytes ranges into a HTTP Range header value.
:param ranges: A list of _CoalescedOffset
:param tail_amount: The amount to get from the end of the file.
:return: HTTP range header string.
At least a non-empty ranges *or* a tail_amount must be
provided.
"""
strings = []
for offset in ranges:
strings.append('%d-%d' % (offset.start,
offset.start + offset.length - 1))
if tail_amount:
strings.append('-%d' % tail_amount)
return ','.join(strings)
def _redirected_to(self, source, target):
"""Returns a transport suitable to re-issue a redirected request.
:param source: The source url as returned by the server.
:param target: The target url as returned by the server.
The redirection can be handled only if the relpath involved is not
renamed by the redirection.
:returns: A transport or None.
"""
def relpath(abspath):
"""Returns the path relative to our base.
The constraints are weaker than the real relpath method because the
abspath is coming from the server and may slightly differ from our
base. We don't check the scheme, host, port, user, password parts,
relying on the caller to give us a proper url (i.e. one returned by
the server mirroring the one we sent).
"""
(scheme,
user, password,
host, port,
path) = self._split_url(abspath)
pl = len(self._path)
return path[pl:].strip('/')
relpath = relpath(source)
if not target.endswith(relpath):
# The final part of the url has been renamed, we can't handle the
# redirection.
return None
new_transport = None
(scheme,
user, password,
host, port,
path) = self._split_url(target)
# Recalculate base path. This is needed to ensure that when the
# redirected tranport will be used to re-try whatever request was
# redirected, we end up with the same url
base_path = path[:-len(relpath)]
if scheme in ('http', 'https'):
# Same protocol family (i.e. http[s]), we will preserve the same
# http client implementation when a redirection occurs from one to
# the other (otherwise users may be surprised that bzr switches
# from one implementation to the other, and devs may suffer
# debugging it).
if (scheme == self._unqualified_scheme
and host == self._host
and port == self._port
and (user is None or user == self._user)):
# If a user is specified, it should match, we don't care about
# passwords, wrong passwords will be rejected anyway.
new_transport = self.clone(base_path)
else:
# Rebuild the url preserving the scheme qualification and the
# credentials (if they don't apply, the redirected to server
# will tell us, but if they do apply, we avoid prompting the
# user)
redir_scheme = scheme + '+' + self._impl_name
new_url = self._unsplit_url(redir_scheme,
self._user, self._password,
host, port,
base_path)
new_transport = get_transport(new_url)
else:
# Redirected to a different protocol
new_url = self._unsplit_url(scheme,
user, password,
host, port,
base_path)
new_transport = get_transport(new_url)
return new_transport
# TODO: May be better located in smart/medium.py with the other
# SmartMedium classes
class SmartClientHTTPMedium(medium.SmartClientMedium):
def __init__(self, http_transport):
super(SmartClientHTTPMedium, self).__init__(http_transport.base)
# We don't want to create a circular reference between the http
# transport and its associated medium. Since the transport will live
# longer than the medium, the medium keep only a weak reference to its
# transport.
self._http_transport_ref = weakref.ref(http_transport)
def get_request(self):
return SmartClientHTTPMediumRequest(self)
def should_probe(self):
return True
def remote_path_from_transport(self, transport):
# Strip the optional 'bzr+' prefix from transport so it will have the
# same scheme as self.
transport_base = transport.base
if transport_base.startswith('bzr+'):
transport_base = transport_base[4:]
rel_url = urlutils.relative_url(self.base, transport_base)
return urllib.unquote(rel_url)
def send_http_smart_request(self, bytes):
try:
# Get back the http_transport hold by the weak reference
t = self._http_transport_ref()
code, body_filelike = t._post(bytes)
if code != 200:
raise InvalidHttpResponse(
t._remote_path('.bzr/smart'),
'Expected 200 response code, got %r' % (code,))
except (errors.InvalidHttpResponse, errors.ConnectionReset), e:
raise errors.SmartProtocolError(str(e))
return body_filelike
def _report_activity(self, bytes, direction):
"""See SmartMedium._report_activity.
Does nothing; the underlying plain HTTP transport will report the
activity that this medium would report.
"""
pass
# TODO: May be better located in smart/medium.py with the other
# SmartMediumRequest classes
class SmartClientHTTPMediumRequest(medium.SmartClientMediumRequest):
"""A SmartClientMediumRequest that works with an HTTP medium."""
def __init__(self, client_medium):
medium.SmartClientMediumRequest.__init__(self, client_medium)
self._buffer = ''
def _accept_bytes(self, bytes):
self._buffer += bytes
def _finished_writing(self):
data = self._medium.send_http_smart_request(self._buffer)
self._response_body = data
def _read_bytes(self, count):
"""See SmartClientMediumRequest._read_bytes."""
return self._response_body.read(count)
def _read_line(self):
line, excess = medium._get_line(self._response_body.read)
if excess != '':
raise AssertionError(
'_get_line returned excess bytes, but this mediumrequest '
'cannot handle excess. (%r)' % (excess,))
return line
def _finished_reading(self):
"""See SmartClientMediumRequest._finished_reading."""
pass
|