1
# Copyright (C) 2005 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
16
"""Implementation of Transport over http.
20
from cStringIO import StringIO
21
import urllib, urllib2
23
from warnings import warn
25
from bzrlib.transport import Transport, Server
26
from bzrlib.errors import (TransportNotPossible, NoSuchFile,
27
TransportError, ConnectionError)
28
from bzrlib.errors import BzrError, BzrCheckError
29
from bzrlib.branch import Branch
30
from bzrlib.trace import mutter
33
def extract_auth(url, password_manager):
35
Extract auth parameters from am HTTP/HTTPS url and add them to the given
36
password manager. Return the url, minus those auth parameters (which
39
assert url.startswith('http://') or url.startswith('https://')
40
scheme, host = url.split('//', 1)
42
host, path = host.split('/', 1)
48
auth, host = host.split('@', 1)
50
username, password = auth.split(':', 1)
52
username, password = auth, None
54
host, port = host.split(':', 1)
56
# FIXME: if password isn't given, should we ask for it?
57
if password is not None:
58
username = urllib.unquote(username)
59
password = urllib.unquote(password)
60
password_manager.add_password(None, host, username, password)
61
url = scheme + '//' + host + port + path
66
mutter("get_url %s" % url)
67
manager = urllib2.HTTPPasswordMgrWithDefaultRealm()
68
url = extract_auth(url, manager)
69
auth_handler = urllib2.HTTPBasicAuthHandler(manager)
70
opener = urllib2.build_opener(auth_handler)
71
url_f = opener.open(url)
74
class HttpTransport(Transport):
75
"""This is the transport agent for http:// access.
77
TODO: Implement pipelined versions of all of the *_multi() functions.
80
def __init__(self, base):
81
"""Set the base path where files will be stored."""
82
assert base.startswith('http://') or base.startswith('https://')
85
super(HttpTransport, self).__init__(base)
86
# In the future we might actually connect to the remote host
87
# rather than using get_url
88
# self._connection = None
89
(self._proto, self._host,
90
self._path, self._parameters,
91
self._query, self._fragment) = urlparse.urlparse(self.base)
93
def should_cache(self):
94
"""Return True if the data pulled across should be cached locally.
98
def clone(self, offset=None):
99
"""Return a new HttpTransport with root at self.base + offset
100
For now HttpTransport does not actually connect, so just return
101
a new HttpTransport object.
104
return HttpTransport(self.base)
106
return HttpTransport(self.abspath(offset))
108
def abspath(self, relpath):
109
"""Return the full url to the given relative path.
110
This can be supplied with a string or a list
112
assert isinstance(relpath, basestring)
113
if isinstance(relpath, basestring):
114
relpath_parts = relpath.split('/')
116
# TODO: Don't call this with an array - no magic interfaces
117
relpath_parts = relpath[:]
118
if len(relpath_parts) > 1:
119
if relpath_parts[0] == '':
120
raise ValueError("path %r within branch %r seems to be absolute"
121
% (relpath, self._path))
122
if relpath_parts[-1] == '':
123
raise ValueError("path %r within branch %r seems to be a directory"
124
% (relpath, self._path))
125
basepath = self._path.split('/')
126
if len(basepath) > 0 and basepath[-1] == '':
127
basepath = basepath[:-1]
128
for p in relpath_parts:
130
if len(basepath) == 0:
131
# In most filesystems, a request for the parent
132
# of root, just returns root.
135
elif p == '.' or p == '':
139
# Possibly, we could use urlparse.urljoin() here, but
140
# I'm concerned about when it chooses to strip the last
141
# portion of the path, and when it doesn't.
142
path = '/'.join(basepath)
143
return urlparse.urlunparse((self._proto,
144
self._host, path, '', '', ''))
146
def has(self, relpath):
147
"""Does the target location exist?
149
TODO: HttpTransport.has() should use a HEAD request,
150
not a full GET request.
152
TODO: This should be changed so that we don't use
153
urllib2 and get an exception, the code path would be
154
cleaner if we just do an http HEAD request, and parse
159
path = self.abspath(relpath)
161
# Without the read and then close()
162
# we tend to have busy sockets.
166
except urllib2.URLError, e:
167
mutter('url error code: %s for has url: %r', e.code, path)
172
mutter('io error: %s %s for has url: %r',
173
e.errno, errno.errorcode.get(e.errno), path)
174
if e.errno == errno.ENOENT:
176
raise TransportError(orig_error=e)
178
def get(self, relpath, decode=False):
179
"""Get the file at the given relative path.
181
:param relpath: The relative path to the file
185
path = self.abspath(relpath)
187
except urllib2.HTTPError, e:
188
mutter('url error code: %s for has url: %r', e.code, path)
190
raise NoSuchFile(path, extra=e)
192
except (BzrError, IOError), e:
193
if hasattr(e, 'errno'):
194
mutter('io error: %s %s for has url: %r',
195
e.errno, errno.errorcode.get(e.errno), path)
196
if e.errno == errno.ENOENT:
197
raise NoSuchFile(path, extra=e)
198
raise ConnectionError(msg = "Error retrieving %s: %s"
199
% (self.abspath(relpath), str(e)),
202
def put(self, relpath, f, mode=None):
203
"""Copy the file-like or string object into the location.
205
:param relpath: Location to put the contents, relative to base.
206
:param f: File-like or string object.
208
raise TransportNotPossible('http PUT not supported')
210
def mkdir(self, relpath, mode=None):
211
"""Create a directory at the given path."""
212
raise TransportNotPossible('http does not support mkdir()')
214
def rmdir(self, relpath):
215
"""See Transport.rmdir."""
216
raise TransportNotPossible('http does not support rmdir()')
218
def append(self, relpath, f):
219
"""Append the text in the file-like object into the final
222
raise TransportNotPossible('http does not support append()')
224
def copy(self, rel_from, rel_to):
225
"""Copy the item at rel_from to the location at rel_to"""
226
raise TransportNotPossible('http does not support copy()')
228
def copy_to(self, relpaths, other, mode=None, pb=None):
229
"""Copy a set of entries from self into another Transport.
231
:param relpaths: A list/generator of entries to be copied.
233
TODO: if other is LocalTransport, is it possible to
234
do better than put(get())?
236
# At this point HttpTransport might be able to check and see if
237
# the remote location is the same, and rather than download, and
238
# then upload, it could just issue a remote copy_this command.
239
if isinstance(other, HttpTransport):
240
raise TransportNotPossible('http cannot be the target of copy_to()')
242
return super(HttpTransport, self).copy_to(relpaths, other, mode=mode, pb=pb)
244
def move(self, rel_from, rel_to):
245
"""Move the item at rel_from to the location at rel_to"""
246
raise TransportNotPossible('http does not support move()')
248
def delete(self, relpath):
249
"""Delete the item at relpath"""
250
raise TransportNotPossible('http does not support delete()')
252
def is_readonly(self):
253
"""See Transport.is_readonly."""
257
"""See Transport.listable."""
260
def stat(self, relpath):
261
"""Return the stat information for a file.
263
raise TransportNotPossible('http does not support stat()')
265
def lock_read(self, relpath):
266
"""Lock the given file for shared (read) access.
267
:return: A lock object, which should be passed to Transport.unlock()
269
# The old RemoteBranch ignore lock for reading, so we will
270
# continue that tradition and return a bogus lock object.
271
class BogusLock(object):
272
def __init__(self, path):
276
return BogusLock(relpath)
278
def lock_write(self, relpath):
279
"""Lock the given file for exclusive (write) access.
280
WARNING: many transports do not support this, so trying avoid using it
282
:return: A lock object, which should be passed to Transport.unlock()
284
raise TransportNotPossible('http does not support lock_write()')
287
#---------------- test server facilities ----------------
288
import BaseHTTPServer, SimpleHTTPServer, socket, time
292
class WebserverNotAvailable(Exception):
296
class BadWebserverPath(ValueError):
298
return 'path %s is not in %s' % self.args
301
class TestingHTTPRequestHandler(SimpleHTTPServer.SimpleHTTPRequestHandler):
303
def log_message(self, format, *args):
304
self.server.test_case.log("webserver - %s - - [%s] %s",
305
self.address_string(),
306
self.log_date_time_string(),
309
def handle_one_request(self):
310
"""Handle a single HTTP request.
312
You normally don't need to override this method; see the class
313
__doc__ string for information on how to handle specific HTTP
314
commands such as GET and POST.
317
for i in xrange(1,11): # Don't try more than 10 times
319
self.raw_requestline = self.rfile.readline()
320
except socket.error, e:
321
if e.args[0] in (errno.EAGAIN, errno.EWOULDBLOCK):
322
# omitted for now because some tests look at the log of
323
# the server and expect to see no errors. see recent
324
# email thread. -- mbp 20051021.
325
## self.log_message('EAGAIN (%d) while reading from raw_requestline' % i)
331
if not self.raw_requestline:
332
self.close_connection = 1
334
if not self.parse_request(): # An error code has been sent, just exit
336
mname = 'do_' + self.command
337
if not hasattr(self, mname):
338
self.send_error(501, "Unsupported method (%r)" % self.command)
340
method = getattr(self, mname)
343
class TestingHTTPServer(BaseHTTPServer.HTTPServer):
344
def __init__(self, server_address, RequestHandlerClass, test_case):
345
BaseHTTPServer.HTTPServer.__init__(self, server_address,
347
self.test_case = test_case
350
class HttpServer(Server):
351
"""A test server for http transports."""
353
_HTTP_PORTS = range(13000, 0x8000)
355
def _http_start(self):
357
for port in self._HTTP_PORTS:
359
httpd = TestingHTTPServer(('localhost', port),
360
TestingHTTPRequestHandler,
362
except socket.error, e:
363
if e.args[0] == errno.EADDRINUSE:
365
print >>sys.stderr, "Cannot run webserver :-("
371
raise WebserverNotAvailable("Cannot run webserver :-( "
372
"no free ports in range %s..%s" %
373
(_HTTP_PORTS[0], _HTTP_PORTS[-1]))
375
self._http_base_url = 'http://localhost:%s/' % port
376
self._http_starting.release()
377
httpd.socket.settimeout(0.1)
379
while self._http_running:
381
httpd.handle_request()
382
except socket.timeout:
385
def _get_remote_url(self, path):
386
path_parts = path.split(os.path.sep)
387
if os.path.isabs(path):
388
if path_parts[:len(self._local_path_parts)] != \
389
self._local_path_parts:
390
raise BadWebserverPath(path, self.test_dir)
391
remote_path = '/'.join(path_parts[len(self._local_path_parts):])
393
remote_path = '/'.join(path_parts)
395
self._http_starting.acquire()
396
self._http_starting.release()
397
return self._http_base_url + remote_path
399
def log(self, *args, **kwargs):
400
"""Capture Server log output."""
401
self.logs.append(args[3])
404
"""See bzrlib.transport.Server.setUp."""
405
self._home_dir = os.getcwdu()
406
self._local_path_parts = self._home_dir.split(os.path.sep)
407
self._http_starting = threading.Lock()
408
self._http_starting.acquire()
409
self._http_running = True
410
self._http_base_url = None
411
self._http_thread = threading.Thread(target=self._http_start)
412
self._http_thread.setDaemon(True)
413
self._http_thread.start()
414
self._http_proxy = os.environ.get("http_proxy")
415
if self._http_proxy is not None:
416
del os.environ["http_proxy"]
420
"""See bzrlib.transport.Server.tearDown."""
421
self._http_running = False
422
self._http_thread.join()
423
if self._http_proxy is not None:
425
os.environ["http_proxy"] = self._http_proxy
428
"""See bzrlib.transport.Server.get_url."""
429
return self._get_remote_url(self._home_dir)
431
def get_bogus_url(self):
432
"""See bzrlib.transport.Server.get_bogus_url."""
433
return 'http://jasldkjsalkdjalksjdkljasd'
436
def get_test_permutations():
437
"""Return the permutations to be used in testing."""
438
warn("There are no HTTPS transport provider tests yet.")
439
return [(HttpTransport, HttpServer),
added
removed
bzrlib/transport/http.py
