1
# Copyright (C) 2006 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
20
from SimpleHTTPServer import SimpleHTTPRequestHandler
31
from bzrlib.transport import Server
32
from bzrlib.transport.local import LocalURLServer
35
class WebserverNotAvailable(Exception):
39
class BadWebserverPath(ValueError):
41
return 'path %s is not in %s' % self.args
44
class TestingHTTPRequestHandler(SimpleHTTPRequestHandler):
46
def log_message(self, format, *args):
47
self.server.test_case.log('webserver - %s - - [%s] %s "%s" "%s"',
48
self.address_string(),
49
self.log_date_time_string(),
51
self.headers.get('referer', '-'),
52
self.headers.get('user-agent', '-'))
54
def handle_one_request(self):
55
"""Handle a single HTTP request.
57
You normally don't need to override this method; see the class
58
__doc__ string for information on how to handle specific HTTP
59
commands such as GET and POST.
62
for i in xrange(1,11): # Don't try more than 10 times
64
self.raw_requestline = self.rfile.readline()
65
except socket.error, e:
66
if e.args[0] in (errno.EAGAIN, errno.EWOULDBLOCK):
67
# omitted for now because some tests look at the log of
68
# the server and expect to see no errors. see recent
69
# email thread. -- mbp 20051021.
70
## self.log_message('EAGAIN (%d) while reading from raw_requestline' % i)
76
if not self.raw_requestline:
77
self.close_connection = 1
79
if not self.parse_request(): # An error code has been sent, just exit
81
mname = 'do_' + self.command
82
if getattr(self, mname, None) is None:
83
self.send_error(501, "Unsupported method (%r)" % self.command)
85
method = getattr(self, mname)
88
_range_regexp = re.compile(r'^(?P<start>\d+)-(?P<end>\d+)$')
89
_tail_regexp = re.compile(r'^-(?P<tail>\d+)$')
91
def parse_ranges(self, ranges_header):
92
"""Parse the range header value and returns ranges and tail.
94
RFC2616 14.35 says that syntactically invalid range
95
specifiers MUST be ignored. In that case, we return 0 for
96
tail and [] for ranges.
100
if not ranges_header.startswith('bytes='):
101
# Syntactically invalid header
104
ranges_header = ranges_header[len('bytes='):]
105
for range_str in ranges_header.split(','):
106
# FIXME: RFC2616 says end is optional and default to file_size
107
range_match = self._range_regexp.match(range_str)
108
if range_match is not None:
109
start = int(range_match.group('start'))
110
end = int(range_match.group('end'))
112
# Syntactically invalid range
114
ranges.append((start, end))
116
tail_match = self._tail_regexp.match(range_str)
117
if tail_match is not None:
118
tail = int(tail_match.group('tail'))
120
# Syntactically invalid range
124
def send_range_content(self, file, start, length):
126
self.wfile.write(file.read(length))
128
def get_single_range(self, file, file_size, start, end):
129
self.send_response(206)
130
length = end - start + 1
131
self.send_header('Accept-Ranges', 'bytes')
132
self.send_header("Content-Length", "%d" % length)
134
self.send_header("Content-Type", 'application/octet-stream')
135
self.send_header("Content-Range", "bytes %d-%d/%d" % (start,
139
self.send_range_content(file, start, length)
141
def get_multiple_ranges(self, file, file_size, ranges):
142
self.send_response(206)
143
self.send_header('Accept-Ranges', 'bytes')
144
boundary = "%d" % random.randint(0,0x7FFFFFFF)
145
self.send_header("Content-Type",
146
"multipart/byteranges; boundary=%s" % boundary)
148
for (start, end) in ranges:
149
self.wfile.write("--%s\r\n" % boundary)
150
self.send_header("Content-type", 'application/octet-stream')
151
self.send_header("Content-Range", "bytes %d-%d/%d" % (start,
155
self.send_range_content(file, start, end - start + 1)
156
self.wfile.write("--%s\r\n" % boundary)
160
"""Serve a GET request.
162
Handles the Range header.
165
path = self.translate_path(self.path)
166
ranges_header_value = self.headers.get('Range')
167
if ranges_header_value is None or os.path.isdir(path):
168
# Let the mother class handle most cases
169
return SimpleHTTPRequestHandler.do_GET(self)
172
# Always read in binary mode. Opening files in text
173
# mode may cause newline translations, making the
174
# actual size of the content transmitted *less* than
175
# the content-length!
176
file = open(path, 'rb')
178
self.send_error(404, "File not found")
181
file_size = os.fstat(file.fileno())[6]
182
tail, ranges = self.parse_ranges(ranges_header_value)
183
# Normalize tail into ranges
185
ranges.append((file_size - tail, file_size))
187
self._satisfiable_ranges = True
189
self._satisfiable_ranges = False
191
def check_range(range_specifier):
192
start, end = range_specifier
193
# RFC2616 14.35, ranges are invalid if start >= file_size
194
if start >= file_size:
195
self._satisfiable_ranges = False # Side-effect !
197
# RFC2616 14.35, end values should be truncated
198
# to file_size -1 if they exceed it
199
end = min(end, file_size - 1)
202
ranges = map(check_range, ranges)
204
if not self._satisfiable_ranges:
205
# RFC2616 14.16 and 14.35 says that when a server
206
# encounters unsatisfiable range specifiers, it
207
# SHOULD return a 416.
209
# FIXME: We SHOULD send a Content-Range header too,
210
# but the implementation of send_error does not
211
# allows that. So far.
212
self.send_error(416, "Requested range not satisfiable")
216
(start, end) = ranges[0]
217
self.get_single_range(file, file_size, start, end)
219
self.get_multiple_ranges(file, file_size, ranges)
222
if sys.platform == 'win32':
223
# On win32 you cannot access non-ascii filenames without
224
# decoding them into unicode first.
225
# However, under Linux, you can access bytestream paths
226
# without any problems. If this function was always active
227
# it would probably break tests when LANG=C was set
228
def translate_path(self, path):
229
"""Translate a /-separated PATH to the local filename syntax.
231
For bzr, all url paths are considered to be utf8 paths.
232
On Linux, you can access these paths directly over the bytestream
233
request, but on win32, you must decode them, and access them
236
# abandon query parameters
237
path = urlparse.urlparse(path)[2]
238
path = posixpath.normpath(urllib.unquote(path))
239
path = path.decode('utf-8')
240
words = path.split('/')
241
words = filter(None, words)
244
drive, word = os.path.splitdrive(word)
245
head, word = os.path.split(word)
246
if word in (os.curdir, os.pardir): continue
247
path = os.path.join(path, word)
251
class TestingHTTPServer(BaseHTTPServer.HTTPServer):
252
def __init__(self, server_address, RequestHandlerClass, test_case):
253
BaseHTTPServer.HTTPServer.__init__(self, server_address,
255
self.test_case = test_case
258
class HttpServer(Server):
259
"""A test server for http transports.
261
Subclasses can provide a specific request handler.
264
# used to form the url that connects to this server
265
_url_protocol = 'http'
267
# Subclasses can provide a specific request handler
268
def __init__(self, request_handler=TestingHTTPRequestHandler):
269
Server.__init__(self)
270
self.request_handler = request_handler
272
def _get_httpd(self):
273
return TestingHTTPServer(('localhost', 0),
274
self.request_handler,
277
def _http_start(self):
279
httpd = self._get_httpd()
280
host, self.port = httpd.socket.getsockname()
281
self._http_base_url = '%s://localhost:%s/' % (self._url_protocol,
283
self._http_starting.release()
284
httpd.socket.settimeout(0.1)
286
while self._http_running:
288
httpd.handle_request()
289
except socket.timeout:
292
def _get_remote_url(self, path):
293
path_parts = path.split(os.path.sep)
294
if os.path.isabs(path):
295
if path_parts[:len(self._local_path_parts)] != \
296
self._local_path_parts:
297
raise BadWebserverPath(path, self.test_dir)
298
remote_path = '/'.join(path_parts[len(self._local_path_parts):])
300
remote_path = '/'.join(path_parts)
302
return self._http_base_url + remote_path
304
def log(self, format, *args):
305
"""Capture Server log output."""
306
self.logs.append(format % args)
308
def setUp(self, backing_transport_server=None):
309
"""See bzrlib.transport.Server.setUp.
311
:param backing_transport_server: The transport that requests over this
312
protocol should be forwarded to. Note that this is currently not
315
# XXX: TODO: make the server back onto vfs_server rather than local
317
assert backing_transport_server is None or \
318
isinstance(backing_transport_server, LocalURLServer), \
319
"HTTPServer currently assumes local transport, got %s" % \
320
backing_transport_server
321
self._home_dir = os.getcwdu()
322
self._local_path_parts = self._home_dir.split(os.path.sep)
323
self._http_starting = threading.Lock()
324
self._http_starting.acquire()
325
self._http_running = True
326
self._http_base_url = None
327
self._http_thread = threading.Thread(target=self._http_start)
328
self._http_thread.setDaemon(True)
329
self._http_thread.start()
330
# Wait for the server thread to start (i.e release the lock)
331
self._http_starting.acquire()
332
self._http_starting.release()
336
"""See bzrlib.transport.Server.tearDown."""
337
self._http_running = False
338
self._http_thread.join()
341
"""See bzrlib.transport.Server.get_url."""
342
return self._get_remote_url(self._home_dir)
344
def get_bogus_url(self):
345
"""See bzrlib.transport.Server.get_bogus_url."""
346
# this is chosen to try to prevent trouble with proxies, weird dns,
348
return 'http://127.0.0.1:1/'
351
class HttpServer_urllib(HttpServer):
352
"""Subclass of HttpServer that gives http+urllib urls.
354
This is for use in testing: connections to this server will always go
355
through urllib where possible.
358
# urls returned by this server should require the urllib client impl
359
_url_protocol = 'http+urllib'
362
class HttpServer_PyCurl(HttpServer):
363
"""Subclass of HttpServer that gives http+pycurl urls.
365
This is for use in testing: connections to this server will always go
366
through pycurl where possible.
369
# We don't care about checking the pycurl availability as
370
# this server will be required only when pycurl is present
372
# urls returned by this server should require the pycurl client impl
373
_url_protocol = 'http+pycurl'
added
removed
bzrlib/tests/HttpServer.py
