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
34
class WebserverNotAvailable(Exception):
38
class BadWebserverPath(ValueError):
40
return 'path %s is not in %s' % self.args
43
class TestingHTTPRequestHandler(SimpleHTTPRequestHandler):
45
def log_message(self, format, *args):
46
tcs = self.server.test_case_server
47
tcs.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)
159
"""Serve a GET request.
161
Handles the Range header.
164
path = self.translate_path(self.path)
165
ranges_header_value = self.headers.get('Range')
166
if ranges_header_value is None or os.path.isdir(path):
167
# Let the mother class handle most cases
168
return SimpleHTTPRequestHandler.do_GET(self)
171
# Always read in binary mode. Opening files in text
172
# mode may cause newline translations, making the
173
# actual size of the content transmitted *less* than
174
# the content-length!
175
file = open(path, 'rb')
177
self.send_error(404, "File not found")
180
file_size = os.fstat(file.fileno())[6]
181
tail, ranges = self.parse_ranges(ranges_header_value)
182
# Normalize tail into ranges
184
ranges.append((file_size - tail, file_size))
186
self._satisfiable_ranges = True
188
self._satisfiable_ranges = False
190
def check_range(range_specifier):
191
start, end = range_specifier
192
# RFC2616 14.35, ranges are invalid if start >= file_size
193
if start >= file_size:
194
self._satisfiable_ranges = False # Side-effect !
196
# RFC2616 14.35, end values should be truncated
197
# to file_size -1 if they exceed it
198
end = min(end, file_size - 1)
201
ranges = map(check_range, ranges)
203
if not self._satisfiable_ranges:
204
# RFC2616 14.16 and 14.35 says that when a server
205
# encounters unsatisfiable range specifiers, it
206
# SHOULD return a 416.
208
# FIXME: We SHOULD send a Content-Range header too,
209
# but the implementation of send_error does not
210
# allows that. So far.
211
self.send_error(416, "Requested range not satisfiable")
215
(start, end) = ranges[0]
216
self.get_single_range(file, file_size, start, end)
218
self.get_multiple_ranges(file, file_size, ranges)
221
if sys.platform == 'win32':
222
# On win32 you cannot access non-ascii filenames without
223
# decoding them into unicode first.
224
# However, under Linux, you can access bytestream paths
225
# without any problems. If this function was always active
226
# it would probably break tests when LANG=C was set
227
def translate_path(self, path):
228
"""Translate a /-separated PATH to the local filename syntax.
230
For bzr, all url paths are considered to be utf8 paths.
231
On Linux, you can access these paths directly over the bytestream
232
request, but on win32, you must decode them, and access them
235
# abandon query parameters
236
path = urlparse.urlparse(path)[2]
237
path = posixpath.normpath(urllib.unquote(path))
238
path = path.decode('utf-8')
239
words = path.split('/')
240
words = filter(None, words)
243
drive, word = os.path.splitdrive(word)
244
head, word = os.path.split(word)
245
if word in (os.curdir, os.pardir): continue
246
path = os.path.join(path, word)
250
class TestingHTTPServer(BaseHTTPServer.HTTPServer):
252
def __init__(self, server_address, RequestHandlerClass,
254
BaseHTTPServer.HTTPServer.__init__(self, server_address,
256
# test_case_server can be used to communicate between the
257
# tests and the server (or the request handler and the
258
# server), allowing dynamic behaviors to be defined from
260
self.test_case_server = test_case_server
263
class HttpServer(Server):
264
"""A test server for http transports.
266
Subclasses can provide a specific request handler.
269
# used to form the url that connects to this server
270
_url_protocol = 'http'
272
# Subclasses can provide a specific request handler
273
def __init__(self, request_handler=TestingHTTPRequestHandler):
274
Server.__init__(self)
275
self.request_handler = request_handler
276
self.host = 'localhost'
280
def _get_httpd(self):
281
if self._httpd is None:
282
self._httpd = TestingHTTPServer((self.host, self.port),
283
self.request_handler,
285
host, self.port = self._httpd.socket.getsockname()
288
def _http_start(self):
289
httpd = self._get_httpd()
290
self._http_base_url = '%s://%s:%s/' % (self._url_protocol,
293
self._http_starting.release()
294
httpd.socket.settimeout(0.1)
296
while self._http_running:
298
httpd.handle_request()
299
except socket.timeout:
302
def _get_remote_url(self, path):
303
path_parts = path.split(os.path.sep)
304
if os.path.isabs(path):
305
if path_parts[:len(self._local_path_parts)] != \
306
self._local_path_parts:
307
raise BadWebserverPath(path, self.test_dir)
308
remote_path = '/'.join(path_parts[len(self._local_path_parts):])
310
remote_path = '/'.join(path_parts)
312
return self._http_base_url + remote_path
314
def log(self, format, *args):
315
"""Capture Server log output."""
316
self.logs.append(format % args)
319
"""See bzrlib.transport.Server.setUp."""
320
self._home_dir = os.getcwdu()
321
self._local_path_parts = self._home_dir.split(os.path.sep)
322
self._http_starting = threading.Lock()
323
self._http_starting.acquire()
324
self._http_running = True
325
self._http_base_url = None
326
self._http_thread = threading.Thread(target=self._http_start)
327
self._http_thread.setDaemon(True)
328
self._http_thread.start()
329
# Wait for the server thread to start (i.e release the lock)
330
self._http_starting.acquire()
331
self._http_starting.release()
335
"""See bzrlib.transport.Server.tearDown."""
336
self._http_running = False
337
self._http_thread.join()
340
"""See bzrlib.transport.Server.get_url."""
341
return self._get_remote_url(self._home_dir)
343
def get_bogus_url(self):
344
"""See bzrlib.transport.Server.get_bogus_url."""
345
# this is chosen to try to prevent trouble with proxies, weird dns,
347
return 'http://127.0.0.1:1/'
350
class HttpServer_urllib(HttpServer):
351
"""Subclass of HttpServer that gives http+urllib urls.
353
This is for use in testing: connections to this server will always go
354
through urllib where possible.
357
# urls returned by this server should require the urllib client impl
358
_url_protocol = 'http+urllib'
361
class HttpServer_PyCurl(HttpServer):
362
"""Subclass of HttpServer that gives http+pycurl urls.
364
This is for use in testing: connections to this server will always go
365
through pycurl where possible.
368
# We don't care about checking the pycurl availability as
369
# this server will be required only when pycurl is present
371
# urls returned by this server should require the pycurl client impl
372
_url_protocol = 'http+pycurl'