1 # web.py -- WSGI smart-http server
2 # Copryight (C) 2010 Google, Inc.
4 # This program is free software; you can redistribute it and/or
5 # modify it under the terms of the GNU General Public License
6 # as published by the Free Software Foundation; version 2
7 # or (at your option) any later version of the License.
9 # This program is distributed in the hope that it will be useful,
10 # but WITHOUT ANY WARRANTY; without even the implied warranty of
11 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 # GNU General Public License for more details.
14 # You should have received a copy of the GNU General Public License
15 # along with this program; if not, write to the Free Software
16 # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston,
19 """HTTP server for dulwich that implements the git smart HTTP protocol."""
21 from cStringIO import StringIO
26 from urlparse import parse_qs
28 from dulwich.misc import parse_qs
29 from dulwich.server import (
35 HTTP_NOT_FOUND = '404 Not Found'
36 HTTP_FORBIDDEN = '403 Forbidden'
39 def date_time_string(timestamp=None):
40 # Based on BaseHTTPServer.py in python2.5
41 weekdays = ['Mon', 'Tue', 'Wed', 'Thu', 'Fri', 'Sat', 'Sun']
43 'Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun',
44 'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec']
46 timestamp = time.time()
47 year, month, day, hh, mm, ss, wd, y, z = time.gmtime(timestamp)
48 return '%s, %02d %3s %4d %02d:%02d:%02d GMD' % (
49 weekdays[wd], day, months[month], year, hh, mm, ss)
53 """Extract the URL prefix from a regex match.
55 :param mat: A regex match object.
56 :returns: The URL prefix, defined as the text before the match in the
57 original string. Normalized to start with one leading slash and end with
60 return '/' + mat.string[:mat.start()].strip('/')
63 def get_repo(backend, mat):
64 """Get a Repo instance for the given backend and URL regex match."""
65 return backend.open_repository(url_prefix(mat))
68 def send_file(req, f, content_type):
69 """Send a file-like object to the request output.
71 :param req: The HTTPGitRequest object to send output to.
72 :param f: An open file-like object to send; will be closed.
73 :param content_type: The MIME type for the file.
74 :yield: The contents of the file.
77 yield req.not_found('File not found')
81 req.respond(HTTP_OK, content_type)
88 yield req.not_found('Error reading file')
93 def get_text_file(req, backend, mat):
95 return send_file(req, get_repo(backend, mat).get_named_file(mat.group()),
99 def get_loose_object(req, backend, mat):
100 sha = mat.group(1) + mat.group(2)
101 object_store = get_repo(backend, mat).object_store
102 if not object_store.contains_loose(sha):
103 yield req.not_found('Object not found')
106 data = object_store[sha].as_legacy_object()
108 yield req.not_found('Error reading object')
110 req.respond(HTTP_OK, 'application/x-git-loose-object')
114 def get_pack_file(req, backend, mat):
116 return send_file(req, get_repo(backend, mat).get_named_file(mat.group()),
117 'application/x-git-packed-objects')
120 def get_idx_file(req, backend, mat):
122 return send_file(req, get_repo(backend, mat).get_named_file(mat.group()),
123 'application/x-git-packed-objects-toc')
126 default_services = {'git-upload-pack': UploadPackHandler,
127 'git-receive-pack': ReceivePackHandler}
128 def get_info_refs(req, backend, mat, services=None):
130 services = default_services
131 params = parse_qs(req.environ['QUERY_STRING'])
132 service = params.get('service', [None])[0]
133 if service and not req.dumb:
134 handler_cls = services.get(service, None)
135 if handler_cls is None:
136 yield req.forbidden('Unsupported service %s' % service)
139 req.respond(HTTP_OK, 'application/x-%s-advertisement' % service)
141 dummy_input = StringIO() # GET request, handler doesn't need to read
142 handler = handler_cls(backend, [url_prefix(mat)],
143 dummy_input.read, output.write,
144 stateless_rpc=True, advertise_refs=True)
145 handler.proto.write_pkt_line('# service=%s\n' % service)
146 handler.proto.write_pkt_line(None)
148 yield output.getvalue()
151 # TODO: select_getanyfile() (see http-backend.c)
153 req.respond(HTTP_OK, 'text/plain')
154 repo = get_repo(backend, mat)
155 refs = repo.get_refs()
156 for name in sorted(refs.iterkeys()):
157 # get_refs() includes HEAD as a special case, but we don't want to
165 yield '%s\t%s\n' % (sha, name)
166 peeled_sha = repo.get_peeled(name)
167 if peeled_sha != sha:
168 yield '%s\t%s^{}\n' % (peeled_sha, name)
171 def get_info_packs(req, backend, mat):
173 req.respond(HTTP_OK, 'text/plain')
174 for pack in get_repo(backend, mat).object_store.packs:
175 yield 'P pack-%s.pack\n' % pack.name()
178 class _LengthLimitedFile(object):
179 """Wrapper class to limit the length of reads from a file-like object.
181 This is used to ensure EOF is read from the wsgi.input object once
182 Content-Length bytes are read. This behavior is required by the WSGI spec
183 but not implemented in wsgiref as of 2.5.
185 def __init__(self, input, max_bytes):
187 self._bytes_avail = max_bytes
189 def read(self, size=-1):
190 if self._bytes_avail <= 0:
192 if size == -1 or size > self._bytes_avail:
193 size = self._bytes_avail
194 self._bytes_avail -= size
195 return self._input.read(size)
197 # TODO: support more methods as necessary
200 def handle_service_request(req, backend, mat, services=None):
202 services = default_services
203 service = mat.group().lstrip('/')
204 handler_cls = services.get(service, None)
205 if handler_cls is None:
206 yield req.forbidden('Unsupported service %s' % service)
209 req.respond(HTTP_OK, 'application/x-%s-response' % service)
212 input = req.environ['wsgi.input']
213 # This is not necessary if this app is run from a conforming WSGI server.
214 # Unfortunately, there's no way to tell that at this point.
215 # TODO: git may used HTTP/1.1 chunked encoding instead of specifying
217 if 'CONTENT_LENGTH' in req.environ:
218 input = _LengthLimitedFile(input, int(req.environ['CONTENT_LENGTH']))
219 handler = handler_cls(backend, [url_prefix(mat)], input.read, output.write,
222 yield output.getvalue()
225 class HTTPGitRequest(object):
226 """Class encapsulating the state of a single git HTTP request.
228 :ivar environ: the WSGI environment for the request.
231 def __init__(self, environ, start_response, dumb=False):
232 self.environ = environ
234 self._start_response = start_response
235 self._cache_headers = []
238 def add_header(self, name, value):
239 """Add a header to the response."""
240 self._headers.append((name, value))
242 def respond(self, status=HTTP_OK, content_type=None, headers=None):
243 """Begin a response with the given status and other headers."""
245 self._headers.extend(headers)
247 self._headers.append(('Content-Type', content_type))
248 self._headers.extend(self._cache_headers)
250 self._start_response(status, self._headers)
252 def not_found(self, message):
253 """Begin a HTTP 404 response and return the text of a message."""
254 self._cache_headers = []
255 self.respond(HTTP_NOT_FOUND, 'text/plain')
258 def forbidden(self, message):
259 """Begin a HTTP 403 response and return the text of a message."""
260 self._cache_headers = []
261 self.respond(HTTP_FORBIDDEN, 'text/plain')
265 """Set the response to never be cached by the client."""
266 self._cache_headers = [
267 ('Expires', 'Fri, 01 Jan 1980 00:00:00 GMT'),
268 ('Pragma', 'no-cache'),
269 ('Cache-Control', 'no-cache, max-age=0, must-revalidate'),
272 def cache_forever(self):
273 """Set the response to be cached forever by the client."""
275 self._cache_headers = [
276 ('Date', date_time_string(now)),
277 ('Expires', date_time_string(now + 31536000)),
278 ('Cache-Control', 'public, max-age=31536000'),
282 class HTTPGitApplication(object):
283 """Class encapsulating the state of a git WSGI application.
285 :ivar backend: the Backend object backing this application
289 ('GET', re.compile('/HEAD$')): get_text_file,
290 ('GET', re.compile('/info/refs$')): get_info_refs,
291 ('GET', re.compile('/objects/info/alternates$')): get_text_file,
292 ('GET', re.compile('/objects/info/http-alternates$')): get_text_file,
293 ('GET', re.compile('/objects/info/packs$')): get_info_packs,
294 ('GET', re.compile('/objects/([0-9a-f]{2})/([0-9a-f]{38})$')): get_loose_object,
295 ('GET', re.compile('/objects/pack/pack-([0-9a-f]{40})\\.pack$')): get_pack_file,
296 ('GET', re.compile('/objects/pack/pack-([0-9a-f]{40})\\.idx$')): get_idx_file,
298 ('POST', re.compile('/git-upload-pack$')): handle_service_request,
299 ('POST', re.compile('/git-receive-pack$')): handle_service_request,
302 def __init__(self, backend, dumb=False):
303 self.backend = backend
306 def __call__(self, environ, start_response):
307 path = environ['PATH_INFO']
308 method = environ['REQUEST_METHOD']
309 req = HTTPGitRequest(environ, start_response, self.dumb)
310 # environ['QUERY_STRING'] has qs args
312 for smethod, spath in self.services.iterkeys():
313 if smethod != method:
315 mat = spath.search(path)
317 handler = self.services[smethod, spath]
320 return req.not_found('Sorry, that method is not supported')
321 return handler(req, self.backend, mat)