2 from twisted.internet import reactor, defer, protocol
3 from twisted.internet.protocol import ClientFactory
4 from twisted.web2.client.interfaces import IHTTPClientManager
5 from twisted.web2.client.http import ProtocolError, ClientRequest, HTTPClientProtocol
6 from twisted.trial import unittest
7 from zope.interface import implements
8 from twisted.web2 import stream as stream_mod, http, http_headers, responsecode
10 class HTTPClientManager(ClientFactory):
11 """A manager for all HTTP requests to a single site.
13 Controls all requests that got to a single site (host and port).
14 This includes buffering requests until they can be sent and reconnecting
15 in the even of the connection being closed.
19 implements(IHTTPClientManager)
21 def __init__(self, host, port):
27 self.connecting = False
28 self.request_queue = []
29 self.response_queue = []
34 assert(self.closed and not self.connecting)
35 self.connecting = True
36 d = protocol.ClientCreator(reactor, HTTPClientProtocol, self).connectTCP(self.host, self.port)
37 d.addCallback(self.connected)
39 def connected(self, proto):
41 self.connecting = False
47 self.proto.transport.loseConnection()
50 return not self.busy and not self.request_queue and not self.response_queue
52 def submitRequest(self, request):
53 request.deferRequest = defer.Deferred()
54 self.request_queue.append(request)
56 return request.deferRequest
58 def processQueue(self):
59 if not self.request_queue:
66 if self.busy and not self.pipeline:
68 if self.response_queue and not self.pipeline:
71 req = self.request_queue.pop(0)
72 self.response_queue.append(req)
73 req.deferResponse = self.proto.submitRequest(req, False)
74 req.deferResponse.addCallback(self.requestComplete)
75 req.deferResponse.addErrback(self.requestError)
77 def requestComplete(self, resp):
78 req = self.response_queue.pop(0)
79 req.deferRequest.callback(resp)
81 def requestError(self, error):
82 req = self.response_queue.pop(0)
83 req.deferRequest.errback(error)
85 def clientBusy(self, proto):
88 def clientIdle(self, proto):
92 def clientPipelining(self, proto):
96 def clientGone(self, proto):
97 for req in self.response_queue:
98 req.deferRequest.errback(ProtocolError('lost connection'))
100 self.pipeline = False
102 self.connecting = False
103 self.response_queue = []
105 if self.request_queue:
108 class HTTPDownloader:
109 """Manages all the HTTP connections to various sites."""
114 def get(self, host, port, request):
115 site = host + ":" + str(port)
116 if site not in self.clients:
117 self.clients[site] = HTTPClientManager(host, port)
118 return self.clients[site].submitRequest(request)
121 for site in self.clients:
122 self.clients[site].close()
125 class TestClientManager(unittest.TestCase):
126 """Unit tests for the HTTPClientManager."""
131 def gotResp(self, resp, num, expect):
132 self.failUnless(resp.code >= 200 and resp.code < 300, "Got a non-200 response: %r" % resp.code)
133 if expect is not None:
134 self.failUnless(resp.stream.length == expect, "Length was incorrect, got %r, expected %r" % (resp.stream.length, expect))
139 stream_mod.readStream(resp.stream, print_).addCallback(printdone)
141 def test_download(self):
142 host = 'www.camrdale.org'
143 self.client = HTTPClientManager(host, 80)
145 lastDefer = defer.Deferred()
147 d = self.client.submitRequest(ClientRequest("GET", '/robots.txt', {'Host':host}, None))
148 d.addCallback(self.gotResp, 1, 309)
149 d.addBoth(lastDefer.callback)
153 host = 'www.camrdale.org'
154 self.client = HTTPClientManager(host, 80)
156 lastDefer = defer.Deferred()
158 d = self.client.submitRequest(ClientRequest("HEAD", '/robots.txt', {'Host':host}, None))
159 d.addCallback(self.gotResp, 1, 0)
160 d.addBoth(lastDefer.callback)
163 def test_multiple_downloads(self):
164 host = 'www.camrdale.org'
165 self.client = HTTPClientManager(host, 80)
167 lastDefer = defer.Deferred()
169 def newRequest(path, num, expect, last=False):
170 d = self.client.submitRequest(ClientRequest("GET", path, {'Host':host}, None))
171 d.addCallback(self.gotResp, num, expect)
173 d.addCallback(lastDefer.callback)
175 newRequest("/", 1, 3433)
176 newRequest("/blog/", 2, 37121)
177 newRequest("/camrdale.html", 3, 2234)
178 self.pending_calls.append(reactor.callLater(1, newRequest, '/robots.txt', 4, 309))
179 self.pending_calls.append(reactor.callLater(10, newRequest, '/wikilink.html', 5, 3084))
180 self.pending_calls.append(reactor.callLater(30, newRequest, '/sitemap.html', 6, 4750))
181 self.pending_calls.append(reactor.callLater(31, newRequest, '/PlanetLab.html', 7, 2783))
182 self.pending_calls.append(reactor.callLater(32, newRequest, '/openid.html', 8, 2525))
183 self.pending_calls.append(reactor.callLater(32, newRequest, '/subpage.html', 9, 2381))
184 self.pending_calls.append(reactor.callLater(62, newRequest, '/sitemap2.rss', 0, 302362, True))
187 def test_range(self):
188 host = 'www.camrdale.org'
189 self.client = HTTPClientManager(host, 80)
191 lastDefer = defer.Deferred()
193 d = self.client.submitRequest(ClientRequest("GET", '/robots.txt', {'Host':host, 'Range': ('bytes', [(100, 199)])}, None))
194 d.addCallback(self.gotResp, 1, 100)
195 d.addBoth(lastDefer.callback)
199 for p in self.pending_calls:
202 self.pending_calls = []
207 class TestDownloader(unittest.TestCase):
208 """Unit tests for the HTTPDownloader."""
213 def gotResp(self, resp, num, expect):
214 self.failUnless(resp.code >= 200 and resp.code < 300, "Got a non-200 response: %r" % resp.code)
215 if expect is not None:
216 self.failUnless(resp.stream.length == expect, "Length was incorrect, got %r, expected %r" % (resp.stream.length, expect))
221 stream_mod.readStream(resp.stream, print_).addCallback(printdone)
223 def test_download(self):
224 self.manager = HTTPDownloader()
226 lastDefer = defer.Deferred()
228 host = 'www.camrdale.org'
229 d = self.manager.get(host, 80, ClientRequest("GET", '/robots.txt', {'Host':host}, None))
230 d.addCallback(self.gotResp, 1, 309)
231 d.addBoth(lastDefer.callback)
235 self.manager = HTTPDownloader()
237 lastDefer = defer.Deferred()
239 host = 'www.camrdale.org'
240 d = self.manager.get(host, 80, ClientRequest("HEAD", '/robots.txt', {'Host':host}, None))
241 d.addCallback(self.gotResp, 1, 0)
242 d.addBoth(lastDefer.callback)
245 def test_multiple_downloads(self):
246 self.manager = HTTPDownloader()
248 lastDefer = defer.Deferred()
250 def newRequest(host, path, num, expect, last=False):
251 d = self.manager.get(host, 80, ClientRequest("GET", path, {'Host':host}, None))
252 d.addCallback(self.gotResp, num, expect)
254 d.addCallback(lastDefer.callback)
256 newRequest('www.camrdale.org', "/", 1, 3433)
257 newRequest('www.camrdale.org', "/blog/", 2, 37121)
258 newRequest('www.google.ca', "/", 3, None)
259 self.pending_calls.append(reactor.callLater(1, newRequest, 'www.sfu.ca', '/', 4, None))
260 self.pending_calls.append(reactor.callLater(10, newRequest, 'www.camrdale.org', '/wikilink.html', 5, 3084))
261 self.pending_calls.append(reactor.callLater(30, newRequest, 'www.camrdale.org', '/sitemap.html', 6, 4750))
262 self.pending_calls.append(reactor.callLater(31, newRequest, 'www.sfu.ca', '/studentcentral/index.html', 7, None))
263 self.pending_calls.append(reactor.callLater(32, newRequest, 'www.camrdale.org', '/openid.html', 8, 2525))
264 self.pending_calls.append(reactor.callLater(32, newRequest, 'www.camrdale.org', '/subpage.html', 9, 2381))
265 self.pending_calls.append(reactor.callLater(62, newRequest, 'www.google.ca', '/intl/en/options/', 0, None, True))
268 def test_range(self):
269 self.manager = HTTPDownloader()
271 lastDefer = defer.Deferred()
273 host = 'www.camrdale.org'
274 d = self.manager.get(host, 80, ClientRequest("GET", '/robots.txt', {'Host':host, 'Range': ('bytes', [(100, 199)])}, None))
275 d.addCallback(self.gotResp, 1, 100)
276 d.addBoth(lastDefer.callback)
280 for p in self.pending_calls:
283 self.pending_calls = []
285 self.manager.closeAll()