Remove the originated time from the DHT value storage.
[quix0rs-apt-p2p.git] / apt_dht_Khashmir / DHT.py
1
2 from datetime import datetime
3 import os, sha, random
4
5 from twisted.internet import defer, reactor
6 from twisted.internet.abstract import isIPAddress
7 from twisted.python import log
8 from twisted.trial import unittest
9 from zope.interface import implements
10
11 from apt_dht.interfaces import IDHT
12 from khashmir import Khashmir
13
14 khashmir_dir = 'apt-dht-Khashmir'
15
16 class DHTError(Exception):
17     """Represents errors that occur in the DHT."""
18
19 class DHT:
20     
21     implements(IDHT)
22     
23     def __init__(self):
24         self.config = None
25         self.cache_dir = ''
26         self.bootstrap = []
27         self.bootstrap_node = False
28         self.joining = None
29         self.joined = False
30         self.outstandingJoins = 0
31         self.foundAddrs = []
32         self.storing = {}
33         self.retrieving = {}
34         self.retrieved = {}
35     
36     def loadConfig(self, config, section):
37         """See L{apt_dht.interfaces.IDHT}."""
38         self.config_parser = config
39         self.section = section
40         self.config = {}
41         self.cache_dir = os.path.join(self.config_parser.get(section, 'cache_dir'), khashmir_dir)
42         if not os.path.exists(self.cache_dir):
43             os.makedirs(self.cache_dir)
44         self.bootstrap = self.config_parser.getstringlist(section, 'BOOTSTRAP')
45         self.bootstrap_node = self.config_parser.getboolean(section, 'BOOTSTRAP_NODE')
46         for k in self.config_parser.options(section):
47             if k in ['K', 'HASH_LENGTH', 'CONCURRENT_REQS', 'STORE_REDUNDANCY', 
48                      'MAX_FAILURES', 'PORT']:
49                 self.config[k] = self.config_parser.getint(section, k)
50             elif k in ['CHECKPOINT_INTERVAL', 'MIN_PING_INTERVAL', 
51                        'BUCKET_STALENESS', 'KEINITIAL_DELAY', 'KE_DELAY', 'KE_AGE']:
52                 self.config[k] = self.config_parser.gettime(section, k)
53             elif k in ['SPEW']:
54                 self.config[k] = self.config_parser.getboolean(section, k)
55             else:
56                 self.config[k] = self.config_parser.get(section, k)
57     
58     def join(self):
59         """See L{apt_dht.interfaces.IDHT}."""
60         if self.config is None:
61             raise DHTError, "configuration not loaded"
62         if self.joining:
63             raise DHTError, "a join is already in progress"
64
65         self.khashmir = Khashmir(self.config, self.cache_dir)
66         
67         self.joining = defer.Deferred()
68         for node in self.bootstrap:
69             host, port = node.rsplit(':', 1)
70             port = int(port)
71             if isIPAddress(host):
72                 self._join_gotIP(host, port)
73             else:
74                 reactor.resolve(host).addCallback(self._join_gotIP, port)
75         
76         return self.joining
77
78     def _join_gotIP(self, ip, port):
79         """Called after an IP address has been found for a single bootstrap node."""
80         self.outstandingJoins += 1
81         self.khashmir.addContact(ip, port, self._join_single, self._join_error)
82     
83     def _join_single(self, addr):
84         """Called when a single bootstrap node has been added."""
85         self.outstandingJoins -= 1
86         if addr:
87             self.foundAddrs.append(addr)
88         if addr or self.outstandingJoins <= 0:
89             self.khashmir.findCloseNodes(self._join_complete, self._join_complete)
90         log.msg('Got back from bootstrap node: %r' % (addr,))
91     
92     def _join_error(self, failure = None):
93         """Called when a single bootstrap node has failed."""
94         self.outstandingJoins -= 1
95         log.msg("bootstrap node could not be reached")
96         if self.outstandingJoins <= 0:
97             self.khashmir.findCloseNodes(self._join_complete, self._join_complete)
98
99     def _join_complete(self, result):
100         """Called when the tables have been initialized with nodes."""
101         if not self.joined and len(result) > 0:
102             self.joined = True
103         if self.joining and self.outstandingJoins <= 0:
104             df = self.joining
105             self.joining = None
106             if self.joined or self.bootstrap_node:
107                 self.joined = True
108                 df.callback(self.foundAddrs)
109             else:
110                 df.errback(DHTError('could not find any nodes to bootstrap to'))
111         
112     def getAddrs(self):
113         return self.foundAddrs
114         
115     def leave(self):
116         """See L{apt_dht.interfaces.IDHT}."""
117         if self.config is None:
118             raise DHTError, "configuration not loaded"
119         
120         if self.joined or self.joining:
121             if self.joining:
122                 self.joining.errback(DHTError('still joining when leave was called'))
123                 self.joining = None
124             self.joined = False
125             self.khashmir.shutdown()
126         
127     def getValue(self, key):
128         """See L{apt_dht.interfaces.IDHT}."""
129         if self.config is None:
130             raise DHTError, "configuration not loaded"
131         if not self.joined:
132             raise DHTError, "have not joined a network yet"
133
134         d = defer.Deferred()
135         if key not in self.retrieving:
136             self.khashmir.valueForKey(key, self._getValue)
137         self.retrieving.setdefault(key, []).append(d)
138         return d
139         
140     def _getValue(self, key, result):
141         if result:
142             self.retrieved.setdefault(key, []).extend(result)
143         else:
144             final_result = []
145             if key in self.retrieved:
146                 final_result = self.retrieved[key]
147                 del self.retrieved[key]
148             for i in range(len(self.retrieving[key])):
149                 d = self.retrieving[key].pop(0)
150                 d.callback(final_result)
151             del self.retrieving[key]
152
153     def storeValue(self, key, value):
154         """See L{apt_dht.interfaces.IDHT}."""
155         if self.config is None:
156             raise DHTError, "configuration not loaded"
157         if not self.joined:
158             raise DHTError, "have not joined a network yet"
159
160         if key in self.storing and value in self.storing[key]:
161             raise DHTError, "already storing that key with the same value"
162
163         d = defer.Deferred()
164         self.khashmir.storeValueForKey(key, value, self._storeValue)
165         self.storing.setdefault(key, {})[value] = d
166         return d
167     
168     def _storeValue(self, key, value, result):
169         if key in self.storing and value in self.storing[key]:
170             if len(result) > 0:
171                 self.storing[key][value].callback(result)
172             else:
173                 self.storing[key][value].errback(DHTError('could not store value %s in key %s' % (value, key)))
174             del self.storing[key][value]
175             if len(self.storing[key].keys()) == 0:
176                 del self.storing[key]
177
178 class TestSimpleDHT(unittest.TestCase):
179     """Unit tests for the DHT."""
180     
181     timeout = 2
182     DHT_DEFAULTS = {'PORT': 9977, 'K': 8, 'HASH_LENGTH': 160,
183                     'CHECKPOINT_INTERVAL': 900, 'CONCURRENT_REQS': 4,
184                     'STORE_REDUNDANCY': 3, 'MAX_FAILURES': 3,
185                     'MIN_PING_INTERVAL': 900,'BUCKET_STALENESS': 3600,
186                     'KEINITIAL_DELAY': 15, 'KE_DELAY': 1200,
187                     'KE_AGE': 3600, 'SPEW': False, }
188
189     def setUp(self):
190         self.a = DHT()
191         self.b = DHT()
192         self.a.config = self.DHT_DEFAULTS.copy()
193         self.a.config['PORT'] = 4044
194         self.a.bootstrap = ["127.0.0.1:4044"]
195         self.a.bootstrap_node = True
196         self.a.cache_dir = '/tmp'
197         self.b.config = self.DHT_DEFAULTS.copy()
198         self.b.config['PORT'] = 4045
199         self.b.bootstrap = ["127.0.0.1:4044"]
200         self.b.cache_dir = '/tmp'
201         
202     def test_bootstrap_join(self):
203         d = self.a.join()
204         return d
205         
206     def node_join(self, result):
207         d = self.b.join()
208         return d
209     
210     def test_join(self):
211         self.lastDefer = defer.Deferred()
212         d = self.a.join()
213         d.addCallback(self.node_join)
214         d.addCallback(self.lastDefer.callback)
215         return self.lastDefer
216
217     def value_stored(self, result, value):
218         self.stored -= 1
219         if self.stored == 0:
220             self.get_values()
221         
222     def store_values(self, result):
223         self.stored = 3
224         d = self.a.storeValue(sha.new('4045').digest(), str(4045*3))
225         d.addCallback(self.value_stored, 4045)
226         d = self.a.storeValue(sha.new('4044').digest(), str(4044*2))
227         d.addCallback(self.value_stored, 4044)
228         d = self.b.storeValue(sha.new('4045').digest(), str(4045*2))
229         d.addCallback(self.value_stored, 4045)
230
231     def check_values(self, result, values):
232         self.checked -= 1
233         self.failUnless(len(result) == len(values))
234         for v in result:
235             self.failUnless(v in values)
236         if self.checked == 0:
237             self.lastDefer.callback(1)
238     
239     def get_values(self):
240         self.checked = 4
241         d = self.a.getValue(sha.new('4044').digest())
242         d.addCallback(self.check_values, [str(4044*2)])
243         d = self.b.getValue(sha.new('4044').digest())
244         d.addCallback(self.check_values, [str(4044*2)])
245         d = self.a.getValue(sha.new('4045').digest())
246         d.addCallback(self.check_values, [str(4045*2), str(4045*3)])
247         d = self.b.getValue(sha.new('4045').digest())
248         d.addCallback(self.check_values, [str(4045*2), str(4045*3)])
249
250     def test_store(self):
251         from twisted.internet.base import DelayedCall
252         DelayedCall.debug = True
253         self.lastDefer = defer.Deferred()
254         d = self.a.join()
255         d.addCallback(self.node_join)
256         d.addCallback(self.store_values)
257         return self.lastDefer
258
259     def tearDown(self):
260         self.a.leave()
261         try:
262             os.unlink(self.a.khashmir.store.db)
263         except:
264             pass
265         self.b.leave()
266         try:
267             os.unlink(self.b.khashmir.store.db)
268         except:
269             pass
270
271 class TestMultiDHT(unittest.TestCase):
272     
273     timeout = 60
274     num = 20
275     DHT_DEFAULTS = {'PORT': 9977, 'K': 8, 'HASH_LENGTH': 160,
276                     'CHECKPOINT_INTERVAL': 900, 'CONCURRENT_REQS': 4,
277                     'STORE_REDUNDANCY': 3, 'MAX_FAILURES': 3,
278                     'MIN_PING_INTERVAL': 900,'BUCKET_STALENESS': 3600,
279                     'KEINITIAL_DELAY': 15, 'KE_DELAY': 1200,
280                     'KE_AGE': 3600, 'SPEW': False, }
281
282     def setUp(self):
283         self.l = []
284         self.startport = 4081
285         for i in range(self.num):
286             self.l.append(DHT())
287             self.l[i].config = self.DHT_DEFAULTS.copy()
288             self.l[i].config['PORT'] = self.startport + i
289             self.l[i].bootstrap = ["127.0.0.1:" + str(self.startport)]
290             self.l[i].cache_dir = '/tmp'
291         self.l[0].bootstrap_node = True
292         
293     def node_join(self, result, next_node):
294         d = self.l[next_node].join()
295         if next_node + 1 < len(self.l):
296             d.addCallback(self.node_join, next_node + 1)
297         else:
298             d.addCallback(self.lastDefer.callback)
299     
300     def test_join(self):
301         self.timeout = 2
302         self.lastDefer = defer.Deferred()
303         d = self.l[0].join()
304         d.addCallback(self.node_join, 1)
305         return self.lastDefer
306         
307     def store_values(self, result, i = 0, j = 0):
308         if j > i:
309             j -= i+1
310             i += 1
311         if i == len(self.l):
312             self.get_values()
313         else:
314             d = self.l[j].storeValue(sha.new(str(self.startport+i)).digest(), str((self.startport+i)*(j+1)))
315             d.addCallback(self.store_values, i, j+1)
316     
317     def get_values(self, result = None, check = None, i = 0, j = 0):
318         if result is not None:
319             self.failUnless(len(result) == len(check))
320             for v in result:
321                 self.failUnless(v in check)
322         if j >= len(self.l):
323             j -= len(self.l)
324             i += 1
325         if i == len(self.l):
326             self.lastDefer.callback(1)
327         else:
328             d = self.l[i].getValue(sha.new(str(self.startport+j)).digest())
329             check = []
330             for k in range(self.startport+j, (self.startport+j)*(j+1)+1, self.startport+j):
331                 check.append(str(k))
332             d.addCallback(self.get_values, check, i, j + random.randrange(1, min(len(self.l), 10)))
333
334     def store_join(self, result, next_node):
335         d = self.l[next_node].join()
336         if next_node + 1 < len(self.l):
337             d.addCallback(self.store_join, next_node + 1)
338         else:
339             d.addCallback(self.store_values)
340     
341     def test_store(self):
342         from twisted.internet.base import DelayedCall
343         DelayedCall.debug = True
344         self.lastDefer = defer.Deferred()
345         d = self.l[0].join()
346         d.addCallback(self.store_join, 1)
347         return self.lastDefer
348
349     def tearDown(self):
350         for i in self.l:
351             try:
352                 i.leave()
353                 os.unlink(i.khashmir.store.db)
354             except:
355                 pass