]> git.mxchange.org Git - quix0rs-apt-p2p.git/blobdiff - apt_dht_Khashmir/DHT.py
Use compact encoding of peer info for downloads.
[quix0rs-apt-p2p.git] / apt_dht_Khashmir / DHT.py
index 1c4416b5bc86c785d0953866b28e5f552683985b..23b2755313f0d1f91e7f08de7829674bb37ae614 100644 (file)
@@ -1,12 +1,18 @@
 
-import os
+from datetime import datetime
+import os, sha, random
 
-from twisted.internet import defer
+from twisted.internet import defer, reactor
+from twisted.internet.abstract import isIPAddress
+from twisted.python import log
 from twisted.trial import unittest
 from zope.interface import implements
 
 from apt_dht.interfaces import IDHT
 from khashmir import Khashmir
+from bencode import bencode, bdecode
+
+khashmir_dir = 'apt-dht-Khashmir'
 
 class DHTError(Exception):
     """Represents errors that occur in the DHT."""
@@ -22,13 +28,20 @@ class DHT:
         self.bootstrap_node = False
         self.joining = None
         self.joined = False
+        self.outstandingJoins = 0
+        self.foundAddrs = []
+        self.storing = {}
+        self.retrieving = {}
+        self.retrieved = {}
     
     def loadConfig(self, config, section):
         """See L{apt_dht.interfaces.IDHT}."""
         self.config_parser = config
         self.section = section
-        self.config = []
-        self.cache_dir = self.config_parser.get('DEFAULT', 'cache_dir')
+        self.config = {}
+        self.cache_dir = os.path.join(self.config_parser.get(section, 'cache_dir'), khashmir_dir)
+        if not os.path.exists(self.cache_dir):
+            os.makedirs(self.cache_dir)
         self.bootstrap = self.config_parser.getstringlist(section, 'BOOTSTRAP')
         self.bootstrap_node = self.config_parser.getboolean(section, 'BOOTSTRAP_NODE')
         for k in self.config_parser.options(section):
@@ -38,15 +51,17 @@ class DHT:
             elif k in ['CHECKPOINT_INTERVAL', 'MIN_PING_INTERVAL', 
                        'BUCKET_STALENESS', 'KEINITIAL_DELAY', 'KE_DELAY', 'KE_AGE']:
                 self.config[k] = self.config_parser.gettime(section, k)
+            elif k in ['SPEW']:
+                self.config[k] = self.config_parser.getboolean(section, k)
             else:
                 self.config[k] = self.config_parser.get(section, k)
-        if 'PORT' not in self.config:
-            self.config['PORT'] = self.config_parser.getint('DEFAULT', 'PORT')
     
     def join(self):
         """See L{apt_dht.interfaces.IDHT}."""
         if self.config is None:
             raise DHTError, "configuration not loaded"
+        if self.joining:
+            raise DHTError, "a join is already in progress"
 
         self.khashmir = Khashmir(self.config, self.cache_dir)
         
@@ -54,29 +69,59 @@ class DHT:
         for node in self.bootstrap:
             host, port = node.rsplit(':', 1)
             port = int(port)
-            self.khashmir.addContact(host, port, self._join_single)
+            if isIPAddress(host):
+                self._join_gotIP(host, port)
+            else:
+                reactor.resolve(host).addCallback(self._join_gotIP, port)
         
         return self.joining
+
+    def _join_gotIP(self, ip, port):
+        """Called after an IP address has been found for a single bootstrap node."""
+        self.outstandingJoins += 1
+        self.khashmir.addContact(ip, port, self._join_single, self._join_error)
     
-    def _join_single(self):
+    def _join_single(self, addr):
         """Called when a single bootstrap node has been added."""
-        self.khashmir.findCloseNodes(self._join_complete)
+        self.outstandingJoins -= 1
+        if addr:
+            self.foundAddrs.append(addr)
+        if addr or self.outstandingJoins <= 0:
+            self.khashmir.findCloseNodes(self._join_complete, self._join_complete)
+        log.msg('Got back from bootstrap node: %r' % (addr,))
     
+    def _join_error(self, failure = None):
+        """Called when a single bootstrap node has failed."""
+        self.outstandingJoins -= 1
+        log.msg("bootstrap node could not be reached")
+        if self.outstandingJoins <= 0:
+            self.khashmir.findCloseNodes(self._join_complete, self._join_complete)
+
     def _join_complete(self, result):
         """Called when the tables have been initialized with nodes."""
-        if not self.joined:
+        if not self.joined and len(result) > 0:
             self.joined = True
-            if len(result) > 0 or self.bootstrap_node:
-                self.joining.callback(result)
+        if self.joining and self.outstandingJoins <= 0:
+            df = self.joining
+            self.joining = None
+            if self.joined or self.bootstrap_node:
+                self.joined = True
+                df.callback(self.foundAddrs)
             else:
-                self.joining.errback(DHTError('could not find any nodes to bootstrap to'))
+                df.errback(DHTError('could not find any nodes to bootstrap to'))
+        
+    def getAddrs(self):
+        return self.foundAddrs
         
     def leave(self):
         """See L{apt_dht.interfaces.IDHT}."""
         if self.config is None:
             raise DHTError, "configuration not loaded"
         
-        if self.joined:
+        if self.joined or self.joining:
+            if self.joining:
+                self.joining.errback(DHTError('still joining when leave was called'))
+                self.joining = None
             self.joined = False
             self.khashmir.shutdown()
         
@@ -88,9 +133,24 @@ class DHT:
             raise DHTError, "have not joined a network yet"
 
         d = defer.Deferred()
-        self.khashmir.valueForKey(key, d.callback)
+        if key not in self.retrieving:
+            self.khashmir.valueForKey(key, self._getValue)
+        self.retrieving.setdefault(key, []).append(d)
         return d
         
+    def _getValue(self, key, result):
+        if result:
+            self.retrieved.setdefault(key, []).extend([bdecode(r) for r in result])
+        else:
+            final_result = []
+            if key in self.retrieved:
+                final_result = self.retrieved[key]
+                del self.retrieved[key]
+            for i in range(len(self.retrieving[key])):
+                d = self.retrieving[key].pop(0)
+                d.callback(final_result)
+            del self.retrieving[key]
+
     def storeValue(self, key, value):
         """See L{apt_dht.interfaces.IDHT}."""
         if self.config is None:
@@ -98,7 +158,25 @@ class DHT:
         if not self.joined:
             raise DHTError, "have not joined a network yet"
 
-        self.khashmir.storeValueForKey(key, value)
+        bvalue = bencode(value)
+
+        if key in self.storing and bvalue in self.storing[key]:
+            raise DHTError, "already storing that key with the same value"
+
+        d = defer.Deferred()
+        self.khashmir.storeValueForKey(key, bvalue, self._storeValue)
+        self.storing.setdefault(key, {})[bvalue] = d
+        return d
+    
+    def _storeValue(self, key, bvalue, result):
+        if key in self.storing and bvalue in self.storing[key]:
+            if len(result) > 0:
+                self.storing[key][bvalue].callback(result)
+            else:
+                self.storing[key][bvalue].errback(DHTError('could not store value %s in key %s' % (bvalue, key)))
+            del self.storing[key][bvalue]
+            if len(self.storing[key].keys()) == 0:
+                del self.storing[key]
 
 class TestSimpleDHT(unittest.TestCase):
     """Unit tests for the DHT."""
@@ -109,7 +187,7 @@ class TestSimpleDHT(unittest.TestCase):
                     'STORE_REDUNDANCY': 3, 'MAX_FAILURES': 3,
                     'MIN_PING_INTERVAL': 900,'BUCKET_STALENESS': 3600,
                     'KEINITIAL_DELAY': 15, 'KE_DELAY': 1200,
-                    'KE_AGE': 3600, }
+                    'KE_AGE': 3600, 'SPEW': False, }
 
     def setUp(self):
         self.a = DHT()
@@ -138,33 +216,75 @@ class TestSimpleDHT(unittest.TestCase):
         d.addCallback(self.node_join)
         d.addCallback(self.lastDefer.callback)
         return self.lastDefer
+
+    def value_stored(self, result, value):
+        self.stored -= 1
+        if self.stored == 0:
+            self.get_values()
         
+    def store_values(self, result):
+        self.stored = 3
+        d = self.a.storeValue(sha.new('4045').digest(), str(4045*3))
+        d.addCallback(self.value_stored, 4045)
+        d = self.a.storeValue(sha.new('4044').digest(), str(4044*2))
+        d.addCallback(self.value_stored, 4044)
+        d = self.b.storeValue(sha.new('4045').digest(), str(4045*2))
+        d.addCallback(self.value_stored, 4045)
+
+    def check_values(self, result, values):
+        self.checked -= 1
+        self.failUnless(len(result) == len(values))
+        for v in result:
+            self.failUnless(v in values)
+        if self.checked == 0:
+            self.lastDefer.callback(1)
+    
+    def get_values(self):
+        self.checked = 4
+        d = self.a.getValue(sha.new('4044').digest())
+        d.addCallback(self.check_values, [str(4044*2)])
+        d = self.b.getValue(sha.new('4044').digest())
+        d.addCallback(self.check_values, [str(4044*2)])
+        d = self.a.getValue(sha.new('4045').digest())
+        d.addCallback(self.check_values, [str(4045*2), str(4045*3)])
+        d = self.b.getValue(sha.new('4045').digest())
+        d.addCallback(self.check_values, [str(4045*2), str(4045*3)])
+
+    def test_store(self):
+        from twisted.internet.base import DelayedCall
+        DelayedCall.debug = True
+        self.lastDefer = defer.Deferred()
+        d = self.a.join()
+        d.addCallback(self.node_join)
+        d.addCallback(self.store_values)
+        return self.lastDefer
+
     def tearDown(self):
         self.a.leave()
         try:
-            os.unlink(self.a.khashmir.db)
+            os.unlink(self.a.khashmir.store.db)
         except:
             pass
         self.b.leave()
         try:
-            os.unlink(self.b.khashmir.db)
+            os.unlink(self.b.khashmir.store.db)
         except:
             pass
 
 class TestMultiDHT(unittest.TestCase):
     
-    timeout = 10
+    timeout = 60
     num = 20
     DHT_DEFAULTS = {'PORT': 9977, 'K': 8, 'HASH_LENGTH': 160,
                     'CHECKPOINT_INTERVAL': 900, 'CONCURRENT_REQS': 4,
                     'STORE_REDUNDANCY': 3, 'MAX_FAILURES': 3,
                     'MIN_PING_INTERVAL': 900,'BUCKET_STALENESS': 3600,
                     'KEINITIAL_DELAY': 15, 'KE_DELAY': 1200,
-                    'KE_AGE': 3600, }
+                    'KE_AGE': 3600, 'SPEW': False, }
 
     def setUp(self):
         self.l = []
-        self.startport = 4088
+        self.startport = 4081
         for i in range(self.num):
             self.l.append(DHT())
             self.l[i].config = self.DHT_DEFAULTS.copy()
@@ -181,15 +301,58 @@ class TestMultiDHT(unittest.TestCase):
             d.addCallback(self.lastDefer.callback)
     
     def test_join(self):
+        self.timeout = 2
         self.lastDefer = defer.Deferred()
         d = self.l[0].join()
         d.addCallback(self.node_join, 1)
         return self.lastDefer
         
+    def store_values(self, result, i = 0, j = 0):
+        if j > i:
+            j -= i+1
+            i += 1
+        if i == len(self.l):
+            self.get_values()
+        else:
+            d = self.l[j].storeValue(sha.new(str(self.startport+i)).digest(), str((self.startport+i)*(j+1)))
+            d.addCallback(self.store_values, i, j+1)
+    
+    def get_values(self, result = None, check = None, i = 0, j = 0):
+        if result is not None:
+            self.failUnless(len(result) == len(check))
+            for v in result:
+                self.failUnless(v in check)
+        if j >= len(self.l):
+            j -= len(self.l)
+            i += 1
+        if i == len(self.l):
+            self.lastDefer.callback(1)
+        else:
+            d = self.l[i].getValue(sha.new(str(self.startport+j)).digest())
+            check = []
+            for k in range(self.startport+j, (self.startport+j)*(j+1)+1, self.startport+j):
+                check.append(str(k))
+            d.addCallback(self.get_values, check, i, j + random.randrange(1, min(len(self.l), 10)))
+
+    def store_join(self, result, next_node):
+        d = self.l[next_node].join()
+        if next_node + 1 < len(self.l):
+            d.addCallback(self.store_join, next_node + 1)
+        else:
+            d.addCallback(self.store_values)
+    
+    def test_store(self):
+        from twisted.internet.base import DelayedCall
+        DelayedCall.debug = True
+        self.lastDefer = defer.Deferred()
+        d = self.l[0].join()
+        d.addCallback(self.store_join, 1)
+        return self.lastDefer
+
     def tearDown(self):
         for i in self.l:
             try:
                 i.leave()
-                os.unlink(i.khashmir.db)
+                os.unlink(i.khashmir.store.db)
             except:
                 pass