PyBitmessage-2021-04-27/src/inventory.py

142 lines
5.8 KiB
Python
Raw Normal View History

import collections
import random
from threading import current_thread, RLock
import time
from helper_sql import *
from singleton import Singleton
@Singleton
class Inventory(collections.MutableMapping):
def __init__(self):
super(self.__class__, self).__init__()
self._inventory = {} #of objects (like msg payloads and pubkey payloads) Does not include protocol headers (the first 24 bytes of each packet).
self.numberOfInventoryLookupsPerformed = 0
self._streams = collections.defaultdict(set) # key = streamNumer, value = a set which holds the inventory object hashes that we are aware of. This is used whenever we receive an inv message from a peer to check to see what items are new to us. We don't delete things out of it; instead, the singleCleaner thread clears and refills it every couple hours.
self.lock = RLock() # Guarantees that two receiveDataThreads don't receive and process the same message concurrently (probably sent by a malicious individual)
self.InventoryItem = collections.namedtuple('InventoryItem', 'type stream payload expires tag')
def __contains__(self, hash):
with self.lock:
self.numberOfInventoryLookupsPerformed += 1
if hash in self._inventory:
return True
return bool(sqlQuery('SELECT 1 FROM inventory WHERE hash=?', hash))
def __getitem__(self, hash):
with self.lock:
if hash in self._inventory:
return self._inventory[hash]
rows = sqlQuery('SELECT objecttype, streamnumber, payload, expirestime, tag FROM inventory WHERE hash=?', hash)
if not rows:
raise KeyError(hash)
return self.InventoryItem(*rows[0])
def __setitem__(self, hash, value):
with self.lock:
value = self.InventoryItem(*value)
self._inventory[hash] = value
self._streams[value.stream].add(hash)
def __delitem__(self, hash):
raise NotImplementedError
def __iter__(self):
with self.lock:
hashes = self._inventory.keys()[:]
hashes += (hash for hash, in sqlQuery('SELECT hash FROM inventory'))
return hashes.__iter__()
def __len__(self):
with self.lock:
return len(self._inventory) + sqlQuery('SELECT count(*) FROM inventory')[0][0]
def by_type_and_tag(self, type, tag):
with self.lock:
values = [value for value in self._inventory.values() if value.type == type and value.tag == tag]
values += (self.InventoryItem(*value) for value in sqlQuery('SELECT objecttype, streamnumber, payload, expirestime, tag FROM inventory WHERE objecttype=? AND tag=?', type, tag))
return values
def hashes_by_stream(self, stream):
with self.lock:
return self._streams[stream]
def unexpired_hashes_by_stream(self, stream):
with self.lock:
t = int(time.time())
hashes = [hash for hash, value in self._inventory.items() if value.stream == stream and value.expires > t]
hashes += (payload for payload, in sqlQuery('SELECT hash FROM inventory WHERE streamnumber=? AND expirestime>?', stream, t))
return hashes
def flush(self):
with self.lock: # If you use both the inventoryLock and the sqlLock, always use the inventoryLock OUTSIDE of the sqlLock.
with SqlBulkExecute() as sql:
for hash, value in self._inventory.items():
sql.execute('INSERT INTO inventory VALUES (?, ?, ?, ?, ?, ?)', hash, *value)
self._inventory.clear()
def clean(self):
with self.lock:
sqlExecute('DELETE FROM inventory WHERE expirestime<?',int(time.time()) - (60 * 60 * 3))
self._streams.clear()
for hash, value in self.items():
self._streams[value.stream].add(hash)
@Singleton
class Missing(object):
def __init__(self):
super(self.__class__, self).__init__()
self.lock = RLock()
self.hashes = {}
def add(self, objectHash):
with self.lock:
if not objectHash in self.hashes:
self.hashes[objectHash] = {'peers':[], 'requested':0}
self.hashes[objectHash]['peers'].append(current_thread().peer)
def len(self):
with self.lock:
return len(self.hashes)
def pull(self, count=1):
if count < 1:
raise ValueError("Must be at least one")
with self.lock:
now = time.time()
since = now - 300 # once every 5 minutes
try:
matchingHashes = {k:v for k, v in self.hashes.iteritems() if current_thread().peer in self.hashes[k]['peers'] and self.hashes[k]['requested'] < since}
if count > len(matchingHashes):
count = len(matchingHashes)
objectHashes = random.sample(matchingHashes, count)
except (IndexError, KeyError): # list is empty
return None
for objectHash in objectHashes:
try:
self.hashes[objectHash]['peers'].remove(current_thread().peer)
except ValueError:
pass
if len(self.hashes[objectHash]['peers']) == 0:
self.delete(objectHash)
else:
self.hashes[objectHash]['requested'] = now
return objectHashes
def delete(self, objectHash):
with self.lock:
if objectHash in self.hashes:
del self.hashes[objectHash]
def threadEnd(self):
with self.lock:
for objectHash in self.hashes:
try:
self.hashes[objectHash]['peers'].remove(current_thread().peer)
except ValueError:
pass
# current_thread().peer