1 from json import dumps, loads
4 import ast, time, threading, hashlib
5 from Queue import Queue
6 import traceback, sys, os, random
10 Hash = lambda x: hashlib.sha256(hashlib.sha256(x).digest()).digest()
11 hash_encode = lambda x: x[::-1].encode('hex')
12 hash_decode = lambda x: x.decode('hex')[::-1]
17 return s.decode('hex')[::-1].encode('hex')
20 def int_to_hex(i, length=1):
21 s = hex(i)[2:].rstrip('L')
22 s = "0"*(2*length - len(s)) + s
25 def header_to_string(res):
26 pbh = res.get('prev_block_hash')
27 if pbh is None: pbh = '0'*64
28 s = int_to_hex(res.get('version'),4) \
30 + rev_hex(res.get('merkle_root')) \
31 + int_to_hex(int(res.get('timestamp')),4) \
32 + int_to_hex(int(res.get('bits')),4) \
33 + int_to_hex(int(res.get('nonce')),4)
36 def header_from_string( s):
37 hex_to_int = lambda s: eval('0x' + s[::-1].encode('hex'))
39 h['version'] = hex_to_int(s[0:4])
40 h['prev_block_hash'] = hash_encode(s[4:36])
41 h['merkle_root'] = hash_encode(s[36:68])
42 h['timestamp'] = hex_to_int(s[68:72])
43 h['bits'] = hex_to_int(s[72:76])
44 h['nonce'] = hex_to_int(s[76:80])
50 from processor import Processor, print_log
52 class BlockchainProcessor(Processor):
54 def __init__(self, config, shared):
55 Processor.__init__(self)
58 self.up_to_date = False
59 self.watched_addresses = []
60 self.history_cache = {}
62 self.cache_lock = threading.Lock()
63 self.headers_data = ''
65 self.mempool_addresses = {}
66 self.mempool_hist = {}
67 self.mempool_hashes = []
68 self.mempool_lock = threading.Lock()
70 self.address_queue = Queue()
71 self.dbpath = config.get('leveldb', 'path')
73 self.dblock = threading.Lock()
75 self.db = leveldb.LevelDB(self.dbpath)
77 traceback.print_exc(file=sys.stdout)
80 self.bitcoind_url = 'http://%s:%s@%s:%s/' % (
81 config.get('bitcoind','user'),
82 config.get('bitcoind','password'),
83 config.get('bitcoind','host'),
84 config.get('bitcoind','port'))
89 self.sent_header = None
93 hist = self.deserialize(self.db.Get('height'))
94 self.last_hash, self.height, _ = hist[0]
95 print_log( "hist", hist )
97 #traceback.print_exc(file=sys.stdout)
98 print_log('initializing database')
100 self.last_hash = '000000000019d6689c085ae165831e934ff763ae46a2a6c172b3f1b60a8ce26f'
103 self.init_headers(self.height)
105 threading.Timer(0, lambda: self.catch_up(sync=False)).start()
106 while not shared.stopped() and not self.up_to_date:
110 print "keyboard interrupt: stopping threads"
114 print_log( "blockchain is up to date." )
116 threading.Timer(10, self.main_iteration).start()
120 def bitcoind(self, method, params=[]):
121 postdata = dumps({"method": method, 'params': params, 'id':'jsonrpc'})
122 respdata = urllib.urlopen(self.bitcoind_url, postdata).read()
124 if r['error'] != None:
125 raise BaseException(r['error'])
126 return r.get('result')
129 def serialize(self, h):
131 for txid, txpos, height in h:
132 s += txid + int_to_hex(txpos, 4) + int_to_hex(height, 4)
133 return s.decode('hex')
136 def deserialize(self, s):
139 txid = s[0:32].encode('hex')
140 txpos = int( rev_hex( s[32:36].encode('hex') ), 16 )
141 height = int( rev_hex( s[36:40].encode('hex') ), 16 )
142 h.append( ( txid, txpos, height ) )
147 def block2header(self, b):
148 return {"block_height":b.get('height'), "version":b.get('version'), "prev_block_hash":b.get('previousblockhash'),
149 "merkle_root":b.get('merkleroot'), "timestamp":b.get('time'), "bits":int(b.get('bits'),16), "nonce":b.get('nonce')}
152 def get_header(self, height):
153 block_hash = self.bitcoind('getblockhash', [height])
154 b = self.bitcoind('getblock', [block_hash])
155 return self.block2header(b)
158 def init_headers(self, db_height):
159 self.chunk_cache = {}
160 self.headers_filename = os.path.join( self.dbpath, 'blockchain_headers')
162 if os.path.exists(self.headers_filename):
163 height = os.path.getsize(self.headers_filename)/80 - 1 # the current height
165 prev_hash = self.hash_header(self.read_header(height))
169 open(self.headers_filename,'wb').close()
173 if height < db_height:
174 print_log( "catching up missing headers:", height, db_height)
177 while height != db_height:
179 header = self.get_header(height)
181 assert prev_hash == header.get('prev_block_hash')
182 self.write_header(header, sync=False)
183 prev_hash = self.hash_header(header)
184 if height%1000==0: print_log("headers file:",height)
185 except KeyboardInterrupt:
192 def hash_header(self, header):
193 return rev_hex(Hash(header_to_string(header).decode('hex')).encode('hex'))
196 def read_header(self, block_height):
197 if os.path.exists(self.headers_filename):
198 f = open(self.headers_filename,'rb')
199 f.seek(block_height*80)
203 h = header_from_string(h)
207 def read_chunk(self, index):
208 f = open(self.headers_filename,'rb')
209 f.seek(index*2016*80)
210 chunk = f.read(2016*80)
212 return chunk.encode('hex')
215 def write_header(self, header, sync=True):
216 if not self.headers_data:
217 self.headers_offset = header.get('block_height')
219 self.headers_data += header_to_string(header).decode('hex')
220 if sync or len(self.headers_data) > 40*100:
223 def pop_header(self):
224 # we need to do this only if we have not flushed
225 if self.headers_data:
226 self.headers_data = self.headers_data[:-40]
228 def flush_headers(self):
229 if not self.headers_data: return
230 f = open(self.headers_filename,'rb+')
231 f.seek(self.headers_offset*80)
232 f.write(self.headers_data)
234 self.headers_data = ''
237 def get_chunk(self, i):
238 # store them on disk; store the current chunk in memory
239 chunk = self.chunk_cache.get(i)
241 chunk = self.read_chunk(i)
242 self.chunk_cache[i] = chunk
246 def get_transaction(self, txid, block_height=-1, is_coinbase = False):
247 raw_tx = self.bitcoind('getrawtransaction', [txid, 0, block_height])
248 vds = deserialize.BCDataStream()
249 vds.write(raw_tx.decode('hex'))
250 out = deserialize.parse_Transaction(vds, is_coinbase)
254 def get_history(self, addr, cache_only=False):
255 with self.cache_lock: hist = self.history_cache.get( addr )
256 if hist is not None: return hist
257 if cache_only: return -1
261 hist = self.deserialize(self.db.Get(addr))
267 # should not be necessary
268 hist.sort( key=lambda tup: tup[1])
269 # check uniqueness too...
272 with self.mempool_lock:
273 for txid in self.mempool_hist.get(addr,[]):
274 hist.append((txid, 0, 0))
276 hist = map(lambda x: {'tx_hash':x[0], 'height':x[2]}, hist)
277 # add something to distinguish between unused and empty addresses
278 if hist == [] and is_known: hist = ['*']
280 with self.cache_lock: self.history_cache[addr] = hist
284 def get_status(self, addr, cache_only=False):
285 tx_points = self.get_history(addr, cache_only)
286 if cache_only and tx_points == -1: return -1
288 if not tx_points: return None
289 if tx_points == ['*']: return '*'
292 status += tx.get('tx_hash') + ':%d:' % tx.get('height')
293 return hashlib.sha256( status ).digest().encode('hex')
296 def get_merkle(self, tx_hash, height):
298 block_hash = self.bitcoind('getblockhash', [height])
299 b = self.bitcoind('getblock', [block_hash])
300 tx_list = b.get('tx')
301 tx_pos = tx_list.index(tx_hash)
303 merkle = map(hash_decode, tx_list)
304 target_hash = hash_decode(tx_hash)
306 while len(merkle) != 1:
307 if len(merkle)%2: merkle.append( merkle[-1] )
310 new_hash = Hash( merkle[0] + merkle[1] )
311 if merkle[0] == target_hash:
312 s.append( hash_encode( merkle[1]))
313 target_hash = new_hash
314 elif merkle[1] == target_hash:
315 s.append( hash_encode( merkle[0]))
316 target_hash = new_hash
321 return {"block_height":height, "merkle":s, "pos":tx_pos}
326 def add_to_history(self, addr, tx_hash, tx_pos, tx_height):
329 s = (tx_hash + int_to_hex(tx_pos, 4) + int_to_hex(tx_height, 4)).decode('hex')
331 serialized_hist = self.batch_list[addr]
333 l = len(serialized_hist)/40
334 for i in range(l-1, -1, -1):
335 item = serialized_hist[40*i:40*(i+1)]
336 item_height = int( rev_hex( item[36:40].encode('hex') ), 16 )
337 if item_height < tx_height:
338 serialized_hist = serialized_hist[0:40*(i+1)] + s + serialized_hist[40*(i+1):]
341 serialized_hist = s + serialized_hist
343 self.batch_list[addr] = serialized_hist
346 txo = (tx_hash + int_to_hex(tx_pos, 4)).decode('hex')
347 self.batch_txio[txo] = addr
350 def remove_from_history(self, tx_hash, tx_pos):
352 txi = (tx_hash + int_to_hex(tx_pos, 4)).decode('hex')
354 addr = self.batch_txio[txi]
356 raise BaseException(tx_hash, tx_pos)
357 print "WARNING: cannot find address for", (tx_hash, tx_pos)
360 serialized_hist = self.batch_list[addr]
362 l = len(serialized_hist)/40
364 item = serialized_hist[40*i:40*(i+1)]
365 if item[0:36] == txi:
366 height = int( rev_hex( item[36:40].encode('hex') ), 16 )
367 serialized_hist = serialized_hist[0:40*i] + serialized_hist[40*(i+1):]
370 raise BaseException("prevout not found", addr, hist, tx_hash, tx_pos)
372 self.batch_list[addr] = serialized_hist
376 def deserialize_block(self, block):
377 txlist = block.get('tx')
378 tx_hashes = [] # ordered txids
379 txdict = {} # deserialized tx
381 for raw_tx in txlist:
382 tx_hash = hash_encode(Hash(raw_tx.decode('hex')))
383 tx_hashes.append(tx_hash)
384 vds = deserialize.BCDataStream()
385 vds.write(raw_tx.decode('hex'))
386 tx = deserialize.parse_Transaction(vds, is_coinbase)
389 return tx_hashes, txdict
391 def get_undo_info(self, height):
392 s = self.db.Get("undo%d"%(height%100))
395 def write_undo_info(self, batch, height, undo_info):
396 batch.Put("undo%d"%(height%100), repr(undo_info))
399 def import_block(self, block, block_hash, block_height, sync, revert=False):
401 self.batch_list = {} # address -> history
402 self.batch_txio = {} # transaction i/o -> address
407 # deserialize transactions
409 tx_hashes, txdict = self.deserialize_block(block)
414 # read addresses of tx outputs
415 for tx_hash, tx in txdict.items():
416 for x in tx.get('outputs'):
417 txo = (tx_hash + int_to_hex(x.get('index'), 4)).decode('hex')
418 self.batch_txio[txo] = x.get('address')
420 # read addresses of tx inputs
421 for tx in txdict.values():
422 for x in tx.get('inputs'):
423 txi = (x.get('prevout_hash') + int_to_hex(x.get('prevout_n'), 4)).decode('hex')
424 inputs_to_read.append(txi)
426 inputs_to_read.sort()
427 for txi in inputs_to_read:
429 addr = self.db.Get(txi)
431 # the input could come from the same block
433 self.batch_txio[txi] = addr
434 addr_to_read.append(addr)
437 # read histories of addresses
438 for txid, tx in txdict.items():
439 for x in tx.get('outputs'):
440 addr_to_read.append(x.get('address'))
443 for addr in addr_to_read:
445 self.batch_list[addr] = self.db.Get(addr)
447 self.batch_list[addr] = ''
451 undo_info = self.get_undo_info(block_height)
452 print "undo", block_height, undo_info
458 for txid in tx_hashes: # must be ordered
463 for x in tx.get('inputs'):
464 prevout_height, prevout_addr = self.remove_from_history( x.get('prevout_hash'), x.get('prevout_n'))
465 undo.append( (prevout_height, prevout_addr) )
466 undo_info[txid] = undo
468 for x in tx.get('outputs'):
469 self.add_to_history( x.get('address'), txid, x.get('index'), block_height)
472 for x in tx.get('outputs'):
473 self.remove_from_history( txid, x.get('index'))
476 for x in tx.get('inputs'):
477 prevout_height, prevout_addr = undo_info.get(txid)[i]
480 # read the history into batch list
481 self.batch_list[prevout_addr] = self.db.Get(prevout_addr)
482 # re-add them to the history
483 self.add_to_history( prevout_addr, x.get('prevout_hash'), x.get('prevout_n'), prevout_height)
484 print "new hist", self.deserialize(self.batch_list[prevout_addr])
491 batch = leveldb.WriteBatch()
492 for addr, serialized_hist in self.batch_list.items():
493 batch.Put(addr, serialized_hist)
494 l = len(serialized_hist)
499 for txio, addr in self.batch_txio.items():
500 batch.Put(txio, addr)
501 # delete spent inputs
502 for txi in inputs_to_read:
506 if not revert: self.write_undo_info(batch, block_height, undo_info)
509 batch.Put('height', self.serialize( [(block_hash, block_height, 0)] ) )
512 self.db.Write(batch, sync = sync)
515 if t3 - t0 > 10 and not sync:
516 print_log("block", block_height,
517 "parse:%0.2f "%(t00 - t0),
518 "read:%0.2f "%(t1 - t00),
519 "proc:%.2f "%(t2-t1),
520 "write:%.2f "%(t3-t2),
521 "max:", max_len, max_addr)
523 for addr in self.batch_list.keys(): self.invalidate_cache(addr)
527 def add_request(self, request):
528 # see if we can get if from cache. if not, add to queue
529 if self.process( request, cache_only = True) == -1:
530 self.queue.put(request)
534 def process(self, request, cache_only = False):
535 #print "abe process", request
537 message_id = request['id']
538 method = request['method']
539 params = request.get('params',[])
543 if method == 'blockchain.numblocks.subscribe':
546 elif method == 'blockchain.headers.subscribe':
549 elif method == 'blockchain.address.subscribe':
552 result = self.get_status(address, cache_only)
553 self.watch_address(address)
554 except BaseException, e:
555 error = str(e) + ': ' + address
556 print_log( "error:", error )
558 elif method == 'blockchain.address.subscribe2':
561 result = self.get_status(address, cache_only)
562 self.watch_address(address)
563 except BaseException, e:
564 error = str(e) + ': ' + address
565 print_log( "error:", error )
567 elif method == 'blockchain.address.get_history2':
570 result = self.get_history( address, cache_only )
571 except BaseException, e:
572 error = str(e) + ': ' + address
573 print_log( "error:", error )
575 elif method == 'blockchain.block.get_header':
581 result = self.get_header( height )
582 except BaseException, e:
583 error = str(e) + ': %d'% height
584 print_log( "error:", error )
586 elif method == 'blockchain.block.get_chunk':
592 result = self.get_chunk( index )
593 except BaseException, e:
594 error = str(e) + ': %d'% index
595 print_log( "error:", error)
597 elif method == 'blockchain.transaction.broadcast':
598 txo = self.bitcoind('sendrawtransaction', params)
599 print_log( "sent tx:", txo )
602 elif method == 'blockchain.transaction.get_merkle':
608 tx_height = params[1]
609 result = self.get_merkle(tx_hash, tx_height)
610 except BaseException, e:
611 error = str(e) + ': ' + tx_hash
612 print_log( "error:", error )
614 elif method == 'blockchain.transaction.get':
618 result = self.bitcoind('getrawtransaction', [tx_hash, 0, height] )
619 except BaseException, e:
620 error = str(e) + ': ' + tx_hash
621 print_log( "error:", error )
624 error = "unknown method:%s"%method
626 if cache_only and result == -1: return -1
629 response = { 'id':message_id, 'error':error }
630 self.push_response(response)
632 response = { 'id':message_id, 'result':result }
633 self.push_response(response)
636 def watch_address(self, addr):
637 if addr not in self.watched_addresses:
638 self.watched_addresses.append(addr)
642 def catch_up(self, sync = True):
646 while not self.shared.stopped():
649 info = self.bitcoind('getinfo')
650 bitcoind_height = info.get('blocks')
651 bitcoind_block_hash = self.bitcoind('getblockhash', [bitcoind_height])
652 if self.last_hash == bitcoind_block_hash:
653 self.up_to_date = True
657 self.up_to_date = False
658 next_block_hash = self.bitcoind('getblockhash', [self.height+1])
659 next_block = self.bitcoind('getblock', [next_block_hash, 1])
661 revert = (random.randint(1, 1000)!=1) if self.is_test else False
662 if (next_block.get('previousblockhash') == self.last_hash) and not revert:
664 self.import_block(next_block, next_block_hash, self.height+1, sync)
665 self.height = self.height + 1
666 self.write_header(self.block2header(next_block), sync)
667 self.last_hash = next_block_hash
669 if (self.height)%100 == 0 and not sync:
671 print_log( "catch_up: block %d (%.3fs)"%( self.height, t2 - t1 ) )
675 # revert current block
676 block = self.bitcoind('getblock', [self.last_hash, 1])
677 print_log( "blockchain reorg", self.height, block.get('previousblockhash'), self.last_hash )
678 self.import_block(block, self.last_hash, self.height, sync, revert=True)
682 self.height = self.height -1
684 # read previous header from disk
685 self.header = self.read_header(self.height)
686 self.last_hash = self.hash_header(self.header)
689 self.header = self.block2header(self.bitcoind('getblock', [self.last_hash]))
694 def memorypool_update(self):
696 mempool_hashes = self.bitcoind('getrawmempool')
698 for tx_hash in mempool_hashes:
699 if tx_hash in self.mempool_hashes: continue
701 tx = self.get_transaction(tx_hash)
704 for x in tx.get('inputs'):
705 txi = (x.get('prevout_hash') + int_to_hex(x.get('prevout_n'), 4)).decode('hex')
707 addr = self.db.Get(txi)
710 l = self.mempool_addresses.get(tx_hash, [])
713 self.mempool_addresses[tx_hash] = l
715 for x in tx.get('outputs'):
716 addr = x.get('address')
717 l = self.mempool_addresses.get(tx_hash, [])
720 self.mempool_addresses[tx_hash] = l
722 self.mempool_hashes.append(tx_hash)
724 # remove older entries from mempool_hashes
725 self.mempool_hashes = mempool_hashes
727 # remove deprecated entries from mempool_addresses
728 for tx_hash, addresses in self.mempool_addresses.items():
729 if tx_hash not in self.mempool_hashes:
730 self.mempool_addresses.pop(tx_hash)
733 new_mempool_hist = {}
734 for tx_hash, addresses in self.mempool_addresses.items():
735 for addr in addresses:
736 h = new_mempool_hist.get(addr, [])
739 new_mempool_hist[addr] = h
741 for addr in new_mempool_hist.keys():
742 if addr in self.mempool_hist.keys():
743 if self.mempool_hist[addr] != new_mempool_hist[addr]:
744 self.invalidate_cache(addr)
746 self.invalidate_cache(addr)
748 with self.mempool_lock:
749 self.mempool_hist = new_mempool_hist
753 def invalidate_cache(self, address):
754 with self.cache_lock:
755 if self.history_cache.has_key(address):
756 print_log( "cache: invalidating", address )
757 self.history_cache.pop(address)
759 if address in self.watched_addresses:
760 self.address_queue.put(address)
764 def main_iteration(self):
766 if self.shared.stopped():
767 print_log( "blockchain processor terminating" )
775 self.memorypool_update()
777 # print "mempool:", len(self.mempool_addresses), len(self.mempool_hist), "%.3fs"%(t3 - t2)
780 if self.sent_height != self.height:
781 self.sent_height = self.height
782 self.push_response({ 'id': None, 'method':'blockchain.numblocks.subscribe', 'params':[self.height] })
784 if self.sent_header != self.header:
785 print_log( "blockchain: %d (%.3fs)"%( self.height, t2 - t1 ) )
786 self.sent_header = self.header
787 self.push_response({ 'id': None, 'method':'blockchain.headers.subscribe', 'params':[self.header] })
791 addr = self.address_queue.get(False)
794 if addr in self.watched_addresses:
795 status = self.get_status( addr )
796 self.push_response({ 'id': None, 'method':'blockchain.address.subscribe', 'params':[addr, status] })
797 self.push_response({ 'id': None, 'method':'blockchain.address.subscribe2', 'params':[addr, status] })
800 if not self.shared.stopped():
801 threading.Timer(10, self.main_iteration).start()
803 print_log( "blockchain processor terminating" )