def extract_addresses_from_txlist(tx_hashes_tx, _getrawtransaction_batch):
"""
helper for extract_addresses, seperated so we can pass in a mocked _getrawtransaction_batch for test purposes
"""
logger.debug('extract_addresses_from_txlist, txs: %d' % (len(tx_hashes_tx.keys()), ))
tx_hashes_addresses = {}
tx_inputs_hashes = set() # use set to avoid duplicates
for tx_hash, tx in tx_hashes_tx.items():
tx_hashes_addresses[tx_hash] = set()
for vout in tx['vout']:
if 'addresses' in vout['scriptPubKey']:
tx_hashes_addresses[tx_hash].update(tuple(vout['scriptPubKey']['addresses']))
tx_inputs_hashes.update([vin['txid'] for vin in tx['vin']])
logger.debug('extract_addresses, input TXs: %d' % (len(tx_inputs_hashes), ))
# chunk txs to avoid huge memory spikes
for tx_inputs_hashes_chunk in util.chunkify(list(tx_inputs_hashes), config.BACKEND_RAW_TRANSACTIONS_CACHE_SIZE):
raw_transactions = _getrawtransaction_batch(tx_inputs_hashes_chunk, verbose=True)
for tx_hash, tx in tx_hashes_tx.items():
for vin in tx['vin']:
vin_tx = raw_transactions.get(vin['txid'], None)
if not vin_tx:
continue
vout = vin_tx['vout'][vin['vout']]
if 'addresses' in vout['scriptPubKey']:
tx_hashes_addresses[tx_hash].update(tuple(vout['scriptPubKey']['addresses']))
return tx_hashes_addresses, tx_hashes_tx
def extract_addresses(txhash_list):
logger.debug('extract_addresses, txs: %d' % (len(txhash_list), ))
tx_hashes_tx = getrawtransaction_batch(txhash_list, verbose=True)
tx_hashes_addresses = {}
tx_inputs_hashes = set() # use set to avoid duplicates
for tx_hash, tx in tx_hashes_tx.items():
tx_hashes_addresses[tx_hash] = set()
for vout in tx['vout']:
if 'addresses' in vout['scriptPubKey']:
tx_hashes_addresses[tx_hash].update(tuple(vout['scriptPubKey']['addresses']))
tx_inputs_hashes.update([vin['txid'] for vin in tx['vin']])
logger.debug('extract_addresses, input TXs: %d' % (len(tx_inputs_hashes), ))
# chunk txs to avoid huge memory spikes
for tx_inputs_hashes_chunk in util.chunkify(list(tx_inputs_hashes), config.BACKEND_RAW_TRANSACTIONS_CACHE_SIZE):
raw_transactions = getrawtransaction_batch(tx_inputs_hashes_chunk, verbose=True)
for tx_hash, tx in tx_hashes_tx.items():
for vin in tx['vin']:
vin_tx = raw_transactions.get(vin['txid'], None)
if not vin_tx:
continue
vout = vin_tx['vout'][vin['vout']]
if 'addresses' in vout['scriptPubKey']:
tx_hashes_addresses[tx_hash].update(tuple(vout['scriptPubKey']['addresses']))
return tx_hashes_addresses, tx_hashes_tx
def rpc_batch(request_list):
responses = collections.deque()
def make_call(chunk):
#send a list of requests to bitcoind to be executed
#note that this is list executed serially, in the same thread in bitcoind
#e.g. see: https://github.com/bitcoin/bitcoin/blob/master/src/rpcserver.cpp#L939
responses.extend(rpc_call(chunk))
chunks = util.chunkify(request_list, config.RPC_BATCH_SIZE)
with concurrent.futures.ThreadPoolExecutor(max_workers=config.BACKEND_RPC_BATCH_NUM_WORKERS) as executor:
for chunk in chunks:
executor.submit(make_call, chunk)
return list(responses)
def rpc_batch(request_list):
responses = collections.deque()
def make_call(chunk):
#send a list of requests to bitcoind to be executed
#note that this is list executed serially, in the same thread in bitcoind
#e.g. see: https://github.com/bitcoin/bitcoin/blob/master/src/rpcserver.cpp#L939
responses.extend(rpc_call(chunk))
chunks = util.chunkify(request_list, config.RPC_BATCH_SIZE)
with concurrent.futures.ThreadPoolExecutor(max_workers=config.BACKEND_RPC_BATCH_NUM_WORKERS) as executor:
for chunk in chunks:
executor.submit(make_call, chunk)
return list(responses)
def getrawtransaction_batch(txhash_list,
verbose=False,
skip_missing=False,
_retry=0):
_logger = logger.getChild("getrawtransaction_batch")
if len(txhash_list) > config.BACKEND_RAW_TRANSACTIONS_CACHE_SIZE:
#don't try to load in more than BACKEND_RAW_TRANSACTIONS_CACHE_SIZE entries in a single call
txhash_list_chunks = util.chunkify(
txhash_list, config.BACKEND_RAW_TRANSACTIONS_CACHE_SIZE)
txes = {}
for txhash_list_chunk in txhash_list_chunks:
txes.update(
getrawtransaction_batch(txhash_list_chunk,
verbose=verbose,
skip_missing=skip_missing))
return txes
tx_hash_call_id = {}
payload = []
noncached_txhashes = set()
txhash_list = set(txhash_list)
# payload for transactions not in cache
for tx_hash in txhash_list:
if tx_hash not in raw_transactions_cache:
call_id = binascii.hexlify(os.urandom(5)).decode('utf8')
payload.append({
"method": 'getrawtransaction',
"params": [tx_hash, 1],
"jsonrpc": "2.0",
"id": call_id
})
noncached_txhashes.add(tx_hash)
tx_hash_call_id[call_id] = tx_hash
#refresh any/all cache entries that already exist in the cache,
# so they're not inadvertently removed by another thread before we can consult them
#(this assumes that the size of the working set for any given workload doesn't exceed the max size of the cache)
for tx_hash in txhash_list.difference(noncached_txhashes):
raw_transactions_cache.refresh(tx_hash)
_logger.debug(
"getrawtransaction_batch: txhash_list size: {} / raw_transactions_cache size: {} / # getrawtransaction calls: {}"
.format(len(txhash_list), len(raw_transactions_cache), len(payload)))
# populate cache
if len(payload) > 0:
batch_responses = rpc_batch(payload)
for response in batch_responses:
if 'error' not in response or response['error'] is None:
tx_hex = response['result']
tx_hash = tx_hash_call_id[response['id']]
raw_transactions_cache[tx_hash] = tx_hex
elif skip_missing and 'error' in response and response['error'][
'code'] == -5:
raw_transactions_cache[tx_hash] = None
logging.debug(
'Missing TX with no raw info skipped (txhash: {}): {}'.
format(tx_hash_call_id.get(response.get('id', '??'), '??'),
response['error']))
else:
#TODO: this seems to happen for bogus transactions? Maybe handle it more gracefully than just erroring out?
raise BackendRPCError('{} (txhash:: {})'.format(
response['error'],
tx_hash_call_id.get(response.get('id', '??'), '??')))
# get transactions from cache
result = {}
for tx_hash in txhash_list:
try:
if verbose:
result[tx_hash] = raw_transactions_cache[tx_hash]
else:
result[tx_hash] = raw_transactions_cache[tx_hash][
'hex'] if raw_transactions_cache[
tx_hash] is not None else None
except KeyError as e: #shows up most likely due to finickyness with addrindex not always returning results that we need...
_logger.warning(
"tx missing in rawtx cache: {} -- txhash_list size: {}, hash: {} / raw_transactions_cache size: {} / # rpc_batch calls: {} / txhash in noncached_txhashes: {} / txhash in txhash_list: {} -- list {}"
.format(
e, len(txhash_list),
hashlib.md5(json.dumps(
list(txhash_list)).encode()).hexdigest(),
len(raw_transactions_cache), len(payload), tx_hash
in noncached_txhashes, tx_hash in txhash_list,
list(txhash_list.difference(noncached_txhashes))))
if _retry < GETRAWTRANSACTION_MAX_RETRIES: #try again
time.sleep(
0.05 * (_retry + 1)
) # Wait a bit, hitting the index non-stop may cause it to just break down... TODO: Better handling
r = getrawtransaction_batch([tx_hash],
verbose=verbose,
skip_missing=skip_missing,
_retry=_retry + 1)
result[tx_hash] = r[tx_hash]
else:
raise #already tried again, give up
return result
def getrawtransaction_batch(txhash_list, verbose=False, skip_missing=False, _retry=0):
_logger = logger.getChild("getrawtransaction_batch")
if len(txhash_list) > config.BACKEND_RAW_TRANSACTIONS_CACHE_SIZE:
#don't try to load in more than BACKEND_RAW_TRANSACTIONS_CACHE_SIZE entries in a single call
txhash_list_chunks = util.chunkify(txhash_list, config.BACKEND_RAW_TRANSACTIONS_CACHE_SIZE)
txes = {}
for txhash_list_chunk in txhash_list_chunks:
txes.update(getrawtransaction_batch(txhash_list_chunk, verbose=verbose, skip_missing=skip_missing))
return txes
tx_hash_call_id = {}
payload = []
noncached_txhashes = set()
txhash_list = set(txhash_list)
# payload for transactions not in cache
for tx_hash in txhash_list:
if tx_hash not in raw_transactions_cache:
call_id = binascii.hexlify(os.urandom(5)).decode('utf8')
payload.append({
"method": 'getrawtransaction',
"params": [tx_hash, 1],
"jsonrpc": "2.0",
"id": call_id
})
noncached_txhashes.add(tx_hash)
tx_hash_call_id[call_id] = tx_hash
#refresh any/all cache entries that already exist in the cache,
# so they're not inadvertently removed by another thread before we can consult them
#(this assumes that the size of the working set for any given workload doesn't exceed the max size of the cache)
for tx_hash in txhash_list.difference(noncached_txhashes):
raw_transactions_cache.refresh(tx_hash)
_logger.debug("getrawtransaction_batch: txhash_list size: {} / raw_transactions_cache size: {} / # getrawtransaction calls: {}".format(
len(txhash_list), len(raw_transactions_cache), len(payload)))
# populate cache
if len(payload) > 0:
batch_responses = rpc_batch(payload)
for response in batch_responses:
if 'error' not in response or response['error'] is None:
tx_hex = response['result']
tx_hash = tx_hash_call_id[response['id']]
raw_transactions_cache[tx_hash] = tx_hex
elif skip_missing and 'error' in response and response['error']['code'] == -5:
raw_transactions_cache[tx_hash] = None
logging.debug('Missing TX with no raw info skipped (txhash: {}): {}'.format(
tx_hash_call_id.get(response.get('id', '??'), '??'), response['error']))
else:
#TODO: this seems to happen for bogus transactions? Maybe handle it more gracefully than just erroring out?
raise BackendRPCError('{} (txhash:: {})'.format(response['error'], tx_hash_call_id.get(response.get('id', '??'), '??')))
# get transactions from cache
result = {}
for tx_hash in txhash_list:
try:
if verbose:
result[tx_hash] = raw_transactions_cache[tx_hash]
else:
result[tx_hash] = raw_transactions_cache[tx_hash]['hex'] if raw_transactions_cache[tx_hash] is not None else None
except KeyError as e: #shows up most likely due to finickyness with addrindex not always returning results that we need...
_logger.warning("tx missing in rawtx cache: {} -- txhash_list size: {}, hash: {} / raw_transactions_cache size: {} / # rpc_batch calls: {} / txhash in noncached_txhashes: {} / txhash in txhash_list: {} -- list {}".format(
e, len(txhash_list), hashlib.md5(json.dumps(list(txhash_list)).encode()).hexdigest(), len(raw_transactions_cache), len(payload),
tx_hash in noncached_txhashes, tx_hash in txhash_list, list(txhash_list.difference(noncached_txhashes)) ))
if _retry < GETRAWTRANSACTION_MAX_RETRIES: #try again
time.sleep(0.05 * (_retry + 1)) # Wait a bit, hitting the index non-stop may cause it to just break down... TODO: Better handling
r = getrawtransaction_batch([tx_hash], verbose=verbose, skip_missing=skip_missing, _retry=_retry+1)
result[tx_hash] = r[tx_hash]
else:
raise #already tried again, give up
return result
def getrawtransaction_batch(txhash_list, verbose=False, _recursing=False):
if len(txhash_list) > config.BACKEND_RAW_TRANSACTIONS_CACHE_SIZE:
#don't try to load in more than BACKEND_RAW_TRANSACTIONS_CACHE_SIZE entries in a single call
txhash_list_chunks = util.chunkify(txhash_list, config.BACKEND_RAW_TRANSACTIONS_CACHE_SIZE)
txes = {}
for txhash_list_chunk in txhash_list_chunks:
txes.update(getrawtransaction_batch(txhash_list_chunk, verbose=verbose))
return txes
tx_hash_call_id = {}
payload = []
noncached_txhashes = set()
txhash_list = set(txhash_list)
# payload for transactions not in cache
for tx_hash in txhash_list:
if tx_hash not in raw_transactions_cache:
call_id = binascii.hexlify(os.urandom(5)).decode('utf8')
payload.append({
"method": 'getrawtransaction',
"params": [tx_hash, 1],
"jsonrpc": "2.0",
"id": call_id
})
noncached_txhashes.add(tx_hash)
tx_hash_call_id[call_id] = tx_hash
#refresh any/all cache entries that already exist in the cache,
# so they're not inadvertently removed by another thread before we can consult them
#(this assumes that the size of the working set for any given workload doesn't exceed the max size of the cache)
for tx_hash in txhash_list.difference(noncached_txhashes):
raw_transactions_cache.refresh(tx_hash)
logger.debug("getrawtransaction_batch: txhash_list size: {} / raw_transactions_cache size: {} / # getrawtransaction calls: {}".format(
len(txhash_list), len(raw_transactions_cache), len(payload)))
# populate cache
added_entries_to_cache = []
if len(payload) > 0:
batch_responses = rpc_batch(payload)
for response in batch_responses:
if 'error' not in response or response['error'] is None:
tx_hex = response['result']
tx_hash = tx_hash_call_id[response['id']]
raw_transactions_cache[tx_hash] = tx_hex
added_entries_to_cache.append(tx_hash) #for debugging
else:
#TODO: this seems to happen for bogus transactions? Maybe handle it more gracefully than just erroring out?
raise BackendRPCError('{} (txhash:: {})'.format(response['error'], tx_hash_call_id.get(response.get('id', '??'), '??')))
# get transactions from cache
result = {}
for tx_hash in txhash_list:
try:
if verbose:
result[tx_hash] = raw_transactions_cache[tx_hash]
else:
result[tx_hash] = raw_transactions_cache[tx_hash]['hex']
except KeyError: #shows up most likely due to finickyness with addrindex not always returning results that we need...
logger.debug("tx missing in rawtx cache: {} -- txhash_list size: {}, hash: {} / raw_transactions_cache size: {} / # rpc_batch calls: {} / txhash in noncached_txhashes: {} / txhash in txhash_list: {} -- list {}".format(
e, len(txhash_list), hashlib.md5(json.dumps(list(txhash_list)).encode()).hexdigest(), len(raw_transactions_cache), len(payload),
tx_hash in noncached_txhashes, tx_hash in txhash_list, list(txhash_list.difference(noncached_txhashes)) ))
if not _recursing: #try again
r = getrawtransaction_batch([tx_hash], verbose=verbose, _recursing=True)
result[tx_hash] = r[tx_hash]
else:
raise #already tried again, give up
return result
