Speed up script parsing for ~3% faster throughput

Also improves the coin abstraction

lib/coins.py

@@ -17,9 +17,9 @@ import struct
 import sys
 
 from lib.hash import Base58, hash160, double_sha256, hash_to_str
-from lib.script import ScriptPubKey
+from lib.script import ScriptPubKey, Script
 from lib.tx import Deserializer
-from lib.util import subclasses
+from lib.util import cachedproperty, subclasses
 
 
 class CoinError(Exception):
@@ -47,6 +47,19 @@ class Coin(object):
         raise CoinError('unknown coin {} and network {} combination'
                         .format(name, net))
 
+    @cachedproperty
+    def hash168_handlers(cls):
+        return ScriptPubKey.PayToHandlers(
+            address = cls.P2PKH_hash168_from_hash160,
+            script_hash = cls.P2SH_hash168_from_hash160,
+            pubkey = cls.P2PKH_hash168_from_pubkey,
+            unknown = lambda x : None,
+        )
+
+    @classmethod
+    def hash168_from_script(cls, script):
+        return ScriptPubKey.pay_to(script, cls.hash168_handlers)
+
     @staticmethod
     def lookup_xverbytes(verbytes):
         '''Return a (is_xpub, coin_class) pair given xpub/xprv verbytes.'''
@@ -75,11 +88,18 @@ class Coin(object):
         return Base58.encode_check(hash168)
 
     @classmethod
-    def P2PKH_address_from_hash160(cls, hash_bytes):
+    def P2PKH_hash168_from_hash160(cls, hash160):
+        assert len(hash160) == 20
+        return bytes([cls.P2PKH_VERBYTE]) + hash160
+
+    @classmethod
+    def P2PKH_hash168_from_pubkey(cls, pubkey):
+        return cls.P2PKH_hash168_from_hash160(hash160(pubkey))
+
+    @classmethod
+    def P2PKH_address_from_hash160(cls, hash160):
         '''Return a P2PKH address given a public key.'''
-        assert len(hash_bytes) == 20
+        return Base58.encode_check(cls.P2PKH_hash168_from_hash160(hash160))
-        payload = bytes([cls.P2PKH_VERBYTE]) + hash_bytes
-        return Base58.encode_check(payload)
 
     @classmethod
     def P2PKH_address_from_pubkey(cls, pubkey):
@@ -87,11 +107,14 @@ class Coin(object):
         return cls.P2PKH_address_from_hash160(hash160(pubkey))
 
     @classmethod
-    def P2SH_address_from_hash160(cls, pubkey_bytes):
+    def P2SH_hash168_from_hash160(cls, hash160):
-        '''Return a coin address given a public key.'''
+        assert len(hash160) == 20
-        assert len(hash_bytes) == 20
+        return bytes([cls.P2SH_VERBYTE]) + hash160
-        payload = bytes([cls.P2SH_VERBYTE]) + hash_bytes
+
-        return Base58.encode_check(payload)
+    @classmethod
+    def P2SH_address_from_hash160(cls, hash160):
+        '''Return a coin address given a hash160.'''
+        return Base58.encode_check(cls.P2SH_hash168_from_hash160(hash160))
 
     @classmethod
     def multisig_address(cls, m, pubkeys):

lib/script.py

@@ -8,8 +8,8 @@
 '''Script-related classes and functions.'''
 
 
-from binascii import hexlify
 import struct
+from collections import namedtuple
 
 from lib.enum import Enumeration
 from lib.hash import hash160
@@ -133,60 +133,38 @@ class ScriptSig(object):
 
 
 class ScriptPubKey(object):
-    '''A script from a tx output that gives conditions necessary for
+    '''A class for handling a tx output script that gives conditions
-    spending.'''
+    necessary for spending.
+    '''
 
-    TO_ADDRESS, TO_P2SH, TO_PUBKEY, TO_UNKNOWN = range(4)
     TO_ADDRESS_OPS = [OpCodes.OP_DUP, OpCodes.OP_HASH160, -1,
                       OpCodes.OP_EQUALVERIFY, OpCodes.OP_CHECKSIG]
     TO_P2SH_OPS = [OpCodes.OP_HASH160, -1, OpCodes.OP_EQUAL]
     TO_PUBKEY_OPS = [-1, OpCodes.OP_CHECKSIG]
 
-    def __init__(self, script, coin, kind, hash168, pubkey=None):
+    PayToHandlers = namedtuple('PayToHandlers',
-        self.script = script
+                               'address script_hash pubkey unknown')
-        self.coin = coin
-        self.kind = kind
-        self.hash168 = hash168
-        if pubkey:
-            self.pubkey = pubkey
-
-    @cachedproperty
-    def address(self):
-        if self.kind == ScriptPubKey.TO_P2SH:
-            return self.coin.P2SH_address_from_hash160(self.hash168[1:])
-        if self.hash160:
-            return self.coin.P2PKH_address_from_hash160(self.hash168[1:])
-        return ''
-
-    @classmethod
-    def from_script(cls, script, coin):
-        '''Returns an instance of this class.  Uncrecognised scripts return
-        an object of kind TO_UNKNOWN.'''
-        try:
-            return cls.parse_script(script, coin)
-        except ScriptError:
-            return cls(script, coin, cls.TO_UNKNOWN, None)
 
     @classmethod
-    def parse_script(cls, script, coin):
+    def pay_to(cls, script, handlers):
-        '''Returns an instance of this class.  Raises on unrecognised
+        '''Parse a script, invoke the appropriate handler and
-        scripts.'''
+        return the result.
+
+        One of the following handlers is invoked:
+           handlers.address(hash160)
+           handlers.script_hash(hash160)
+           handlers.pubkey(pubkey)
+           handlers.unknown(None)
+        '''
         ops, datas = Script.get_ops(script)
 
         if Script.match_ops(ops, cls.TO_ADDRESS_OPS):
-            return cls(script, coin, cls.TO_ADDRESS,
+            return handlers.address(datas[2])
-                       bytes([coin.P2PKH_VERBYTE]) + datas[2])
-
         if Script.match_ops(ops, cls.TO_P2SH_OPS):
-            return cls(script, coin, cls.TO_P2SH,
+            return handlers.script_hash(datas[1])
-                       bytes([coin.P2SH_VERBYTE]) + datas[1])
-
         if Script.match_ops(ops, cls.TO_PUBKEY_OPS):
-            pubkey = datas[0]
+            return handlers.pubkey(datas[0])
-            return cls(script, coin, cls.TO_PUBKEY,
+        return handlers.unknown(None)
-                       bytes([coin.P2PKH_VERBYTE]) + hash160(pubkey), pubkey)
-
-        raise ScriptError('unknown script pubkey pattern')
 
     @classmethod
     def P2SH_script(cls, hash160):
@@ -317,4 +295,4 @@ class Script(object):
                 print(name)
             else:
                 print('{} {} ({:d} bytes)'
-                      .format(name, hexlify(data).decode('ascii'), len(data)))
+                      .format(name, data.hex(), len(data)))

lib/util.py

@@ -31,16 +31,11 @@ class cachedproperty(object):
         self.f = f
 
     def __get__(self, obj, type):
-        if obj is None:
+        obj = obj or type
-            return self
         value = self.f(obj)
-        obj.__dict__[self.f.__name__] = value
+        setattr(obj, self.f.__name__, value)
         return value
 
-    def __set__(self, obj, value):
-        raise AttributeError('cannot set {} on {}'
-                             .format(self.f.__name__, obj))
-
 
 def deep_getsizeof(obj):
     """Find the memory footprint of a Python object.

server/block_processor.py

@@ -20,7 +20,6 @@ from functools import partial
 from server.cache import FSCache, UTXOCache, NO_CACHE_ENTRY
 from server.daemon import DaemonError
 from lib.hash import hash_to_str
-from lib.script import ScriptPubKey
 from lib.tx import Deserializer
 from lib.util import chunks, LoggedClass
 from server.storage import open_db
@@ -205,12 +204,11 @@ class MemPool(LoggedClass):
 
         # The mempool is unordered, so process all outputs first so
         # that looking for inputs has full info.
-        parse_script = ScriptPubKey.from_script
+        script_hash168 = self.bp.coin.hash168_from_script
-        coin = self.bp.coin
         utxo_lookup = self.bp.utxo_cache.lookup
 
         def txout_pair(txout):
-            return (parse_script(txout.pk_script, coin).hash168, txout.value)
+            return (script_hash168(txout.pk_script), txout.value)
 
         for hex_hash, tx in new_txs.items():
             txout_pairs = tuple(txout_pair(txout) for txout in tx.outputs)
@@ -745,8 +743,7 @@ class BlockProcessor(LoggedClass):
         # Use local vars for speed in the loops
         history = self.history
         tx_num = self.tx_count
-        coin = self.coin
+        script_hash168 = self.coin.hash168_from_script
-        parse_script = ScriptPubKey.from_script
         pack = struct.pack
 
         for tx, tx_hash in zip(txs, tx_hashes):
@@ -763,7 +760,7 @@ class BlockProcessor(LoggedClass):
             # Add the new UTXOs
             for idx, txout in enumerate(tx.outputs):
                 # Get the hash168.  Ignore scripts we can't grok.
-                hash168 = parse_script(txout.pk_script, coin).hash168
+                hash168 = script_hash168(txout.pk_script)
                 if hash168:
                     hash168s.add(hash168)
                     put_utxo(tx_hash + pack('<H', idx),