From 66441024417e324b41b951ee6a8622e0d8aa6850 Mon Sep 17 00:00:00 2001
From: Neil Booth <kyuupichan@gmail.com>
Date: Mon, 10 Oct 2016 11:17:26 +0900
Subject: [PATCH] Use hash168 to distinguish address types in DB

---
 lib/coins.py  | 46 ++++++++++++++++++-------------
 lib/script.py | 17 +++++++-----
 lib/util.py   | 18 ------------
 query.py      |  8 +++---
 server/db.py  | 76 +++++++++++++++++++++++++--------------------------
 5 files changed, 79 insertions(+), 86 deletions(-)

diff --git a/lib/coins.py b/lib/coins.py
index 6c5c7b6..8afed4d 100644
--- a/lib/coins.py
+++ b/lib/coins.py
@@ -16,7 +16,7 @@ class CoinError(Exception):
 
 
 class Coin(object):
-    '''Base class of coin hierarchy'''
+    '''Base class of coin hierarchy.'''
 
     # Not sure if these are coin-specific
     HEADER_LEN = 80
@@ -52,59 +52,67 @@ class Coin(object):
         raise CoinError("version bytes unrecognised")
 
     @classmethod
-    def address_to_hash160(cls, addr):
-        '''Returns a hash160 given an address'''
+    def address_to_hash168(cls, addr):
+        '''Return a 21-byte hash given an address.
+
+        This is the hash160 prefixed by the address version byte.
+        '''
         result = Base58.decode_check(addr)
         if len(result) != 21:
             raise CoinError('invalid address: {}'.format(addr))
-        return result[1:]
+        return result
 
     @classmethod
     def P2PKH_address_from_hash160(cls, hash_bytes):
-        '''Returns a P2PKH address given a public key'''
+        '''Return a P2PKH address given a public key.'''
         assert len(hash_bytes) == 20
         payload = bytes([cls.P2PKH_VERBYTE]) + hash_bytes
         return Base58.encode_check(payload)
 
     @classmethod
     def P2PKH_address_from_pubkey(cls, pubkey):
-        '''Returns a coin address given a public key'''
+        '''Return a coin address given a public key.'''
         return cls.P2PKH_address_from_hash160(hash160(pubkey))
 
     @classmethod
     def P2SH_address_from_hash160(cls, pubkey_bytes):
-        '''Returns a coin address given a public key'''
+        '''Return a coin address given a public key.'''
         assert len(hash_bytes) == 20
         payload = bytes([cls.P2SH_VERBYTE]) + hash_bytes
         return Base58.encode_check(payload)
 
     @classmethod
     def multisig_address(cls, m, pubkeys):
-        '''Returns the P2SH address for an M of N multisig transaction.  Pass
-        the N pubkeys of which M are needed to sign it.  If generating
-        an address for a wallet, it is the caller's responsibility to
-        sort them to ensure order does not matter for, e.g., wallet
-        recovery.'''
+        '''Return the P2SH address for an M of N multisig transaction.
+
+        Pass the N pubkeys of which M are needed to sign it.  If
+        generating an address for a wallet, it is the caller's
+        responsibility to sort them to ensure order does not matter
+        for, e.g., wallet recovery.
+        '''
         script = cls.pay_to_multisig_script(m, pubkeys)
         payload = bytes([cls.P2SH_VERBYTE]) + hash160(pubkey_bytes)
         return Base58.encode_check(payload)
 
     @classmethod
     def pay_to_multisig_script(cls, m, pubkeys):
-        '''Returns a P2SH multisig script for an M of N multisig
-        transaction.'''
+        '''Return a P2SH script for an M of N multisig transaction.'''
         return ScriptPubKey.multisig_script(m, pubkeys)
 
     @classmethod
     def pay_to_pubkey_script(cls, pubkey):
-        '''Returns a pubkey script that pays to pubkey.  The input is the
-        raw pubkey bytes (length 33 or 65).'''
+        '''Return a pubkey script that pays to a pubkey.
+
+        Pass the raw pubkey bytes (length 33 or 65).
+        '''
         return ScriptPubKey.P2PK_script(pubkey)
 
     @classmethod
     def pay_to_address_script(cls, address):
-        '''Returns a pubkey script that pays to pubkey hash.  Input is the
-        address (either P2PKH or P2SH) in base58 form.'''
+        '''Return a pubkey script that pays to a pubkey hash.
+
+        Pass the address (either P2PKH or P2SH) in base58 form.
+        '''
         raw = Base58.decode_check(address)
 
         # Require version byte plus hash160.
@@ -121,7 +129,7 @@ class Coin(object):
 
     @classmethod
     def prvkey_WIF(privkey_bytes, compressed):
-        "The private key encoded in Wallet Import Format"
+        "Return the private key encoded in Wallet Import Format."
         payload = bytearray([cls.WIF_BYTE]) + privkey_bytes
         if compressed:
             payload.append(0x01)
diff --git a/lib/script.py b/lib/script.py
index adfb23f..0239a1d 100644
--- a/lib/script.py
+++ b/lib/script.py
@@ -131,20 +131,20 @@ class ScriptPubKey(object):
     TO_P2SH_OPS = [OpCodes.OP_HASH160, -1, OpCodes.OP_EQUAL]
     TO_PUBKEY_OPS = [-1, OpCodes.OP_CHECKSIG]
 
-    def __init__(self, script, coin, kind, hash160, pubkey=None):
+    def __init__(self, script, coin, kind, hash168, pubkey=None):
         self.script = script
         self.coin = coin
         self.kind = kind
-        self.hash160 = hash160
+        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.hash160)
+            return self.coin.P2SH_address_from_hash160(self.hash168[1:])
         if self.hash160:
-            return self.coin.P2PKH_address_from_hash160(self.hash160)
+            return self.coin.P2PKH_address_from_hash160(self.hash168[1:])
         return ''
 
     @classmethod
@@ -163,14 +163,17 @@ class ScriptPubKey(object):
         ops, datas = Script.get_ops(script)
 
         if Script.match_ops(ops, cls.TO_ADDRESS_OPS):
-            return cls(script, coin, cls.TO_ADDRESS, datas[2])
+            return cls(script, coin, cls.TO_ADDRESS,
+                       bytes([coin.P2PKH_VERBYTE]) + datas[2])
 
         if Script.match_ops(ops, cls.TO_P2SH_OPS):
-            return cls(script, coin, cls.TO_P2SH, datas[1])
+            return cls(script, coin, cls.TO_P2SH,
+                       bytes([coin.P2SH_VERBYTE]) + datas[1])
 
         if Script.match_ops(ops, cls.TO_PUBKEY_OPS):
             pubkey = datas[0]
-            return cls(script, coin, cls.TO_PUBKEY, hash160(pubkey), pubkey)
+            return cls(script, coin, cls.TO_PUBKEY,
+                       bytes([coin.P2PKH_VERBYTE]) + hash160(pubkey), pubkey)
 
         raise ScriptError('unknown script pubkey pattern')
 
diff --git a/lib/util.py b/lib/util.py
index d23598a..40f9e2f 100644
--- a/lib/util.py
+++ b/lib/util.py
@@ -5,24 +5,6 @@
 import sys
 
 
-class Log(object):
-    '''Logging base class'''
-
-    VERBOSE = True
-
-    def diagnostic_name(self):
-        return self.__class__.__name__
-
-    def log(self, *msgs):
-        if Log.VERBOSE:
-            print('[{}]: '.format(self.diagnostic_name()), *msgs,
-                  file=sys.stdout, flush=True)
-
-    def log_error(self, *msg):
-        print('[{}]: ERROR: {}'.format(self.diagnostic_name()), *msgs,
-              file=sys.stderr, flush=True)
-
-
 # Method decorator.  To be used for calculations that will always
 # deliver the same result.  The method cannot take any arguments
 # and should be accessed as an attribute.
diff --git a/query.py b/query.py
index 63cdc72..23ed768 100644
--- a/query.py
+++ b/query.py
@@ -23,21 +23,21 @@ def main():
         limit = 10
     for addr in sys.argv[argc:]:
         print('Address: ', addr)
-        hash160 = coin.address_to_hash160(addr)
+        hash168 = coin.address_to_hash168(addr)
         n = None
-        for n, (tx_hash, height) in enumerate(db.get_history(hash160, limit)):
+        for n, (tx_hash, height) in enumerate(db.get_history(hash168, limit)):
             print('History #{:d}: hash: {} height: {:d}'
                   .format(n + 1, bytes(reversed(tx_hash)).hex(), height))
         if n is None:
             print('No history')
         n = None
-        for n, utxo in enumerate(db.get_utxos(hash160, limit)):
+        for n, utxo in enumerate(db.get_utxos(hash168, limit)):
             print('UTXOs #{:d}: hash: {} pos: {:d} height: {:d} value: {:d}'
                   .format(n, bytes(reversed(utxo.tx_hash)).hex(),
                           utxo.tx_pos, utxo.height, utxo.value))
         if n is None:
             print('No UTXOs')
-        balance = db.get_balance(hash160)
+        balance = db.get_balance(hash168)
         print('Balance: {} {}'.format(coin.decimal_value(balance),
                                       coin.SHORTNAME))
 
diff --git a/server/db.py b/server/db.py
index a166cac..d053975 100644
--- a/server/db.py
+++ b/server/db.py
@@ -256,56 +256,56 @@ class DB(object):
         self.history.pop(None, None)
 
         flush_id = struct.pack('>H', self.flush_count)
-        for hash160, hist in self.history.items():
-            key = b'H' + hash160 + flush_id
+        for hash168, hist in self.history.items():
+            key = b'H' + hash168 + flush_id
             batch.put(key, array.array('I', hist).tobytes())
 
         self.history = defaultdict(list)
 
-    def get_hash160(self, tx_hash, idx, delete=True):
+    def get_hash168(self, tx_hash, idx, delete=True):
         key = b'h' + tx_hash[:ADDR_TX_HASH_LEN] + struct.pack('<H', idx)
         data = self.get(key)
         if data is None:
             return None
 
-        if len(data) == 24:
+        if len(data) == 25:
             if delete:
                 self.delete(key)
-            return data[:20]
+            return data[:21]
 
-        assert len(data) % 24 == 0
+        assert len(data) % 25 == 0
         self.hcolls += 1
         if self.hcolls % 1000 == 0:
-            self.logger.info('{} total hash160 compressed key collisions'
+            self.logger.info('{} total hash168 compressed key collisions'
                              .format(self.hcolls))
-        for n in range(0, len(data), 24):
-            (tx_num, ) = struct.unpack('<I', data[n+20:n+24])
+        for n in range(0, len(data), 25):
+            (tx_num, ) = struct.unpack('<I', data[n+21 : n+25])
             my_hash, height = self.get_tx_hash(tx_num)
             if my_hash == tx_hash:
                 if delete:
-                    self.put(key, data[:n] + data[n + 24:])
-                return data[n:n+20]
-        else:
-            raise Exception('could not resolve hash160 collision')
+                    self.put(key, data[:n] + data[n+25:])
+                return data[n : n+21]
+
+        raise Exception('could not resolve hash168 collision')
 
     def spend_utxo(self, prevout):
-        hash160 = self.get_hash160(prevout.hash, prevout.n)
-        if hash160 is None:
+        hash168 = self.get_hash168(prevout.hash, prevout.n)
+        if hash168 is None:
             # This indicates a successful spend of a non-standard script
             # self.logger.info('ignoring spend of non-standard UTXO {}/{:d} '
             #                  'at height {:d}'
             #                  .format(bytes(reversed(prevout.hash)).hex(),
             #                          prevout.n, self.height))
             return None
-
-        key = (b'u' + hash160 + prevout.hash[:UTXO_TX_HASH_LEN]
+        key = (b'u' + hash168 + prevout.hash[:UTXO_TX_HASH_LEN]
                + struct.pack('<H', prevout.n))
         data = self.get(key)
         if data is None:
+            # Uh-oh, this should not happen.  It may be recoverable...
             self.logger.error('found no UTXO for {} / {:d} key {}'
                              .format(bytes(reversed(prevout.hash)).hex(),
                                      prevout.n, bytes(key).hex()))
-            return hash160
+            return hash168
 
         if len(data) == 12:
             (tx_num, ) = struct.unpack('<I', data[:4])
@@ -324,35 +324,35 @@ class DB(object):
             data = data[:n] + data[n + 12:]
             self.put(key, data)
 
-        return hash160
+        return hash168
 
     def put_utxo(self, tx_hash, idx, txout):
         pk = ScriptPubKey.from_script(txout.pk_script, self.coin)
-        if not pk.hash160:
+        if not pk.hash168:
             return None
 
         pack = struct.pack
         idxb = pack('<H', idx)
         txcb = pack('<I', self.tx_count)
 
-        # First write the hash160 lookup
+        # First write the hash168 lookup
         key = b'h' + tx_hash[:ADDR_TX_HASH_LEN] + idxb
-        # b'' avoids this annoyance: https://bugs.python.org/issue13298
-        value = b''.join([pk.hash160, txcb])
+        # b''.join avoids this: https://bugs.python.org/issue13298
+        value = b''.join((pk.hash168, txcb))
         prior_value = self.get(key)
         if prior_value:   # Should almost never happen
             value += prior_value
         self.put(key, value)
 
         # Next write the UTXO
-        key = b'u' + pk.hash160 + tx_hash[:UTXO_TX_HASH_LEN] + idxb
+        key = b'u' + pk.hash168 + tx_hash[:UTXO_TX_HASH_LEN] + idxb
         value = txcb + pack('<Q', txout.value)
         prior_value = self.get(key)
         if prior_value:   # Should almost never happen
             value += prior_value
         self.put(key, value)
 
-        return pk.hash160
+        return pk.hash168
 
     def open_file(self, filename, truncate=False, create=False):
         try:
@@ -420,16 +420,16 @@ class DB(object):
             self.flush()
 
     def process_tx(self, tx_hash, tx):
-        hash160s = set()
+        hash168s = set()
         if not tx.is_coinbase:
             for txin in tx.inputs:
-                hash160s.add(self.spend_utxo(txin.prevout))
+                hash168s.add(self.spend_utxo(txin.prevout))
 
         for idx, txout in enumerate(tx.outputs):
-            hash160s.add(self.put_utxo(tx_hash, idx, txout))
+            hash168s.add(self.put_utxo(tx_hash, idx, txout))
 
-        for hash160 in hash160s:
-            self.history[hash160].append(self.tx_count)
+        for hash168 in hash168s:
+            self.history[hash168].append(self.tx_count)
 
         self.tx_count += 1
 
@@ -458,7 +458,7 @@ class DB(object):
         assert isinstance(limit, int) and limit >= 0
         return limit
 
-    def get_history(self, hash160, limit=1000):
+    def get_history(self, hash168, limit=1000):
         '''Generator that returns an unpruned, sorted list of (tx_hash,
         height) tuples of transactions that touched the address,
         earliest in the blockchain first.  Includes both spending and
@@ -466,7 +466,7 @@ class DB(object):
         Set limit to None to get them all.
         '''
         limit = self.resolve_limit(limit)
-        prefix = b'H' + hash160
+        prefix = b'H' + hash168
         for key, hist in self.db.iterator(prefix=prefix):
             a = array.array('I')
             a.frombytes(hist)
@@ -476,18 +476,18 @@ class DB(object):
                 yield self.get_tx_hash(tx_num)
                 limit -= 1
 
-    def get_balance(self, hash160):
+    def get_balance(self, hash168):
         '''Returns the confirmed balance of an address.'''
-        return sum(utxo.value for utxo in self.get_utxos(hash160, limit=None))
+        return sum(utxo.value for utxo in self.get_utxos(hash168, limit=None))
 
-    def get_utxos(self, hash160, limit=1000):
+    def get_utxos(self, hash168, limit=1000):
         '''Generator that yields all UTXOs for an address sorted in no
         particular order.  By default yields at most 1000 entries.
         Set limit to None to get them all.
         '''
         limit = self.resolve_limit(limit)
         unpack = struct.unpack
-        prefix = b'u' + hash160
+        prefix = b'u' + hash168
         utxos = []
         for k, v in self.db.iterator(prefix=prefix):
             (tx_pos, ) = unpack('<H', k[-2:])
@@ -501,7 +501,7 @@ class DB(object):
                 yield UTXO(tx_num, tx_pos, tx_hash, height, value)
                 limit -= 1
 
-    def get_utxos_sorted(self, hash160):
+    def get_utxos_sorted(self, hash168):
         '''Returns all the UTXOs for an address sorted by height and
         position in the block.'''
-        return sorted(self.get_utxos(hash160, limit=None))
+        return sorted(self.get_utxos(hash168, limit=None))