NVC protocol v. 60011 support
[p2pool.git] / p2pool / bitcoin / data.py
index a16a920..74cb94b 100644 (file)
 from __future__ import division
 
-import struct
-import cStringIO as StringIO
 import hashlib
+import random
 import warnings
 
-from . import base58
-from p2pool.util import bases, expiring_dict, math
+import p2pool
+from p2pool.util import math, pack
 
-class EarlyEnd(Exception):
-    pass
+def hash256(data):
+    return pack.IntType(256).unpack(hashlib.sha256(hashlib.sha256(data).digest()).digest())
 
-class LateEnd(Exception):
-    pass
+def hash160(data):
+    return pack.IntType(160).unpack(hashlib.new('ripemd160', hashlib.sha256(data).digest()).digest())
 
-class Type(object):
-    # the same data can have only one unpacked representation, but multiple packed binary representations
-    
-    #def __hash__(self):
-    #    return hash(tuple(self.__dict__.items()))
-    
-    #def __eq__(self, other):
-    #    if not isinstance(other, Type):
-    #        raise NotImplementedError()
-    #    return self.__dict__ == other.__dict__
-    
-    def _unpack(self, data):
-        f = StringIO.StringIO(data)
-        
-        obj = self.read(f)
-        
-        if f.tell() != len(data):
-            raise LateEnd('underread ' + repr((self, data)))
-        
-        return obj
-    
-    def unpack(self, data):
-        obj = self._unpack(data)
-        
-        if __debug__:
-            data2 = self._pack(obj)
-            if data2 != data:
-                assert self._unpack(data2) == obj
-        
-        return obj
-    
-    def _pack(self, obj):
-        f = StringIO.StringIO()
-        
-        self.write(f, obj)
-        
-        data = f.getvalue()
-        
-        return data
-    
-    def pack(self, obj):
-        data = self._pack(obj)
-        assert self._unpack(data) == obj
-                
-        return data
-    
-    
-    def pack_base58(self, obj):
-        return base58.base58_encode(self.pack(obj))
-    
-    def unpack_base58(self, base58_data):
-        return self.unpack(base58.base58_decode(base58_data))
-        
-    
-    def hash160(self, obj):
-        return ShortHashType().unpack(hashlib.new('ripemd160', hashlib.sha256(self.pack(obj)).digest()).digest())
-    
-    def hash256(self, obj):
-        return HashType().unpack(hashlib.sha256(hashlib.sha256(self.pack(obj)).digest()).digest())
-
-class VarIntType(Type):
-    # redundancy doesn't matter here because bitcoin and p2pool both reencode before hashing
-    def read(self, file):
-        data = file.read(1)
-        if len(data) != 1:
-            raise EarlyEnd()
-        first = ord(data)
-        if first < 0xfd:
-            return first
-        elif first == 0xfd:
-            desc = '<H'
-        elif first == 0xfe:
-            desc = '<I'
-        elif first == 0xff:
-            desc = '<Q'
-        else:
-            raise AssertionError()
-        length = struct.calcsize(desc)
-        data = file.read(length)
-        if len(data) != length:
-            raise EarlyEnd()
-        return struct.unpack(desc, data)[0]
-    
-    def write(self, file, item):
-        if item < 0xfd:
-            file.write(struct.pack('<B', item))
-        elif item <= 0xffff:
-            file.write(struct.pack('<BH', 0xfd, item))
-        elif item <= 0xffffffff:
-            file.write(struct.pack('<BI', 0xfe, item))
-        elif item <= 0xffffffffffffffff:
-            file.write(struct.pack('<BQ', 0xff, item))
-        else:
-            raise ValueError('int too large for varint')
+def scrypt(data):
+    return pack.IntType(256).unpack(__import__('ltc_scrypt').getPoWHash(data))
 
-class VarStrType(Type):
-    _inner_size = VarIntType()
-    
-    def read(self, file):
-        length = self._inner_size.read(file)
-        res = file.read(length)
-        if len(res) != length:
-            raise EarlyEnd('var str not long enough %r' % ((length, len(res), res),))
-        return res
-    
-    def write(self, file, item):
-        self._inner_size.write(file, len(item))
-        file.write(item)
-
-class FixedStrType(Type):
-    def __init__(self, length):
-        self.length = length
-    
-    def read(self, file):
-        res = file.read(self.length)
-        if len(res) != self.length:
-            raise EarlyEnd('early EOF!')
-        return res
-    
-    def write(self, file, item):
-        if len(item) != self.length:
-            raise ValueError('incorrect length!')
-        file.write(item)
-
-class EnumType(Type):
-    def __init__(self, inner, values):
+class ChecksummedType(pack.Type):
+    def __init__(self, inner, checksum_func=lambda data: hashlib.sha256(hashlib.sha256(data).digest()).digest()[:4]):
         self.inner = inner
-        self.values = values
-        
-        self.keys = {}
-        for k, v in values.iteritems():
-            if v in self.keys:
-                raise ValueError('duplicate value in values')
-            self.keys[v] = k
-    
-    def read(self, file):
-        return self.keys[self.inner.read(file)]
-    
-    def write(self, file, item):
-        self.inner.write(file, self.values[item])
-
-class HashType(Type):
-    def read(self, file):
-        data = file.read(256//8)
-        if len(data) != 256//8:
-            raise EarlyEnd('incorrect length!')
-        return int(data[::-1].encode('hex'), 16)
-    
-    def write(self, file, item):
-        if not 0 <= item < 2**256:
-            raise ValueError("invalid hash value")
-        if item != 0 and item < 2**160:
-            warnings.warn("very low hash value - maybe you meant to use ShortHashType? %x" % (item,))
-        file.write(('%064x' % (item,)).decode('hex')[::-1])
-
-class ShortHashType(Type):
-    def read(self, file):
-        data = file.read(160//8)
-        if len(data) != 160//8:
-            raise EarlyEnd('incorrect length!')
-        return int(data[::-1].encode('hex'), 16)
-    
-    def write(self, file, item):
-        if item >= 2**160:
-            raise ValueError("invalid hash value")
-        file.write(('%040x' % (item,)).decode('hex')[::-1])
-
-class ListType(Type):
-    _inner_size = VarIntType()
+        self.checksum_func = checksum_func
     
-    def __init__(self, type):
-        self.type = type
-    
-    def read(self, file):
-        length = self._inner_size.read(file)
-        return [self.type.read(file) for i in xrange(length)]
-    
-    def write(self, file, item):
-        self._inner_size.write(file, len(item))
-        for subitem in item:
-            self.type.write(file, subitem)
-
-class FastLittleEndianUnsignedInteger(Type):
     def read(self, file):
-        data = map(ord, file.read(4))
-        return data[0] + (data[1] << 8) + (data[2] << 16) + (data[3] << 24)
-    
-    def write(self, file, item):
-        StructType("<I").write(file, item)
-
-class StructType(Type):
-    def __init__(self, desc):
-        self.desc = desc
-        self.length = struct.calcsize(self.desc)
-    
-    def read(self, file):
-        data = file.read(self.length)
-        if len(data) != self.length:
-            raise EarlyEnd()
-        res, = struct.unpack(self.desc, data)
-        return res
-    
-    def write(self, file, item):
-        data = struct.pack(self.desc, item)
-        if struct.unpack(self.desc, data)[0] != item:
-            # special test because struct doesn't error on some overflows
-            raise ValueError("item didn't survive pack cycle (%r)" % (item,))
-        file.write(data)
-
-class IPV6AddressType(Type):
-    def read(self, file):
-        data = file.read(16)
-        if len(data) != 16:
-            raise EarlyEnd()
-        if data[:12] != '00000000000000000000ffff'.decode('hex'):
-            raise ValueError("ipv6 addresses not supported yet")
-        return '.'.join(str(ord(x)) for x in data[12:])
-    
-    def write(self, file, item):
-        bits = map(int, item.split('.'))
-        if len(bits) != 4:
-            raise ValueError("invalid address: %r" % (bits,))
-        data = '00000000000000000000ffff'.decode('hex') + ''.join(chr(x) for x in bits)
-        assert len(data) == 16, len(data)
-        file.write(data)
-
-class ComposedType(Type):
-    def __init__(self, fields):
-        self.fields = fields
-    
-    def read(self, file):
-        item = {}
-        for key, type_ in self.fields:
-            item[key] = type_.read(file)
-        return item
-    
-    def write(self, file, item):
-        for key, type_ in self.fields:
-            type_.write(file, item[key])
-
-class ChecksummedType(Type):
-    def __init__(self, inner):
-        self.inner = inner
-    
-    def read(self, file):
-        obj = self.inner.read(file)
+        obj, file = self.inner.read(file)
         data = self.inner.pack(obj)
         
-        if file.read(4) != hashlib.sha256(hashlib.sha256(data).digest()).digest()[:4]:
-            raise ValueError("invalid checksum")
+        calculated_checksum = self.checksum_func(data)
+        checksum, file = pack.read(file, len(calculated_checksum))
+        if checksum != calculated_checksum:
+            raise ValueError('invalid checksum')
         
-        return obj
+        return obj, file
     
     def write(self, file, item):
         data = self.inner.pack(item)
-        file.write(data)
-        file.write(hashlib.sha256(hashlib.sha256(data).digest()).digest()[:4])
-
-class FloatingIntegerType(Type):
-    # redundancy doesn't matter here because bitcoin checks binary bits against its own computed bits
-    # so it will always be encoded 'normally' in blocks (they way bitcoin does it)
-    _inner = StructType("<I")
-    _inner = FastLittleEndianUnsignedInteger()
+        return (file, data), self.checksum_func(data)
+
+class FloatingInteger(object):
+    __slots__ = ['bits', '_target']
     
-    def read(self, file):
-        bits = self._inner.read(file)
-        target = self._bits_to_target(bits)
-        if __debug__:
-            if self._target_to_bits(target) != bits:
-                raise ValueError("bits in non-canonical form")
-        return target
+    @classmethod
+    def from_target_upper_bound(cls, target):
+        n = math.natural_to_string(target)
+        if n and ord(n[0]) >= 128:
+            n = '\x00' + n
+        bits2 = (chr(len(n)) + (n + 3*chr(0))[:3])[::-1]
+        bits = pack.IntType(32).unpack(bits2)
+        return cls(bits)
+    
+    def __init__(self, bits, target=None):
+        self.bits = bits
+        self._target = None
+        if target is not None and self.target != target:
+            raise ValueError('target does not match')
+    
+    @property
+    def target(self):
+        res = self._target
+        if res is None:
+            res = self._target = math.shift_left(self.bits & 0x00ffffff, 8 * ((self.bits >> 24) - 3))
+        return res
     
-    def write(self, file, item):
-        self._inner.write(file, self._target_to_bits(item))
+    def __hash__(self):
+        return hash(self.bits)
     
-    def truncate_to(self, x):
-        return self._bits_to_target(self._target_to_bits(x, _check=False))
-        
-    def _bits_to_target(self, bits2):
-        target = math.shift_left(bits2 & 0x00ffffff, 8 * ((bits2 >> 24) - 3))
-        assert target == self._bits_to_target1(struct.pack("<I", bits2))
-        assert self._target_to_bits(target, _check=False) == bits2
-        return target
+    def __eq__(self, other):
+        return self.bits == other.bits
+    
+    def __ne__(self, other):
+        return not (self == other)
     
-    def _bits_to_target1(self, bits):
-        bits = bits[::-1]
-        length = ord(bits[0])
-        return bases.string_to_natural((bits[1:] + "\0"*length)[:length])
+    def __cmp__(self, other):
+        assert False
+    
+    def __repr__(self):
+        return 'FloatingInteger(bits=%s, target=%s)' % (hex(self.bits), hex(self.target))
 
-    def _target_to_bits(self, target, _check=True):
-        n = bases.natural_to_string(target)
-        if n and ord(n[0]) >= 128:
-            n = "\x00" + n
-        bits2 = (chr(len(n)) + (n + 3*chr(0))[:3])[::-1]
-        bits = struct.unpack("<I", bits2)[0]
-        if _check:
-            if self._bits_to_target(bits) != target:
-                raise ValueError(repr((target, self._bits_to_target(bits, _check=False))))
-        return bits
-
-class PossiblyNone(Type):
-    def __init__(self, none_value, inner):
-        self.none_value = none_value
-        self.inner = inner
+class FloatingIntegerType(pack.Type):
+    _inner = pack.IntType(32)
     
     def read(self, file):
-        value = self.inner.read(file)
-        return None if value == self.none_value else value
+        bits, file = self._inner.read(file)
+        return FloatingInteger(bits), file
     
     def write(self, file, item):
-        if item == self.none_value:
-            raise ValueError("none_value used")
-        self.inner.write(file, self.none_value if item is None else item)
-
-address_type = ComposedType([
-    ('services', StructType('<Q')),
-    ('address', IPV6AddressType()),
-    ('port', StructType('>H')),
+        return self._inner.write(file, item.bits)
+
+address_type = pack.ComposedType([
+    ('services', pack.IntType(64)),
+    ('address', pack.IPV6AddressType()),
+    ('port', pack.IntType(16, 'big')),
 ])
 
-tx_type = ComposedType([
-    ('version', StructType('<I')),
-    ('tx_ins', ListType(ComposedType([
-        ('previous_output', PossiblyNone(dict(hash=0, index=2**32 - 1), ComposedType([
-            ('hash', HashType()),
-            ('index', StructType('<I')),
+tx_type = pack.ComposedType([
+    ('version', pack.IntType(32)),
+    ('timestamp', pack.IntType(32)), # txn timestamp
+    ('tx_ins', pack.ListType(pack.ComposedType([
+        ('previous_output', pack.PossiblyNoneType(dict(hash=0, index=2**32 - 1), pack.ComposedType([
+            ('hash', pack.IntType(256)),
+            ('index', pack.IntType(32)),
         ]))),
-        ('script', VarStrType()),
-        ('sequence', PossiblyNone(2**32 - 1, StructType('<I'))),
+        ('script', pack.VarStrType()),
+        ('sequence', pack.PossiblyNoneType(2**32 - 1, pack.IntType(32))),
     ]))),
-    ('tx_outs', ListType(ComposedType([
-        ('value', StructType('<Q')),
-        ('script', VarStrType()),
+    ('tx_outs', pack.ListType(pack.ComposedType([
+        ('value', pack.IntType(64)),
+        ('script', pack.VarStrType()),
     ]))),
-    ('lock_time', StructType('<I')),
+    ('lock_time', pack.IntType(32)),
+])
+
+merkle_link_type = pack.ComposedType([
+    ('branch', pack.ListType(pack.IntType(256))),
+    ('index', pack.IntType(32)),
 ])
 
-block_header_type = ComposedType([
-    ('version', StructType('<I')),
-    ('previous_block', PossiblyNone(0, HashType())),
-    ('merkle_root', HashType()),
-    ('timestamp', StructType('<I')),
-    ('target', FloatingIntegerType()),
-    ('nonce', StructType('<I')),
+merkle_tx_type = pack.ComposedType([
+    ('tx', tx_type),
+    ('block_hash', pack.IntType(256)),
+    ('merkle_link', merkle_link_type),
 ])
 
-block_type = ComposedType([
+block_header_type = pack.ComposedType([
+    ('version', pack.IntType(32)),
+    ('previous_block', pack.PossiblyNoneType(0, pack.IntType(256))),
+    ('merkle_root', pack.IntType(256)),
+    ('timestamp', pack.IntType(32)),
+    ('bits', FloatingIntegerType()),
+    ('nonce', pack.IntType(32)),
+])
+
+block_type = pack.ComposedType([
     ('header', block_header_type),
-    ('txs', ListType(tx_type)),
+    ('txs', pack.ListType(tx_type)),
+    ('signature', pack.VarStrType()), # header signature field
+])
+
+# merged mining
+
+aux_pow_type = pack.ComposedType([
+    ('merkle_tx', merkle_tx_type),
+    ('merkle_link', merkle_link_type),
+    ('parent_block_header', block_header_type),
+])
+
+aux_pow_coinbase_type = pack.ComposedType([
+    ('merkle_root', pack.IntType(256, 'big')),
+    ('size', pack.IntType(32)),
+    ('nonce', pack.IntType(32)),
 ])
 
+def make_auxpow_tree(chain_ids):
+    for size in (2**i for i in xrange(31)):
+        if size < len(chain_ids):
+            continue
+        res = {}
+        for chain_id in chain_ids:
+            pos = (1103515245 * chain_id + 1103515245 * 12345 + 12345) % size
+            if pos in res:
+                break
+            res[pos] = chain_id
+        else:
+            return res, size
+    raise AssertionError()
+
+# merkle trees
 
-merkle_record_type = ComposedType([
-    ('left', HashType()),
-    ('right', HashType()),
+merkle_record_type = pack.ComposedType([
+    ('left', pack.IntType(256)),
+    ('right', pack.IntType(256)),
 ])
 
-def merkle_hash(tx_list):
-    if not tx_list:
+def merkle_hash(hashes):
+    if not hashes:
         return 0
-    hash_list = map(tx_type.hash256, tx_list)
+    hash_list = list(hashes)
     while len(hash_list) > 1:
-        hash_list = [merkle_record_type.hash256(dict(left=left, right=left if right is None else right))
-            for left, right in zip(hash_list[::2], hash_list[1::2] + [None])]
+        hash_list = [hash256(merkle_record_type.pack(dict(left=left, right=right)))
+            for left, right in zip(hash_list[::2], hash_list[1::2] + [hash_list[::2][-1]])]
     return hash_list[0]
 
+def calculate_merkle_link(hashes, index):
+    # XXX optimize this
+    
+    hash_list = [(lambda _h=h: _h, i == index, []) for i, h in enumerate(hashes)]
+    
+    while len(hash_list) > 1:
+        hash_list = [
+            (
+                lambda _left=left, _right=right: hash256(merkle_record_type.pack(dict(left=_left(), right=_right()))),
+                left_f or right_f,
+                (left_l if left_f else right_l) + [dict(side=1, hash=right) if left_f else dict(side=0, hash=left)],
+            )
+            for (left, left_f, left_l), (right, right_f, right_l) in
+                zip(hash_list[::2], hash_list[1::2] + [hash_list[::2][-1]])
+        ]
+    
+    res = [x['hash']() for x in hash_list[0][2]]
+    
+    assert hash_list[0][1]
+    if p2pool.DEBUG:
+        new_hashes = [random.randrange(2**256) if x is None else x
+            for x in hashes]
+        assert check_merkle_link(new_hashes[index], dict(branch=res, index=index)) == merkle_hash(new_hashes)
+    assert index == sum(k*2**i for i, k in enumerate([1-x['side'] for x in hash_list[0][2]]))
+    
+    return dict(branch=res, index=index)
+
+def check_merkle_link(tip_hash, link):
+    if link['index'] >= 2**len(link['branch']):
+        raise ValueError('index too large')
+    return reduce(lambda c, (i, h): hash256(merkle_record_type.pack(
+        dict(left=h, right=c) if (link['index'] >> i) & 1 else
+        dict(left=c, right=h)
+    )), enumerate(link['branch']), tip_hash)
+
+# targets
+
 def target_to_average_attempts(target):
+    assert 0 <= target and isinstance(target, (int, long)), target
+    if target >= 2**256: warnings.warn('target >= 2**256!')
     return 2**256//(target + 1)
 
+def average_attempts_to_target(average_attempts):
+    assert average_attempts > 0
+    return min(int(2**256/average_attempts - 1 + 0.5), 2**256-1)
+
+def target_to_difficulty(target):
+    assert 0 <= target and isinstance(target, (int, long)), target
+    if target >= 2**256: warnings.warn('target >= 2**256!')
+    return (0xffff0000 * 2**(256-64) + 1)/(target + 1)
+
+def difficulty_to_target(difficulty):
+    assert difficulty >= 0
+    if difficulty == 0: return 2**256-1
+    return min(int((0xffff0000 * 2**(256-64) + 1)/difficulty - 1 + 0.5), 2**256-1)
+
 # human addresses
 
-human_address_type = ChecksummedType(ComposedType([
-    ('version', StructType("<B")),
-    ('pubkey_hash', ShortHashType()),
-]))
+base58_alphabet = '123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz'
+
+def base58_encode(bindata):
+    bindata2 = bindata.lstrip(chr(0))
+    return base58_alphabet[0]*(len(bindata) - len(bindata2)) + math.natural_to_string(math.string_to_natural(bindata2), base58_alphabet)
 
-pubkey_type = FixedStrType(65)
+def base58_decode(b58data):
+    b58data2 = b58data.lstrip(base58_alphabet[0])
+    return chr(0)*(len(b58data) - len(b58data2)) + math.natural_to_string(math.string_to_natural(b58data2, base58_alphabet))
+
+human_address_type = ChecksummedType(pack.ComposedType([
+    ('version', pack.IntType(8)),
+    ('pubkey_hash', pack.IntType(160)),
+]))
 
 def pubkey_hash_to_address(pubkey_hash, net):
-    return human_address_type.pack_base58(dict(version=net.BITCOIN_ADDRESS_VERSION, pubkey_hash=pubkey_hash))
+    return base58_encode(human_address_type.pack(dict(version=net.ADDRESS_VERSION, pubkey_hash=pubkey_hash)))
 
 def pubkey_to_address(pubkey, net):
-    return pubkey_hash_to_address(pubkey_type.hash160(pubkey), net)
+    return pubkey_hash_to_address(hash160(pubkey), net)
 
 def address_to_pubkey_hash(address, net):
-    x = human_address_type.unpack_base58(address)
-    if x['version'] != net.BITCOIN_ADDRESS_VERSION:
+    x = human_address_type.unpack(base58_decode(address))
+    if x['version'] != net.ADDRESS_VERSION:
         raise ValueError('address not for this net!')
     return x['pubkey_hash']
 
-# network definitions
+def address_to_script(address, net):
+    x = human_address_type.unpack(base58_decode(address))
 
-class Mainnet(object):
-    BITCOIN_P2P_PREFIX = 'f9beb4d9'.decode('hex')
-    BITCOIN_P2P_PORT = 8333
-    BITCOIN_ADDRESS_VERSION = 0
+    print x['pubkey_hash']
 
-class Testnet(object):
-    BITCOIN_P2P_PREFIX = 'fabfb5da'.decode('hex')
-    BITCOIN_P2P_PORT = 18333
-    BITCOIN_ADDRESS_VERSION = 111
+    if x['version'] != net.ADDRESS_VERSION:
+        raise ValueError('address not for this net!')
+    return '\x76\xa9' + ('\x14' + pack.IntType(160).pack(x['pubkey_hash'])) + '\x88\xac'
+
+
+# transactions
+
+def pubkey_to_script2(pubkey):
+    assert len(pubkey) <= 75
+    return (chr(len(pubkey)) + pubkey) + '\xac'
+
+def pubkey_hash_to_script2(pubkey_hash):
+    return '\x76\xa9' + ('\x14' + pack.IntType(160).pack(pubkey_hash)) + '\x88\xac'
+
+def script2_to_address(script2, net):
+    try:
+        pubkey = script2[1:-1]
+        script2_test = pubkey_to_script2(pubkey)
+    except:
+        pass
+    else:
+        if script2_test == script2:
+            return pubkey_to_address(pubkey, net)
+    
+    try:
+        pubkey_hash = pack.IntType(160).unpack(script2[3:-2])
+        script2_test2 = pubkey_hash_to_script2(pubkey_hash)
+    except:
+        pass
+    else:
+        if script2_test2 == script2:
+            return pubkey_hash_to_address(pubkey_hash, net)
+
+def script2_to_human(script2, net):
+    try:
+        pubkey = script2[1:-1]
+        script2_test = pubkey_to_script2(pubkey)
+    except:
+        pass
+    else:
+        if script2_test == script2:
+            return 'Pubkey. Address: %s' % (pubkey_to_address(pubkey, net),)
+    
+    try:
+        pubkey_hash = pack.IntType(160).unpack(script2[3:-2])
+        script2_test2 = pubkey_hash_to_script2(pubkey_hash)
+    except:
+        pass
+    else:
+        if script2_test2 == script2:
+            return 'Address. Address: %s' % (pubkey_hash_to_address(pubkey_hash, net),)
+    
+    return 'Unknown. Script: %s'  % (script2.encode('hex'),)