store hashes in the database, to optimize space. use util.py
[electrum-server.git] / backends / bitcoind / blockchain_processor.py
index 7a6194a..7891f68 100644 (file)
@@ -6,47 +6,8 @@ from Queue import Queue
 import traceback, sys, os, random
 
 
-
-Hash = lambda x: hashlib.sha256(hashlib.sha256(x).digest()).digest()
-hash_encode = lambda x: x[::-1].encode('hex')
-hash_decode = lambda x: x.decode('hex')[::-1]
-
-
-
-def rev_hex(s):
-    return s.decode('hex')[::-1].encode('hex')
-
-
-def int_to_hex(i, length=1):
-    s = hex(i)[2:].rstrip('L')
-    s = "0"*(2*length - len(s)) + s
-    return rev_hex(s)
-
-def header_to_string(res):
-    pbh = res.get('prev_block_hash')
-    if pbh is None: pbh = '0'*64
-    s = int_to_hex(res.get('version'),4) \
-        + rev_hex(pbh) \
-        + rev_hex(res.get('merkle_root')) \
-        + int_to_hex(int(res.get('timestamp')),4) \
-        + int_to_hex(int(res.get('bits')),4) \
-        + int_to_hex(int(res.get('nonce')),4)
-    return s
-
-def header_from_string( s):
-    hex_to_int = lambda s: eval('0x' + s[::-1].encode('hex'))
-    h = {}
-    h['version'] = hex_to_int(s[0:4])
-    h['prev_block_hash'] = hash_encode(s[4:36])
-    h['merkle_root'] = hash_encode(s[36:68])
-    h['timestamp'] = hex_to_int(s[68:72])
-    h['bits'] = hex_to_int(s[72:76])
-    h['nonce'] = hex_to_int(s[76:80])
-    return h
-
-
-
-
+from util import Hash, hash_encode, hash_decode, rev_hex, int_to_hex
+from util import bc_address_to_hash_160, hash_160_to_bc_address, header_to_string, header_from_string
 from processor import Processor, print_log
 
 class BlockchainProcessor(Processor):
@@ -174,7 +135,7 @@ class BlockchainProcessor(Processor):
             print_log( "catching up missing headers:", height, db_height)
 
         try:
-            while height != db_height:
+            while height < db_height:
                 height = height + 1
                 header = self.get_header(height)
                 if height>1: 
@@ -258,7 +219,8 @@ class BlockchainProcessor(Processor):
 
         with self.dblock:
             try:
-                hist = self.deserialize(self.db.Get(addr))
+                hash_160 = bc_address_to_hash_160(addr)
+                hist = self.deserialize(self.db.Get(hash_160))
                 is_known = True
             except: 
                 hist = []
@@ -347,16 +309,16 @@ class BlockchainProcessor(Processor):
         self.batch_txio[txo] = addr
 
 
-    def remove_from_history(self, tx_hash, tx_pos):
+    def remove_from_history(self, addr, tx_hash, tx_pos):
                     
         txi = (tx_hash + int_to_hex(tx_pos, 4)).decode('hex')
-        try:
-            addr = self.batch_txio[txi]
-        except:
-            raise BaseException(tx_hash, tx_pos)
-            print "WARNING: cannot find address for", (tx_hash, tx_pos)
-            return
 
+        if addr is None:
+            try:
+                addr = self.batch_txio[txi]
+            except:
+                raise BaseException(tx_hash, tx_pos)
+        
         serialized_hist = self.batch_list[addr]
 
         l = len(serialized_hist)/40
@@ -367,6 +329,7 @@ class BlockchainProcessor(Processor):
                 serialized_hist = serialized_hist[0:40*i] + serialized_hist[40*(i+1):]
                 break
         else:
+            hist = self.deserialize(serialized_hist)
             raise BaseException("prevout not found", addr, hist, tx_hash, tx_pos)
 
         self.batch_list[addr] = serialized_hist
@@ -393,7 +356,8 @@ class BlockchainProcessor(Processor):
         return eval(s)
 
     def write_undo_info(self, batch, height, undo_info):
-        batch.Put("undo%d"%(height%100), repr(undo_info))
+        if self.is_test or height > self.bitcoind_height - 100:
+            batch.Put("undo%d"%(height%100), repr(undo_info))
 
 
     def import_block(self, block, block_hash, block_height, sync, revert=False):
@@ -401,7 +365,8 @@ class BlockchainProcessor(Processor):
         self.batch_list = {}  # address -> history
         self.batch_txio = {}  # transaction i/o -> address
 
-        inputs_to_read = []
+        block_inputs = []
+        block_outputs = []
         addr_to_read = []
 
         # deserialize transactions
@@ -410,21 +375,16 @@ class BlockchainProcessor(Processor):
 
         t00 = time.time()
 
-        if revert:
-            # read addresses of tx outputs
-            for tx_hash, tx in txdict.items():
-                for x in tx.get('outputs'):
-                    txo = (tx_hash + int_to_hex(x.get('index'), 4)).decode('hex')
-                self.batch_txio[txo] = x.get('address')
-        else:
+
+        if not revert:
             # read addresses of tx inputs
             for tx in txdict.values():
                 for x in tx.get('inputs'):
                     txi = (x.get('prevout_hash') + int_to_hex(x.get('prevout_n'), 4)).decode('hex')
-                    inputs_to_read.append(txi)
+                    block_inputs.append(txi)
 
-            inputs_to_read.sort()
-            for txi in inputs_to_read:
+            block_inputs.sort()
+            for txi in block_inputs:
                 try:
                     addr = self.db.Get(txi)
                 except:
@@ -433,11 +393,17 @@ class BlockchainProcessor(Processor):
                 self.batch_txio[txi] = addr
                 addr_to_read.append(addr)
 
-
+        else:
+            for txid, tx in txdict.items():
+                for x in tx.get('outputs'):
+                    txo = (txid + int_to_hex(x.get('index'), 4)).decode('hex')
+                    block_outputs.append(txo)
+            
         # read histories of addresses
         for txid, tx in txdict.items():
             for x in tx.get('outputs'):
-                addr_to_read.append(x.get('address'))
+                hash_160 = bc_address_to_hash_160(x.get('address'))
+                addr_to_read.append(hash_160)
 
         addr_to_read.sort()
         for addr in addr_to_read:
@@ -449,28 +415,31 @@ class BlockchainProcessor(Processor):
 
         if revert: 
             undo_info = self.get_undo_info(block_height)
-            print "undo", block_height, undo_info
+            # print "undo", block_height, undo_info
         else: undo_info = {}
 
         # process
         t1 = time.time()
 
+        if revert: tx_hashes = tx_hashes[::-1]
         for txid in tx_hashes: # must be ordered
             tx = txdict[txid]
             if not revert:
 
                 undo = []
                 for x in tx.get('inputs'):
-                    prevout_height, prevout_addr = self.remove_from_history( x.get('prevout_hash'), x.get('prevout_n'))
+                    prevout_height, prevout_addr = self.remove_from_history( None, x.get('prevout_hash'), x.get('prevout_n'))
                     undo.append( (prevout_height, prevout_addr) )
                 undo_info[txid] = undo
 
                 for x in tx.get('outputs'):
-                    self.add_to_history( x.get('address'), txid, x.get('index'), block_height)
+                    hash_160 = bc_address_to_hash_160(x.get('address'))
+                    self.add_to_history( hash_160, txid, x.get('index'), block_height)
                     
             else:
                 for x in tx.get('outputs'):
-                    self.remove_from_history( txid, x.get('index'))
+                    hash_160 = bc_address_to_hash_160(x.get('address'))
+                    self.remove_from_history( hash_160, txid, x.get('index'))
 
                 i = 0
                 for x in tx.get('inputs'):
@@ -478,10 +447,12 @@ class BlockchainProcessor(Processor):
                     i += 1
 
                     # read the history into batch list
-                    self.batch_list[prevout_addr] = self.db.Get(prevout_addr)
+                    if self.batch_list.get(prevout_addr) is None:
+                        self.batch_list[prevout_addr] = self.db.Get(prevout_addr)
+
                     # re-add them to the history
                     self.add_to_history( prevout_addr, x.get('prevout_hash'), x.get('prevout_n'), prevout_height)
-                    print "new hist", self.deserialize(self.batch_list[prevout_addr])
+                    print_log( "new hist for", hash_160_to_bc_address(prevout_addr), self.deserialize(self.batch_list[prevout_addr]) )
 
         # write
         max_len = 0
@@ -496,14 +467,23 @@ class BlockchainProcessor(Processor):
                 max_len = l
                 max_addr = addr
 
-        for txio, addr in self.batch_txio.items():
-            batch.Put(txio, addr)
-        # delete spent inputs
-        for txi in inputs_to_read:
-            batch.Delete(txi)
+        if not revert:
+            # add new created outputs
+            for txio, addr in self.batch_txio.items():
+                batch.Put(txio, addr)
+            # delete spent inputs
+            for txi in block_inputs:
+                batch.Delete(txi)
+            # add undo info 
+            self.write_undo_info(batch, block_height, undo_info)
+        else:
+            # restore spent inputs
+            for txio, addr in self.batch_txio.items():
+                batch.Put(txio, addr)
+            # delete spent outputs
+            for txo in block_outputs:
+                batch.Delete(txo)
 
-        # add undo info 
-        if not revert: self.write_undo_info(batch, block_height, undo_info)
 
         # add the max
         batch.Put('height', self.serialize( [(block_hash, block_height, 0)] ) )
@@ -647,8 +627,8 @@ class BlockchainProcessor(Processor):
 
             # are we done yet?
             info = self.bitcoind('getinfo')
-            bitcoind_height = info.get('blocks')
-            bitcoind_block_hash = self.bitcoind('getblockhash', [bitcoind_height])
+            self.bitcoind_height = info.get('blocks')
+            bitcoind_block_hash = self.bitcoind('getblockhash', [self.bitcoind_height])
             if self.last_hash == bitcoind_block_hash: 
                 self.up_to_date = True
                 break
@@ -658,7 +638,9 @@ class BlockchainProcessor(Processor):
             next_block_hash = self.bitcoind('getblockhash', [self.height+1])
             next_block = self.bitcoind('getblock', [next_block_hash, 1])
 
-            revert = (random.randint(1, 1000)!=1) if self.is_test else False
+            # fixme: this is unsafe, if we revert when the undo info is not yet written 
+            revert = (random.randint(1, 100)==1) if self.is_test else False        
+
             if (next_block.get('previousblockhash') == self.last_hash) and not revert:
 
                 self.import_block(next_block, next_block_hash, self.height+1, sync)