fix #3899; and more aggressively catch exceptions in tx.deserialize()

lib/tests/test_transaction.py

@@ -235,6 +235,11 @@ class TestTransaction(unittest.TestCase):
         tx = transaction.Transaction('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')
         self.assertEqual('e9933221a150f78f9f224899f8568ff6422ffcc28ca3d53d87936368ff7c4b1d', tx.txid())
 
+    # input: p2sh, not multisig
+    def test_txid_regression_issue_3899(self):
+        tx = transaction.Transaction('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')
+        self.assertEqual('f570d5d1e965ee61bcc7005f8fefb1d3abbed9d7ddbe035e2a68fa07e5fc4a0d', tx.txid())
+
 
 class NetworkMock(object):
 

lib/transaction.py

@@ -32,6 +32,8 @@ from .util import print_error, profiler
 from . import bitcoin
 from .bitcoin import *
 import struct
+import traceback
+import sys
 
 #
 # Workalike python implementation of Bitcoin's CDataStream class.
@@ -303,7 +305,8 @@ def parse_scriptSig(d, _bytes):
         decoded = [ x for x in script_GetOp(_bytes) ]
     except Exception as e:
         # coinbase transactions raise an exception
-        print_error("cannot find address in input script", bh2u(_bytes))
+        print_error("parse_scriptSig: cannot find address in input script (coinbase?)",
+                    bh2u(_bytes))
         return
 
     match = [ opcodes.OP_PUSHDATA4 ]
@@ -334,9 +337,9 @@ def parse_scriptSig(d, _bytes):
             d['pubkeys'] = ["(pubkey)"]
         return
 
-    # non-generated TxIn transactions push a signature
+    # p2pkh TxIn transactions push a signature
-    # (seventy-something bytes) and then their public key
+    # (71-73 bytes) and then their public key
-    # (65 bytes) onto the stack:
+    # (33 or 65 bytes) onto the stack:
     match = [ opcodes.OP_PUSHDATA4, opcodes.OP_PUSHDATA4 ]
     if match_decoded(decoded, match):
         sig = bh2u(decoded[0][1])
@@ -345,7 +348,8 @@ def parse_scriptSig(d, _bytes):
             signatures = parse_sig([sig])
             pubkey, address = xpubkey_to_address(x_pubkey)
         except:
-            print_error("cannot find address in input script", bh2u(_bytes))
+            print_error("parse_scriptSig: cannot find address in input script (p2pkh?)",
+                        bh2u(_bytes))
             return
         d['type'] = 'p2pkh'
         d['signatures'] = signatures
@@ -357,19 +361,26 @@ def parse_scriptSig(d, _bytes):
 
     # p2sh transaction, m of n
     match = [ opcodes.OP_0 ] + [ opcodes.OP_PUSHDATA4 ] * (len(decoded) - 1)
-    if not match_decoded(decoded, match):
+    if match_decoded(decoded, match):
-        print_error("cannot find address in input script", bh2u(_bytes))
+        x_sig = [bh2u(x[1]) for x in decoded[1:-1]]
+        try:
+            m, n, x_pubkeys, pubkeys, redeemScript = parse_redeemScript(decoded[-1][1])
+        except NotRecognizedRedeemScript:
+            # we could still guess:
+            # d['address'] = hash160_to_p2sh(hash_160(decoded[-1][1]))
+            return
+        # write result in d
+        d['type'] = 'p2sh'
+        d['num_sig'] = m
+        d['signatures'] = parse_sig(x_sig)
+        d['x_pubkeys'] = x_pubkeys
+        d['pubkeys'] = pubkeys
+        d['redeemScript'] = redeemScript
+        d['address'] = hash160_to_p2sh(hash_160(bfh(redeemScript)))
         return
-    x_sig = [bh2u(x[1]) for x in decoded[1:-1]]
+
-    m, n, x_pubkeys, pubkeys, redeemScript = parse_redeemScript(decoded[-1][1])
+    print_error("parse_scriptSig: cannot find address in input script (unknown)",
-    # write result in d
+                bh2u(_bytes))
-    d['type'] = 'p2sh'
-    d['num_sig'] = m
-    d['signatures'] = parse_sig(x_sig)
-    d['x_pubkeys'] = x_pubkeys
-    d['pubkeys'] = pubkeys
-    d['redeemScript'] = redeemScript
-    d['address'] = hash160_to_p2sh(hash_160(bfh(redeemScript)))
 
 
 def parse_redeemScript(s):
@@ -380,7 +391,7 @@ def parse_redeemScript(s):
     op_n = opcodes.OP_1 + n - 1
     match_multisig = [ op_m ] + [opcodes.OP_PUSHDATA4]*n + [ op_n, opcodes.OP_CHECKMULTISIG ]
     if not match_decoded(dec2, match_multisig):
-        print_error("cannot find address in input script", bh2u(s))
+        print_error("parse_redeemScript: not multisig", bh2u(s))
         raise NotRecognizedRedeemScript()
     x_pubkeys = [bh2u(x[1]) for x in dec2[1:-2]]
     pubkeys = [safe_parse_pubkey(x) for x in x_pubkeys]
@@ -436,7 +447,11 @@ def parse_input(vds):
         d['num_sig'] = 0
         if scriptSig:
             d['scriptSig'] = bh2u(scriptSig)
-            parse_scriptSig(d, scriptSig)
+            try:
+                parse_scriptSig(d, scriptSig)
+            except BaseException:
+                traceback.print_exc(file=sys.stderr)
+                print_error('failed to parse scriptSig', bh2u(scriptSig))
         else:
             d['scriptSig'] = ''
 
@@ -465,25 +480,40 @@ def parse_witness(vds, txin):
     # between p2wpkh and p2wsh; we do this based on number of witness items,
     # hence (FIXME) p2wsh with n==2 (maybe n==1 ?) will probably fail.
     # If v==0 and n==2, we need parent scriptPubKey to distinguish between p2wpkh and p2wsh.
-    if txin['type'] == 'coinbase':
+    try:
-        pass
+        if txin['type'] == 'coinbase':
-    elif txin['type'] == 'p2wsh-p2sh' or n > 2:
+            pass
-        try:
+        elif txin['type'] == 'p2wsh-p2sh' or n > 2:
-            m, n, x_pubkeys, pubkeys, witnessScript = parse_redeemScript(bfh(w[-1]))
+            try:
-        except NotRecognizedRedeemScript:
+                m, n, x_pubkeys, pubkeys, witnessScript = parse_redeemScript(bfh(w[-1]))
+            except NotRecognizedRedeemScript:
+                raise UnknownTxinType()
+            txin['signatures'] = parse_sig(w[1:-1])
+            txin['num_sig'] = m
+            txin['x_pubkeys'] = x_pubkeys
+            txin['pubkeys'] = pubkeys
+            txin['witnessScript'] = witnessScript
+            if not txin.get('scriptSig'):  # native segwit script
+                txin['type'] = 'p2wsh'
+                txin['address'] = bitcoin.script_to_p2wsh(txin['witnessScript'])
+        elif txin['type'] == 'p2wpkh-p2sh' or n == 2:
+            txin['num_sig'] = 1
+            txin['x_pubkeys'] = [w[1]]
+            txin['pubkeys'] = [safe_parse_pubkey(w[1])]
+            txin['signatures'] = parse_sig([w[0]])
+            if not txin.get('scriptSig'):  # native segwit script
+                txin['type'] = 'p2wpkh'
+                txin['address'] = bitcoin.public_key_to_p2wpkh(bfh(txin['pubkeys'][0]))
+        else:
             raise UnknownTxinType()
-        txin['signatures'] = parse_sig(w[1:-1])
+    except UnknownTxinType:
-        txin['num_sig'] = m
+        txin['type'] = 'unknown'
-        txin['x_pubkeys'] = x_pubkeys
+        # FIXME: GUI might show 'unknown' address (e.g. for a non-multisig p2wsh)
-        txin['pubkeys'] = pubkeys
+    except BaseException:
-        txin['witnessScript'] = witnessScript
+        txin['type'] = 'unknown'
-    elif txin['type'] == 'p2wpkh-p2sh' or n == 2:
+        traceback.print_exc(file=sys.stderr)
-        txin['num_sig'] = 1
+        print_error('failed to parse witness', txin.get('witness'))
-        txin['x_pubkeys'] = [w[1]]
+
-        txin['pubkeys'] = [safe_parse_pubkey(w[1])]
-        txin['signatures'] = parse_sig([w[0]])
-    else:
-        raise UnknownTxinType()
 
 def parse_output(vds, i):
     d = {}
@@ -513,20 +543,7 @@ def deserialize(raw):
     if is_segwit:
         for i in range(n_vin):
             txin = d['inputs'][i]
-            try:
+            parse_witness(vds, txin)
-                parse_witness(vds, txin)
-            except UnknownTxinType:
-                txin['type'] = 'unknown'
-                # FIXME: GUI might show 'unknown' address (e.g. for a non-multisig p2wsh)
-                continue
-            # segwit-native script
-            if not txin.get('scriptSig'):
-                if txin['num_sig'] == 1:
-                    txin['type'] = 'p2wpkh'
-                    txin['address'] = bitcoin.public_key_to_p2wpkh(bfh(txin['pubkeys'][0]))
-                else:
-                    txin['type'] = 'p2wsh'
-                    txin['address'] = bitcoin.script_to_p2wsh(txin['witnessScript'])
     d['lockTime'] = vds.read_uint32()
     return d