Reject altered leading zero bytes in BTC addresses

Author: Plastic-time

src/validators/crypto_addresses/btc_address.py

@@ -9,16 +9,20 @@
 
 
 def _decode_base58(addr: str):
-    """Decode base58."""
+    """Decode base58, preserving each leading one as a zero byte."""
     alphabet = "123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz"
-    return sum((58**enm) * alphabet.index(idx) for enm, idx in enumerate(addr[::-1]))
+    decoded = sum((58**enm) * alphabet.index(idx) for enm, idx in enumerate(addr[::-1]))
+    zero_bytes = len(addr) - len(addr.lstrip("1"))
+    return b"\0" * zero_bytes + decoded.to_bytes((decoded.bit_length() + 7) // 8, "big")
 
 
 def _validate_old_btc_address(addr: str):
     """Validate P2PKH and P2SH type address."""
     if len(addr) not in range(25, 35):
         return False
-    decoded_bytes = _decode_base58(addr).to_bytes(25, "big")
+    decoded_bytes = _decode_base58(addr)
+    if len(decoded_bytes) != 25:
+        return False
     header, checksum = decoded_bytes[:-4], decoded_bytes[-4:]
     return checksum == sha256(sha256(header).digest()).digest()[:4]
 

tests/crypto_addresses/test_btc_address.py

@@ -12,6 +12,10 @@
     [
         # P2PKH (Pay-to-PubkeyHash) type
         "1BvBMSEYstWetqTFn5Au4m4GFg7xJaNVN2",
+        # P2PKH with zero bytes in the payload, encoded as leading ones
+        "1111111111111111111114oLvT2",
+        # Legacy testnet P2PKH
+        "mipcBbFg9gMiCh81Kj8tqqdgoZub1ZJRfn",
         # P2SH (Pay to script hash) type
         "3J98t1WpEZ73CNmQviecrnyiWrnqRhWNLy",
         # Bech32/segwit type
@@ -41,3 +45,18 @@ def test_returns_true_on_valid_btc_address(value: str):
 def test_returns_failed_validation_on_invalid_btc_address(value: str):
     """Test returns failed validation on invalid btc address."""
     assert isinstance(btc_address(value), ValidationError)
+
+
+@pytest.mark.parametrize(
+    "value",
+    [
+        # Removing the version byte's leading one must not be silently padded.
+        "BvBMSEYstWetqTFn5Au4m4GFg7xJaNVN2",
+        # Both remain within the existing character-count bounds.
+        "111111111111111111114oLvT2",
+        "11111111111111111111114oLvT2",
+    ],
+)
+def test_rejects_changed_leading_zero_bytes(value: str):
+    """Leading ones encode bytes and cannot be added or removed."""
+    assert isinstance(btc_address(value), ValidationError)