Normalizes EIP-712 fixed-length bytes in Ragger client

Also some type-hinting and simplified the integer encoding

client/src/ledger_app_clients/ethereum/eip712/InputData.py

@@ -4,7 +4,8 @@
 import signal
 import sys
 import copy
-from typing import Any, Callable, Optional
+from typing import Any, Callable, Optional, Union
+import struct
 
 from ledger_app_clients.ethereum import keychain
 from ledger_app_clients.ethereum.client import EthAppClient, EIP712FieldType
@@ -118,69 +119,63 @@ def send_struct_def_field(typename, keyname):
     return (typename, type_enum, typesize, array_lvls)
 
 
-def encode_integer(value, typesize):
-    data = bytearray()
-
+def encode_integer(value: Union[str | int], typesize: int) -> bytes:
     # Some are already represented as integers in the JSON, but most as strings
     if isinstance(value, str):
-        base = 10
-        if value.startswith("0x"):
-            base = 16
-        value = int(value, base)
+        value = int(value, 0)
 
     if value == 0:
-        data.append(0)
+        data = b'\x00'
     else:
-        if value < 0:  # negative number, send it as unsigned
-            mask = 0
-            for i in range(typesize):  # make a mask as big as the typesize
-                mask = (mask << 8) | 0xff
-            value &= mask
-        while value > 0:
-            data.append(value & 0xff)
-            value >>= 8
-        data.reverse()
+        # biggest uint type accepted by struct.pack
+        uint64_mask = 0xffffffffffffffff
+        data = struct.pack(">QQQQ",
+                           (value >> 192) & uint64_mask,
+                           (value >> 128) & uint64_mask,
+                           (value >> 64) & uint64_mask,
+                           value & uint64_mask)
+        data = data[len(data) - typesize:]
+        data = data.lstrip(b'\x00')
     return data
 
 
-def encode_int(value, typesize):
+def encode_int(value: str, typesize: int) -> bytes:
     return encode_integer(value, typesize)
 
 
-def encode_uint(value, typesize):
+def encode_uint(value: str, typesize: int) -> bytes:
     return encode_integer(value, typesize)
 
 
-def encode_hex_string(value, size):
-    data = bytearray()
-    value = value[2:]  # skip 0x
-    byte_idx = 0
-    while byte_idx < size:
-        data.append(int(value[(byte_idx * 2):(byte_idx * 2 + 2)], 16))
-        byte_idx += 1
-    return data
+def encode_hex_string(value: str, size: int) -> bytes:
+    assert value.startswith("0x")
+    value = value[2:]
+    if len(value) < (size * 2):
+        value = value.rjust(size * 2, "0")
+    assert len(value) == (size * 2)
+    return bytes.fromhex(value)
 
 
-def encode_address(value, typesize):
+def encode_address(value: str, typesize: int) -> bytes:
     return encode_hex_string(value, 20)
 
 
-def encode_bool(value, typesize):
-    return encode_integer(value, typesize)
+def encode_bool(value: str, typesize: int) -> bytes:
+    return encode_integer(value, 1)
 
 
-def encode_string(value, typesize):
+def encode_string(value: str, typesize: int) -> bytes:
     data = bytearray()
     for char in value:
         data.append(ord(char))
     return data
 
 
-def encode_bytes_fix(value, typesize):
+def encode_bytes_fix(value: str, typesize: int) -> bytes:
     return encode_hex_string(value, typesize)
 
 
-def encode_bytes_dyn(value, typesize):
+def encode_bytes_dyn(value: str, typesize: int) -> bytes:
     # length of the value string
     # - the length of 0x (2)
     # / by the length of one byte in a hex string (2)