Merge pull request #3 from jamesob/jamesob-25-03-0.0.2

0.0.2: base58, memzero

Author: jamesob

CHANGELOG.md

@@ -0,0 +1,6 @@
+## 0.0.2
+
+- Changed `bip32_serialze` str_len to a pointer which returns the final length of the
+  base58-encoded out string.
+- Added some precautionary `sodium_memzero()` calls.
+- Made `bip32_b58_encode()` and `bip32_b58_decode()` a public part of the API.

bip32.c

@@ -66,6 +66,7 @@ int bip32_from_seed(bip32_key *key, const unsigned char *seed, size_t seed_len)
     memcpy(key->chain_code, output + BIP32_PRIVKEY_SIZE, BIP32_CHAINCODE_SIZE);
 
 exit:
+    sodium_memzero(output, crypto_auth_hmacsha512_BYTES);
     secp256k1_context_destroy(ctx);
     return retcode;
 }
@@ -164,6 +165,9 @@ int bip32_index_derive(bip32_key *target, const bip32_key *source, uint32_t inde
 
     bip32_hmac_sha512(output, source->chain_code, BIP32_CHAINCODE_SIZE, hmac_msg, hmac_msg_len);
 
+    // hmac_msg potentially contains privkey bytes.
+    sodium_memzero(hmac_msg, hmac_msg_len);
+
     memcpy(target->chain_code, output + BIP32_PRIVKEY_SIZE, BIP32_CHAINCODE_SIZE);
 
     if (source->is_private) {
@@ -222,19 +226,6 @@ int bip32_index_derive(bip32_key *target, const bip32_key *source, uint32_t inde
     return retcode;
 }
 
-
-// Returns true if invalid path characters are detected in a path string.
-static bool has_invalid_path_characters(const char* str) {
-    const char* valid = "m/0123456789hH'pP";
-    while (*str) {
-        if (!strchr(valid, *str)) {
-            return true;
-        }
-        str++;
-    }
-    return false;
-}
-
 int bip32_derive_from_str(bip32_key* target, const char* source, const char* path) {
     if (!target || !source || !path || strncmp(path, "m", 1) != 0) {
         return 0;
@@ -250,6 +241,7 @@ int bip32_derive_from_str(bip32_key* target, const char* source, const char* pat
         strncmp(source, "xpub", 4) == 0 ||
         strncmp(source, "tpub", 4) == 0) {
         if (!bip32_deserialize(&basekey, source, strlen(source))) {
+            sodium_memzero(&basekey, sizeof(bip32_key));
             return 0;
         }
     }
@@ -261,6 +253,7 @@ int bip32_derive_from_str(bip32_key* target, const char* source, const char* pat
             return 0;
         }
         if (!bip32_from_seed(&basekey, seedbytes, bin_len)) {
+            sodium_memzero(&basekey, sizeof(bip32_key));
             return 0;
         }
     } else {
@@ -269,6 +262,7 @@ int bip32_derive_from_str(bip32_key* target, const char* source, const char* pat
 
     if (bip32_derive(&basekey, path)) {
         memcpy(target, &basekey, sizeof(bip32_key));
+        sodium_memzero(&basekey, sizeof(bip32_key));
         return 1;
     }
     return 0;
@@ -284,6 +278,18 @@ int bip32_derive_from_seed(bip32_key* target, const unsigned char* seed, size_t
     return 0;
 }
 
+// Returns true if invalid path characters are detected in a path string.
+static bool has_invalid_path_characters(const char* str) {
+    const char* valid = "m/0123456789hH'pP";
+    while (*str) {
+        if (!strchr(valid, *str)) {
+            return true;
+        }
+        str++;
+    }
+    return false;
+}
+
 // Do an in-place derivation on `key`.
 int bip32_derive(bip32_key* key, const char* path) {
     if (!path || strncmp(path, "m", 1) != 0 || has_invalid_path_characters(path)) {
@@ -314,6 +320,7 @@ int bip32_derive(bip32_key* key, const char* path) {
         if (bip32_index_derive(key, &tmp, path_index) != 1) {
             return 0;
         }
+        sodium_memzero(&tmp, sizeof(bip32_key));
         p = strchr(end, '/');
     }
 
@@ -323,7 +330,7 @@ int bip32_derive(bip32_key* key, const char* path) {
 #define SER_SIZE 78
 #define SER_PLUS_CHECKSUM_SIZE (SER_SIZE + 4)
 
-int bip32_serialize(const bip32_key *key, char *str, size_t str_len) {
+int bip32_serialize(const bip32_key *key, char *str, size_t* str_len) {
     unsigned char data[SER_PLUS_CHECKSUM_SIZE];
     uint32_t version;
 
@@ -362,7 +369,10 @@ int bip32_serialize(const bip32_key *key, char *str, size_t str_len) {
     bip32_sha256_double(hash, data, 78);
     memcpy(data + SER_SIZE, hash, 4);
 
-    return b58enc(str, &str_len, data, SER_PLUS_CHECKSUM_SIZE);
+    bool b58_ok = bip32_b58_encode(str, str_len, data, SER_PLUS_CHECKSUM_SIZE);
+    sodium_memzero(data, SER_PLUS_CHECKSUM_SIZE);
+
+    return b58_ok ? 1 : 0;
 }
 
 #define BIP32_BASE58_BYTES_LEN 82
@@ -371,7 +381,8 @@ int bip32_deserialize(bip32_key *key, const char *str, const size_t str_len) {
     unsigned char data[BIP32_BASE58_BYTES_LEN];
     size_t data_len = BIP32_BASE58_BYTES_LEN;
 
-    if (!b58tobin(data, &data_len, str, str_len) || data_len != BIP32_BASE58_BYTES_LEN) {
+    if (!bip32_b58_decode(data, &data_len, str, str_len) || data_len != BIP32_BASE58_BYTES_LEN) {
+        sodium_memzero(data, BIP32_BASE58_BYTES_LEN);
         return 0;
     }
 
@@ -426,16 +437,22 @@ int bip32_deserialize(bip32_key *key, const char *str, const size_t str_len) {
 
     if (key->is_private) {
         if (data[45] != 0) {
+            sodium_memzero(data, BIP32_BASE58_BYTES_LEN);
             secp256k1_context_destroy(ctx);
             return 0;
         }
+
         memcpy(key->key.privkey, data + 46, BIP32_PRIVKEY_SIZE);
+        sodium_memzero(data, BIP32_BASE58_BYTES_LEN);
+
         if (!secp256k1_ec_seckey_verify(ctx, key->key.privkey)) {
             secp256k1_context_destroy(ctx);
             return 0;
         }
     } else {
         memcpy(key->key.pubkey, data + 45, BIP32_PUBKEY_SIZE);
+        sodium_memzero(data, BIP32_BASE58_BYTES_LEN);
+
         secp256k1_pubkey pubkey;
         if (!secp256k1_ec_pubkey_parse(ctx, &pubkey, key->key.pubkey, BIP32_PUBKEY_SIZE)) {
             secp256k1_context_destroy(ctx);
@@ -484,3 +501,11 @@ void bip32_hmac_sha512(
     crypto_auth_hmacsha512_update(&state, msg, msg_len);
     crypto_auth_hmacsha512_final(&state, hmac_out);
 }
+
+bool bip32_b58_encode(char* str_out, size_t* out_size, const unsigned char* data, size_t data_size) {
+    return b58enc(str_out, out_size, data, data_size);
+}
+
+bool bip32_b58_decode(unsigned char* bin_out, size_t* out_size, const char* str_in, size_t str_size) {
+    return b58tobin(bin_out, out_size, str_in, str_size);
+}

bip32.h

@@ -1,3 +1,7 @@
+/**
+ * This API is generally designed to mimic the libsecp256k1 library, in terms of
+ * argument order and return conventions.
+ */
 #include <stdbool.h>
 #include <stdint.h>
 #include <stdio.h>
@@ -60,11 +64,14 @@ int bip32_derive_from_str(bip32_key *target, const char* source, const char* pat
  */
 int bip32_derive(bip32_key *target, const char* path);
 
-/** Serialize a BIP32 key to its base58 string representation.
+/** Serialize a BIP32 key to its base58 string representation. Writes the resulting
+ * string to `str`, and the length of the resulting string to `str_len`.
+ *
+ * `str_len` must initially be set to the maximum length of the `str` buffer.
  *
  * Returns 1 if successful.
  */
-int bip32_serialize(const bip32_key *key, char *str, size_t str_len);
+int bip32_serialize(const bip32_key *key, char *str, size_t* str_len);
 
 /** Deserialize a BIP32 key from its base58 string representation.
  *
@@ -106,6 +113,18 @@ void bip32_hmac_sha512(
     size_t msg_len
 );
 
+/** Encode some bytes to a base58 string.
+ *
+ * Returns true if successful.
+ */
+bool bip32_b58_encode(char* str_out, size_t* out_size, const unsigned char* data, size_t data_size);
+
+/** Decode some bytes from a base58 string.
+ *
+ * Returns true if successful.
+ */
+bool bip32_b58_decode(unsigned char* bin_out, size_t* out_size, const char* str_in, size_t str_size);
+
 #ifdef __cplusplus
 }
 #endif

examples/cli.c

@@ -15,7 +15,8 @@ int main(int argc, char *argv[]) {
        return 1;
    }
 
-   if (!bip32_serialize(&key, serialized, sizeof(serialized))) {
+   size_t out_size;
+   if (!bip32_serialize(&key, serialized, &out_size)) {
        fprintf(stderr, "Serialization failed\n");
        return 1;
    }

examples/py/bindings.py

@@ -3,9 +3,17 @@
 from functools import lru_cache
 from pathlib import Path
 from ctypes import (
-    c_uint8, c_uint32, c_size_t, c_char_p, c_ubyte, c_void_p,
-    Structure, Union, POINTER, create_string_buffer
-)
+    c_uint8,
+    c_uint32,
+    c_size_t,
+    c_char_p,
+    c_ubyte,
+    c_void_p,
+    Structure,
+    Union,
+    POINTER,
+    create_string_buffer,
+    byref)
 
 
 @lru_cache
@@ -20,7 +28,9 @@ def get_bip32_module():
     bip32_lib.bip32_init.argtypes = [POINTER(BIP32Key)]
     bip32_lib.bip32_init.restype = None
 
-    bip32_lib.bip32_derive_from_seed.argtypes = [POINTER(BIP32Key), POINTER(c_ubyte), c_size_t, c_char_p]
+    bip32_lib.bip32_derive_from_seed.argtypes = [
+        POINTER(BIP32Key), POINTER(c_ubyte), c_size_t, c_char_p
+    ]
     bip32_lib.bip32_derive_from_seed.restype = ctypes.c_int
 
     bip32_lib.bip32_derive_from_str.argtypes = [POINTER(BIP32Key), c_char_p, c_char_p]
@@ -29,7 +39,9 @@ def get_bip32_module():
     bip32_lib.bip32_derive.argtypes = [POINTER(BIP32Key), c_char_p]
     bip32_lib.bip32_derive.restype = ctypes.c_int
 
-    bip32_lib.bip32_serialize.argtypes = [POINTER(BIP32Key), c_char_p, c_size_t]
+    bip32_lib.bip32_serialize.argtypes = [
+        POINTER(BIP32Key), c_char_p, POINTER(c_size_t)
+    ]
     bip32_lib.bip32_serialize.restype = ctypes.c_bool
 
     bip32_lib.bip32_deserialize.argtypes = [POINTER(BIP32Key), c_char_p, c_size_t]
@@ -38,14 +50,22 @@ def get_bip32_module():
     bip32_lib.bip32_get_public.argtypes = [POINTER(BIP32Key), POINTER(BIP32Key)]
     bip32_lib.bip32_get_public.restype = ctypes.c_int
 
+    bip32_lib.bip32_b58_encode.argtypes = [
+        c_char_p, POINTER(c_size_t), POINTER(c_ubyte), c_size_t
+    ]
+    bip32_lib.bip32_b58_encode.restype = ctypes.c_bool
+
+    bip32_lib.bip32_b58_decode.argtypes = [
+        POINTER(c_ubyte), POINTER(c_size_t), c_char_p, c_size_t
+    ]
+    bip32_lib.bip32_b58_decode.restype = ctypes.c_bool
+
     return bip32_lib
 
 
 class KeyUnion(Union):
-    _fields_ = [
-        ('privkey', c_uint8 * 32),
-        ('pubkey', c_uint8 * 33)
-    ]
+    _fields_ = [('privkey', c_uint8 * 32), ('pubkey', c_uint8 * 33)]
+
 
 class BIP32Key(Structure):
     _fields_ = [
@@ -65,6 +85,7 @@ def print(self):
 
 
 class BIP32:
+
     def __init__(self):
         self.key = BIP32Key()
         self.bip32_lib = get_bip32_module()
@@ -80,7 +101,9 @@ def derive(self, path: str) -> 'BIP32':
 
     def serialize(self):
         buf = create_string_buffer(200)  # Standard BIP32 serialization length
-        if not self.bip32_lib.bip32_serialize(self.key, buf, len(buf)):
+        out_len = c_size_t(len(buf))
+
+        if not self.bip32_lib.bip32_serialize(self.key, buf, byref(out_len)):
             raise ValueError("Serialization failed")
         return buf.value.decode()
 
@@ -106,7 +129,8 @@ def derive(source: str, path: str = 'm') -> BIP32:
 
     """
     b = BIP32()
-    if not get_bip32_module().bip32_derive_from_str(b.key, source.encode(), path.encode()):
+    if not get_bip32_module().bip32_derive_from_str(
+            b.key, source.encode(), path.encode()):
         raise ValueError("failed")
     return b
 
@@ -115,6 +139,35 @@ def derive_from_seed(seed: bytes, path: str = 'm') -> BIP32:
     b = BIP32()
     c_seed = ctypes.c_char_p(seed)
     seed_ptr = ctypes.cast(c_seed, POINTER(c_ubyte))
-    if not get_bip32_module().bip32_derive_from_seed(b.key, seed_ptr, len(seed), path.encode()):
+    if not get_bip32_module().bip32_derive_from_seed(
+            b.key, seed_ptr, len(seed), path.encode()):
         raise ValueError("failed")
     return b
+
+
+def b58_encode(inp: bytes) -> str:
+    data_len = len(inp)
+    data_arr = (c_ubyte * data_len)(*inp)
+
+    out_size = c_size_t(data_len * 2)
+    str_out = ctypes.create_string_buffer(out_size.value)
+
+    if not get_bip32_module().bip32_b58_encode(
+            str_out, byref(out_size), data_arr, data_len):
+        raise ValueError("base58 encoding failed")
+
+    return str_out.value[:out_size.value].decode('utf-8')
+
+
+def b58_decode(in_str: str) -> bytes:
+    str_bytes = c_char_p(in_str.encode('utf-8'))
+    str_len = len(in_str)
+
+    out_size = c_size_t(str_len * 2)
+    bin_out = (c_ubyte * out_size.value)()
+
+    if not get_bip32_module().bip32_b58_decode(
+            bin_out, byref(out_size), str_bytes, str_len):
+        raise ValueError("base58 decoding failed")
+
+    return bytes(bin_out[-out_size.value:])

examples/py/test_fuzz_cross_impl.py

@@ -18,7 +18,7 @@
 import pytest
 from hypothesis import given, strategies as st, target, settings
 
-from bindings import derive, derive_from_seed
+from bindings import derive, derive_from_seed, b58_decode, b58_encode
 
 log = logging.getLogger(__name__)
 logging.basicConfig()
@@ -181,5 +181,32 @@ def test_xpub_impls(bip32_path):
     assert ours == pys
 
 
+
+@given(b58_data=st.binary(min_size=0, max_size=1000))
+@settings(max_examples=1000)
+def test_base58(b58_data: bytes):
+    if b58_data and len(b58_data) >= 2:
+        # TODO: figure out why the base58 impl is failing on example b':'
+        assert b58_decode(b58_encode(b58_data)) == b58_data
+
+
+def test_base58_known_vectors():
+    cases = [
+        (bytes.fromhex(""), ""),
+        (bytes.fromhex("00"), "1"),
+        (bytes.fromhex("0000"), "11"),
+        (bytes.fromhex("68656c6c6f20776f726c64"), "StV1DL6CwTryKyV"),
+        (bytes.fromhex("0068656c6c6f20776f726c64"), "1StV1DL6CwTryKyV"),
+        (bytes.fromhex("000068656c6c6f20776f726c64"), "11StV1DL6CwTryKyV"),
+    ]
+
+    for raw, encoded in cases:
+        if raw:  # Skip empty input for encoding test
+            assert b58_encode(raw) == encoded
+
+        if encoded:  # Skip empty input for decoding test
+            assert b58_decode(encoded) == raw
+
+
 if __name__ == "__main__":
     pytest.main([__file__, "-v", "--capture=no", "--hypothesis-show-statistics", "-x"] + sys.argv[1:])