bip93: use generalized-HRP draft vectors

Author: BenWestgate

tests/data/bip93_vectors.py

@@ -79,6 +79,31 @@
     "codex32_peev": "cl10peevst6cqh0wu7p5ssjyf4z4ez42ks9jlt3zneju9uuypr2hddak6tlqsjhsks4laxts8q",
 }
 
+# BIP-93 generalized-HRP draft, bitcoin/bips bip93-generalize-hrp @ 01374bf.
+BIP93_GENERAL_HRP = {
+    "share_a": "test_vector12spana320zyxwvutsrqpnmlkjhgfedca320zyxwvutsrqpnmlkjhgfedca304ppsqh4l7v3dh",
+    "share_c": "test_vector12spancacdefghjklmnpqrstuvwxyz023acdefghjklmnpqrstuvwxyz023jxmjy7q9xl7d3ul",
+    "derived_d": "test_vector12spandll4f8jlh4e5vdvuldlfxu2jhdnll4f8jlh4e5vdvuldlfxu2jhdnx3pe3yqzdl7p30y",
+    "secret_s": "test_vector12spans6xqguzttxkeqnjsjzv4jv3nz5k6xqguzttxkeqnjsjzv4jv3nz5kdf37v5qm2l7l3az",
+    "valid_83": (
+        "a83characterlongcodex32humanreadablepartforsecretsharewiththedigit1andthelettersbio"
+        "10keyss9mfjjk25y05e3nq"
+    ),
+    "invalid_gap_regular": (
+        "test_vector12spans6xqguzttxkeqnjsjzv4jv3nz5k6xqguzttxkeqnjsjzv4jv3nz5kp0f0tlr2cxh0t"
+    ),
+    "invalid_gap_long": (
+        "test_vector12spans6xqguzttxkeqnjsjzv4jv3nz5k6xqguzttxkeqnjsjzv4jv3nz5c2m7mxmk8s5qecp"
+    ),
+    "invalid_uppercase_hrp_checksum": (
+        "CL10PEEVST6CQH0WU7P5SSJYF4Z4EZ42KS9JLT3ZNEJU9UUYPR2HDDAK6TLQS3J5AYX4Y08079"
+    ),
+    "invalid_84": (
+        "a84characterslongcodex32humanreadablepartforsecretsharewiththedigit1andthelettersbio"
+        "10keyss9hqxu3pxsnkdkth"
+    ),
+}
+
 
 VALID_CODEX32 = [
     "A12UEL5LLGCHJ4UJCQVHG",
@@ -116,8 +141,6 @@
     + "12fauxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx836hdd09mhkhkhx",  # HRP character out of range
     "\x80"
     + "12fauxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxlhf5ywnkmk4r3tc",  # HRP character out of range
-    # HRP exceeds 83 characters (also exceeds the overall max length)
-    "a1003characterslonghumanreadablepartthatcontainsthenumber1,theexcludedcharactersbio,andeveryus-asciicharacterin[33-126]!\"#$%&'()*+,-./0123456789:;<=>?@[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~!\"#$%&'()*+,-./0123456789:;<=>?@[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~!\"#$%&'()*+,-./0123456789:;<=>?@[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~!\"#$%&'()*+,-./0123456789:;<=>?@[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~!\"#$%&'()*+,-./0123456789:;<=>?@[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~!\"#$%&'()*+,-./0123456789:;<=>?@[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~!\"#$%&'()*+,-./0123456789:;<=>?@[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~!\"#$%&'()*+,-./0123456789:;<=>?@[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~!\"#$%&'()*+,-./0123456789:;<=>?@[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~!\"#$%&'()*+,-./0123456789:;<=>?@[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~!\"#$%&'()*+,-./0123456789:;<=>?@[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~!\"#$%&'()*+,-./0123456789:;<=>?@[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~!\"#$%&'()*+,-./0123456789:;<=>?@[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~12fauxxru38cppmlpu0t6l",
     "y12bfauxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxt3y5fewy4gnw2hs",  # Invalid data character
     "lt12ifauxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxjxd0ehq868vm3zl",  # Invalid data character
     "in12fauxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxvgegljrsvs5w9q",  # Too short checksum

tests/test_generic_hrp.py

@@ -7,6 +7,7 @@
 
 import pytest
 from _codex32_oracle import oracle_encode
+from data.bip93_vectors import BIP93_GENERAL_HRP
 
 from codex32 import (
     CorrectionContext,
@@ -18,7 +19,7 @@
     recover_secret,
 )
 from codex32.cli import main, ms_main
-from codex32.errors import InvalidLength, MismatchedHrp, MismatchedProfile
+from codex32.errors import InvalidChecksum, InvalidLength, MismatchedHrp, MismatchedProfile
 from tools._wallet_test_vectors import STUB_FINGERPRINT
 
 UNKNOWN = {
@@ -180,16 +181,31 @@ def fingerprint(_secret: object) -> bytes:
     assert _invoke(ms_main, ["share", "d", "--plain"], basis)[0] == 2
 
 
-def test_83_character_hrp_may_span_printable_ascii() -> None:
-    pool = [chr(code) for code in range(33, 127) if not chr(code).isupper()]
-    hrp = "".join(pool[index % len(pool)] for index in range(83))
-    assert "1" in hrp  # the separator is the last "1", so an HRP may contain one
-    assert parse_codex32(oracle_encode(hrp, "0testsq")).hrp == hrp
+def test_bip93_generalized_hrp_share_vectors() -> None:
+    share_a = parse_codex32(BIP93_GENERAL_HRP["share_a"])
+    share_c = parse_codex32(BIP93_GENERAL_HRP["share_c"])
+    assert share_a.hrp == share_c.hrp == "test_vector"
+    assert recover_secret([share_a, share_c]).text == BIP93_GENERAL_HRP["secret_s"]
+    assert derive_share([share_a, share_c], "d").text == BIP93_GENERAL_HRP["derived_d"]
 
 
-def test_hrp_over_83_characters_is_rejected_at_the_bip173_limit() -> None:
-    within_limit = oracle_encode("z" * 83, "0testsq")
-    over_limit = oracle_encode("z" * 84, "0testsq")
-    assert parse_codex32(within_limit).hrp == "z" * 83
+def test_bip93_generalized_hrp_boundary_vectors() -> None:
+    valid = BIP93_GENERAL_HRP["valid_83"]
+    invalid = BIP93_GENERAL_HRP["invalid_84"]
+    artifact = parse_codex32(valid)
+
+    assert len(artifact.hrp) == 83
+    assert artifact.hrp == valid.rsplit("1", 1)[0]
+    with pytest.raises(InvalidLength):
+        parse_codex32(invalid)
+
+
+@pytest.mark.parametrize("key", ("invalid_gap_regular", "invalid_gap_long"))
+def test_bip93_generalized_hrp_gap_vectors_are_invalid(key: str) -> None:
     with pytest.raises(InvalidLength):
-        parse_codex32(over_limit)
+        parse_codex32(BIP93_GENERAL_HRP[key])
+
+
+def test_bip93_generalized_hrp_checksum_uses_lowercase_hrp() -> None:
+    with pytest.raises(InvalidChecksum):
+        parse_codex32(BIP93_GENERAL_HRP["invalid_uppercase_hrp_checksum"])

tests/test_profiles.py

@@ -118,17 +118,6 @@ def test_expanded_codeword_upper_bound() -> None:
         _checksum_for_encoded_length("ms", len(bech32_decode(oversized)[1]))
 
 
-def test_long_checksum_residue_stays_correct_past_its_period() -> None:
-    # Probes the polynomial itself with u5 values, not a codex32 string: the
-    # arithmetic still closes beyond the code's 1023-symbol period, so length is
-    # what verify() has to enforce, not the residue.
-    body = bech32_hrp_expand("ms") + [0] * 1100
-    codeword = body + _CODEX32_LONG.create(body)
-    assert len(codeword) > 1023
-    assert _CODEX32_LONG.polymod(codeword) == _CODEX32_LONG.constant
-    assert _CODEX32_LONG.verify(codeword) is False
-
-
 def test_checksum_is_verified_before_unknown_hrp_dispatch() -> None:
     valid_generic = _oracle_encode("zz", "0tests" + "q" * 26)
     artifact = parse_codex32(valid_generic)