fix: enforce canonical factor contract json

2026-09-01 08:35:37 +08:00
parent b853cc0d4c
commit 003d182e49
2 changed files with 61 additions and 0 deletions
@@ -263,6 +263,57 @@ def test_golden_contracts_are_content_addressed_round_trippable_and_deeply_immut
        snapshot.snapshot_id = "rhdsv1:sha256:" + "0" * 64  # type: ignore[misc]


+@pytest.mark.parametrize("variant", ["whitespace", "key_order"])
+def test_contract_decoders_reject_non_canonical_json(variant: str) -> None:
+    fixture = _golden()
+    snapshot, foundation = _snapshot_and_foundation(fixture)
+    definition = _golden_definition()
+    factor_set = FactorSetRef.create(
+        **_factor_set_arguments(
+            fixture=fixture,
+            snapshot=snapshot,
+            foundation=foundation,
+            definition=definition,
+        )
+    )
+    binding = LegacyFactorBinding.create(
+        definition=definition,
+        legacy_factor_id="factor:demo-momentum",
+        legacy_version="1.0.0",
+        legacy_definition_sha256="b" * 64,
+        legacy_dataset_schema_version="1.0.0",
+        canonical_input_schema_digest=definition.input_schema_digest,
+        correspondence_evidence_digest="sha256:" + "5" * 64,
+    )
+
+    def non_canonical(value: str) -> str:
+        if variant == "whitespace":
+            return value + "\n"
+        loaded = json.loads(value)
+        reversed_items = dict(reversed(tuple(loaded.items())))
+        return json.dumps(reversed_items, ensure_ascii=False, separators=(",", ":"))
+
+    decoders = (
+        lambda value: FactorDefinition.from_json(value),
+        lambda value: FactorSetRef.from_json(
+            value,
+            definitions=(definition,),
+            dataset_snapshot=snapshot,
+            foundation=foundation,
+        ),
+        lambda value: LegacyFactorBinding.from_json(value, definition=definition),
+    )
+    for decoder, encoded in zip(
+        decoders,
+        (definition.to_json(), factor_set.to_json(), binding.to_json()),
+        strict=True,
+    ):
+        with pytest.raises(FactorContractError) as exc_info:
+            decoder(non_canonical(encoded))
+        assert exc_info.value.code is ContractErrorCode.INVALID_FORMAT
+        assert exc_info.value.path == "$"
+
+
 def test_factor_definition_identity_is_order_independent_where_semantics_are_unordered() -> None:
    first_input = FactorInput("prices", "sha256:" + "6" * 64, ("close",))
    second_input = FactorInput("volumes", "sha256:" + "7" * 64, ("volume",))