CultLib/packages/cultcache-py/tests/test_union_parity.py
Meta Vens 6afa8ed9c4 cultcache-py, -rs and -ts: unit-arm tests assert against the bytes the C# reference wrote
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-14 20:33:24 +02:00

752 lines
37 KiB
Python

from __future__ import annotations
import collections
import dataclasses
import shutil
import tempfile
import traceback
import types
import unittest
from pathlib import Path
import msgpack
from cultcache_py import (
MAX_PAYLOAD_DEPTH,
MAX_SLOT_KEY,
CultCache,
CultCachePayloadDepthError,
DatabaseEntryField,
SlotCodec,
SingleFileMessagePackBackingStore,
UnionArm,
database_entry_field,
define_database_entry_type,
define_document_type,
define_record_codec,
define_union,
leaves as t,
)
VECTORS = Path(__file__).resolve().parents[3] / "tests" / "vectors" / "union-parity"
_GOLDEN = "GameCult.Caching.Tests.UnionParityVectorTests+"
_node = t.nullable(t.lazy(lambda: node))
composition = define_record_codec(
"Composition",
[database_entry_field("name", 0, codec=t.string), database_entry_field("body", 1, codec=_node)],
type_name=_GOLDEN + "Composition",
)
box = define_record_codec(
"Box",
[
database_entry_field("id", 0, codec=t.string),
database_entry_field("children", 1, codec=t.list(t.lazy(lambda: node))),
],
type_name=_GOLDEN + "Box",
)
text = define_record_codec(
"Text",
[database_entry_field("id", 0, codec=t.string), database_entry_field("content", 1, codec=t.string)],
type_name=_GOLDEN + "Text",
)
use = define_record_codec(
"Use",
[
database_entry_field("id", 0, codec=t.string),
database_entry_field("composition", 1, codec=composition),
database_entry_field("slots", 2, codec=t.list(t.nullable(t.lazy(lambda: node)))),
],
type_name=_GOLDEN + "Use",
)
image = define_record_codec(
"Image",
[
database_entry_field("id", 0, codec=t.string),
database_entry_field("uri", 1, codec=t.string),
database_entry_field("alt", 3, codec=t.string),
],
type_name=_GOLDEN + "Image",
)
wrap = define_record_codec(
"Wrap",
[database_entry_field("inner", 0, codec=_node), database_entry_field("id", 1, codec=t.string)],
type_name=_GOLDEN + "Wrap",
)
node = define_union(
"Node",
{0: ("Box", box), 1: ("Text", text), 2: ("Use", use), 5: ("Image", image), 6: ("Wrap", wrap)},
type_name=_GOLDEN + "Node",
)
# The golden's Outcome: the arms of a list carry an element id, so Denied, Floor and Ceiling are one-slot records.
_id_only = lambda name: define_record_codec(name, [database_entry_field("id", 0, codec=t.string)]) # noqa: E731
outcome = define_union(
"Outcome",
{
0: (
"Pending",
define_record_codec(
"Pending",
[database_entry_field("message", 0, codec=t.string), database_entry_field("id", 1, codec=t.string)],
),
),
1: (
"Completed",
define_record_codec(
"Completed",
[database_entry_field("source_version", 0, codec=t.int32), database_entry_field("id", 1, codec=t.string)],
),
),
2: ("Denied", _id_only("Denied")),
-2147483648: ("Floor", _id_only("Floor")),
2147483647: ("Ceiling", _id_only("Ceiling")),
},
type_name=_GOLDEN + "Outcome",
)
# The codec's unit arms (no value, written [key, []]), kept apart from the golden: a document list's arms carry ids.
unit_outcome = define_union(
"Outcome",
{
0: ("Pending", define_record_codec("Pending", [database_entry_field("message", 0, codec=t.string)])),
2: ("Denied", None),
-2147483648: ("Floor", None),
2147483647: ("Ceiling", None),
},
type_name=_GOLDEN + "UnitOutcome",
)
tool = define_record_codec(
"Tool",
[database_entry_field("name", 0, codec=t.string), database_entry_field("uses", 1, codec=t.int32)],
type_name=_GOLDEN + "Tool",
)
part = define_record_codec(
"Part",
[database_entry_field("name", 0, codec=t.string), database_entry_field("count", 1, codec=t.int32)],
type_name=_GOLDEN + "Part",
)
gem = define_record_codec("Gem", [database_entry_field("name", 0, codec=t.string)], type_name=_GOLDEN + "Gem")
item = define_union("Item", {0: ("Gem", gem), 1: ("Tool", tool), 2: ("Part", part)}, type_name=_GOLDEN + "Item")
crafted = define_union("CraftedItem", {0: ("Tool", tool), 1: ("Part", part)}, type_name=_GOLDEN + "CraftedItem")
surface_document = define_database_entry_type(
"tests.union_surface",
[
database_entry_field("id", 0, codec=t.string),
database_entry_field("version", 1, codec=t.int32),
database_entry_field("root", 2, codec=_node),
database_entry_field("outcomes", 3, codec=t.list(outcome)),
database_entry_field("part", 4, codec=t.nullable(item)),
database_entry_field("crafted", 5, codec=t.nullable(crafted)),
database_entry_field("missing", 6, codec=_node),
],
name="id",
)
def _surface_value() -> dict:
return {
"id": "surface",
"version": 3,
"root": UnionArm(
"Box",
{
"id": "0000000000c0",
"children": [
UnionArm("Text", {"id": "0000000000c1", "content": "hello"}),
UnionArm(
"Use",
{
"id": "0000000000c2",
"composition": {
"name": "comp",
"body": UnionArm(
"Box",
{
"id": "0000000000d0",
"children": [
UnionArm("Text", {"id": "0000000000d1", "content": "x"}),
UnionArm("Image", {"id": "0000000000d2", "uri": "uri://a", "alt": "alt"}),
],
},
),
},
"slots": [
UnionArm("Text", {"id": "0000000000e1", "content": "slot"}),
None,
UnionArm("Box", {"id": "0000000000e2", "children": []}),
],
},
),
UnionArm("Image", {"id": "0000000000c3", "uri": "uri://b", "alt": ""}),
],
},
),
"outcomes": [
UnionArm("Pending", {"message": "wait", "id": "0000000000a1"}),
UnionArm("Completed", {"source_version": 7, "id": "0000000000a2"}),
UnionArm("Denied", {"id": "0000000000a3"}),
],
"part": UnionArm("Tool", {"name": "hammer", "uses": 12}),
"crafted": UnionArm("Part", {"name": "gear", "count": 4}),
"missing": None,
}
def _vector(name: str) -> bytes:
return (VECTORS / name).read_bytes()
class UnionSurfaceVectorTests(unittest.TestCase):
def test_surface_payload_decodes_to_the_value_and_re_encodes_byte_identical(self) -> None:
payload = _vector("union-surface.msgpack")
decoded = surface_document.decode_payload(payload)
self.assertEqual(decoded, _surface_value())
self.assertEqual(surface_document.encode_payload(decoded), payload)
def test_the_csharp_written_store_reads_and_the_cache_rewrites_the_same_payload(self) -> None:
with tempfile.TemporaryDirectory() as tmp:
path = Path(tmp) / "union-surface.cc"
shutil.copyfile(VECTORS / "union-surface.cc", path)
cache = (
CultCache.builder()
.register_document_type(surface_document)
.add_generic_store(SingleFileMessagePackBackingStore(path))
.build()
)
cache.pull_all_backing_stores()
self.assertEqual(cache.get(surface_document, "surface"), _surface_value())
fresh = Path(tmp) / "fresh.cc"
writer = (
CultCache.builder()
.register_document_type(surface_document)
.add_generic_store(SingleFileMessagePackBackingStore(fresh))
.build()
)
writer.pull_all_backing_stores()
writer.put(surface_document, "surface", _surface_value())
[envelope] = SingleFileMessagePackBackingStore(fresh).pull_all()
self.assertEqual(envelope.payload, _vector("union-surface.msgpack"))
def test_the_missing_slot_is_a_node_and_agrees_with_the_csharp_written_root(self) -> None:
payload = _vector("union-surface.msgpack")
slots = msgpack.unpackb(payload, raw=False)
slots[6] = slots[2]
decoded = surface_document.decode_payload(msgpack.packb(slots, use_bin_type=True))
self.assertEqual(decoded["missing"], _surface_value()["root"])
self.assertEqual(surface_document.encode_payload(decoded), msgpack.packb(slots, use_bin_type=True))
with self.assertRaises(ValueError):
surface_document.encode_payload({**_surface_value(), "missing": UnionArm("Part", {"name": "gear", "count": 4})})
def test_the_store_vector_payload_equals_the_surface_payload(self) -> None:
[envelope] = SingleFileMessagePackBackingStore(VECTORS / "union-surface.cc").pull_all()
self.assertEqual(envelope.payload, _vector("union-surface.msgpack"))
def test_every_outcome_in_the_manifest_is_held(self) -> None:
outcomes = msgpack.unpackb(_vector("outcomes.msgpack"), raw=False)
self.assertEqual(len(outcomes), 9)
refused = read_only = 0
for file, outcome_kind, union, key, detail in outcomes:
payload = _vector(file)
if outcome_kind == "refused":
refused += 1
with self.assertRaises(ValueError, msg=file) as caught:
surface_document.decode_payload(payload)
message = str(caught.exception)
self.assertIn(union, message, file)
if detail == "unknown-arm":
self.assertIn(f"arm key {key} ", message, file)
else:
self.assertEqual(detail, "arity")
self.assertIn("malformed", message, file)
else:
read_only += 1
self.assertEqual(outcome_kind, "read-only", file)
self.assertEqual(surface_document.decode_payload(payload), _surface_value(), file)
self.assertEqual((refused, read_only), (5, 4))
class UnionCodecRuleTests(unittest.TestCase):
CANARY = "canary-7f3a91"
def test_a_union_refuses_nil_both_ways_unless_it_is_wrapped_in_nullable(self) -> None:
for refuse in (lambda: node.encode(None), lambda: node.decode(None)):
with self.assertRaises(ValueError) as caught:
refuse()
self.assertEqual(str(caught.exception), "Union 'Node': expected an arm, found nil")
maybe = t.nullable(node)
self.assertIsNone(maybe.encode(None))
self.assertIsNone(maybe.decode(None))
self.assertEqual(maybe.decode(maybe.encode(UnionArm("Text", {"id": "a", "content": "b"}))), UnionArm("Text", {"id": "a", "content": "b"}))
with self.assertRaises(ValueError) as write:
surface_document.encode_payload({**_surface_value(), "outcomes": [None]})
self.assertEqual(
str(write.exception),
"tests.union_surface slot 3 (outcomes): element 0: Union 'Outcome': expected an arm, found nil",
)
def test_unit_arm_is_written_with_empty_slots_and_reads_back(self) -> None:
# The bytes are the C# reference's (tests/vectors/union-parity/unit-arm-denied.msgpack), not spelled here.
reference = _vector("unit-arm-denied.msgpack")
self.assertEqual(msgpack.packb(unit_outcome.encode(UnionArm("Denied"))), reference)
self.assertEqual(unit_outcome.decode(msgpack.unpackb(reference, raw=False)), UnionArm("Denied", None))
def test_unit_arm_refuses_a_value_to_carry(self) -> None:
with self.assertRaises(ValueError) as caught:
unit_outcome.encode(UnionArm("Denied", {"message": self.CANARY}))
self.assertEqual(str(caught.exception), "Union 'Outcome' arm 2 is a unit arm and carries no value")
def test_arm_keys_are_written_as_declared(self) -> None:
self.assertEqual(node.encode(UnionArm("Image", {"id": "a", "uri": "b", "alt": "c"})), [5, ["a", "b", None, "c"]])
self.assertEqual(node.encode(UnionArm("Text", {"id": "a", "content": "b"})), [1, ["a", "b"]])
def test_encode_refuses_an_arm_the_union_does_not_declare_without_echoing_it(self) -> None:
with self.assertRaises(ValueError) as caught:
crafted.encode(UnionArm("Gem", {"name": self.CANARY}))
self.assertEqual(
str(caught.exception), "Union 'CraftedItem' was given a value that is not one of its arms"
)
with self.assertRaises(ValueError):
node.encode({"id": "a"})
def test_unknown_key_names_the_union_the_key_and_the_declared_keys_never_the_payload(self) -> None:
with self.assertRaises(ValueError) as caught:
node.decode([4, [self.CANARY]])
message = str(caught.exception)
self.assertEqual(message, "Union 'Node' does not declare arm key 4 (declared: 0, 1, 2, 5, 6)")
def test_a_malformed_union_value_is_refused_without_echoing_it(self) -> None:
for raw in (
[1],
[1, ["a"], "extra"],
[1, self.CANARY],
self.CANARY,
{"k": self.CANARY},
[self.CANARY, []],
[True, []],
[1.0, []],
):
with self.subTest(raw=type(raw).__name__):
with self.assertRaises(ValueError) as caught:
node.decode(raw)
message = str(caught.exception)
self.assertTrue(message.startswith("Union 'Node' value is malformed: "), message)
self.assertNotIn(self.CANARY, message)
def test_extra_trailing_arm_slots_are_read(self) -> None:
self.assertEqual(
node.decode([1, ["a", "b", "extra"]]),
UnionArm("Text", {"id": "a", "content": "b"}),
)
def test_a_list_slot_requires_an_array_and_keeps_nil_elements_of_a_nullable_item(self) -> None:
slots = [UnionArm("Text", {"id": "a", "content": "b"}), None]
inner = {"name": "c", "body": None}
encoded = use.encode({"id": "u", "composition": inner, "slots": slots})
self.assertEqual(encoded, ["u", ["c", None], [[1, ["a", "b"]], None]])
self.assertEqual(use.decode(encoded)["slots"], slots)
with self.assertRaises(ValueError) as caught:
use.decode(["u", ["c", None], self.CANARY])
self.assertEqual(str(caught.exception), "Use slot 2 (slots): expected an array, found string")
with self.assertRaises(ValueError) as nil_element:
box.decode(["b", [None]])
self.assertEqual(
str(nil_element.exception),
"Box slot 1 (children): element 0: Union 'Node': expected an arm, found nil",
)
def test_a_list_of_nested_records_keeps_a_nil_element_as_csharp_writes_it(self) -> None:
row = define_record_codec(
"Row", [("id", 0), database_entry_field("cells", 1, codec=t.list(t.nullable(text)))]
)
cells = [None, {"id": "c", "content": "x"}, None]
encoded = row.encode({"id": "r", "cells": cells})
self.assertEqual(encoded, ["r", [None, ["c", "x"], None]])
self.assertEqual(row.decode(encoded), {"id": "r", "cells": cells})
def test_a_unit_arm_with_extra_slots_is_read_past(self) -> None:
self.assertEqual(unit_outcome.decode([2, ["extra"]]), UnionArm("Denied"))
def test_union_arms_are_immutable_hashable_values(self) -> None:
arm = UnionArm("Denied")
with self.assertRaises(dataclasses.FrozenInstanceError):
arm.name = "Pending" # type: ignore[misc]
self.assertEqual(len({UnionArm("Denied"), UnionArm("Denied"), UnionArm("Pending", "wait")}), 2)
def test_a_short_record_is_read_with_nil_for_the_slots_it_lacks_as_the_reference_reads_it(self) -> None:
self.assertEqual(node.decode([1, ["only-id"]]), UnionArm("Text", {"id": "only-id", "content": None}))
self.assertEqual(node.decode([0, ["sapphire"]]), UnionArm("Box", {"id": "sapphire", "children": None}))
def test_definition_rules(self) -> None:
cases = [
("Empty", {}, "Union 'Empty' requires at least one arm"),
("Dup", {0: ("A", None), 1: ("A", None)}, "Union 'Dup' has duplicate arm names"),
("Wide", {2**31: ("A", None)}, "Union 'Wide' arm keys must be integers within Int32"),
("Narrow", {-(2**31) - 1: ("A", None)}, "Union 'Narrow' arm keys must be integers within Int32"),
("Flag", {True: ("A", None), "x": ("B", None)}, "Union 'Flag' arm keys must be integers within Int32"),
]
for union_name, arms, expected in cases:
with self.subTest(union=union_name):
with self.assertRaises(ValueError) as caught:
define_union(union_name, arms)
self.assertEqual(str(caught.exception), expected)
for union_name, key in (("TrueKey", True), ("FalseKey", False)):
with self.assertRaises(ValueError) as caught:
define_union(union_name, {key: ("A", None)})
self.assertEqual(str(caught.exception), f"Union {union_name!r} arm keys must be integers within Int32")
edge = define_union("Edge", {-(2**31): ("Low", None), -1: ("Negative", None), 2**31 - 1: ("High", None)})
for arm_name, key in (("Low", -(2**31)), ("Negative", -1), ("High", 2**31 - 1)):
self.assertEqual(edge.encode(UnionArm(arm_name)), [key, []])
self.assertEqual(edge.decode([key, []]), UnionArm(arm_name))
for loose in (lambda: node, node.encode, "node", object):
with self.assertRaises(ValueError) as caught:
database_entry_field("loose", 1, codec=loose) # type: ignore[arg-type]
self.assertIn("must be a SlotCodec", str(caught.exception))
with self.assertRaises(ValueError):
define_union("Loose", {0: ("A", lambda: node)}) # type: ignore[dict-item]
class RecordCodecTests(unittest.TestCase):
def test_a_default_fills_a_slot_past_the_payloads_end_and_a_callable_default_is_called(self) -> None:
record = define_record_codec(
"Defaulted",
[("id", 0), ("label", 1, "none"), ("tags", 2, list), ("plain", 3, 0)],
)
self.assertEqual(record.decode(["a"]), {"id": "a", "label": "none", "tags": [], "plain": 0})
self.assertEqual(record.decode(["a", "x"]), {"id": "a", "label": "x", "tags": [], "plain": 0})
self.assertEqual(record.decode(["a", None, None, None]), {"id": "a", "label": None, "tags": None, "plain": None})
self.assertEqual(record.decode(["a", "x", ["t"], 5]), {"id": "a", "label": "x", "tags": ["t"], "plain": 5})
def test_a_document_type_names_its_members_and_marks_the_name_and_index_slots(self) -> None:
document = define_database_entry_type(
"tests.members",
[("id", 0), ("source_version", 1), ("owner", 2)],
name="id",
indexes={"by_owner": "owner"},
)
members = {member.slot: member for member in document.catalog_entry().members}
self.assertEqual(
[members[slot].member_name for slot in (0, 1, 2)], ["Id", "SourceVersion", "Owner"]
)
self.assertEqual([members[slot].is_name for slot in (0, 1, 2)], [True, False, False])
self.assertEqual([members[slot].index_alias for slot in (0, 1, 2)], [None, None, "by_owner"])
def test_a_document_without_a_string_name_marks_no_member_as_the_name(self) -> None:
document = define_database_entry_type("tests.named", [("id", 0)], name=lambda value: value["id"])
self.assertFalse(document.catalog_entry().members[0].is_name)
def test_a_record_that_is_not_an_array_is_refused_naming_the_type_only(self) -> None:
for refuse in (lambda r: r.decode({"id": "canary-7f3a91"}), lambda r: r.decode("canary-7f3a91")):
record = define_record_codec("Plain", [("id", 0)])
with self.assertRaises(ValueError) as caught:
refuse(record)
self.assertEqual(str(caught.exception), "DatabaseEntry document type 'Plain' expected a MessagePack array")
self.assertNotIn("canary-7f3a91", str(caught.exception))
def test_field_tuples_of_the_wrong_length_are_refused(self) -> None:
for fields in ([("id",)], [("id", 0, 1, 2)]):
with self.assertRaises(ValueError) as caught:
define_record_codec("Bad", fields)
self.assertEqual(
str(caught.exception), "DatabaseEntry field tuples must be (name, key) or (name, key, default)"
)
self.assertEqual(define_record_codec("Pair", [("id", 0)]).decode(["a"]), {"id": "a"})
self.assertEqual(define_record_codec("Triple", [("id", 1, "d")]).decode([None]), {"id": "d"})
def test_duplicate_and_empty_field_lists_are_refused_naming_the_type(self) -> None:
with self.assertRaises(ValueError) as caught:
define_record_codec("Twice", [("a", 0), ("b", 0)])
self.assertEqual(str(caught.exception), "DatabaseEntry document type 'Twice' has duplicate field keys")
with self.assertRaises(ValueError) as empty:
define_record_codec("Nothing", [])
self.assertEqual(str(empty.exception), "DatabaseEntry document types require at least one field")
def test_a_plain_field_helper_makes_an_untyped_slot(self) -> None:
plain = define_record_codec("Plain", [database_entry_field("id", 0), database_entry_field("label", 1)])
self.assertEqual(plain.decode(plain.encode({"id": "a", "label": "b"})), {"id": "a", "label": "b"})
self.assertEqual(plain.decode(["a"]), {"id": "a", "label": None})
field = database_entry_field("id", 0)
self.assertIs(field.codec, t.any)
def test_a_directly_constructed_field_requires_a_default_and_is_untyped(self) -> None:
with self.assertRaises(TypeError):
DatabaseEntryField(name="id", key=0) # type: ignore[call-arg]
field = DatabaseEntryField(name="id", key=0, default=dataclasses.MISSING)
self.assertIs(field.codec, t.any)
def test_slot_codecs_and_fields_are_immutable(self) -> None:
with self.assertRaises(dataclasses.FrozenInstanceError):
node.encode = None # type: ignore[misc]
with self.assertRaises(dataclasses.FrozenInstanceError):
database_entry_field("id", 0).key = 1 # type: ignore[misc]
def test_slot_keys_are_integers_within_zero_and_the_cap_however_the_field_is_made(self) -> None:
message = "DatabaseEntry field keys must be integers within 0..65535"
self.assertEqual(MAX_SLOT_KEY, 65535)
database_entry_field("zero", 0)
top = define_record_codec("Top", [("last", MAX_SLOT_KEY)])
self.assertEqual(len(top.encode({"last": "x"})), MAX_SLOT_KEY + 1)
self.assertEqual(top.decode(top.encode({"last": "x"})), {"last": "x"})
for make in (
lambda key: database_entry_field("bad", key),
lambda key: DatabaseEntryField(name="bad", key=key, default=dataclasses.MISSING),
lambda key: define_record_codec("Bad", [("bad", key)]),
lambda key: define_record_codec("Bad", [("bad", key, "d")]),
):
for key in (-1, MAX_SLOT_KEY + 1, 2**31, True, 1.0, "1"):
with self.subTest(key=key):
with self.assertRaises(ValueError) as caught:
make(key)
self.assertEqual(str(caught.exception), message)
def test_slot_order_follows_keys_not_declaration_order(self) -> None:
record = define_record_codec("Ordered", [("b", 2), ("a", 0)])
self.assertEqual(record.encode({"a": "x", "b": "y"}), ["x", None, "y"])
CANARY = "CANARY-union-golden-4c2a"
def _wrap_chain(wraps: int, leaf: UnionArm) -> UnionArm:
value = leaf
for _ in range(wraps):
value = UnionArm("Wrap", {"inner": value, "id": "0000000000b1"})
return value
class BoundsVectorTests(unittest.TestCase):
def test_every_bounds_outcome_is_held(self) -> None:
entries = msgpack.unpackb(_vector("bounds-outcomes.msgpack"), raw=False)
self.assertEqual(entries[0], ["limit", MAX_PAYLOAD_DEPTH])
counts = {"accepted": 0, "refused": 0, "read-only": 0}
for file, outcome_kind, union, key, detail in entries[1:]:
counts[outcome_kind] += 1
payload = _vector(file)
if outcome_kind == "accepted":
self.assertEqual(
surface_document.encode_payload(surface_document.decode_payload(payload)), _vector(detail), file
)
elif outcome_kind == "read-only":
self.assertEqual(surface_document.decode_payload(payload), _surface_value(), file)
else:
with self.assertRaises(ValueError, msg=file) as caught:
surface_document.decode_payload(payload)
message = str(caught.exception)
self.assertNotIn(CANARY, message, file)
if detail == "depth":
self.assertIsInstance(caught.exception, CultCachePayloadDepthError, file)
self.assertIn(surface_document.type, message, file)
self.assertIn("nests 257 containers deep", message, file)
self.assertIn("the limit is 256", message, file)
self.assertIn("refused on read", message, file)
continue
self.assertNotIsInstance(caught.exception, CultCachePayloadDepthError, file)
self.assertIn(f"Union '{union}' ", message, file)
if detail == "key-out-of-range":
self.assertIn("outside Int32", message, file)
self.assertNotIn("does not declare", message, file)
elif detail == "key-not-integer":
self.assertIn("non-integer", message, file)
else:
self.assertEqual(detail, "nil-slots")
self.assertIn(f"arm {key} has nil slots", message, file)
self.assertEqual(counts, {"accepted": 4, "refused": 7, "read-only": 1})
def test_the_value_at_the_limit_writes_the_vector_and_the_value_past_it_is_refused_on_write(self) -> None:
at_limit = {**_surface_value(), "root": _wrap_chain(126, UnionArm("Box", {"id": "0000000000f0", "children": []}))}
self.assertEqual(surface_document.encode_payload(at_limit), _vector("depth-at-limit.msgpack"))
past = {**_surface_value(), "root": _wrap_chain(127, UnionArm("Text", {"id": "0000000000f0", "content": CANARY}))}
with self.assertRaises(CultCachePayloadDepthError) as caught:
surface_document.encode_payload(past)
message = str(caught.exception)
self.assertEqual((caught.exception.depth, caught.exception.limit), (257, 256))
self.assertIn("the limit is 256", message)
self.assertIn("refused on write", message)
self.assertNotIn(CANARY, message)
def test_a_cache_put_of_the_value_past_the_limit_refuses_and_stores_nothing(self) -> None:
past = {**_surface_value(), "root": _wrap_chain(127, UnionArm("Box", {"id": "0000000000f0", "children": []}))}
with tempfile.TemporaryDirectory() as tmp:
path = Path(tmp) / "store.cc"
cache = (
CultCache.builder()
.register_document_type(surface_document)
.add_generic_store(SingleFileMessagePackBackingStore(path))
.build()
)
with self.assertRaises(CultCachePayloadDepthError):
cache.put(surface_document, "surface", past)
self.assertIsNone(cache.get(surface_document, "surface"))
def test_a_payload_beyond_the_c_unpackers_own_stack_is_the_depth_refusal(self) -> None:
deep = b"\x91" * 2000 + b"\xc0"
with self.assertRaises(CultCachePayloadDepthError) as caught:
surface_document.decode_payload(deep)
self.assertEqual(str(caught.exception).count("refused on read"), 1)
huge = b"\x98" + b"\xc0" * 7 + b"\x91" * 100000 + b"\xc0"
with self.assertRaises(CultCachePayloadDepthError):
surface_document.decode_payload(huge)
def test_a_deep_value_is_refused_by_the_depth_error_not_a_recursion_error(self) -> None:
tree = define_document_type("tests.deep_tree", encode=lambda value: value)
with self.assertRaises(CultCachePayloadDepthError) as caught:
tree.encode_payload(_nest(5000))
self.assertIn("refused on write", str(caught.exception))
document = define_document_type("tests.json_tree")
with self.assertRaises(CultCachePayloadDepthError) as read:
document.decode_payload(b"[" * 300 + b"]" * 300)
self.assertIn("refused on read", str(read.exception))
def test_a_codec_that_exhausts_the_interpreter_stack_keeps_its_own_recursion_error_on_each_side(self) -> None:
def boom(value: object) -> object:
return boom(value)
document = define_document_type("tests.boom", encode=boom, decode=boom)
for run in (lambda: document.encode_payload("x"), lambda: document.decode_payload(b'"x"')):
try:
run()
except RecursionError as caught:
self.assertNotIsInstance(caught, CultCachePayloadDepthError)
self.assertIn("boom", [frame.name for frame in traceback.extract_tb(caught.__traceback__)])
else:
self.fail("the codec's RecursionError was swallowed")
with self.assertRaises(CultCachePayloadDepthError) as stack:
surface_document.decode_payload(bytes([0x91]) * 2000 + bytes([0xC0]))
self.assertEqual(
str(stack.exception),
"tests.union_surface payload nests 257 containers deep; the limit is 256 (refused on read).",
)
def test_a_json_payload_past_the_interpreters_own_stack_keeps_the_decoders_recursion_error(self) -> None:
document = define_document_type("tests.json_tree")
with self.assertRaises(RecursionError) as caught:
document.decode_payload(b"[" * 100000)
self.assertNotIsInstance(caught.exception, CultCachePayloadDepthError)
def test_a_refusal_reports_the_depth_where_the_walk_stopped_whatever_the_payloads_real_depth(self) -> None:
document = define_document_type("tests.json_tree", encode=lambda value: value)
for depth in (257, 300, 900):
with self.assertRaises(CultCachePayloadDepthError) as write:
document.encode_payload(_nest(depth))
self.assertEqual((write.exception.depth, write.exception.limit), (257, 256), depth)
with self.assertRaises(CultCachePayloadDepthError) as read:
document.decode_payload(b"[" * depth + b"]" * depth)
self.assertEqual((read.exception.depth, read.exception.limit), (257, 256), depth)
def test_the_validator_sees_the_value_on_write_and_the_decoded_value_on_read(self) -> None:
seen: list[object] = []
document = define_document_type("tests.validated", validate=seen.append)
document.decode_payload(document.encode_payload({"a": 1}))
self.assertEqual(seen, [{"a": 1}, {"a": 1}])
def test_a_cycle_is_refused_as_depth(self) -> None:
loop: list = []
loop.append(loop)
document = define_document_type("tests.cycle", encode=lambda value: value)
with self.assertRaises(CultCachePayloadDepthError):
document.encode_payload(loop)
def _nest(depth: int, mixed: bool = False) -> object:
value: object = "leaf"
for level in range(depth):
value = {"k": value} if mixed and level % 2 == 0 else [value]
return value
def _as_tuple(tree: object) -> object:
return tuple(_as_tuple(child) for child in tree) if isinstance(tree, list) else tree
class PayloadDepthRuleTests(unittest.TestCase):
"""The nesting rule at its seam, on definitions that are not unions: the same routine serves any payload codec."""
def test_the_limit_is_256_and_depth_256_is_accepted_on_read_and_write_while_257_is_refused(self) -> None:
self.assertEqual(MAX_PAYLOAD_DEPTH, 256)
document = define_document_type("tests.json_tree", encode=lambda value: value)
for mixed in (False, True):
at_limit = _nest(256, mixed)
self.assertEqual(document.decode_payload(document.encode_payload(at_limit)), at_limit)
past = _nest(257, mixed)
with self.assertRaises(CultCachePayloadDepthError):
document.encode_payload(past)
with self.assertRaises(CultCachePayloadDepthError):
document.decode_payload(document.payload_encoder(past))
def test_a_tuple_and_a_container_dict_key_count_as_containers(self) -> None:
document = define_document_type("tests.tuple_tree", encode=lambda value: value, payload_encoder=lambda tree: b"ok")
self.assertEqual(document.encode_payload(_as_tuple(_nest(256))), b"ok")
with self.assertRaises(CultCachePayloadDepthError):
document.encode_payload(_as_tuple(_nest(257)))
self.assertEqual(document.encode_payload({_as_tuple(_nest(255)): 1}), b"ok")
with self.assertRaises(CultCachePayloadDepthError):
document.encode_payload({_as_tuple(_nest(256)): 1})
def test_every_mapping_sequence_and_set_counts_not_only_list_tuple_and_dict(self) -> None:
wrappers = {
"deque": lambda inner: collections.deque([inner]),
"mappingproxy": lambda inner: types.MappingProxyType({"k": inner}),
"frozenset": lambda inner: frozenset([inner]),
"userlist": lambda inner: collections.UserList([inner]),
}
for name, wrap in wrappers.items():
document = define_document_type(f"tests.{name}", encode=lambda value: value, payload_encoder=lambda tree: b"ok")
at_limit: object = "leaf"
for _ in range(256):
at_limit = wrap(at_limit)
self.assertEqual(document.encode_payload(at_limit), b"ok", name)
with self.assertRaises(CultCachePayloadDepthError, msg=name):
document.encode_payload(wrap(at_limit))
decoding = define_document_type(f"tests.{name}.read", payload_decoder=lambda payload, w=wrap: w(at_limit))
with self.assertRaises(CultCachePayloadDepthError, msg=name):
decoding.decode_payload(b"x")
def test_text_and_bytes_are_leaves_however_long(self) -> None:
document = define_document_type("tests.leaves", encode=lambda value: value, payload_encoder=lambda tree: b"ok")
self.assertEqual(document.encode_payload([["x" * 100000, b"y" * 100000, bytearray(5), memoryview(b"z")]]), b"ok")
def test_a_scalar_payload_has_no_depth(self) -> None:
document = define_document_type("tests.scalar", encode=lambda value: value)
for scalar in ("x", 5, None):
self.assertEqual(document.decode_payload(document.encode_payload(scalar)), scalar)
def test_the_refusal_names_the_type_the_depth_and_the_limit_and_never_the_payload(self) -> None:
document = define_document_type("tests.json_tree", encode=lambda value: value)
value: object = CANARY
for _ in range(257):
value = [value]
with self.assertRaises(CultCachePayloadDepthError) as caught:
document.encode_payload(value)
self.assertEqual(
str(caught.exception),
"tests.json_tree payload nests 257 containers deep; the limit is 256 (refused on write).",
)
self.assertIsInstance(caught.exception, ValueError)
class UnionDecodeBoundsTests(unittest.TestCase):
def test_the_arm_key_refusals_name_the_union_and_say_what_is_wrong(self) -> None:
cases = [
([2**31, []], "Union 'Node' value is malformed: the arm key is outside Int32"),
([-(2**31) - 1, []], "Union 'Node' value is malformed: the arm key is outside Int32"),
([1.0, []], "Union 'Node' value is malformed: the arm key is non-integer"),
([True, []], "Union 'Node' value is malformed: the arm key is non-integer"),
([1, None], "Union 'Node' arm 1 has nil slots"),
([1, CANARY], "Union 'Node' value is malformed: expected [armKey, [armSlots]]"),
]
for raw, expected in cases:
with self.subTest(key=raw[0], slots=type(raw[1]).__name__):
with self.assertRaises(ValueError) as caught:
node.decode(raw)
self.assertEqual(str(caught.exception), expected)
def test_the_ends_of_the_int32_key_domain_are_unit_arms_that_read_and_write(self) -> None:
floor = _vector("unit-arm-floor.msgpack")
ceiling = _vector("unit-arm-ceiling.msgpack")
self.assertEqual(unit_outcome.decode(msgpack.unpackb(floor, raw=False)), UnionArm("Floor"))
self.assertEqual(msgpack.packb(unit_outcome.encode(UnionArm("Floor"))), floor)
self.assertEqual(unit_outcome.decode(msgpack.unpackb(ceiling, raw=False)), UnionArm("Ceiling"))
self.assertEqual(msgpack.packb(unit_outcome.encode(UnionArm("Ceiling"))), ceiling)
if __name__ == "__main__":
unittest.main()