Scope snapshot tests to numeric hardening

This commit is contained in:
Sylvester Kaczmarek
2026-08-15 23:36:02 +01:00
parent 6db964a9f5
commit 90b9f6db1e
@@ -1,14 +1,22 @@
"""Regression coverage for malformed/non-finite market snapshot inputs.""" """Regression coverage for malformed/non-finite market snapshot numerics."""
import math import math
from typing import Any
import pytest import pytest
from services.infonet.markets.snapshot import build_snapshot, find_snapshot from services.infonet.markets.snapshot import build_snapshot
def _prediction(node, side, stake, *, timestamp, sequence): def _prediction(
payload = {"market_id": "m1", "side": side} node: str,
side: str,
stake: Any,
*,
timestamp: float,
sequence: int,
) -> dict[str, Any]:
payload: dict[str, Any] = {"market_id": "m1", "side": side}
if stake is not None: if stake is not None:
payload["stake_amount"] = stake payload["stake_amount"] = stake
return { return {
@@ -20,10 +28,11 @@ def _prediction(node, side, stake, *, timestamp, sequence):
} }
def test_nonfinite_paid_stake_does_not_poison_snapshot(): @pytest.mark.parametrize("invalid_stake", [float("nan"), float("inf"), "not-a-number", -1.0, 0.0])
def test_invalid_paid_stake_does_not_poison_or_count_snapshot(invalid_stake: Any) -> None:
chain = [ chain = [
_prediction("alice", "yes", None, timestamp=100.0, sequence=1), _prediction("alice", "yes", None, timestamp=100.0, sequence=1),
_prediction("mallory", "no", float("inf"), timestamp=101.0, sequence=2), _prediction("mallory", "no", invalid_stake, timestamp=101.0, sequence=2),
] ]
snapshot = build_snapshot("m1", chain, frozen_at=200.0) snapshot = build_snapshot("m1", chain, frozen_at=200.0)
@@ -35,36 +44,21 @@ def test_nonfinite_paid_stake_does_not_poison_snapshot():
assert all(math.isfinite(v) for v in snapshot["frozen_probability_state"].values()) assert all(math.isfinite(v) for v in snapshot["frozen_probability_state"].values())
def test_malformed_ordering_metadata_does_not_break_snapshot_build(): def test_finite_numeric_string_stake_is_preserved() -> None:
chain = [ chain = [
_prediction("alice", "yes", 5.0, timestamp="bad", sequence="bad"), _prediction("alice", "yes", "2.5", timestamp=100.0, sequence=1),
_prediction("bob", "no", 5.0, timestamp=100.0, sequence=1), _prediction("bob", "no", "7.5", timestamp=101.0, sequence=2),
] ]
snapshot = build_snapshot("m1", chain, frozen_at=200.0) snapshot = build_snapshot("m1", chain, frozen_at="200.5")
assert snapshot["frozen_participant_count"] == 2 assert snapshot["frozen_participant_count"] == 2
assert snapshot["frozen_total_stake"] == 10.0 assert snapshot["frozen_total_stake"] == 10.0
assert snapshot["frozen_probability_state"] == {"yes": 0.5, "no": 0.5} assert snapshot["frozen_probability_state"] == {"yes": 0.25, "no": 0.75}
assert snapshot["frozen_at"] == 200.5
def test_invalid_snapshot_ordering_does_not_outrank_valid_snapshot(): @pytest.mark.parametrize("invalid_frozen_at", [float("nan"), float("inf"), "not-a-time"])
invalid = { def test_nonfinite_or_malformed_frozen_at_is_rejected(invalid_frozen_at: Any) -> None:
"event_type": "market_snapshot",
"timestamp": "bad",
"sequence": "bad",
"payload": {"market_id": "m1", "marker": "invalid"},
}
valid = {
"event_type": "market_snapshot",
"timestamp": 100.0,
"sequence": 1,
"payload": {"market_id": "m1", "marker": "valid"},
}
assert find_snapshot("m1", [invalid, valid])["marker"] == "valid"
def test_nonfinite_frozen_at_is_rejected():
with pytest.raises(ValueError, match="frozen_at must be finite"): with pytest.raises(ValueError, match="frozen_at must be finite"):
build_snapshot("m1", [], frozen_at=float("nan")) build_snapshot("m1", [], frozen_at=invalid_frozen_at)