mirror of
https://github.com/norandom/OpenUltraSAST.git
synced 2026-10-16 11:49:28 +00:00
Marius CiepluchandClaude Opus 5.5
f5180a14ec
learn/retrieve.py: stage 1 is a per-instrument Gower distance over the profile's allow-listed features with explicit missing (both ran: mean feature distance; differing states 1; same non-ran state 0; null in one 1; priors dropped with priors=off), n1 = 40 with seeded tie-breaks; stage 2 re-ranks by r = lam(1 - D) + (1 - lam)cos, signals_only without the target's vector; the balance rule takes k_ret = 6 (<= k/2 per label, <= 2 per group, <= 2 other families; `short` recorded). eligible() is the only way into a prompt: not the target's group, not an evaluated group, not the held-out source or framework, and outside deployment not a near-duplicate (cos >= 0.98). assert_boundary repeats the check on rendered ids; demonstrations of a held-out framework swap to their alternates; deployment_fold excludes a scanned repository that is a corpus group; seed_memory never hands a case its own group. learn/folds.py: compile split (25% of groups per primary family, 60/40 boot/val), outer grouped 5-fold stratified by family and label mix, leave-one-source-out, leave-one-framework-out (insufficient data below 10 groups), assert_disjoint and a folds digest; all seeded. Tests: test_learn_retrieve.py (hand-computed Gower, lam 0/0.5, balance, near-duplicates, test_boundary over outer/LOSO/LOFO folds with demonstrations), test_learn_folds.py. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01EmM1j46qgKMh3zqD1QMwEo
65 lines
2.6 KiB
Python
65 lines
2.6 KiB
Python
"""Shared fixtures for the decision-engine tests: example rows with valid feature records, and scripted clients."""
|
|
|
|
from __future__ import annotations
|
|
|
|
import hashlib
|
|
from collections.abc import Mapping, Sequence
|
|
from typing import Any
|
|
|
|
from openultrasast.learn.examples import Example
|
|
from openultrasast.learn.features import NOT_APPLICABLE, Part, engine_part, quick_part, record
|
|
|
|
|
|
def x_record(*, hits: int = 0, engine: str = "ran", findings: int = 0, language: str = "python", prior_hits: int = 0) -> dict[str, Any]:
|
|
"""A valid ``static`` record: ``hits`` enabled quick hits, the engine in state ``engine`` with ``findings``."""
|
|
quick = quick_part([("r", "enabled")] * hits, {})
|
|
quick = Part("ran", {**quick.values, "qr.prior_hits": prior_hits}, None)
|
|
parts: dict[str, Part] = {"quick": quick, "delta": NOT_APPLICABLE, "facts": NOT_APPLICABLE}
|
|
if engine == "ran":
|
|
parts["engine"] = engine_part([{"rung": "suspicion"}] * findings, completion=1.0, degraded=False)
|
|
elif engine != "none":
|
|
parts["engine"] = Part(engine)
|
|
return record("a.py::f", "injection", language, "static", parts)
|
|
|
|
|
|
def example(
|
|
name: str,
|
|
*,
|
|
group: str,
|
|
label: int,
|
|
family: str = "injection",
|
|
source: str = "pairs",
|
|
frameworks: Sequence[str] = (),
|
|
hits: int = 0,
|
|
engine: str = "ran",
|
|
findings: int = 0,
|
|
license: str = "",
|
|
) -> Example:
|
|
rec = x_record(hits=hits, engine=engine, findings=findings)
|
|
return Example(
|
|
id=hashlib.sha256(name.encode()).hexdigest(), group=group, source=source, frameworks=tuple(frameworks), family=family, label=label,
|
|
language="python", profile="static", unit="pin", x=rec["x"], instruments=rec["instruments"], roles=(),
|
|
excerpt_sha=hashlib.sha256(f"excerpt:{name}".encode()).hexdigest(), license=license,
|
|
) # fmt: skip
|
|
|
|
|
|
def corpus(groups: int = 12, per_group: int = 4) -> list[Example]:
|
|
"""``groups`` repository groups over three sources and two frameworks, each with positives and negatives."""
|
|
out: list[Example] = []
|
|
for g in range(groups):
|
|
source = ("pairs", "population-v1", "population-v2")[g % 3]
|
|
frameworks = ("flask",) if g % 4 == 0 else ("django",) if g % 4 == 1 else ()
|
|
family = ("injection", "path")[g % 2]
|
|
for i in range(per_group):
|
|
out.append(
|
|
example(
|
|
f"g{g}-e{i}", group=f"org{g}/repo{g}", label=i % 2, family=family, source=source, frameworks=frameworks,
|
|
hits=(g + i) % 5, findings=i % 3,
|
|
)
|
|
) # fmt: skip
|
|
return out
|
|
|
|
|
|
def by_id(examples: Sequence[Example]) -> Mapping[str, Example]:
|
|
return {e.id: e for e in examples}
|