portfolio-optimiser-claude/tests/test_method_spec_loadbearing.py

"""Spec-integrity seam for the method spec (method-spec §11, row "Spec integrity").

The sibling of ``test_ingest_spec_loadbearing.py``: this repo is built from
``shared/method-spec.md`` alone (pulled unchanged from commons), and this test keeps that
contract honest — it goes RED when the spec goes missing, names a concrete agent toolkit
(the framework-neutrality rule), loses a structural section/step marker, or stops
documenting a contract field this implementation consumes (§12 completeness).

Precision note (lesson from the ingest guard's detach spot-check): a substring-anywhere
assertion does not detach when only the §12 table row is removed but the field is still
mentioned in running text. The field assertions here are therefore scoped to the §12
cross-check block and require the backticked form, so removing a table row alone turns
the guard red. All assertions are presence-only (no forbidden-new-sections semantics),
so a commons amendment that ADDS sections or fields keeps the guard green.

Red-proofs run against a mutated COPY of the spec in ``tmp_path`` — never against
``shared/`` itself.
"""

from __future__ import annotations

from pathlib import Path

import pytest

SPEC = Path(__file__).resolve().parents[1] / "shared" / "method-spec.md"

# Concrete agent toolkits / vendor stacks the framework-neutral spec MUST NOT name
# (same list as the ingest-spec guard).
_FORBIDDEN_TOOLKITS = (
    "claude",
    "anthropic",
    "openai",
    "gpt",
    "gemini",
    "llama",
    "langchain",
    "autogen",
    "crewai",
    "semantic kernel",
    "microsoft agent framework",
    "agent sdk",
    "bedrock",
    "vertex",
    "foundry",
    "maf",
)

# Structural markers the spec must keep: the twelve normative sections, the eight loop
# steps, and RFC 2119 normative language.
_STRUCTURE_MARKERS = tuple(
    [f"## {n}." for n in range(1, 13)] + [f"### Step {n}" for n in range(1, 9)] + ["MUST"]
)

# Every §12 contract field this implementation consumes (src-consumed per the 2026-07-16
# review's consume-list; the last four — outcome, modelled_saving_nok, gap_source,
# context_key — are golden/test-consumed only, guarded all the same).
_CONTRACT_FIELDS = (
    "decision",
    "marker",
    "rationale",
    "id",
    "proposal_features",
    "affected_codes",
    "measure_type",
    "claimed_saving_nok",
    "description",
    "project_id",
    "measure",
    "affected_items",
    "code",
    "quantity",
    "unit_cost",
    "assumptions",
    "validates",
    "nominal_feasible",
    "p10",
    "p50",
    "p90",
    "realization_rate",
    "expected_actual_saving_nok",
    "type",
    "approved",
    "rejected",
    "approved_with_adjustment",
    "outcome",
    "modelled_saving_nok",
    "gap_source",
    "context_key",
)


def _named_toolkits(text: str) -> list[str]:
    low = text.lower()
    return [tok for tok in _FORBIDDEN_TOOLKITS if tok in low]


def _missing_markers(text: str) -> list[str]:
    return [marker for marker in _STRUCTURE_MARKERS if marker not in text]


def _section12_block(text: str) -> str:
    start = text.find("## 12.")
    if start == -1:
        return ""
    end = text.find("\n## ", start + 1)
    return text[start:] if end == -1 else text[start:end]


def _undocumented_fields(text: str) -> list[str]:
    block = _section12_block(text)
    return [field for field in _CONTRACT_FIELDS if f"`{field}`" not in block]


# --- The guard itself (against the real spec) ---------------------------------------


def test_spec_is_present() -> None:
    # RED if the spec goes missing (the method stops being implementable from spec alone).
    assert SPEC.is_file(), "method-spec.md missing — subtree pull the commons contract"


def test_spec_keeps_structure_markers() -> None:
    missing = _missing_markers(SPEC.read_text(encoding="utf-8"))
    assert not missing, f"structural markers gone from the spec: {missing}"


def test_spec_names_no_agent_toolkit() -> None:
    present = _named_toolkits(SPEC.read_text(encoding="utf-8"))
    assert not present, f"framework-neutral spec names a concrete toolkit: {present}"


@pytest.mark.parametrize("field", _CONTRACT_FIELDS)
def test_spec_documents_contract_field(field: str) -> None:
    undocumented = _undocumented_fields(SPEC.read_text(encoding="utf-8"))
    assert field not in undocumented, (
        f"contract field {field!r} is no longer documented (backticked) in the §12 block"
    )


# --- Red-proofs: the guard MUST fail on a detached spec (mutated copy, never shared/) --


def test_guard_red_when_spec_missing(tmp_path: Path) -> None:
    assert not (tmp_path / "method-spec.md").is_file()


@pytest.mark.parametrize("toolkit", _FORBIDDEN_TOOLKITS)
def test_guard_red_when_toolkit_injected(tmp_path: Path, toolkit: str) -> None:
    mutated = SPEC.read_text(encoding="utf-8") + f"\n\nBuilt on {toolkit}.\n"
    copy = tmp_path / "method-spec.md"
    copy.write_text(mutated, encoding="utf-8")
    assert toolkit in _named_toolkits(copy.read_text(encoding="utf-8"))


@pytest.mark.parametrize("field", _CONTRACT_FIELDS)
def test_guard_red_when_field_removed_from_section12(tmp_path: Path, field: str) -> None:
    text = SPEC.read_text(encoding="utf-8")
    block = _section12_block(text)
    mutated = text.replace(block, block.replace(f"`{field}`", ""))
    copy = tmp_path / "method-spec.md"
    copy.write_text(mutated, encoding="utf-8")
    assert field in _undocumented_fields(copy.read_text(encoding="utf-8"))