Point 2 of the sweep, enumerated rather than assumed. STATE's total was right and
its distribution was not: 86 hits confirmed (`assert not X` 41 / `== []` 42 /
`== {}` 2 / `== set()` 1), but per file measured `test_cli_paritet` 13 (STATE said
19), `test_preflight` 10 (11), `test_step7` 4 (6).
AST triage split the 86: 55 hits sit in 50 tests whose assertions are ALL
negative; the other 31 already have a positive sibling assert in the same test.
Two negative results worth recording, because they bound the remaining work:
- The `test_preflight` "clears" family (`_check_credentials(...) == []` and
friends) is NOT vacuous. Each sits beside a sibling in the same class that
asserts refusals are non-empty, so a no-op checker turns the sibling red.
Class-level pairing is a real control; these need no change.
- `test_method_spec_loadbearing.py` already models the right pattern for
detectors — explicit `test_guard_red_when_*` red-proofs against a mutated COPY.
This commit fixes the class that had no control at all: static/AST guards that
assert an absence without ever showing the scanner can detect a presence.
1. TAUTOLOGICAL RED-PROOFS (both spec guards). `test_guard_red_when_spec_missing`
asserted a file is absent from a fresh `tmp_path` — true by construction of the
fixture, and it never called the guard it is named for. It would have stayed
green with `test_spec_is_present` deleted outright. Both now exercise the same
`_spec_is_present` predicate the guard calls, in both directions.
2. MISSING RED-PROOF. `test_spec_keeps_structure_markers` had none, unlike its
toolkit and contract-field siblings: with `_STRUCTURE_MARKERS` emptied or
`_missing_markers` stubbed to `[]` it reported green forever. Added
`test_guard_red_when_marker_removed`, parametrized over all 21 markers.
3. BLIND IMPORT SCANNERS (costsim x2, okf, preflight, notify). Every one asserted
`not names & {forbidden}` or `outside == set()` with nothing showing `names`
was non-empty — an empty scan satisfies them exactly as well as real purity.
`test_okf_is_pure_stdlib`'s subset check is likewise trivially true of the
empty set, so it did not guard its neighbour either. Each now asserts a
known-present module first. The notify guard gets the strongest form
available: it proves the detector DOES match a network import inside the seam,
so the matcher itself is shown to work rather than only its silence.
Value-proved, not merely detach-proved. Seven vacuity mutations run against the
NEW tests: all seven RED, each dying on the intended control line. The same
mutations run against the PRE-CHANGE tests (session edits stashed): all five
applicable ones GREEN — blind to the vacuity they were meant to catch. Green
before, red after, same mutation, is the value-proof.
Harness held original bytes in memory, restored in `finally`, sha256-verified
every restore, and checked each run ACTUALLY RAN (a wrong test id yields rc!=0
and mimics red). `git status` clean before and after.
Remaining in the class and NOT closed here: ~45 all-negative tests, mostly CLI
refusal (`calls == []` after a refused invocation) and empty-default
(`missing dir -> []`). Listed in STATE, not silently dropped.
Suite 690 -> 711.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DJmse16bEkaSBtvXhncEUc
199 lines
8.1 KiB
Python
199 lines
8.1 KiB
Python
"""Spec-integrity seam for the ingest spec (ingest-spec §11).
|
|
|
|
The D7 analog of MAF's I1 framework-guard: this repo consumes ``shared/ingest-spec.md``
|
|
UNCHANGED from commons, and this test keeps the contract honest — it goes RED when the
|
|
spec goes missing, names a concrete agent toolkit (the framework-neutrality rule), or
|
|
stops documenting a contract field. It is the load-bearing guard the ingest layer relies
|
|
on to keep being implementable "from this spec alone".
|
|
|
|
Form mirrored from the sibling ``test_method_spec_loadbearing.py``: every predicate takes
|
|
the spec TEXT as an argument, so the detach-proofs are tests in the suite rather than a
|
|
one-off spot-check that dies with the session. 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" / "ingest-spec.md"
|
|
|
|
# Concrete agent toolkits / vendor stacks the framework-neutral spec MUST NOT name.
|
|
_FORBIDDEN_TOOLKITS = (
|
|
"claude",
|
|
"anthropic",
|
|
"openai",
|
|
"gpt",
|
|
"gemini",
|
|
"llama",
|
|
"langchain",
|
|
"autogen",
|
|
"crewai",
|
|
"semantic kernel",
|
|
"microsoft agent framework",
|
|
"agent sdk",
|
|
"bedrock",
|
|
"vertex",
|
|
"foundry",
|
|
"maf",
|
|
)
|
|
|
|
# The §12 cross-check table is the ANCHOR, and the spec appoints it itself: "Every field
|
|
# of the machine-readable contracts, mapped to its normative section (completeness is
|
|
# enforced by the spec-integrity test)" — this test is that enforcer. Asserting over the
|
|
# whole spec text instead would be green-but-dead: prose saturates the field names (§1's
|
|
# honesty rule alone carries `generated: true` twice), so no amendment dropping a row
|
|
# could ever turn it red.
|
|
_CROSS_CHECK_HEADING = "## 12. Cross-check table"
|
|
|
|
# Every field of the machine-readable contracts the D7 implementation depends on — the
|
|
# spec's §12 cross-check table must keep documenting each (spec-integrity).
|
|
_CONTRACT_FIELDS = (
|
|
"manifest_version",
|
|
"source",
|
|
"bundle_summary",
|
|
"extractions",
|
|
"source_system",
|
|
"source_query",
|
|
"ingested_at",
|
|
"ingest_manifest",
|
|
"generated",
|
|
"okf_type",
|
|
"max_rows",
|
|
"root",
|
|
"connection_ref", # the sql source reference the D7 sql connector (I5) depends on
|
|
)
|
|
|
|
|
|
def _cross_check_table(text: str) -> str:
|
|
"""The §12 section body — heading to end-of-spec or the next section, whichever first."""
|
|
start = text.index(_CROSS_CHECK_HEADING) # RED (ValueError) if §12 is renamed or dropped
|
|
end = text.find("\n## ", start + len(_CROSS_CHECK_HEADING))
|
|
return text[start:] if end == -1 else text[start:end]
|
|
|
|
|
|
def _named_toolkits(text: str) -> list[str]:
|
|
low = text.lower()
|
|
return [tok for tok in _FORBIDDEN_TOOLKITS if tok in low]
|
|
|
|
|
|
def _undocumented_fields(text: str) -> list[str]:
|
|
# The row's FIRST column is the documented-field claim; a field named only in another
|
|
# row's "Contract" prose does not count.
|
|
table = _cross_check_table(text)
|
|
return [field for field in _CONTRACT_FIELDS if f"| `{field}` |" not in table]
|
|
|
|
|
|
def _slice_defects(table: str) -> list[str]:
|
|
"""Defects in the anchor itself — a slice that widened into the surrounding spec."""
|
|
defects = []
|
|
if not table.startswith(_CROSS_CHECK_HEADING):
|
|
defects.append("does not start at the §12 heading")
|
|
if "Honesty rule" in table:
|
|
defects.append("leaked §1 prose — the anchor is not a slice")
|
|
if "\n## " in table:
|
|
defects.append("leaked a following section")
|
|
return defects
|
|
|
|
|
|
def _row_removed(text: str, field: str) -> str:
|
|
table = _cross_check_table(text)
|
|
kept = [ln for ln in table.splitlines(keepends=True) if not ln.startswith(f"| `{field}` |")]
|
|
return text.replace(table, "".join(kept))
|
|
|
|
|
|
def _row_renamed(text: str, field: str) -> str:
|
|
table = _cross_check_table(text)
|
|
return text.replace(table, table.replace(f"| `{field}` |", f"| `{field}_renamed` |"))
|
|
|
|
|
|
# --- The guard itself (against the real spec) ---------------------------------------
|
|
|
|
|
|
def _spec_is_present(path: Path) -> bool:
|
|
"""The presence predicate itself, so the red-proof can exercise THE SAME one."""
|
|
return path.is_file()
|
|
|
|
|
|
def test_spec_is_present() -> None:
|
|
# RED if the spec goes missing (the layer stops being implementable from spec alone).
|
|
assert _spec_is_present(SPEC), "ingest-spec.md missing — subtree pull the commons contract"
|
|
|
|
|
|
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}"
|
|
|
|
|
|
def test_cross_check_slice_is_a_slice_and_not_the_whole_spec() -> None:
|
|
# Guards the anchor itself: a slice that degenerated into the full text would make
|
|
# every row assertion below green-but-dead again, silently. RED if it widens.
|
|
defects = _slice_defects(_cross_check_table(SPEC.read_text(encoding="utf-8")))
|
|
assert not defects, f"the §12 anchor degenerated: {defects}"
|
|
|
|
|
|
@pytest.mark.parametrize("field", _CONTRACT_FIELDS)
|
|
def test_spec_documents_contract_field(field: str) -> None:
|
|
# RED when an amendment drops or renames a §12 row.
|
|
undocumented = _undocumented_fields(SPEC.read_text(encoding="utf-8"))
|
|
assert field not in undocumented, (
|
|
f"contract field {field!r} is no longer a row in the §12 cross-check table"
|
|
)
|
|
|
|
|
|
# --- 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:
|
|
# Was VACUOUS — see the twin in ``test_method_spec_loadbearing.py``: it asserted
|
|
# a file is absent from a fresh ``tmp_path``, true by construction, and never
|
|
# touched the guard it is named for. Now it exercises THE SAME predicate the
|
|
# guard calls, both directions, positive control first.
|
|
assert _spec_is_present(SPEC)
|
|
assert not _spec_is_present(tmp_path / "ingest-spec.md")
|
|
|
|
|
|
@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 / "ingest-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_row_removed_from_cross_check_table(tmp_path: Path, field: str) -> None:
|
|
# M1: the row is gone from §12 — even though prose elsewhere may still name the field.
|
|
copy = tmp_path / "ingest-spec.md"
|
|
copy.write_text(_row_removed(SPEC.read_text(encoding="utf-8"), field), encoding="utf-8")
|
|
assert field in _undocumented_fields(copy.read_text(encoding="utf-8"))
|
|
|
|
|
|
@pytest.mark.parametrize("field", _CONTRACT_FIELDS)
|
|
def test_guard_red_when_row_renamed_in_cross_check_table(tmp_path: Path, field: str) -> None:
|
|
# M2: a detach-proof is not a value-proof — the row still exists, under another name.
|
|
copy = tmp_path / "ingest-spec.md"
|
|
copy.write_text(_row_renamed(SPEC.read_text(encoding="utf-8"), field), encoding="utf-8")
|
|
assert field in _undocumented_fields(copy.read_text(encoding="utf-8"))
|
|
|
|
|
|
def test_guard_red_when_section12_heading_renamed(tmp_path: Path) -> None:
|
|
# M3: fail-closed — no §12 heading means no anchor, and the guard must raise, not
|
|
# silently fall back to a wider (green-but-dead) slice.
|
|
mutated = SPEC.read_text(encoding="utf-8").replace(
|
|
_CROSS_CHECK_HEADING, "## 12. Field reference"
|
|
)
|
|
copy = tmp_path / "ingest-spec.md"
|
|
copy.write_text(mutated, encoding="utf-8")
|
|
with pytest.raises(ValueError):
|
|
_cross_check_table(copy.read_text(encoding="utf-8"))
|
|
|
|
|
|
def test_slice_guard_red_when_anchor_degenerates_to_whole_spec() -> None:
|
|
# M4: the anchor can degenerate. If _cross_check_table ever returned the full text,
|
|
# the row assertions would go green-but-dead again — this proves the slice guard is
|
|
# what catches that, and that it is not itself green by accident.
|
|
defects = _slice_defects(SPEC.read_text(encoding="utf-8"))
|
|
assert defects, "the slice guard accepts the whole spec as the §12 table — it is dead"
|