"""The frozen row-2 denominator stays reproducible (`docs/catalog-census.py`).

Session 1 froze the count of OWASP LLM numbers used with a 2025 or a wrong meaning
at v1.4.1 (`6702d7a`): 95 sites in 23 files, listing hash in `docs/PLAN.md` § T.2.
The census reads that revision from git objects, so the number may not move when
the working tree does. It DOES move if someone edits a classification rule, and
that is the point: a re-judged site must re-freeze the listing deliberately, not
drift the denominator under a later measurement.

Skipped outside a git checkout that holds the revision (an sdist has no history).
"""
from __future__ import annotations

import hashlib
import importlib.util
import subprocess
import sys
from pathlib import Path

import pytest

_ROOT = Path(__file__).resolve().parent.parent
_REV = "6702d7a"
_FROZEN = _ROOT / "docs" / "threat-2026" / "census-6702d7a.tsv"
_FROZEN_SHA256 = "da1aa0fab63636e7e56c520f4856504c156fe4aa223f30b71edd7729fe1da16f"


def _load():
    spec = importlib.util.spec_from_file_location("_docs_catalog_census",
                                                  _ROOT / "docs" / "catalog-census.py")
    module = importlib.util.module_from_spec(spec)
    sys.modules[spec.name] = module
    spec.loader.exec_module(module)
    return module


def _have_rev() -> bool:
    try:
        subprocess.run(["git", "-C", str(_ROOT), "cat-file", "-e", f"{_REV}^{{commit}}"],
                       check=True, capture_output=True)
        return True
    except (OSError, subprocess.CalledProcessError):
        return False


def test_frozen_listing_matches_its_pinned_hash():
    assert hashlib.sha256(_FROZEN.read_bytes()).hexdigest() == _FROZEN_SHA256


@pytest.mark.skipif(not _have_rev(), reason=f"no git history holding {_REV}")
def test_census_at_the_frozen_revision_reproduces_the_listing(tmp_path):
    census = _load()
    out = tmp_path / "census.tsv"
    assert census.main(["--rev", _REV, "--tsv", str(out)]) == 0
    assert out.read_bytes() == _FROZEN.read_bytes()


def test_frozen_denominator_is_95_in_23_files():
    rows = [line.split("\t") for line in _FROZEN.read_text(encoding="utf-8").splitlines()[1:]]
    counted = [r for r in rows if r[0] in ("old", "wrong")]
    assert len(counted) == 95
    assert len({r[1] for r in counted}) == 23
    assert sum(r[0] == "wrong" for r in counted) == 13