"""rerank() must rank an unembedded chunk, not crash on it.

Split from test_model_contract_shapes.py because tools.retrieval imports
psycopg: this module skips on a bare checkout and runs inside the ai-api image,
where the whole suite runs with nothing skipped:

    docker run --rm --entrypoint python yau/ai-api:local -m pytest /app/tests -q
"""

import pytest

pytest.importorskip("psycopg")

from tools.retrieval import Chunk, rerank  # noqa: E402


def chunk(cid: int, text: str, similarity: float | None) -> Chunk:
    return Chunk(
        id=cid, source_file="procedures/X.md", doc_type="procedure",
        doc_number="WRPS-X-001", revision="0", effective_date=None, page=1,
        section_title="S", equipment_id=None, chunk_text=text,
        similarity=similarity,
    )


def test_rerank_survives_a_null_similarity():
    """The bug: `0.75 * None` raised TypeError and 500'd the question.

    A chunk ingested with --no-embed has a NULL embedding, so
    `1 - (NULL <=> vec)` comes back NULL.
    """
    chunks = [chunk(1, "interlock bypass pump", None),
              chunk(2, "unrelated text", 0.4)]
    assert len(rerank(chunks, "interlock bypass procedure", top_n=2)) == 2


def test_unembedded_chunk_still_ranks_on_lexical_overlap():
    """It should degrade to the lexical half, not vanish and not win."""
    strong_words = chunk(1, "interlock bypass procedure pump", None)
    weak_words = chunk(2, "boiler feedwater chemistry", None)
    assert rerank([weak_words, strong_words],
                  "interlock bypass procedure", top_n=1) == [strong_words]


def test_embedded_chunk_outranks_an_unembedded_one_on_equal_text():
    text = "interlock bypass procedure"
    embedded = chunk(1, text, 0.9)
    unembedded = chunk(2, text, None)
    assert rerank([unembedded, embedded], text, top_n=1) == [embedded]




# ---------------------------------------------------------------------------
# Step sections are withheld from the procedural branch
# ---------------------------------------------------------------------------

from tools.retrieval import STEP_SECTION_RE  # noqa: E402


def test_step_sections_are_recognised():
    """These are the sections that ARE the instructions. find_procedure must
    not hand them to the model - an interlock exists because somebody assessed
    a hazard, and a bypass reassembled from fragments is a safety document
    nobody approved."""
    for title in ["3. Procedure", "Steps", "4. Method", "Instructions",
                  "5. Execution", "Work Instruction", "6. Restoration"]:
        assert STEP_SECTION_RE.search(title), title


def test_identifying_sections_are_not_withheld():
    """The header carries the title and authorising role; prerequisites are
    what a procedural answer quotes. Withholding either was the bug."""
    for title in ["DEMO DOCUMENT — NOT A CONTROLLED DOCUMENT",
                  "1. Purpose", "2. Prerequisites", "Scope",
                  "Hazards", "References"]:
        assert not STEP_SECTION_RE.search(title), title


def test_a_missing_section_title_is_not_treated_as_a_step():
    assert not STEP_SECTION_RE.search("")