backend/tests/test_medium_expand.py

"""Medium AI: full past-report paragraph with proofread expansion."""

from __future__ import annotations

from docx import Document

from backend.config import settings
from backend.core import ingest, section_mapper
from backend.core.medium_expand import expand_medium_ai
from backend.core.minimum_weave import map_minimum_reference
from backend.core.rag_store import TIER_REFERENCE, get_rag_store
from backend.models.schema import TemplateSchema


def _write_past_roof_report(path):
    doc = Document()
    doc.add_paragraph("Section E2 - Roof Coverings")
    doc.add_paragraph(
        "The roof covering comprised slate tiles laid to pitched timber rafters. "
        "Inspection was limited to ground level using binoculars."
    )
    doc.save(str(path))


def test_medium_keeps_full_reference_paragraph():
    reference = (
        "The roof covering comprised slate tiles laid to pitched timber rafters. "
        "Inspection was limited to ground level using binoculars."
    )
    out = expand_medium_ai(["slate tile slipped south slope"], reference)
    assert "•" not in out
    assert out.count(".") >= 2
    assert "slate" in out.lower()
    assert "south slope" in out.lower()
    assert "binoculars" in out.lower()
    assert len(out.split()) >= len(reference.split()) * 0.9


def test_medium_at_least_as_long_as_minimum():
    reference = (
        "The roof covering comprised slate tiles laid to pitched timber rafters. "
        "Inspection was limited to ground level using binoculars."
    )
    obs = ["slate tile slipped south slope"]
    schema = TemplateSchema()
    minimum = map_minimum_reference(reference, obs, schema)
    medium = expand_medium_ai(obs, reference, schema=schema)
    assert len(medium.split()) >= len(minimum.split()) * 0.9
    assert medium.count(".") >= minimum.count(".")


def test_medium_mode_full_paragraph_in_report(tmp_path, monkeypatch):
    monkeypatch.setattr(settings, "data_dir", str(tmp_path / "data"))

    tpl = tmp_path / "template.docx"
    doc = Document()
    doc.add_paragraph("D2 Roof coverings")
    doc.save(str(tpl))

    master = tmp_path / "master.docx"
    doc2 = Document()
    doc2.add_paragraph("Section E2 - Roof Coverings")
    doc2.add_paragraph("The roof covering comprises slate laid to pitched timber rafters.")
    doc2.save(str(master))

    past = tmp_path / "past.docx"
    _write_past_roof_report(past)

    tenant = "t-med-ref"
    ingest.ingest_report_template(tenant, tpl)
    ingest.ingest_standard_paragraphs(tenant, master)
    ingest.ingest_reference(tenant, past)
    assert get_rag_store().count(tenant, TIER_REFERENCE) >= 1

    result = section_mapper.generate_report(
        tenant,
        "slate tile slipped south slope",
        interference_level="medium",
    )
    d3 = next(s for s in result.sections if s.section_id == "D2")
    assert d3.status == "OK"
    assert "•" not in d3.text
    assert d3.text.count(".") >= 2
    assert "binoculars" in d3.text.lower()
    assert "south slope" in d3.text.lower()