"""Notes-bounded baseline reduction (verbatim/passthrough path).

When the surveyor's notes do not map onto the retrieved past-report section, the
system must not reproduce that DIFFERENT property's section as-is. These tests pin
the value-level rule: generic principles survive, foreign specifics the notes never
recorded are dropped, page furniture is stripped — independent of any one example.
"""

from __future__ import annotations

from backend.core.paragraph_merge import bound_baseline_to_notes


def _out(baseline: str, notes: list[str]) -> str:
    return bound_baseline_to_notes(baseline, notes).lower()


def test_material_contradiction_dropped_when_notes_say_otherwise():
    # Notes say slate; the other property's baseline says concrete — concrete must go.
    baseline = (
        "The pitched roof covering appears to be of plain concrete tiles. "
        "Roof coverings of this type should be inspected periodically."
    )
    out = _out(baseline, ["imitation slate covering to main roof"])
    assert "concrete" not in out
    assert "inspected periodically" in out  # generic principle kept


def test_note_corroborated_material_is_kept():
    baseline = "The covering is formed of concrete interlocking tiles."
    out = _out(baseline, ["concrete tiles, some slipped"])
    assert "concrete" in out  # notes confirm the material → retained


def test_fabricated_photo_methodology_dropped():
    baseline = (
        "We were able to take a selection of photographs and videos using a camera "
        "secured to a 3-metre extendable pole. "
        "Flat roofs of this kind have a finite service life and will need renewal."
    )
    out = _out(baseline, ["dormer flat roof felt looks aged"])
    assert "photographs" not in out
    assert "extendable pole" not in out
    assert "3-metre" not in out
    assert "finite service life" in out  # general principle kept


def test_existence_finding_dropped_when_notes_silent():
    baseline = (
        "There were hip irons present to provide support for the hip tiles. "
        "Hip irons help prevent hip tiles from slipping should the mortar fail."
    )
    out = _out(baseline, ["ridge line looks straight"])
    assert "there were hip irons present" not in out


def test_no_hip_irons_existence_claim_dropped():
    baseline = "No hip irons were present to provide support to the hip tiles."
    out = _out(baseline, ["roof covering generally sound"])
    assert out.strip() == ""  # the only sentence was a foreign existence claim


def test_position_dropped_unless_in_notes():
    baseline = "The single-storey addition to the rear is of mono-pitched construction."
    assert "rear" not in _out(baseline, ["addition roof mono pitch"])
    assert "rear" in _out(baseline, ["rear addition roof, tiles cracked"])


def test_count_finding_dropped_when_notes_silent_kept_when_present():
    baseline = "One ridge tile is broken and requires replacement."
    assert "one ridge tile" not in _out(baseline, ["chimney pots sound"])
    kept = _out(baseline, ["one ridge tile broken, needs replacing"])
    assert "ridge tile" in kept


def test_page_furniture_stripped():
    baseline = "Condition Rating 2. 27"
    out = bound_baseline_to_notes(baseline, ["covering worn"])
    assert "27" not in out
    # A standalone page-number sentence is removed entirely.
    assert bound_baseline_to_notes("28.", ["covering worn"]).strip() == ""


def test_generic_regulatory_boilerplate_kept():
    baseline = (
        "Building Regulations require that any new roof meets specified thermal "
        "insulation standards for energy efficiency."
    )
    out = _out(baseline, ["nothing relevant to regs in notes"])
    assert "building regulations require" in out


def test_general_principle_kept_when_notes_silent():
    baseline = (
        "Moss and lichen can inhibit the roof's efficiency by trapping moisture, "
        "which can lead to deterioration of roof materials over time."
    )
    out = _out(baseline, ["gutters clear"])
    assert "moss and lichen can inhibit" in out


def test_whole_foreign_section_not_reproduced_verbatim():
    """A dense foreign section with sparse unrelated notes must shrink, not pass through."""
    baseline = (
        "The pitched roof covering appears to be of plain concrete tiles. "
        "The dormer is of flat roof construction finished with bituminised felt. "
        "There were hip irons present to support the hip tiles. "
        "We inspected the roof using binoculars from ground level. "
        "Roofs of this type should be maintained periodically to preserve weatherproofing."
    )
    out = bound_baseline_to_notes(baseline, ["loft hatch stiff"])
    # Foreign fabric/existence/method claims gone …
    assert "concrete" not in out.lower()
    assert "bituminised" not in out.lower()
    assert "hip irons present" not in out.lower()
    assert "binoculars" not in out.lower()
    # … only the general maintenance principle remains.
    assert "maintained periodically" in out.lower()
    assert len(out) < len(baseline)


def test_empty_when_all_specifics_lets_caller_author_from_notes():
    baseline = (
        "The covering is of concrete tiles. "
        "There were hip irons present. "
        "We took photographs using an extendable pole."
    )
    assert bound_baseline_to_notes(baseline, ["roof felt perished"]).strip() == ""