backend/tests/test_baseline_assembly.py

"""REFERENCE baseline assembly and in-place mapping orchestration."""

from __future__ import annotations

from backend.core.paragraph_retriever import assemble_reference_baseline
from backend.core.rag_store import SearchHit, TIER_REFERENCE


def _hit(
    text: str,
    *,
    score: float,
    section_id: str = "D2",
    paragraph_index: int = 1,
    source: str = "past.docx",
) -> SearchHit:
    return SearchHit(
        text=text,
        score=score,
        tier=TIER_REFERENCE,
        is_scrubbed=False,
        section_id=section_id,
        paragraph_index=paragraph_index,
        source_filename=source,
        doc_id=f"reference:{source}",
    )


def test_assemble_reference_baseline_combines_section_paragraphs():
    short = _hit("The roof comprises slate tiles.", score=0.91, paragraph_index=1)
    long = _hit(
        "Inspection was limited to ground level using binoculars. "
        "The underfelt and battens were not visible.",
        score=0.90,
        paragraph_index=2,
    )
    baseline, hits = assemble_reference_baseline(
        [short, long],
        paragraph_section_id="D2",
    )
    assert "slate tiles" in baseline
    assert "binoculars" in baseline
    assert len(baseline.split()) > len(short.text.split())
    assert len(hits) == 2


def test_assemble_reference_baseline_prefers_longer_same_score():
    brief = _hit("Slate roof.", score=0.95, paragraph_index=1)
    verbose = _hit(
        "The roof covering comprised natural slate tiles laid to pitched timber rafters "
        "with underfelt and counter battens. Inspection was from ground level only.",
        score=0.95,
        paragraph_index=1,
        source="verbose.docx",
    )
    baseline, hits = assemble_reference_baseline([brief, verbose], paragraph_section_id="D2")
    assert "timber rafters" in baseline
    assert hits[0].source_filename == "verbose.docx"