backend/tests/test_rag_chunk_bounds.py

from __future__ import annotations

from backend.config import settings
from backend.core.ingest import _chunk_reference_text, _chunk_text
from backend.core.reference_chunker import build_reference_chunks


def test_chunk_text_respects_min_and_max(monkeypatch):
    monkeypatch.setattr(settings, "paragraph_min_chars", 80)
    monkeypatch.setattr(settings, "paragraph_max_chars", 120)
    monkeypatch.setattr(settings, "chunk_overlap", 20)

    short_a = "a" * 40
    short_b = "b" * 40
    long_para = "x" * 250
    text = f"{short_a}\n\n{short_b}\n\n{long_para}"

    chunks = _chunk_text(text)
    assert chunks
    assert all(len(c) <= 120 for c in chunks)
    assert any(len(c) >= 80 for c in chunks)


def test_reference_chunk_keeps_whole_section_under_max(monkeypatch):
    monkeypatch.setattr(settings, "reference_paragraph_max_chars", 8000)
    monkeypatch.setattr(settings, "reference_chunk_overlap", 1500)
    body = "The roof comprises slate tiles laid to pitched timber rafters. " * 40
    text = f"D2 Roof coverings\n\n{body}"
    chunks = build_reference_chunks(text, source_filename="past.docx")
    assert len(chunks) == 1
    assert len(chunks[0].text) <= 8000
    assert "slate tiles" in chunks[0].text


def test_reference_chunk_splits_when_exceeding_max(monkeypatch):
    monkeypatch.setattr(settings, "reference_paragraph_max_chars", 8000)
    monkeypatch.setattr(settings, "reference_chunk_overlap", 1500)
    body = "x" * 9000
    text = f"D2 Roof coverings\n\n{body}"
    chunks = build_reference_chunks(text, source_filename="past.docx")
    assert len(chunks) >= 2
    assert all(len(c.text) <= 8000 for c in chunks)


def test_chunk_reference_text_uses_long_form_limits(monkeypatch):
    monkeypatch.setattr(settings, "paragraph_min_chars", 80)
    monkeypatch.setattr(settings, "reference_paragraph_max_chars", 8000)
    monkeypatch.setattr(settings, "reference_chunk_overlap", 1500)
    para = "y" * 500
    text = f"{para}\n\n{para}"
    chunks = _chunk_reference_text(text)
    assert chunks
    assert all(len(c) <= 8000 for c in chunks)