test_normalization.py

from drawing_intelligence.normalization import (
    enrich_characteristic,
    infer_category,
    records_from_ocr_text,
    validate_characteristics,
)
from drawing_intelligence.schema import Category, Characteristic, ReviewStatus


def test_bilateral_dimension_is_normalized_without_standard_lookup() -> None:
    item = enrich_characteristic(
        Characteristic(
            balloon_id="12",
            requirement="4X ⌀12.00 ±0.05 THRU",
            source_text="4X ⌀12.00 ±0.05 THRU",
        )
    )
    assert item.category == Category.DIAMETER
    assert item.nominal == "12.00"
    assert item.upper_tolerance == "0.05"
    assert item.lower_tolerance == "-0.05"
    assert item.upper_limit == "12.05"
    assert item.lower_limit == "11.95"
    assert item.quantity == 4


def test_thread_and_surface_finish_categories() -> None:
    assert infer_category("M14×1.5-6H") == Category.THREAD
    assert infer_category("Ra 1.6") == Category.SURFACE_FINISH
    assert infer_category("⌖ ⌀0.10 M | A | B | C") == Category.GD_AND_T


def test_ocr_fallback_does_not_fabricate_balloon_mapping() -> None:
    rows = records_from_ocr_text("Ø12 H7\n27: M14×1.5-6H", page=1)
    assert rows[0].balloon_id.startswith("UNASSIGNED")
    assert rows[0].status == ReviewStatus.NEEDS_MAPPING
    assert rows[1].balloon_id == "27"


def test_ai_verified_status_is_reset_for_human_review() -> None:
    item = Characteristic(
        balloon_id="9",
        requirement="R5 ±0.2",
        source_text="R5 ±0.2",
        status=ReviewStatus.VERIFIED,
    )
    validated = validate_characteristics([item])
    assert validated[0].status == ReviewStatus.NEEDS_REVIEW