tests/test_match.py

from __future__ import annotations

import math

import pytest

from app.config import Settings
from app.db import add_catalog_item, init_schema, upsert_vector, VecLoadError, connect
from app.match import band_for, distance_to_similarity, match_line_item
from app.schemas import LineItem, MatchBand, ReceiptExtract
from backends.base import InputType


class FakeEmbed:
    name = "fake"
    dim = 8

    def embed(self, texts: list[str], *, input_type: InputType) -> list[list[float]]:
        del input_type
        out = []
        for text in texts:
            vec = [(b / 255.0) for b in (text.encode("utf-8") + b"\x00" * 8)[:8]]
            if "milk" in text.lower() or "MILK-1" in text:
                vec[0] = 1.0
                vec[1] = 0.95
            n = math.sqrt(sum(x * x for x in vec)) or 1.0
            out.append([x / n for x in vec])
        return out

    def health(self) -> bool:
        return True


def test_bands() -> None:
    assert band_for(0.91, auto=0.88, review=0.72) is MatchBand.auto
    assert band_for(0.80, auto=0.88, review=0.72) is MatchBand.review
    assert band_for(0.10, auto=0.88, review=0.72) is MatchBand.unmatched
    assert abs(distance_to_similarity(0.2) - 0.8) < 1e-9


def test_exact_sku(settings: Settings) -> None:
    try:
        con = connect(settings.db_path)
        init_schema(con, settings)
    except VecLoadError:
        pytest.skip("sqlite-vec not loadable")
    cid = add_catalog_item(con, description="Milk 2%", sku="MILK-1", vendor="HEB")
    embed = FakeEmbed()
    from app.match import embed_catalog_row

    embed_catalog_row(con, settings, embed, cid)
    extract = ReceiptExtract(vendor="HEB", line_items=[])
    item = LineItem(description="2% milk", sku="MILK-1")
    hit = match_line_item(con, settings, embed, extract, item)
    assert hit.band is MatchBand.exact
    assert hit.catalog_id == cid
    con.close()


def test_knn_auto(settings: Settings) -> None:
    try:
        con = connect(settings.db_path)
        init_schema(con, settings)
    except VecLoadError:
        pytest.skip("sqlite-vec not loadable")
    embed = FakeEmbed()
    milk = add_catalog_item(con, description="organic milk", sku="X", vendor="HEB")
    other = add_catalog_item(con, description="bolts", sku="Y", vendor="Ace")
    from app.match import embed_catalog_row

    embed_catalog_row(con, settings, embed, milk)
    embed_catalog_row(con, settings, embed, other)
    extract = ReceiptExtract(vendor="HEB", line_items=[])
    item = LineItem(description="milk")
    hit = match_line_item(con, settings, embed, extract, item)
    assert hit.catalog_id == milk
    assert hit.band in {MatchBand.auto, MatchBand.review}
    con.close()