munch-ease-backend/tests/test_recipes_offline_expected.py

import asyncio
import json
from pathlib import Path
from types import SimpleNamespace

import pytest

from db import connect, create
from recipes import parse_recipe_from_html


SNAP_DIR = Path("tests/sample_files/recipes")


def _iter_expected_files():
    return sorted(SNAP_DIR.glob("*.recipe.json"))


def _load_expected(path: Path):
    # New-style file contains the model JSON directly or null
    return json.loads(path.read_text())


def _expected_subset(expected: dict):
    # expected is serialized with by_alias=True, so image_urls is imageUrls
    name = expected.get("name")
    link = expected.get("link")
    serves = expected.get("serves")
    image_urls = expected.get("imageUrls") or []
    ingredients = expected.get("ingredients") or []
    ing_lines = [ing.get("line") for ing in ingredients]
    return {
        "name": name,
        "link": link,
        "serves": serves,
        "image_urls": image_urls,
        "ingredient_lines": ing_lines,
    }


def _actual_subset(model):
    # Stable, assertion-friendly projection
    return {
        "name": model.name,
        "link": model.link,
        "serves": model.serves,
        "image_urls": list(model.image_urls or []),
        "ingredient_lines": [ing.line for ing in (model.ingredients or [])],
    }


@pytest.mark.parametrize("exp_path", _iter_expected_files())
def test_offline_parse_matches_expected(exp_path: Path):
    async def _run():
        # exp_path is <slug>.recipe.json; Path.stem removes only the last suffix (".json"),
        # leaving ".recipe" in the stem. Strip the trailing ".recipe" to get the HTML slug.
        slug = exp_path.stem
        if slug.endswith(".recipe"):
            slug = slug[: -len(".recipe")]
        expected = _load_expected(exp_path)

        html_path = SNAP_DIR / f"{slug}.html"
        assert html_path.exists(), f"snapshot HTML missing for {slug}"
        html = html_path.read_text()

        # Discover base URL from meta if available
        meta_path = SNAP_DIR / f"{slug}.meta.json"
        base_url = slug
        if meta_path.exists():
            try:
                meta = json.loads(meta_path.read_text())
                base_url = meta.get("final_url") or meta.get("url") or slug
            except Exception:
                pass

        created_by = SimpleNamespace(id=1, display_name="Snapshot Generator")
        conn = await connect()
        await create(conn)
        try:
            model = await parse_recipe_from_html(conn, created_by, base_url, html)
        finally:
            await conn.close()

        if expected is None:
            assert model is None, f"expected no parse for {slug}, but got a model"
            return

        assert model is not None, f"expected a model for {slug}, got None"

        exp_sub = _expected_subset(expected)
        act_sub = _actual_subset(model)

        # Compare key fields; ingredients compare by line only for stability
        assert act_sub == exp_sub

    asyncio.run(_run())