unpopular brand generation

tests/test_brand_store.py

@@ -0,0 +1,274 @@
+"""A brand's own shop as a catalogue source.
+
+WHY THIS SOURCE EXISTS
+----------------------
+`off_bulk.fetch_brand_corpus` is the only thing in the codebase that can
+ENUMERATE a brand's products; everything else verifies one product at a time
+and needs a name as input. For regional South Indian brands Open Food Facts is
+empty - measured 2026-09-10: Udhaiyam 0 hits, Gopuram 0, Tenali Double Horse 0,
+and both of Naga's 2 hits are a British pickle - so their catalogues were
+entirely `qwen2.5:1.5b` invention.
+
+Their storefronts publish the real thing as structured JSON. Measured live:
+Aachi 161 products (159 with a VALID EAN-13), Double Horse 99, Gopuram 122,
+Anil 57 - against 83 for the same brands in Open Food Facts.
+
+WHAT THESE TESTS GUARD
+----------------------
+This source is different in kind from every other one here: it ASSERTS a
+hundred products at once instead of verifying one. So the identity guard is the
+thing that matters most - pointing it at the wrong site imports a hundred of
+somebody else's real products under this brand's name, and they look flawless
+because they are flawless, just not theirs.
+
+The fixtures are trimmed copies of real responses.
+"""
+from __future__ import annotations
+
+import json
+
+import pytest
+
+from app.services import brand_store as bs
+from app.services.brand_registry import get_brand_store_domain, resolve_parent_brand
+
+
+# A real aachifoods.com /products.json record, trimmed.
+SHOPIFY_PRODUCT = {
+    "title": "Soan Papdi Ghee",
+    "product_type": "Snacks",
+    "vendor": "Aachifoods",
+    "images": [{"src": "https://cdn.shopify.com/s/files/1/x/ghee-soan-papdi.webp"}],
+    "variants": [
+        {"title": "200g", "price": "76.00", "sku": "8904209319340", "grams": 215},
+    ],
+}
+
+# A real gopuramproducts.com Store API record, trimmed. Note the price.
+WOO_PRODUCT = {
+    "name": "GOPURAM PREMIUM SAMBRAIN POWDER",
+    "sku": "",
+    "categories": [{"name": "Pooja Products"}],
+    "images": [{"src": "https://gopuramproducts.com/wp-content/uploads/x.jpg"}],
+    "prices": {"price": "5500", "currency_minor_unit": 2},
+}
+
+
+# ---------------------------------------------------------------------------
+# Identity - the guard that matters most
+# ---------------------------------------------------------------------------
+
+def test_a_city_council_is_not_a_food_brand():
+    """THE CASE THIS GUARD EXISTS FOR.
+
+    `retail_presence.resolve_brand_domain("Naga")` really returns
+    `cityofnagacebu.gov.ph` - the website of Naga City in the Philippines -
+    because `_pick_brand_domain` accepted a domain whose stem merely CONTAINS a
+    brand token. Here that would not cost one bad row; it would file a
+    municipal website's contents as a Tamil Nadu brand's product list.
+    """
+    assert bs._store_identity_ok("Naga", "cityofnagacebu.gov.ph", []) is False
+
+
+def test_a_brand_domain_is_the_brand_plus_a_qualifier():
+    """Prefix, not substring - that is the whole distinction.
+    "gopuramproducts".startswith("gopuram") but "cityofnagacebu" does not
+    start with "naga", though it contains it."""
+    for brand, domain in [("Gopuram", "gopuramproducts.com"),
+                          ("Udhaiyam", "udhaiyamdhall.com"),
+                          ("Double Horse", "doublehorse.in"),
+                          ("Naga", "nagafoods.in")]:
+        assert bs._store_identity_ok(brand, domain, []) is True, domain
+
+
+def test_the_vendor_field_can_vouch_for_a_domain_that_does_not():
+    """Shopify states the seller. A brand trading on a domain that does not
+    contain its name is still identifiable."""
+    assert bs._store_identity_ok("Aachi", "someshop.com", ["Aachifoods"]) is True
+    assert bs._store_identity_ok("Aachi", "someshop.com", ["Britannia"]) is False
+
+
+def test_a_rejected_catalogue_is_rejected_whole(monkeypatch):
+    """Half a foreign catalogue is not better than all of it."""
+    monkeypatch.setattr(bs, "fetch_shopify",
+                        lambda d: [dict(SHOPIFY_PRODUCT, vendor="Someone Else")])
+    monkeypatch.setattr(bs, "read_cached", lambda b: None)
+    monkeypatch.setattr(bs, "_write_cache", lambda p, d: None)
+
+    assert bs.fetch_store_catalogue("Naga", "cityofnagacebu.gov.ph", live=True) is None
+
+
+def test_a_curated_domain_skips_the_heuristic(monkeypatch):
+    """A person already checked it. The heuristic cannot recognise every shape
+    a real brand domain takes - "theanilgroup.com" is Anil's and neither starts
+    with nor contains "anil" as a whole token - so applying it to a hand-verified
+    mapping would reject good catalogues for looking unusual."""
+    assert bs._store_identity_ok("Anil", "shop.theanilgroup.com", []) is False
+
+    monkeypatch.setattr(bs, "fetch_shopify", lambda d: [SHOPIFY_PRODUCT])
+    monkeypatch.setattr(bs, "read_cached", lambda b: None)
+    monkeypatch.setattr(bs, "_write_cache", lambda p, d: None)
+
+    rows = bs.fetch_store_catalogue("Anil", "shop.theanilgroup.com",
+                                    live=True, trusted_domain=True)
+    assert rows and len(rows) == 1
+
+
+# ---------------------------------------------------------------------------
+# Field mapping
+# ---------------------------------------------------------------------------
+
+def test_a_shopify_variant_carries_the_pack_size_and_a_real_barcode():
+    rows = bs._shopify_candidates(SHOPIFY_PRODUCT)
+
+    assert len(rows) == 1
+    row = rows[0]
+    assert row["title"] == "Soan Papdi Ghee"
+    assert row["category"] == "Snacks"
+    assert row["size"] == "200g"
+    assert row["price"] == 76.00
+    assert row["image_url"].endswith("ghee-soan-papdi.webp")
+    # A genuine EAN-13 on the Indian 890 GS1 prefix.
+    assert row["barcode"] == "8904209319340"
+
+
+def test_a_shop_internal_sku_is_not_treated_as_a_barcode():
+    """Shops put anything in `sku`. `validate_barcode` decides, so an internal
+    numbering scheme never reaches the barcode column - Gopuram leaves it blank
+    and others use their own codes."""
+    for junk in ("", "GOP-001", "12345"):
+        product = json.loads(json.dumps(SHOPIFY_PRODUCT))
+        product["variants"][0]["sku"] = junk
+        assert bs._shopify_candidates(product)[0]["barcode"] is None, junk
+
+
+def test_default_title_is_not_a_pack_size():
+    """Shopify's placeholder for a product with no variant axis."""
+    product = json.loads(json.dumps(SHOPIFY_PRODUCT))
+    product["variants"][0]["title"] = "Default Title"
+    product["variants"][0].pop("sku")
+
+    assert bs._shopify_candidates(product)[0]["size"] is None
+
+
+def test_woocommerce_prices_are_minor_units():
+    """gopuramproducts.com returns price "5500" with currency_minor_unit 2,
+    which is Rs55.00 and not Rs5,500. Read straight, every price on the site is
+    a hundred times too large - and `validate_price_range` would then reject
+    every row, so the bug would present as "this brand is all wrong" rather
+    than as a units error."""
+    row = bs._woo_candidates(WOO_PRODUCT)[0]
+
+    assert row["price"] == 55.00
+    assert row["category"] == "Pooja Products"
+    assert row["title"] == "GOPURAM PREMIUM SAMBRAIN POWDER"
+
+
+def test_a_utf8_bom_does_not_break_parsing():
+    """WooCommerce really serves one; `resp.json()` raises on it."""
+    class Resp:
+        status_code = 200
+        headers = {"content-type": "application/json; charset=UTF-8"}
+        content = b"\xef\xbb\xbf" + json.dumps([WOO_PRODUCT]).encode()
+        url = "https://example.com"
+
+    parsed = bs._json_or_none(Resp())
+
+    assert isinstance(parsed, list) and parsed[0]["name"].startswith("GOPURAM")
+
+
+# ---------------------------------------------------------------------------
+# The sitemap tier states only what the site states
+# ---------------------------------------------------------------------------
+
+def test_the_sitemap_tier_never_invents_a_size_or_price():
+    """udhaiyamdhall.com is a brochure site: an <h1> and a photograph, no
+    price, no pack size, no variants. Those stay None. A catalogue source that
+    guesses them is the failure this whole area exists to fix."""
+    import inspect
+    source = inspect.getsource(bs.fetch_sitemap)
+
+    assert '"size": None' in source
+    assert '"price": None' in source
+    assert '"barcode": None' in source
+
+
+# ---------------------------------------------------------------------------
+# The registry
+# ---------------------------------------------------------------------------
+
+def test_double_horse_and_tenali_double_horse_stay_apart():
+    """TWO COMPANIES, ONE NAME. Double Horse is Manjilas of Kerala; Tenali
+    Double Horse is an Andhra rice brand.
+
+    `_contains_word` matches whole words in BOTH directions, and "double horse"
+    is a whole word sequence inside "tenali double horse". Registering only one
+    of them moved the collision instead of fixing it - measured: with only the
+    longer key present, resolve_parent_brand("Double Horse") returned
+    "tenali double horse". Each direct entry protects the OTHER brand.
+    """
+    assert resolve_parent_brand("Tenali Double Horse") == "tenali double horse"
+    assert resolve_parent_brand("Double Horse") == "double horse"
+    assert resolve_parent_brand("Tenali Double Horse") != resolve_parent_brand("Double Horse")
+
+
+def test_spelling_variants_reach_the_same_brand():
+    assert resolve_parent_brand("Udhayam") == "udhaiyam"
+    assert resolve_parent_brand("DoubleHorse") == "double horse"
+
+
+def test_a_brand_with_no_verified_store_has_none():
+    """Absent is the safe answer. A guessed domain is not - Naga's search
+    result is a Philippine city council."""
+    assert get_brand_store_domain("Naga") is None
+    assert get_brand_store_domain("Tenali Double Horse") is None
+    assert get_brand_store_domain("Gopuram") == "gopuramproducts.com"
+
+
+def test_existing_brands_are_unaffected():
+    for brand, parent in [("Cadbury Dairy Milk", "cadbury"),
+                          ("Haldiram", "haldirams"),
+                          ("Sunfeast Yippee", "sunfeast")]:
+        assert resolve_parent_brand(brand) == parent
+
+
+# ---------------------------------------------------------------------------
+# Discovery integration
+# ---------------------------------------------------------------------------
+
+def test_a_store_row_outranks_open_food_facts():
+    """A manufacturer's own catalogue is first-party; Open Food Facts is
+    crowd-sourced and, for these brands, absent or wrong."""
+    from app.services import brand_discovery as bd
+
+    assert bd._score(["store"], None) > bd._score(["off"], "openfacts")
+    assert bd._score(["store", "off"], "openfacts") == 1.0
+
+
+def test_a_store_row_survives_the_evidence_gate():
+    """WITHOUT THIS THE WHOLE SOURCE IS INVISIBLE. `require_evidence` drops any
+    row with no `off` source and no evidence, and a store row has neither - it
+    would score 0.25 and never reach the preview."""
+    from app.services import brand_discovery as bd
+    import inspect
+
+    gate = inspect.getsource(bd.discover_brand_products)
+    assert '"store" not in product.sources' in gate
+
+
+def test_discovery_reads_the_store_cache_and_never_fetches(monkeypatch):
+    """Same split as the retail check: the backfill script fetches, discovery
+    reads. A preview must not block on somebody's storefront."""
+    from app.services import brand_discovery as bd
+
+    def explode(*_a, **_kw):
+        raise AssertionError("discovery must not fetch a storefront")
+
+    monkeypatch.setattr(bs, "fetch_shopify", explode)
+    monkeypatch.setattr(bs, "fetch_woocommerce", explode)
+    monkeypatch.setattr(bs, "fetch_sitemap", explode)
+    monkeypatch.setattr(bs, "read_cached", lambda b: [dict(SHOPIFY_PRODUCT, source="store")])
+
+    rows = bd._from_brand_store("Aachi")
+
+    assert rows and rows[0]["source"] == "store"