backend apis updation

tests/test_brand_registry.py

@@ -0,0 +1,113 @@
+"""
+Guard rails for brand -> parent-brand resolution.
+
+`resolve_parent_brand` decides which `brand_*` table, which S3 folder, and
+which seed catalog file a product lands in. An edit to BRAND_ALIASES that
+quietly changes an existing brand's answer would relocate live rows, so the
+first test pins the answer for every realistic input: all alias keys, all
+parent values, and the `brand` field of every bundled seed catalog.
+
+No database, no network - the registry is a plain dict.
+"""
+from __future__ import annotations
+
+import json
+from pathlib import Path
+
+import pytest
+
+from app.services.brand_registry import BRAND_ALIASES, resolve_parent_brand
+from app.services.vector_store import _sanitize_name
+
+SEED_DIR = Path(__file__).resolve().parents[1] / "data" / "seed_catalogs"
+
+
+def _seed_brand_fields() -> list[str]:
+    """The `brand` value of every seed catalog (skipping non-catalog files)."""
+    brands = []
+    for path in sorted(SEED_DIR.glob("*.json")):
+        try:
+            data = json.loads(path.read_text(encoding="utf-8-sig"))
+        except Exception:
+            continue
+        brand = data.get("brand")
+        if brand:
+            brands.append(brand)
+    return brands
+
+
+# Every seed catalog and the table it must continue to feed. `tata` mapping to
+# hindustan_unilever is not a typo: alias "hul tata tea" claims it, which is
+# where all 121 of that file's products already live.
+EXPECTED_SEED_TABLES = {
+    "amul": "brand_amul",
+    "cadbury": "brand_cadbury",
+    "cavinkare": "brand_cavinkare",
+    "coca-cola": "brand_coca_cola",
+    "colgate-palmolive": "brand_colgate_palmolive",
+    "dabur": "brand_dabur",
+    "godrej": "brand_godrej",
+    "grb": "brand_grb",
+    "hindustan unilever": "brand_hindustan_unilever",
+    "lion dates": "brand_lion_dates",
+    "Manna": "brand_manna",
+    "milky mist": "brand_milky_mist",
+    "naga": "brand_naga",
+    "Nestle": "brand_nestle",
+    "p&g": "brand_p_g",
+    "pepsico": "brand_pepsico",
+    "tata": "brand_hindustan_unilever",
+}
+
+
+@pytest.mark.parametrize("alias,parent", sorted(BRAND_ALIASES.items()))
+def test_every_alias_resolves_to_its_declared_parent(alias: str, parent: str) -> None:
+    assert resolve_parent_brand(alias) == parent
+
+
+# Every parent brand resolves to itself except "sunfeast", which the alias
+# "itc sunfeast" claims for ITC. That collapse predates the word-boundary
+# change and is pinned here so it stays a deliberate quirk rather than drift.
+_NON_SELF_PARENTS = {"sunfeast": "itc"}
+
+
+@pytest.mark.parametrize("parent", sorted(set(BRAND_ALIASES.values())))
+def test_parent_brands_resolve_to_their_pinned_answer(parent: str) -> None:
+    assert resolve_parent_brand(parent) == _NON_SELF_PARENTS.get(parent, parent)
+
+
+def test_seed_catalogs_keep_their_current_tables() -> None:
+    """The load-bearing assertion: no seed file changes destination table."""
+    actual = {
+        brand: f"brand_{_sanitize_name(resolve_parent_brand(brand))}"
+        for brand in _seed_brand_fields()
+    }
+    assert actual == EXPECTED_SEED_TABLES
+
+
+def test_known_sub_brands_still_route_to_their_family() -> None:
+    """Word-boundary matching must not break legitimate sub-brand routing."""
+    assert resolve_parent_brand("Dove") == "hindustan unilever"
+    assert resolve_parent_brand("Silk") == "cadbury"
+    assert resolve_parent_brand("Bingo") == "itc"
+    assert resolve_parent_brand("Sunfeast") == "itc"
+
+
+def test_new_brand_is_not_swallowed_by_a_fragment_match() -> None:
+    """
+    "sun" appears inside "hul sunsilk" but is not a word there, so a new brand
+    called Sun must stay Sun. Under the old substring fallback it silently
+    became "hindustan unilever" - the bug this guards.
+    """
+    assert resolve_parent_brand("Sun") == "Sun"
+    assert resolve_parent_brand("Bing") == "Bing"
+
+
+def test_unknown_brand_passes_through_unchanged() -> None:
+    assert resolve_parent_brand("Totally Made Up Brand") == "Totally Made Up Brand"
+
+
+def test_new_brand_round_trips_to_its_own_table() -> None:
+    """ITC is the worked example: it must own brand_itc, not merge elsewhere."""
+    assert resolve_parent_brand("ITC") == "itc"
+    assert _sanitize_name(resolve_parent_brand("ITC")) == "itc"