"""Recorded merges and the match key (rules v2, now v3), with the rows that went wrong. The first rehearsal of `backfill merges` on a copy of the 5 Sep dump (task A9) folded "Nina" into "Ricci Ricci", a PSG "paris" cap into the plain cap, and a Boss Bottled bottle into a Boss Bottled gift set. Each is pinned here; the fold itself was proven on the copy (44 rows forwarded, a second pass changed nothing). """ from app.models import Product from app.services.merges import _conflicts, pick_survivor from app.services.normalize import looks_like_set, match_key from app.services.taxonomy import classify def product(pid, name, brand, *, gtin=None, size=700, attributes=None): return Product(id=pid, name=name, brand=brand, gtin=gtin, size_ml=size, match_key=match_key(brand, name, size), attributes=attributes or {}) class TestMatchKeyVersionTwo: def test_every_spelling_of_a_house_shares_the_key(self): keys = { match_key("Moët & Chandon", "Moët & Chandon Impérial Brut 750ml", 750), match_key("MOET ET CHANDON", "Moet Chandon Imperial Brut 75cl", 750), match_key("Moet and Chandon", "Imperial Brut", 750), } assert keys == {"moet-chandon|imperial||750"} # "brut" is the style every Impérial has assert match_key("Don Julio Tequila", "Don Julio 1942 70cl", 700) == "don-julio|1942||700" assert match_key("Don Julio®", "Don Julio 1942® Tequila 750 ml", 750) == match_key("Don Julio", "Don Julio 1942 Tequila 750ml", 750) def test_a_scent_named_after_the_house_is_not_the_house(self): """Stripping every leading brand word emptied "Nina Ricci Ricci Ricci" and merged it with "Nina".""" assert match_key("Nina Ricci", "Nina Ricci Ricci Ricci", 50) != match_key("Nina Ricci", "Nina 5cl", 50) assert match_key("Nina Ricci", "Nina Ricci Nina 5cl", 50) == match_key("Nina Ricci", "Nina 5cl", 50) def test_only_the_brand_prefix_leaves_the_name(self): """The second "paris" in PSG's "paris" cap tells it from the plain cap.""" assert match_key("Paris Saint-Germain - PSG", 'Paris saint-germain - "paris" cap', None) != \ match_key("Paris Saint-Germain - PSG", "Paris saint germain - cap", None) def test_the_last_brand_word_alone_still_strips(self): assert match_key("GIORGIO ARMANI", "Armani Code 7.5cl", 75) == match_key("GIORGIO ARMANI", "Code 7.5cl", 75) def test_a_name_that_is_only_the_brand_is_the_house_namesake(self): """Version 2 kept the word ("gin-gin|700"); version 3 keys the namesake line as the house with an empty line, which still meets no other house.""" assert match_key("Gin", "Gin", 700) == "gin|||700" assert match_key("Gin", "Gin", 700) != match_key("Gin Mare", "Gin Mare", 700) class TestConflicts: def test_two_barcodes_never_merge(self): group = [product(1, "Dior Homme", "Dior", gtin="3348901484893", size=100), product(2, "Dior Homme", "Dior", gtin="3348901419246", size=100)] assert _conflicts(group) == ["gtin_differs"] def test_one_barcode_or_none_may_merge(self): assert _conflicts([product(1, "Hendrick's Gin 1L", "Hendrick's", gtin="5010327703053", size=1000), product(2, "Gin 1L", "Hendricks", size=1000)]) == [] def test_a_set_never_folds_into_the_bottle(self): """A Boss Bottled 10cl once merged into "Bottled Gift Set" because "gift set" is key noise.""" assert looks_like_set("Bottled Gift Set") and not looks_like_set("Boss Bottled 10cl") assert _conflicts([product(1, "Boss Bottled 10cl", "Hugo Boss", size=100), product(2, "Bottled Gift Set", "Hugo Boss", gtin="3616301623373", size=100)]) == ["set_vs_single"] def test_declared_concentrations_that_differ_keep_the_rows_apart(self): assert _conflicts([product(1, "Sauvage 100ml", "Dior", size=100, attributes={"concentration": "edt"}), product(2, "Sauvage 100ml", "Dior", size=100, attributes={"concentration": "edp"})]) == ["attribute_differs"] assert _conflicts([product(1, "Sauvage 100ml", "Dior", size=100, attributes={"concentration": "edt"}), product(2, "Sauvage 100ml", "Dior", size=100)]) == [] class TestSurvivor: def test_the_barcode_holder_survives_else_the_most_listed_else_the_oldest(self): a, b, c = product(3, "X", "B"), product(2, "X", "B", gtin="5010327703053"), product(1, "X", "B") assert pick_survivor([a, b, c], {3: 5}) is b assert pick_survivor([a, c], {3: 5, 1: 1}) is a assert pick_survivor([a, c], {}) is c class TestUncategorisedBackfillRules: """Launch rows the 3 Sep count listed as uncategorised, and where they now land.""" def test_the_brand_settles_a_name_that_says_nothing(self): assert classify("Don Papa Masskara 70cl", "Don Papa") == "Rum" assert classify("Cîroc Pineapple 1l", "Cîroc") == "Vodka" assert classify("Bombay Sapphire Sunset Special Edition 1l", "Bombay Sapphire") == "Gin" assert classify("Gran Centenario Plata 99cl", "Gran Centenario") == "Tequila & Mezcal" def test_widened_name_rules(self): assert classify("Disaronno Amaretto Originale 1l", "Disaronno") == "Liqueurs & Aperitifs" assert classify("Cream Liqueurs 1L", "Coole Swan") == "Liqueurs & Aperitifs" assert classify("Kosk. Espresso Liq.21% 0.5L", "Koskenkorva") == "Liqueurs & Aperitifs" assert classify("Lancaster Sun Beauty Velvet Milk Spf30 400ml", "Lancaster") == "Skincare" assert classify("OPI Nail Lacquer Berlin There Done That 15ml", "OPI") == "Makeup" assert classify("Kylie Brow Highlighter", "Kylie Cosmetics") == "Makeup" def test_a_house_that_makes_two_categories_is_left_to_the_name(self): assert classify("Torres 10 Imp.Gr.Res. 38% 1L", "Torres") is None assert classify("Milka & Oreo Tablet 300g", "Milka") is None class TestBrandDisplaySpelling: def test_mixed_case_beats_shouting_even_when_rarer(self): from collections import Counter from app.cli import display_spelling assert display_spelling(Counter({"SISLEY": 195, "Sisley": 3})) == "Sisley" assert display_spelling(Counter({"DIOR": 149, "Dior": 4, "Dior Parfums": 1})) == "Dior" assert display_spelling(Counter({"MOET": 2})) == "MOET" assert display_spelling(Counter({"Jameson®": 3, "Jameson": 2})) == "Jameson"