From 6f2aa1bb66d6e53b661a2a9aece8c3d7b98117c5 Mon Sep 17 00:00:00 2001
From: Mateusz Konieczny <matkoniecz@gmail.com>
Date: Thu, 3 Apr 2025 15:33:56 +0200
Subject: [PATCH] improve type matching
---
0_config.py | 170 ++++++++++++++++++++++++++++++++++++++++-
test_matching_logic.py | 63 ++++++++++++++-
2 files changed, 229 insertions(+), 4 deletions(-)
diff --git a/0_config.py b/0_config.py
index acaefc3..88d59cf 100644
--- a/0_config.py
+++ b/0_config.py
@@ -982,6 +982,9 @@ def the_same_feature_type(tags_a, tags_b):
return True
bare_trade = None
other_one = None
+ if tags_a.get("amenity") == "fast_food" and tags_b.get("amenity") == "fast_food":
+ # one of them may have say shop=bakery
+ return True
if tags_a.get("shop") == "trade" and "trade" not in tags_a:
bare_trade = tags_a
other_one = tags_b
@@ -989,7 +992,7 @@ def the_same_feature_type(tags_a, tags_b):
bare_trade = tags_b
other_one = tags_a
if bare_trade != None:
- if other_one.get("shop") in ["trade", "tractor", "agrarian"] or other_one.get("craft") in ["agricultural_engines"]:
+ if other_one.get("shop") in ["trade", "tractor", "agrarian", "tiles", "flooring"] or other_one.get("craft") in ["agricultural_engines"]:
# we want to match say
# shop=trade trade=tiles
# with
@@ -1301,7 +1304,30 @@ def clear_type_conflicts():
'amenity=animal_boarding vs shop=money_lender',
'amenity=animal_boarding vs shop=storage_rental',
'amenity=animal_boarding vs shop=tyres',
+ 'shop=money_lender vs shop=motorcycle_repair',
+ 'shop=money_lender vs shop=paint',
+ 'shop=money_lender vs shop=perfumery',
+ 'shop=money_lender vs shop=shoes',
+ 'shop=money_lender vs shop=sports',
+ 'shop=money_lender vs shop=storage_rental',
+ 'shop=money_lender vs shop=tyres',
+ 'shop=mobile_phone_accessories vs shop=window_blind',
+ 'shop=mobile_phone_accessories vs shop=windows',
+ 'shop=mobile_phone_accessories vs sweet bakery',
+ 'shop=mobile_phone_accessories vs toy_shop',
'amenity=bank vs shop=outdoor',
+ 'shop=health_food vs toy_shop',
+ 'shop=health_food vs shop=tobacco',
+ 'shop=health_food vs shop=travel_agency',
+ 'shop=health_food vs shop=weapons',
+ 'shop=motorcycle vs shop=stationery',
+ 'shop=motorcycle vs shop=storage_rental',
+ 'shop=motorcycle vs shop=travel_agency',
+ 'shop=motorcycle vs shop=window_blind',
+ 'shop=kitchen vs shop=travel_agency',
+ 'shop=kitchen vs shop=tyres',
+ 'shop=kitchen vs shop=travel_agency',
+ 'shop=kitchen vs shop=tyres',
'amenity=atm vs amenity=bank',
'amenity=atm vs convenience/supermarket',
'amenity=bank vs amenity=biergarten',
@@ -3776,7 +3802,52 @@ def clear_type_conflicts():
'amenity=post_office vs shop=bakery',
'amenity=post_office vs craft=cleaning',
'amenity=post_office vs craft=electronics_repair',
+ 'shop=art vs shop=travel_agency',
+ 'shop=art vs shop=travel_agency',
+ 'amenity=post_office vs craft=handicraft',
+ 'amenity=post_office vs craft=handicraft',
+ 'craft=handicraft vs parcel_locker',
+ 'craft=handicraft vs parcel_locker',
+ 'craft=handicraft vs parcel_locker',
+ 'craft=handicraft vs parcel_locker',
+ 'craft=handicraft vs parcel_locker',
+ 'craft=handicraft vs parcel_locker',
+ 'craft=handicraft vs parcel_locker',
+ 'craft=handicraft vs parcel_locker',
+ 'craft=handicraft vs parcel_locker',
+ 'craft=handicraft vs parcel_locker',
+ 'shop=sports vs shop=tailor',
+ 'shop=sports vs shop=tattoo',
+ 'shop=sports vs shop=ticket',
+ 'shop=sports vs shop=tiles',
+ 'shop=sports vs shop=trade',
+ 'shop=sports vs shop=travel_agency',
+ 'shop=sports vs shop=tyres',
+ 'shop=sports vs shop=video_games',
+ 'shop=paint vs shop=pawnbroker',
+ 'shop=paint vs shop=pet',
+ 'shop=paint vs shop=pet_grooming',
+ 'shop=paint vs shop=photo',
+ 'shop=paint vs shop=pottery',
+ 'shop=paint vs shop=psychic',
+ 'shop=paint vs shop=scuba_diving',
+ 'shop=paint vs shop=sewing',
+ 'shop=paint vs shop=swimming_pool',
+ 'shop=paint vs shop=trophy',
+ 'shop=paint vs shop=truck_repair',
+ 'shop=paint vs shop=tyres',
+ 'shop=paint vs shop=window_blind',
+ 'shop=party vs shop=swimming_pool',
'amenity=post_office vs craft=glaziery',
+ 'shop=new_age vs shop=storage_rental',
+ 'shop=newsagent vs shop=optician',
+ 'shop=newsagent vs shop=pawnbroker',
+ 'shop=newsagent vs shop=perfumery',
+ 'shop=newsagent vs shop=pet',
+ 'shop=newsagent vs shop=pet_grooming',
+ 'shop=newsagent vs shop=photo',
+ 'shop=newsagent vs shop=printing',
+ 'shop=newsagent vs shop=radiotechnics',
'amenity=post_office vs craft=hvac',
'amenity=post_office vs craft=metal_construction',
'amenity=post_office vs craft=pottery',
@@ -6654,11 +6725,100 @@ def clear_type_conflicts():
'shop=ticket vs sweet bakery',
'shop=video vs sweet bakery',
'shop=weapons vs toy_shop',
+ 'shop=nuts vs shop=paint',
+ 'shop=optician vs shop=pet_grooming',
+ 'shop=optician vs shop=radiotechnics',
+ 'shop=optician vs shop=seafood',
+ 'shop=optician vs shop=sewing',
+ 'shop=optician vs shop=tattoo',
+ 'shop=optician vs shop=tea',
+ 'shop=optician vs shop=truck',
+ 'shop=mobile_phone_accessories vs shop=tailor',
+ 'shop=mobile_phone_accessories vs shop=tea',
+ 'shop=mobile_phone_accessories vs shop=ticket',
+ 'shop=mobile_phone_accessories vs shop=tiles',
+ 'shop=mobile_phone_accessories vs shop=tobacco',
+ 'shop=mobile_phone_accessories vs shop=trade',
+ 'shop=mobile_phone_accessories vs shop=travel_agency',
+ 'shop=mobile_phone_accessories vs shop=video',
+ 'shop=mobile_phone_accessories vs shop=weapons',
+ 'shop=medical_supply vs toy_shop',
+ 'shop=mall vs shop=mobile_phone_accessories',
+ 'shop=mall vs shop=newsagent',
+ 'shop=mall vs shop=nutrition_supplements',
+ 'shop=mall vs shop=optician',
+ 'shop=mall vs shop=outdoor',
+ 'shop=mall vs shop=paint',
+ 'shop=mall vs shop=perfumery',
+ 'shop=mall vs shop=pet',
+ 'shop=mall vs shop=shoes',
+ 'shop=mall vs shop=sports',
+ 'shop=mall vs shop=stationery',
+ 'shop=mall vs shop=storage_rental',
+ 'shop=mall vs shop=travel_agency',
+ 'shop=mall vs shop=tyres',
+ 'shop=maps vs shop=travel_agency',
+ 'shop=massage vs shop=mobile_phone_accessories',
+ 'shop=massage vs shop=optician',
+ 'shop=massage vs shop=paint',
+ 'shop=massage vs shop=shoes',
+ 'shop=massage vs shop=sports',
+ 'shop=massage vs shop=stationery',
+ 'shop=massage vs shop=travel_agency',
+ 'shop=massage vs shop=tyres',
+ 'shop=lighting vs shop=mobile_phone_accessories',
+ 'shop=lighting vs shop=newsagent',
+ 'shop=lighting vs shop=optician',
+ 'shop=lighting vs shop=perfumery',
+ 'shop=lighting vs shop=shoes',
+ 'shop=lighting vs shop=sports',
+ 'shop=lighting vs shop=travel_agency',
+ 'shop=lighting vs shop=tyres',
+ 'shop=jewelry vs shop=perfumery',
+ 'shop=jewelry vs shop=pet',
+ 'shop=health_food vs shop=paint',
+ 'shop=health_food vs shop=photo',
+ 'shop=hairdresser vs shop=pet_grooming',
+ 'shop=frozen_food vs shop=kitchen',
+ 'shop=frozen_food vs shop=locksmith',
+ 'shop=frozen_food vs shop=mall',
+ 'shop=frozen_food vs shop=optician',
+ 'shop=frozen_food vs shop=paint',
+ 'shop=frozen_food vs shop=travel_agency',
+ 'shop=florist vs shop=greengrocer',
+ 'shop=florist vs shop=hairdresser',
+ 'shop=florist vs shop=household_linen',
+ 'shop=florist vs shop=hvac',
+ 'shop=florist vs shop=jewelry',
+ 'shop=florist vs shop=kiosk',
+ 'shop=florist vs shop=kitchen',
+ 'shop=florist vs shop=mall',
+ 'shop=florist vs shop=medical_supply',
+ 'shop=florist vs shop=newsagent',
+ 'shop=florist vs shop=outdoor',
+ 'shop=florist vs shop=perfumery',
+ 'shop=florist vs shop=repair',
+ 'shop=florist vs shop=sports',
+ 'shop=florist vs shop=tea',
+ 'shop=florist vs shop=tyres',
+ 'shop=flooring vs shop=travel_agency',
+ 'shop=flooring vs shop=tyres',
+ 'shop=flooring vs shop=video',
+ 'shop=flooring vs shop=video_games',
]
def matching_rather_than_type_conflict():
return [
+ 'shop=food vs shop=frozen_food',
+ 'shop=deli vs shop=frozen_food',
+
+ 'shop=health_food vs shop=herbalist',
+
+ 'shop=flooring vs shop=tiles',
+
+ 'shop=safety_equipment vs shop=trade',
+
'shop=cosmetics vs shop=perfumery',
'shop=appliance vs shop=electronics',
@@ -6700,6 +6860,12 @@ def matching_rather_than_type_conflict():
'shop=money_lender vs shop=pawnbroker',
+ 'shop=car_parts vs shop=tyres',
+ 'shop=truck vs shop=tyres',
+
+ 'shop=motorcycle_repair vs shop=tyres',
+ 'shop=motorcycle vs shop=tyres',
+
'shop=motorcycle vs shop=motorcycle_repair',
'shop=photo vs shop=photo_studio',
@@ -6727,8 +6893,6 @@ def matching_rather_than_type_conflict():
'shop=doityourself vs shop=wholesale',
'shop=doityourself vs shop=window_blind',
- 'shop=car_parts vs shop=tyres',
-
'shop=building_materials vs shop=doityourself',
'shop=building_materials vs shop=hardware',
diff --git a/test_matching_logic.py b/test_matching_logic.py
index 1717ebd..6c85812 100644
--- a/test_matching_logic.py
+++ b/test_matching_logic.py
@@ -722,10 +722,71 @@ class CanonicalValueTests(unittest.TestCase):
'name': 'Rituals'
}), True)
-
+ def test_prepared_food_shops_frozen_vs_deli(self):
+ self.assertEqual(config.the_same_feature_type({
+ '@source_uri': 'https://wesolapani.com/shops',
+ '@spider': 'wesola_pani',
+ 'shop': 'frozen_food',
+ 'addr:full': 'Mariana Domagały 45',
+ 'addr:state': 'Lesser Poland Voivodeship',
+ 'addr:country': 'PL',
+ 'phone': '+48 518 841 784',
+ 'brand': 'Wesoła Pani',
+ 'brand:wikidata': 'Q123240454',
+ 'atp_ref': '53',
+ 'opening_hours_in_atp_format': 'Mo-Fr 09:00-21:00; Sa 09:00-20:00',
+ 'name': 'Wesoła Pani'
+}, {
+ 'food': 'prepared_meals',
+ 'name': 'Wesoła Pani',
+ 'opening_hours': 'Mo-Fr 09:00-21:00; Sa 09:00-20:00',
+ 'payment:credit_cards': 'yes',
+ 'payment:debit_cards': 'yes',
+ 'shop': 'deli'
+}), True)
+
+ def test_prepared_food_shops_frozen_vs_food_shop(self):
+ self.assertEqual(config.the_same_feature_type({
+ '@source_uri': 'https://wesolapani.com/shops',
+ '@spider': 'wesola_pani',
+ 'shop': 'frozen_food',
+ 'addr:full': 'Mariana Domagały 45',
+ 'addr:state': 'Lesser Poland Voivodeship',
+ 'addr:country': 'PL',
+ 'phone': '+48 518 841 784',
+ 'brand': 'Wesoła Pani',
+ 'brand:wikidata': 'Q123240454',
+ 'atp_ref': '53',
+ 'opening_hours_in_atp_format': 'Mo-Fr 09:00-21:00; Sa 09:00-20:00',
+ 'name': 'Wesoła Pani'
+}, {
+ 'food': 'prepared_meals',
+ 'name': 'Wesoła Pani',
+ 'opening_hours': 'Mo-Fr 09:00-21:00; Sa 09:00-20:00',
+ 'payment:credit_cards': 'yes',
+ 'payment:debit_cards': 'yes',
+ 'shop': 'food'
+}), True)
+
+ def test_bakery_fast_food_place(self):
+ # (1) amenity=fast_food
+ # (2) amenity=fast_food + shop=bakery
+ # (3) shop=bakery
+ # (1) must match with (2)
+ # (2) must match with (3)
+ # (1) may match with (3) - or may not match, not matching is preferable, I think
+ self.assertEqual(config.the_same_feature_type({'shop': 'bakery', 'amenity': 'fast_food'}, {'shop': 'bakery'}), True)
+ self.assertEqual(config.the_same_feature_type({'shop': 'bakery', 'amenity': 'fast_food'}, {'amenity': 'fast_food'}), True)
+
def test_match_tile_shops_tagged_in_a_different_way(self):
self.assertEqual(config.the_same_feature_type({'shop': 'trade', 'trade': 'tiles'}, {'shop': 'trade'}), True)
self.assertEqual(config.the_same_feature_type({'shop': 'trade', 'trade': 'tiles'}, {'shop': 'tiles'}), True)
+ self.assertEqual(config.the_same_feature_type({'shop': 'trade', 'trade': 'tiles'}, {'shop': 'flooring'}), True)
+ self.assertEqual(config.the_same_feature_type({'shop': 'tiles'}, {'shop': 'flooring'}), True)
+
+ def test_match_unqualified_trade_shop_with_tile_shop(self):
+ self.assertEqual(config.the_same_feature_type({'shop': 'trade'}, {'shop': 'tiles'}), True)
+ self.assertEqual(config.the_same_feature_type({'shop': 'trade'}, {'shop': 'flooring'}), True)
def test_do_not_use_clothes_key_for_usual_clothes_shops(self):
self.assertEqual(config.the_same_feature_type({'shop': 'clothes', 'clothes': 'men'}, {'shop': 'clothes', 'clothes': 'women'}), True)