Prioritize guessed categories, don't add non-informational names

ai_name_convert.py

@@ -29,30 +29,36 @@ def ai_name_convert(pois: Sequence[UmPoi]) -> Sequence[UmPoi]:
         answer = complete(
             ('You are provided with a list of restaurant names, which are sometimes descriptive. '
              'These names originate from a Polish data source. '
-             'Your task is to extract an own name for each restaurant. '
+             'Your task is to extract an own name for each restaurant - if there isn\'t one, you output nothing. '
              'The output will be in CSV format, just like the input.'),
 
-            ('1,lokal gastronomiczny Sao Do Asia Food\n'
-             '2,"BoQ"\n'
-             '3,Restauracja Garden w hotelu Double Tree by Hilton\n'
-             '4,Klub Sosnowy\n'
-             '5,Spiewajaca Lipka\n'
-             '6,Bar Alchemy w hotelu\n'
-             '7,w klubie tenisowym Wilga\n'
-             '8,bar w Hotelu "Boss"\n'
-             '9,ruchomy punkt gastronomiczny (room service, mini bar, VIP Lounge)\n'
-             '10,na terenie tymczasowego targowiska "Kawiarnia Olkuska"'),
+            ('1,mała gastronomia\n'
+             '2,lokal gastronomiczny Sao Do Asia Food\n'
+             '3,"BoQ"\n'
+             '4,Restauracja Garden w hotelu Double Tree by Hilton\n'
+             '5,Klub Sosnowy\n'
+             '6,Barek z Wygrodzonym Ogródkiem\n'
+             '7,Spiewajaca Lipka\n'
+             '8,Bar Alchemy w hotelu "Lisbon"\n'
+             '9,w klubie tenisowym Wilga\n'
+             '10,bar w Hotelu "Boss"\n'
+             '11,bufet dla publiczności w Kasynie\n'
+             '12,ruchomy punkt gastronomiczny (room service, mini bar, VIP Lounge)\n'
+             '13,restauracja Indyjska\n'),
 
-            ('1,Sao Do Asia Food\n'
-             '2,BoQ\n'
-             '3,Garden\n'
-             '4,Klub Sosnowy\n'
-             '5,Śpiewająca Lipka\n'
-             '6,Alchemy\n'
-             '7,Wilga\n'
-             '8,Boss\n'
-             '9,Ruchomy punkt gastronomiczny\n'
-             '10,Kawiarnia Olkuska'),
+            ('1,\n'
+             '2,Sao Do Asia Food\n'
+             '3,BoQ\n'
+             '4,Garden\n'
+             '5,Klub Sosnowy\n'
+             '6,\n'
+             '7,Śpiewająca Lipka\n'
+             '8,Alchemy\n'
+             '9,Wilga\n'
+             '10,Boss\n'
+             '11,\n'
+             '12,\n'
+             '13,\n'),
 
             query)
 

config.py

@@ -77,6 +77,7 @@
 
 UM_GUESS_CATEGORY = {
     'bistro': 'restauracja',
+    'burger': 'burgerownia',
     'cafe': 'kawiarnia',
     'caffe': 'kawiarnia',
     'ciastkarnia': 'cukiernia',
@@ -90,12 +91,14 @@
     'piwny': 'piwiarnia',
     'pizza': 'pizzeria',
     'restaurant': 'restauracja',
-    'sushi': 'restauracja',
+    'ristorante': 'restauracja',
     'tawerna': 'restauracja',
-    'winiarnia': 'winiarnia',
     'zajazd': 'restauracja',
 }
 
+assert not set(UM_CATEGORY_TAGS).intersection(UM_GUESS_CATEGORY), \
+    'UM_CATEGORY_TAGS and UM_GUESS_CATEGORY must be disjoint'
+
 DEFAULT_POI_TAGS = {
     'source': 'mapa.um.warszawa.pl',
 }

main.py

@@ -32,6 +32,9 @@ def main():
     for um_poi, osm_match in zip(um_pois, osm_matches):
         um_poi_name = beautify_name(um_poi.name)
 
+        if not um_poi_name:
+            continue
+
         if osm_match is None:
             missing_pois.append(um_poi)
             print(f'[1/2] ❌ {um_poi_name!r} ({um_poi.category!r})')
@@ -48,6 +51,9 @@ def main():
     for um_poi, osm_match in zip(um_pois, osm_matches):
         um_poi_name = beautify_name(um_poi.name)
 
+        if not um_poi_name:
+            continue
+
         if osm_match is None:
             missing_pois.append(um_poi)
             print(f'[2/2] ❌ {um_poi_name!r} ({um_poi.category!r})')
@@ -56,7 +62,7 @@ def main():
             print(f'[2/2] ✅ {um_poi_name!r} ({um_poi.category!r}) ↔ {osm_match_name!r}')
 
     missing_pois = missing_pois[:LIMIT_CHANGES_PER_CHANGESET]
-    print(f'🛟 Limiting to {LIMIT_CHANGES_PER_CHANGESET} POIs for safety')
+    print(f'🛟 Limiting to {LIMIT_CHANGES_PER_CHANGESET} POIs')
     print(f'📍 Total POIs to be added: {len(missing_pois)}')
 
     if not missing_pois:

osm_search.py

@@ -1,6 +1,5 @@
 import re
 from collections import defaultdict
-from pprint import pprint
 from typing import Sequence
 
 from rapidfuzz import fuzz, process

overpass.py

@@ -8,10 +8,8 @@ def build_pois_query(timeout: int) -> str:
     return (
         f'[out:json][timeout:{timeout}];'
         'relation(id:336075);'  # https://www.openstreetmap.org/relation/336075
-        'map_to_area->.r;'
-        'nwr[amenity](area.r);'
-        'out tags center qt;'
-        'nwr[shop](area.r);'
+        'map_to_area;'
+        'nwr[!highway][name](area);'
         'out tags center qt;'
     )
 

um_fetch.py

@@ -6,7 +6,7 @@
 from rapidfuzz import fuzz, process, utils
 from tenacity import retry, stop_after_attempt, wait_exponential
 
-from config import UM_GUESS_CATEGORY, UM_VALID_CATEGORIES
+from config import UM_CATEGORY_TAGS, UM_GUESS_CATEGORY, UM_VALID_CATEGORIES
 from um_poi import UmPoi
 from utils import get_http_client, nice_hash
 
@@ -47,13 +47,16 @@ def _guess_category(name: str) -> str:
     matches = process.extract(name, _GUESS_CATEGORY_CHOICES,
                               scorer=fuzz.partial_ratio,
                               processor=utils.default_process,
-                              limit=1,
+                              limit=3,
                               score_cutoff=85)
 
     if not matches:
         return ''
 
-    match = matches[0][0]
+    if len(matches) == 1:
+        match = matches[0][0]
+    else:
+        match = max(matches, key=lambda m: len(UM_CATEGORY_TAGS[UM_GUESS_CATEGORY.get(m[0], m[0])]))[0]
 
     return UM_GUESS_CATEGORY.get(match, match)
 
@@ -67,11 +70,15 @@ def um_fetch_restaurants() -> Sequence[UmPoi]:
 
     for p in foiarray:
         lat, lng = _PROJ_TRANSFORMER.transform(p['y'], p['x'])
-        category, name, address = _parse_details(p['name'])
+        um_category, name, address = _parse_details(p['name'])
 
-        if not category:
-            category = _guess_category(name)
+        category = _guess_category(name)
+
+        if category:
             print(f'🧩 Guessed category {category!r} for {name!r}')
+        if not category:
+            category = um_category
+            print(f'🏱 Using UM category {category!r} for {name!r}')
 
         p_id = nice_hash((name, address))