- More small improvments to the parser

Utilities/DetermineOriginGame.py

@@ -348,7 +348,7 @@ def get_locations_from_bulbapedia(pokemon_name, form, cache):
         
         variant = ""
         for sibling in event_header.find_next_siblings():
-            if sibling.name == 'h4':
+            if sibling.name == 'h4' or "held" in sibling.getText(strip=True).lower():
                 break
             if sibling.name == 'h5':
                 variant = sibling.get_text(strip=True)
@@ -376,7 +376,7 @@ def get_locations_from_bulbapedia(pokemon_name, form, cache):
 
     # For Later
     for variant in event_tables:
-        if variant == pokemon_name or (form and form in variant):
+        if (variant == pokemon_name and form is None)or (form and form in variant):
             games_container_rows = event_tables[variant].find_all('tr')
             for game_row in games_container_rows:
                 entries = game_row.find_all('td')
@@ -384,7 +384,9 @@ def get_locations_from_bulbapedia(pokemon_name, form, cache):
                     games_string = entries[0].find('a').get('title')
                     for game in all_games:
                         if game in games_string:
-                            game_locations[game] = "Event"
+                            if game not in game_locations:
+                                game_locations[game] = []
+                            game_locations[game].append("Event")
 
     return game_locations
 
@@ -620,9 +622,9 @@ def check_alternative_sources(pokemon, cache):
         if event_status:
             return "Event", "Event"
         
-        bulb_locations = get_locations_from_bulbapedia(pokemon['base_name'], pokemon['form'], cache)
-        if bulb_locations:
-            return bulb_locations[0], "Bulbapedia"
+        #bulb_locations = get_locations_from_bulbapedia(pokemon['base_name'], pokemon['form'], cache)
+        #if bulb_locations:
+        #    return bulb_locations[0], "Bulbapedia"
 
         # Check generation introduced
         #generation = species_data.get('generation', {}).get('name', '')
@@ -649,7 +651,7 @@ def handle_unknown_encounters(pokemon_list, cache):
 if __name__ == "__main__":
     get_cached_data()
     
-    pokemon_list = read_pokemon_list('pokemon_home_list.csv', limit=151)
+    pokemon_list = read_pokemon_list('pokemon_home_list.csv', limit=3000)
     pokemon_list_with_games = determine_earliest_games(pokemon_list, cache)
     pokemon_list_adjusted = adjust_for_evolution(pokemon_list_with_games, cache)
     pokemon_list_with_locations = add_encounter_locations(pokemon_list_adjusted, cache)

pokemon_home_list.csv

@@ -40,10 +40,10 @@ number,name
 0027,Sandshrew (Alolan Form)
 0028,Sandslash
 0028,Sandslash (Alolan Form)
-0029,Nidoran-f
+0029,Nidoran♀
 0030,Nidorina
 0031,Nidoqueen
-0032,Nidoran-m
+0032,Nidoran♂
 0033,Nidorino
 0034,Nidoking
 0035,Clefairy