Use list of german nouns before geocoding

analyze.py

@@ -119,7 +119,8 @@ def get_geoloc(query):
         locations.append({
             "lat": location["geometry"]["location"]["lat"],
             "lng": location["geometry"]["location"]["lng"],
-            "confidence": confidence_map[location["geometry"]["location_type"]]
+            "confidence": confidence_map[location["geometry"]["location_type"]],
+            "returned_place": location["formatted_address"]
         })
 
     return locations

locator.py

@@ -4,7 +4,11 @@ from analyze import *
 
 print("Start geocoding...")
 start_time = time.time()
-articles = Article.select()
+articles = Article.select().where(Article.id < 701)
+
+# use our list of german nouns for filtering
+with open("german_nouns.txt", "r") as f:
+    german_nouns = f.read().splitlines()
 
 for article in articles:
     potential = get_potential_places(article.place, article.description)
@@ -14,18 +18,21 @@ for article in articles:
 
     for place in places:
         query = " ".join([word for (word, tag) in place])
-        print("Query: {}, Berlin".format(query))
+        if query in german_nouns:
+            print("Skipping {}".format(query))
+        else:
+            print("Query: {}, Berlin".format(query))
 
-        locations = get_geoloc(query)
+            locations = get_geoloc(query)
 
-        # TODO: Only insert matches that have a higher confidence than current
-        # ones
-        for location in locations:
-            location["article"] = article
-            location["match"] = query
-            Location.create(**location)
+            # TODO: Only insert matches that have a higher confidence than current
+            # ones
+            for location in locations:
+                location["article"] = article
+                location["match"] = query
+                Location.create(**location)
 
-        time.sleep(1)
+            time.sleep(1)
 
 time_taken = time.time() - start_time
 print("Geocoded {} articles in {} seconds".format(articles.count(), time_taken))

models.py

@@ -24,6 +24,7 @@ class Location(BaseModel):
     lat = DoubleField()
     lng = DoubleField()
     match = CharField()
+    returned_place = CharField()
     article = ForeignKeyField(Article)
 
 class Category(BaseModel):