Update custom_utils.py

custom_utils.py

@@ -139,7 +139,7 @@ def vector_search_naive(openai_api_key,
         }
     }
 
-    pipeline = [vector_search_stage, get_remove_fields_stage()]
+    pipeline = [vector_search_stage, get_project_fields_stage()]
 
     return invoke_search(db, collection, pipeline)
 
@@ -172,7 +172,7 @@ def vector_search_advanced(openai_api_key,
         }
     }
   
-    pipeline = [vector_search_and_filter_stage, get_remove_fields_stage()] + additional_stages
+    pipeline = [vector_search_and_filter_stage, get_project_fields_stage()] + additional_stages
 
     return invoke_search(db, collection, pipeline)
 
@@ -184,22 +184,83 @@ def get_remove_fields_stage():
 def get_project_fields_stage():
     return {
         "$project": {
-            "_id": 0,  
-            "name": 1,
-            "accommodates": 1,
-            "address.street": 1,
-            "address.government_area": 1, 
-            "address.market": 1,
-            "address.country": 1, 
-            "address.country_code": 1, 
-            "address.location.type": 1, 
-            "address.location.coordinates": 1,  
-            "address.location.is_location_exact": 1,
-            "summary": 1,
-            "space": 1,  
+            "_id": 1, 
+            "id": 1, 
+            "listing_url": 1, 
+            "scrape_id": 1, 
+            "last_scraped": 1, 
+            "source": 1, 
+            "name": 1, 
+            "description": 1, 
             "neighborhood_overview": 1, 
-            "notes": 1, 
-            "score": {"$meta": "vectorSearchScore"} 
+            "picture_url": 1, 
+            "host_id": 1, 
+            "host_url": 1, 
+            "host_name": 1, 
+            "host_since": 1, 
+            "host_location": 1, 
+            "host_about": 1, 
+            "host_response_time": 1, 
+            "host_response_rate": 1, 
+            "host_acceptance_rate": 1, 
+            "host_is_superhost": 1, 
+            "host_thumbnail_url": 1, 
+            "host_picture_url": 1, 
+            "host_neighbourhood": 1, 
+            "host_listings_count": 1, 
+            "host_total_listings_count": 1, 
+            "host_verifications": 1, 
+            "host_has_profile_pic": 1, 
+            "host_identity_verified": 1, 
+            "neighbourhood": 1, 
+            "neighbourhood_cleansed": 1, 
+            "neighbourhood_group_cleansed": 1, 
+            "latitude": 1, 
+            "longitude": 1, 
+            "property_type": 1, 
+            "room_type": 1, 
+            "accommodates": 1, 
+            "bathrooms": 1, 
+            "bathrooms_text": 1, 
+            "bedrooms": 1, 
+            "beds": 1, 
+            "amenities": 1, 
+            "price": 1, 
+            "minimum_nights": 1, 
+            "maximum_nights": 1, 
+            "minimum_minimum_nights": 1, 
+            "maximum_minimum_nights": 1, 
+            "minimum_maximum_nights": 1, 
+            "maximum_maximum_nights": 1, 
+            "minimum_nights_avg_ntm": 1, 
+            "maximum_nights_avg_ntm": 1, 
+            "calendar_updated": 1, 
+            "has_availability": 1, 
+            "availability_30": 1, 
+            "availability_60": 1, 
+            "availability_90": 1, 
+            "availability_365": 1, 
+            "calendar_last_scraped": 1, 
+            "number_of_reviews": 1, 
+            "number_of_reviews_ltm": 1, 
+            "number_of_reviews_l30d": 1, 
+            "first_review": 1, 
+            "last_review": 1, 
+            "review_scores_rating": 1, 
+            "review_scores_accuracy": 1, 
+            "review_scores_cleanliness": 1, 
+            "review_scores_checkin": 1, 
+            "review_scores_communication": 1, 
+            "review_scores_location": 1, 
+            "review_scores_value": 1, 
+            "license": 1, 
+            "instant_bookable": 1, 
+            "calculated_host_listings_count": 1, 
+            "calculated_host_listings_count_entire_homes": 1, 
+            "calculated_host_listings_count_private_rooms": 1, 
+            "calculated_host_listings_count_shared_rooms": 1, 
+            "reviews_per_month": 1,
+            "description_embedding": 0
         }
     }