Grace enhanced dining data

backend/dining/api_wrapper.py

@@ -1,9 +1,11 @@
 import datetime
 import json
+import logging
 from concurrent.futures import ThreadPoolExecutor, as_completed
 
 import requests
 from django.conf import settings
+from django.db.models import Count, Max
 from django.utils import timezone
 from django.utils.timezone import make_aware
 from requests.exceptions import ConnectionError, ConnectTimeout, ReadTimeout
@@ -12,9 +14,38 @@
 from utils.errors import APIError
 
 
+logger = logging.getLogger(__name__)
+
 OPEN_DATA_URL = "https://3scale-public-prod-open-data.apps.k8s.upenn.edu/api/v1/dining/"
 OPEN_DATA_ENDPOINTS = {"VENUES": OPEN_DATA_URL + "venues", "MENUS": OPEN_DATA_URL + "menus"}
 
+# Dining icon ids for parsing the API response cor_icon field into allergen boolean fields
+DINING_ICON_IDS = {
+    "vegetarian": "1",
+    "vegan": "4",
+    "in_balance": "7",
+    "halal": "10",
+    "kosher": "11",
+    "jain": "141",
+    "farm_to_fork": "6",
+    "locally_crafted": "55",
+    "garden_grown": "251",
+    "seafood_watch": "3",
+    "organic": "8",
+    "humane": "18",
+    "raw_undercooked": "228",
+    "peanut": "253",
+    "tree_nut": "254",
+    "sesame": "298",
+    "shellfish": "256",
+    "fish": "255",
+    "soy": "260",
+    "milk": "258",
+    "egg": "259",
+    "ask_us": "262",
+    "wheat_gluten": "257",
+}
+
 
 class DiningAPIWrapper:
     def __init__(self):
@@ -122,29 +153,42 @@ def fetch_menu(self, venue_id, date):
             response.json(),
         )  # also storing venue_id to later access in fetched_menus list
 
-    def load_menu(self, date=timezone.now().date()):
+    def load_menus(self, date=None):
         """
         Loads today's menu
+        Invariant: there should be no duplicate Menus. `load_menus` should delete
+        duplicate menus for all venues for the given date.
+
         NOTE: This method should only be used in load_next_menu.py, which is
         run based on a cron job every day
         """
+        if date is None:
+            date = timezone.now().date()
+
         # Venues without a menu should not be parsed
         skipped_venues = [747, 1163, 1731, 1732, 1733, 1464004, 1464009]
 
         # TODO: Handle API responses during empty menus (holidays)
         venues = [v for v in Venue.objects.all() if v.venue_id not in skipped_venues]
         venue_map = {venue.venue_id: venue for venue in venues}
 
-        # Fetch all menus in parallel to speed up loading time.
+        # Fetch menus in parallel to speed up loading time
         fetched_menus = []
-        with ThreadPoolExecutor(max_workers=8) as executor:  # 8 can be tuned
-            futures = [executor.submit(self.fetch_menu, venue.venue_id, date) for venue in venues]
-            for future in as_completed(futures):
+
+        with ThreadPoolExecutor(max_workers=8) as executor:
+            future_to_venue = {
+                executor.submit(self.fetch_menu, venue.venue_id, date): venue.venue_id
+                for venue in venues
+            }
+
+            for future in as_completed(future_to_venue):
                 try:
                     venue_id, response_json = future.result()
                     fetched_menus.append((venue_id, response_json))
-                except Exception as e:
-                    print(f"Error fetching menu: {e}")
+                except Exception:
+                    logger.exception(
+                        f"Dining: error fetching menu for venue {future_to_venue[future]}"
+                    )
 
         # Process the fetched menus and load them into the database
         for venue_id, response in fetched_menus:
@@ -175,6 +219,9 @@ def load_menu(self, date=timezone.now().date()):
                 # Append stations to dining menu
                 self.load_stations(daypart["stations"], dining_menu)
 
+        # delete duplicate menus
+        self.delete_duplicate_menus(date)
+
     def load_stations(self, station_response, dining_menu):
         for station_data in station_response:
             # TODO: This is inefficient for venues such as Houston Market
@@ -186,23 +233,31 @@ def load_stations(self, station_response, dining_menu):
             station.items.add(*items)
             station.save()
 
+    def _build_dining_item(self, key, value):
+        """
+        Helper function for load_items to build a DiningItem object from the dining API
+        """
+        icon_ids = value["cor_icon"] or {}
+        icon_flags = {
+            field_name: icon_id in icon_ids for field_name, icon_id in DINING_ICON_IDS.items()
+        }
+        return DiningItem(
+            item_id=key,
+            name=value["label"],
+            description=value["description"],
+            ingredients=value["ingredients"],
+            allergens=", ".join(value["cor_icon"].values()) if value["cor_icon"] else "",
+            **icon_flags,
+            nutrition_info=json.dumps(
+                {
+                    x["label"]: f"{x['value']}{x['unit']}"
+                    for x in value["nutrition_details"].values()
+                }
+            ),
+        )
+
     def load_items(self, item_response):
-        item_list = [
-            DiningItem(
-                item_id=key,
-                name=value["label"],
-                description=value["description"],
-                ingredients=value["ingredients"],
-                allergens=", ".join(value["cor_icon"].values()) if value["cor_icon"] else "",
-                nutrition_info=json.dumps(
-                    {
-                        x["label"]: f"{x['value']}{x['unit']}"
-                        for x in value["nutrition_details"].values()
-                    }
-                ),
-            )
-            for key, value in item_response.items()
-        ]
+        item_list = [self._build_dining_item(key, value) for key, value in item_response.items()]
         # Ignore conflicts because possibility of duplicate items
         DiningItem.objects.bulk_create(
             item_list,
@@ -212,3 +267,35 @@ def load_items(self, item_response):
             ],
             unique_fields=[DiningItem._meta.pk.name],
         )
+
+    def delete_duplicate_menus(self, date):
+        """Delete duplicate menus for an exact `date`.
+        Will delete all but the most recently created menus for each dining hall
+        """
+        # Find groups of duplicate menus
+        duplicate_groups = (
+            DiningMenu.objects.values("venue", "date", "start_time", "end_time", "service")
+            .annotate(menu_count=Count("id"), keep_id=Max("id"))
+            .filter(menu_count__gt=1, date=date)
+        )
+
+        # Find all ids to delete
+        ids_to_delete = []
+
+        for group in duplicate_groups:
+            ids = (
+                DiningMenu.objects.filter(
+                    venue=group["venue"],
+                    date=group["date"],
+                    start_time=group["start_time"],
+                    end_time=group["end_time"],
+                    service=group["service"],
+                )
+                .exclude(id=group["keep_id"])
+                .values_list("id", flat=True)
+            )
+            ids_to_delete.extend(ids)
+
+        # Delete all duplicates
+        deleted_count, _ = DiningMenu.objects.filter(id__in=ids_to_delete).delete()
+        return deleted_count

backend/dining/management/commands/load_next_menu.py

@@ -14,21 +14,24 @@ class Command(BaseCommand):
     the next 7 days, including the original date.
     """
 
-    def load_one_menu(self, delta, *args, **kwargs):
+    def load_one_menu(self, today, delta, *args, **kwargs):
         """
         Loads menu for a single day
         """
+        date_to_load = today + datetime.timedelta(days=delta)
+
         d = DiningAPIWrapper()
-        d.load_menu(timezone.now().date() + datetime.timedelta(days=delta))
-        delete_menu_view_cache(timezone.now().date() + datetime.timedelta(days=delta))
-        self.stdout.write(
-            "Loaded new Dining Menu for "
-            + str(timezone.now().date() + datetime.timedelta(days=delta))
-        )
+        d.load_menus(date_to_load)
+        delete_menu_view_cache(date_to_load)
+
+        # Error logging
+        self.stdout.write(f"Loaded new Dining Menu for {date_to_load}")
 
     def handle(self, *args, **kwargs):
         """
         Load menu for the next 7 days
         """
+        today = timezone.now().date()
+
         for i in range(7):
-            self.load_one_menu(i)
+            self.load_one_menu(today, i)

backend/dining/migrations/0007_diningitem_allergen_booleans.py

@@ -0,0 +1,128 @@
+# Generated by Django 5.0.2 on 2026-04-17 00:00
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ("dining", "0006_remove_diningmenu_stations_and_more"),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name="diningitem",
+            name="ask_us",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="egg",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="fish",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="farm_to_fork",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="garden_grown",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="halal",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="humane",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="in_balance",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="jain",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="kosher",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="locally_crafted",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="milk",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="organic",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="peanut",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="raw_undercooked",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="seafood_watch",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="sesame",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="shellfish",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="soy",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="tree_nut",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="vegan",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="vegetarian",
+            field=models.BooleanField(default=False),
+        ),
+        migrations.AddField(
+            model_name="diningitem",
+            name="wheat_gluten",
+            field=models.BooleanField(default=False),
+        ),
+    ]

backend/dining/models.py

@@ -16,11 +16,36 @@ class DiningItem(models.Model):
     name = models.CharField(max_length=255)
     description = models.CharField(max_length=1000, blank=True)
     ingredients = models.CharField(max_length=1000, blank=True)  # comma separated list
-    allergens = models.CharField(max_length=1000, blank=True)  # comma separated list
+    allergens = models.CharField(max_length=1000, blank=True)
     nutrition_info = models.CharField(max_length=1000, blank=True)  # json string.
     # Technically, postgres supports json fields but that involves local postgres
     # instead of sqlite AND we don't need to query on this field
 
+    # Dietary information fields (stored as cor_icons)
+    vegetarian = models.BooleanField(default=False)
+    vegan = models.BooleanField(default=False)
+    in_balance = models.BooleanField(default=False)
+    halal = models.BooleanField(default=False)
+    kosher = models.BooleanField(default=False)
+    jain = models.BooleanField(default=False)
+    farm_to_fork = models.BooleanField(default=False)
+    locally_crafted = models.BooleanField(default=False)
+    garden_grown = models.BooleanField(default=False)
+    seafood_watch = models.BooleanField(default=False)
+    organic = models.BooleanField(default=False)
+    humane = models.BooleanField(default=False)
+    raw_undercooked = models.BooleanField(default=False)
+    peanut = models.BooleanField(default=False)
+    tree_nut = models.BooleanField(default=False)
+    sesame = models.BooleanField(default=False)
+    shellfish = models.BooleanField(default=False)
+    fish = models.BooleanField(default=False)
+    soy = models.BooleanField(default=False)
+    milk = models.BooleanField(default=False)
+    egg = models.BooleanField(default=False)
+    ask_us = models.BooleanField(default=False)
+    wheat_gluten = models.BooleanField(default=False)
+
     def __str__(self):
         return f"{self.name}"