make source_url unique; small update to context_date timezone handling

Author: ericahan22

backend/apps/events/migrations/0034_alter_events_source_url.py

@@ -0,0 +1,18 @@
+# Generated by Django 4.2.7 on 2025-12-04 00:39
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('events', '0033_events_comments_count_events_likes_count_and_more'),
+    ]
+
+    operations = [
+        migrations.AlterField(
+            model_name='events',
+            name='source_url',
+            field=models.TextField(blank=True, help_text="'https://university.edu/events/career-fair'", null=True, unique=True),
+        ),
+    ]

backend/apps/events/models.py

@@ -31,7 +31,10 @@ class Events(models.Model):
     )
 
     source_url = models.TextField(
-        null=True, blank=True, help_text="'https://university.edu/events/career-fair'"
+        null=True,
+        blank=True,
+        unique=True,
+        help_text="'https://university.edu/events/career-fair'",
     )
     source_image_url = models.TextField(
         null=True,

backend/services/openai_service.py

@@ -110,16 +110,20 @@ def extract_events_from_caption(
         - Return an array of JSON objects, each object representing a unique event.
         """
         # Get current date and day of week for context
-        now = datetime.now()
+        toronto_tz = pytz_timezone("America/Toronto")
+        now = datetime.now(toronto_tz)
         current_date = now.strftime("%Y-%m-%d")
         current_day_of_week = now.strftime("%A")
 
         context_datetime = (
             post_created_at if isinstance(post_created_at, datetime) else now
         )
-        context_datetime = pytz_timezone("America/Toronto").localize(
-            context_datetime.replace(tzinfo=None)
-        )
+        
+        if context_datetime.tzinfo is None:
+            context_datetime = toronto_tz.localize(context_datetime)
+        else:
+            context_datetime = context_datetime.astimezone(toronto_tz)
+
         context_date = context_datetime.strftime("%Y-%m-%d")
         context_day = context_datetime.strftime("%A")
         context_time = context_datetime.strftime("%H:%M")

backend/utils/scraping_utils.py

@@ -4,7 +4,7 @@
 from difflib import SequenceMatcher
 from pathlib import Path
 
-from django.db import transaction
+from django.db import transaction, IntegrityError
 from django.utils import timezone
 
 from apps.clubs.models import Clubs
@@ -179,6 +179,9 @@ def insert_event_to_db(event_data, ig_handle, source_url, club_type=None):
                 f"{log_prefix} Created {len(event_dates)} EventDates entries for event {event.id}"
             )
             return True
+        except IntegrityError:
+            logger.warning(f"{log_prefix} Duplicate event detected (IntegrityError) - likely race condition")
+            return "duplicate"
         except Exception as e:
             logger.error(f"{log_prefix} Error inserting event to DB: {e}")
             return False