perf: optimization for homepage referral activity tracking

web/views.py

@@ -286,14 +286,56 @@ def index(request):
         :10
     ]  # Get more and then sort by clicks
 
-    # Add click counts manually since WebRequest.user is a CharField, not a ForeignKey
-    for referrer in top_referrers:
-        # Look for both new format /ref/CODE/ and old format ?ref=CODE
-        ref_code = referrer.referral_code
-        clicks = WebRequest.objects.filter(
-            models.Q(path__contains=f"/ref/{ref_code}/") | models.Q(path__contains=f"?ref={ref_code}")
-        ).count()
-        referrer.total_clicks = clicks
+    # PERF FIX: Batch-fetch all click counts in a single query instead of N separate
+    # queries (one per referrer). This avoids an N+1 query problem that runs
+    # unindexed LIKE scans against a potentially huge WebRequest table on every
+    # homepage load.
+    if top_referrers:
+        all_codes = sorted([r.referral_code for r in top_referrers if r.referral_code], key=len, reverse=True)
+
+        if all_codes:
+            # Build a single combined Q expression for all codes at once
+            combined_q = models.Q()
+            for code in all_codes:
+                # Broadened contains check to catch /ref/CODE with or without trailing slash
+                combined_q |= models.Q(path__contains=f"/ref/{code}") | models.Q(path__contains=f"ref={code}")
+
+            # Fetch unique matching paths and their aggregate counts in one go
+            # Using Sum("count") to support models where rows represent multiple hits
+            matching_requests = (
+                WebRequest.objects.filter(combined_q)
+                .values("path")
+                .annotate(total_hits=models.Sum("count"))
+            )
+
+            from urllib.parse import parse_qs, urlparse
+
+            # Map each referral code to its click count in-memory
+            click_counts = {code: 0 for code in all_codes}
+            for req in matching_requests:
+                path = req["path"]
+                total_hits = req["total_hits"] or 0
+                extracted_code = None
+
+                # Try new format /en/ref/CODE/ or /en/ref/CODE
+                if "/ref/" in path:
+                    # Improved regex to handle trailing slashes, end of path, query params, or fragments
+                    m = re.search(r"/ref/([^/?#]+)(?:/|$|\?|#)", path)
+                    if m:
+                        extracted_code = m.group(1)
+
+                # Try query param format ?ref=CODE
+                if not extracted_code and "ref=" in path:
+                    parsed_url = urlparse(path)
+                    params = parse_qs(parsed_url.query)
+                    if "ref" in params:
+                        extracted_code = params["ref"][0]
+
+                if extracted_code and extracted_code in click_counts:
+                    click_counts[extracted_code] += total_hits
+
+            for referrer in top_referrers:
+                referrer.total_clicks = click_counts.get(referrer.referral_code, 0)
 
     # Re-sort to include click count in ranking
     top_referrers = sorted(