Initial commit

site/analytics/middleware.py

@@ -0,0 +1,106 @@
+import logging
+
+from django.utils.crypto import get_random_string
+
+from analytics.models import Attribution, PageView, UTMVisit
+
+
+logger = logging.getLogger(__name__)
+
+CORRELATION_COOKIE = "ms_cid"
+
+# Portal, auth, health, assets, and transactional public endpoints.
+_SKIP_PREFIXES = (
+    "/portal/",
+    "/admin/",
+    "/accounts/",
+    "/api/",
+    "/healthz",
+    "/static/",
+    "/media/",
+    "/unsubscribe/",
+)
+_SKIP_PATHS = frozenset({"/robots.txt", "/sitemap.xml", "/favicon.ico"})
+
+
+class PublicPageViewMiddleware:
+    """Record successful GET hits on public marketing pages."""
+
+    def __init__(self, get_response):
+        self.get_response = get_response
+
+    def __call__(self, request):
+        response = self.get_response(request)
+        if _should_record_page_view(request, response):
+            try:
+                PageView.objects.create(path=request.path[:512])
+            except Exception:
+                logger.exception("Failed to record page view for %s", request.path)
+        return response
+
+
+def _should_record_page_view(request, response) -> bool:
+    if request.method != "GET":
+        return False
+    if getattr(response, "status_code", 0) != 200:
+        return False
+    path = request.path
+    if path in _SKIP_PATHS:
+        return False
+    return not any(path.startswith(prefix) for prefix in _SKIP_PREFIXES)
+
+
+class UTMTrackingMiddleware:
+    """Capture UTM params into UTMVisit and stash a correlation id cookie."""
+
+    def __init__(self, get_response):
+        self.get_response = get_response
+
+    def __call__(self, request):
+        cid = request.COOKIES.get(CORRELATION_COOKIE) or get_random_string(32)
+        request.utm_correlation_id = cid
+
+        params = request.GET
+        has_utm = any(params.get(k) for k in (
+            "utm_source", "utm_medium", "utm_campaign", "utm_term", "utm_content"
+        ))
+        if has_utm or params.get("utm_source"):
+            UTMVisit.objects.create(
+                correlation_id=cid,
+                path=request.path[:512],
+                referrer=(request.META.get("HTTP_REFERER") or "")[:1024],
+                utm_source=params.get("utm_source", "")[:128],
+                utm_medium=params.get("utm_medium", "")[:128],
+                utm_campaign=params.get("utm_campaign", "")[:128],
+                utm_term=params.get("utm_term", "")[:128],
+                utm_content=params.get("utm_content", "")[:128],
+                user_agent=(request.META.get("HTTP_USER_AGENT") or "")[:512],
+            )
+
+        response = self.get_response(request)
+        if CORRELATION_COOKIE not in request.COOKIES:
+            response.set_cookie(
+                CORRELATION_COOKIE,
+                cid,
+                max_age=60 * 60 * 24 * 30,
+                samesite="Lax",
+            )
+        return response
+
+
+def attribute_lead_from_request(request, lead) -> Attribution | None:
+    cid = getattr(request, "utm_correlation_id", None) or request.COOKIES.get(
+        CORRELATION_COOKIE
+    )
+    visit = None
+    if cid:
+        visit = (
+            UTMVisit.objects.filter(correlation_id=cid).order_by("-created_at").first()
+        )
+    return Attribution.objects.create(
+        lead=lead,
+        visit=visit,
+        utm_source=visit.utm_source if visit else "",
+        utm_medium=visit.utm_medium if visit else "",
+        utm_campaign=visit.utm_campaign if visit else "",
+    )