Stats bug prevention

bots/models.py

@@ -7,7 +7,7 @@
 from django.contrib import admin
 from django.contrib.auth import get_user_model
 from django.db import models, transaction
-from django.db.models import Q, IntegerChoices
+from django.db.models import Q, IntegerChoices, QuerySet
 from django.utils import timezone
 from django.utils.text import Truncator
 from phonenumber_field.modelfields import PhoneNumberField
@@ -875,16 +875,9 @@ class ConvoState(models.IntegerChoices):
 
 
 class ConversationQuerySet(models.QuerySet):
-    def get_unique_users(self) -> models.QuerySet["Conversation"]:
+    def distinct_by_user_id(self) -> QuerySet["Conversation"]:
         """Get unique conversations"""
-        return self.distinct(
-            "fb_page_id",
-            "ig_account_id",
-            "wa_phone_number",
-            "slack_user_id",
-            "twilio_phone_number",
-            "web_user_id",
-        )
+        return self.distinct(*Conversation.user_id_fields)
 
     def to_df(self, tz=pytz.timezone(settings.TIME_ZONE)) -> "pd.DataFrame":
         import pandas as pd
@@ -1095,6 +1088,16 @@ class Conversation(models.Model):
 
     objects = ConversationQuerySet.as_manager()
 
+    user_id_fields = [
+        "fb_page_id",
+        "ig_account_id",
+        "slack_user_id",
+        "web_user_id",
+        "wa_phone_number",
+        "twilio_phone_number",
+        "id",
+    ]
+
     class Meta:
         unique_together = [
             ("slack_channel_id", "slack_user_id", "slack_team_id"),
@@ -1120,17 +1123,22 @@ def __str__(self):
     def get_display_name(self):
         return (
             (self.wa_phone_number and self.wa_phone_number.as_international)
+            or (self.twilio_phone_number and self.twilio_phone_number.as_international)
             or self.ig_username
             or self.fb_page_name
             or " in #".join(
                 filter(None, [self.slack_user_name, self.slack_channel_name])
             )
-            or self.fb_page_id
-            or self.slack_user_id
-            or self.web_user_id
-            or (self.twilio_phone_number and self.twilio_phone_number.as_international)
+            or self.unique_user_id()
         )
 
+    def unique_user_id(self) -> str | None:
+        for col in self.user_id_fields:
+            if col == "id":
+                return self.api_integration_id()
+            if value := getattr(self, col, None):
+                return value
+
     get_display_name.short_description = "User"
 
     def last_active_delta(self) -> datetime.timedelta:
@@ -1166,6 +1174,10 @@ def api_integration_id(self) -> str:
 
 
 class MessageQuerySet(models.QuerySet):
+    def distinct_by_user_id(self) -> QuerySet["Message"]:
+        """Get unique users"""
+        return self.distinct(*Message.convo_user_id_fields)
+
     def to_df(self, tz=pytz.timezone(settings.TIME_ZONE)) -> "pd.DataFrame":
         import pandas as pd
 
@@ -1381,6 +1393,10 @@ class Message(models.Model):
 
     objects = MessageQuerySet.as_manager()
 
+    convo_user_id_fields = [
+        f"conversation__{col}" for col in Conversation.user_id_fields
+    ]
+
     class Meta:
         ordering = ("-created_at",)
         get_latest_by = "created_at"

recipes/VideoBotsStats.py

@@ -37,15 +37,6 @@
 from gooey_ui import RedirectException
 from recipes.VideoBots import VideoBotsPage
 
-ID_COLUMNS = [
-    "conversation__fb_page_id",
-    "conversation__ig_account_id",
-    "conversation__wa_phone_number",
-    "conversation__slack_user_id",
-    "conversation__twilio_phone_number",
-    "conversation__web_user_id",
-]
-
 
 class VideoBotsStatsPage(BasePage):
     title = "Copilot Analytics"  # "Create Interactive Video Bots"
@@ -411,7 +402,7 @@ def calculate_overall_stats(*, bi, run_title, run_url):
         bot_integration=bi
     ).order_by()  # type: ignore
     # due to things like personal convos for slack, each user can have multiple conversations
-    users = conversations.get_unique_users().order_by()
+    users = conversations.distinct_by_user_id().order_by()
     messages: MessageQuerySet = Message.objects.filter(conversation__in=conversations).order_by()  # type: ignore
     user_messages = messages.filter(role=CHATML_ROLE_USER).order_by()
     bot_messages = messages.filter(role=CHATML_ROLE_ASSISTANT).order_by()
@@ -420,19 +411,15 @@ def calculate_overall_stats(*, bi, run_title, run_url):
             conversation__in=users,
             created_at__gte=timezone.now() - timedelta(days=7),
         )
-        .distinct(
-            *ID_COLUMNS,
-        )
+        .distinct_by_user_id()
         .count()
     )
     num_active_users_last_30_days = (
         user_messages.filter(
             conversation__in=users,
             created_at__gte=timezone.now() - timedelta(days=30),
         )
-        .distinct(
-            *ID_COLUMNS,
-        )
+        .distinct_by_user_id()
         .count()
     )
     positive_feedbacks = Feedback.objects.filter(
@@ -444,14 +431,17 @@ def calculate_overall_stats(*, bi, run_title, run_url):
         rating=Feedback.Rating.RATING_THUMBS_DOWN,
     ).count()
     run_link = f'Powered By: <a href="{run_url}" target="_blank">{run_title}</a>'
-    connection_detail = bi.get_display_name()
+    if bi.get_display_name() != bi.name:
+        connection_detail = f"- Connected to: {bi.get_display_name()}"
+    else:
+        connection_detail = ""
     st.markdown(
         f"""
             - Platform: {Platform(bi.platform).name.capitalize()}
             - Created on: {bi.created_at.strftime("%b %d, %Y")}
             - Last Updated: {bi.updated_at.strftime("%b %d, %Y")}
             - {run_link}
-            - Connected to: {connection_detail}
+            {connection_detail}
             * {users.count()} Users
             * {num_active_users_last_7_days} Active Users (Last 7 Days)
             * {num_active_users_last_30_days} Active Users (Last 30 Days)
@@ -485,9 +475,7 @@ def calculate_stats_binned_by_time(*, bi, start_date, end_date, factor, trunc_fn
         .annotate(Convos=Count("conversation_id", distinct=True))
         .annotate(
             Senders=Count(
-                Concat(
-                    *ID_COLUMNS,
-                ),
+                Concat(*Message.convo_user_id_fields),
                 distinct=True,
             )
         )