add sampling

plausible · Mar 12, 2024 · 565d4a9 · 565d4a9
1 parent 487064c
commit 565d4a9
Showing 1 changed file with 21 additions and 30 deletions.
diff --git a/lib/plausible/exports.ex b/lib/plausible/exports.ex
@@ -5,6 +5,7 @@ defmodule Plausible.Exports do
 
   # NOTE: needs tests
 
+  require Plausible
   import Ecto.Query
 
   @doc """
@@ -29,6 +30,16 @@ defmodule Plausible.Exports do
     }
   end
 
+  Plausible.on_full_build do
+    defp sampled(table) do
+      Plausible.Stats.Sampling.add_query_hint(from(table))
+    end
+  else
+    defp sampled(table) do
+      table
+    end
+  end
+
   defmacrop date(timestamp) do
     quote do
       selected_as(fragment("toDate(?)", unquote(timestamp)), :date)
@@ -71,9 +82,7 @@ defmodule Plausible.Exports do
   @spec export_visitors_q(pos_integer) :: Ecto.Query.t()
   def export_visitors_q(site_id) do
     visitors_sessions_q =
-      from s in "sessions_v2",
-        # NOTE: no smapling is used right now
-        # hints: ["SAMPLE", unsafe_fragment(^@sample)],
+      from s in sampled("sessions_v2"),
         where: s.site_id == ^site_id,
         group_by: selected_as(:date),
         select: %{
@@ -86,9 +95,7 @@ defmodule Plausible.Exports do
         }
 
     visitors_events_q =
-      from e in "events_v2",
-        # NOTE: no smapling is used right now
-        # hints: ["SAMPLE", unsafe_fragment(^@sample)],
+      from e in sampled("events_v2"),
         where: e.site_id == ^site_id,
         group_by: selected_as(:date),
         select: %{
@@ -122,9 +129,7 @@ defmodule Plausible.Exports do
 
   @spec export_sources_q(pos_integer) :: Ecto.Query.t()
   def export_sources_q(site_id) do
-    from s in "sessions_v2",
-      # NOTE: no smapling is used right now
-      # hints: ["SAMPLE", unsafe_fragment(^@sample)],
+    from s in sampled("sessions_v2"),
       where: s.site_id == ^site_id,
       group_by: [
         selected_as(:date),
@@ -152,9 +157,7 @@ defmodule Plausible.Exports do
   @spec export_pages_q(pos_integer) :: Ecto.Query.t()
   def export_pages_q(site_id) do
     window_q =
-      from e in "events_v2",
-        # NOTE: no smapling is used right now
-        # hints: ["SAMPLE", unsafe_fragment(^@sample)],
+      from e in sampled("events_v2"),
         where: e.site_id == ^site_id,
         select: %{
           timestamp: e.timestamp,
@@ -197,9 +200,7 @@ defmodule Plausible.Exports do
 
   @spec export_entry_pages_q(pos_integer) :: Ecto.Query.t()
   def export_entry_pages_q(site_id) do
-    from s in "sessions_v2",
-      # NOTE: no smapling is used right now
-      # hints: ["SAMPLE", unsafe_fragment(^@sample)],
+    from s in sampled("sessions_v2"),
       where: s.site_id == ^site_id,
       group_by: [selected_as(:date), s.entry_page],
       order_by: selected_as(:date),
@@ -218,9 +219,7 @@ defmodule Plausible.Exports do
 
   @spec export_exit_pages_q(pos_integer) :: Ecto.Query.t()
   def export_exit_pages_q(site_id) do
-    from s in "sessions_v2",
-      # NOTE: no smapling is used right now
-      # hints: ["SAMPLE", unsafe_fragment(^@sample)],
+    from s in sampled("sessions_v2"),
       where: s.site_id == ^site_id,
       group_by: [selected_as(:date), s.exit_page],
       order_by: selected_as(:date),
@@ -237,9 +236,7 @@ defmodule Plausible.Exports do
 
   @spec export_locations_q(pos_integer) :: Ecto.Query.t()
   def export_locations_q(site_id) do
-    from s in "sessions_v2",
-      # NOTE: no smapling is used right now
-      # hints: ["SAMPLE", unsafe_fragment(^@sample)],
+    from s in sampled("sessions_v2"),
       where: s.site_id == ^site_id,
       where: s.city_geoname_id != 0 and s.country_code != "\0\0" and s.country_code != "ZZ",
       group_by: [selected_as(:date), s.country_code, selected_as(:region), s.city_geoname_id],
@@ -258,9 +255,7 @@ defmodule Plausible.Exports do
 
   @spec export_devices_q(pos_integer) :: Ecto.Query.t()
   def export_devices_q(site_id) do
-    from s in "sessions_v2",
-      # NOTE: no smapling is used right now
-      # hints: ["SAMPLE", unsafe_fragment(^@sample)],
+    from s in sampled("sessions_v2"),
       where: s.site_id == ^site_id,
       group_by: [selected_as(:date), s.screen_size],
       order_by: selected_as(:date),
@@ -276,9 +271,7 @@ defmodule Plausible.Exports do
 
   @spec export_browsers_q(pos_integer) :: Ecto.Query.t()
   def export_browsers_q(site_id) do
-    from s in "sessions_v2",
-      # NOTE: no smapling is used right now
-      # hints: ["SAMPLE", unsafe_fragment(^@sample)],
+    from s in sampled("sessions_v2"),
       where: s.site_id == ^site_id,
       group_by: [selected_as(:date), s.browser],
       order_by: selected_as(:date),
@@ -294,9 +287,7 @@ defmodule Plausible.Exports do
 
   @spec export_operating_systems_q(pos_integer) :: Ecto.Query.t()
   def export_operating_systems_q(site_id) do
-    from s in "sessions_v2",
-      # NOTE: no smapling is used right now
-      # hints: ["SAMPLE", unsafe_fragment(^@sample)],
+    from s in sampled("sessions_v2"),
       where: s.site_id == ^site_id,
       group_by: [selected_as(:date), s.operating_system],
       order_by: selected_as(:date),