[home] Remove duplicate reposts from timeline feed, improve query performance, and increase batch size

app/controllers/api/v1/timelines/home_controller.rb

@@ -24,7 +24,7 @@ class Api::V1::Timelines::HomeController < Api::BaseController
   end
 
   def home_statuses
-    theLimit = params[:max_id].nil? ? 10 : limit_param(DEFAULT_STATUSES_LIMIT)
+    theLimit = params[:max_id].nil? ? 20 : limit_param(DEFAULT_STATUSES_LIMIT)
     account_home_feed.get(
       theLimit,
       params[:max_id],

app/models/home_feed.rb

@@ -19,24 +19,38 @@ class HomeFeed < Feed
     pagination_max = "and s.id < #{max_id}" unless max_id.nil?
     pagination_min = "and s.id > #{min_id}" unless min_id.nil?
     Status.find_by_sql "
-      select s.*
+      with cte as
-        from statuses s
+      (
-        left join statuses r
+        select
-          on s.reblog_of_id = r.id
+          row_number() over (partition by sid.reblog_of_id order by sid.id desc) as rn_dupe,
+          sid.*
+        FROM
+         (select
+          s.id,
+          s.reblog_of_id
+         from statuses s
+         left join statuses r
+           on s.reblog_of_id = r.id
+         where
+          s.created_at > NOW() - INTERVAL '7 days'
+          and s.reply is false
+          and (exists(select ff.target_account_id from follows ff where ff.account_id = #{@id} and ff.target_account_id = s.account_id)
+            or s.account_id = #{@id})
+          and not exists(select mm.target_account_id from mutes mm where mm.account_id = #{@id} and mm.target_account_id in (s.account_id, r.account_id))
+          and not exists(select bb.target_account_id from blocks bb where bb.account_id = #{@id} and bb.target_account_id in (s.account_id, r.account_id))
+          #{pagination_max}
+          #{pagination_min}
+         order by s.created_at desc
+          limit #{limit}
+        ) sid
+        inner join statuses s on sid.id = s.id
+      )
+      select
+        s.*
+      from cte
+      inner join statuses s on cte.id = s.id
       where
-        s.created_at > NOW() - INTERVAL '7 days'
+          (cte.rn_dupe = 1 or cte.reblog_of_id is null)
-        and s.reply is false
-        and (
-          exists(select ff.target_account_id from follows ff
-            where ff.account_id = #{@id} and ff.target_account_id = s.account_id)
-          or s.account_id = #{@id})
-        and not exists(select mm.target_account_id from mutes mm
-          where mm.account_id = #{@id} and mm.target_account_id in (s.account_id, r.account_id))
-        and not exists(select bb.target_account_id from blocks bb
-          where bb.account_id = #{@id} and bb.target_account_id in (s.account_id, r.account_id))
-        #{pagination_max}
-        #{pagination_min}
-      order by s.created_at desc limit #{limit}
     "
   end
 end