pleroma.database fill_old_hashtags: Add month_limit argument

[akkoma] / lib / mix / tasks / pleroma / database.ex

diff --git a/lib/mix/tasks/pleroma/database.ex b/lib/mix/tasks/pleroma/database.ex
index 2f1f3346988c8d9aa1aaec0c57dd3cf0911f959f..a098525038e008b77fcadbfa6fce1aa651295e24 100644 (file)
--- a/lib/mix/tasks/pleroma/database.ex
+++ b/lib/mix/tasks/pleroma/database.ex
@@ -4,11 +4,13 @@
 
 defmodule Mix.Tasks.Pleroma.Database do
   alias Pleroma.Conversation
+  alias Pleroma.Maintenance
   alias Pleroma.Object
   alias Pleroma.Repo
   alias Pleroma.User
   require Logger
   require Pleroma.Constants
+  import Ecto.Query
   import Mix.Pleroma
   use Mix.Task
 
@@ -34,17 +36,7 @@ defmodule Mix.Tasks.Pleroma.Database do
     )
 
     if Keyword.get(options, :vacuum) do
-      Logger.info("Runnning VACUUM FULL.")
-
-      Logger.warn(
-        "Re-packing your entire database may take a while and will consume extra disk space during the process."
-      )
-
-      Repo.query!(
-        "vacuum full;",
-        [],
-        timeout: :infinity
-      )
+      Maintenance.vacuum("full")
     end
   end
 
@@ -56,14 +48,18 @@ defmodule Mix.Tasks.Pleroma.Database do
   def run(["update_users_following_followers_counts"]) do
     start_pleroma()
 
-    User
-    |> Repo.all()
-    |> Enum.each(&User.update_follower_count/1)
+    Repo.transaction(
+      fn ->
+        from(u in User, select: u)
+        |> Repo.stream()
+        |> Stream.each(&User.update_follower_count/1)
+        |> Stream.run()
+      end,
+      timeout: :infinity
+    )
   end
 
   def run(["prune_objects" | args]) do
-    import Ecto.Query
-
     {options, [], []} =
       OptionParser.parse(
         args,
@@ -98,30 +94,18 @@ defmodule Mix.Tasks.Pleroma.Database do
     |> Repo.delete_all(timeout: :infinity)
 
     if Keyword.get(options, :vacuum) do
-      Logger.info("Runnning VACUUM FULL.")
-
-      Logger.warn(
-        "Re-packing your entire database may take a while and will consume extra disk space during the process."
-      )
-
-      Repo.query!(
-        "vacuum full;",
-        [],
-        timeout: :infinity
-      )
+      Maintenance.vacuum("full")
     end
   end
 
   def run(["fix_likes_collections"]) do
-    import Ecto.Query
-
     start_pleroma()
 
     from(object in Object,
       where: fragment("(?)->>'likes' is not null", object.data),
       select: %{id: object.id, likes: fragment("(?)->>'likes'", object.data)}
     )
-    |> Pleroma.RepoStreamer.chunk_stream(100)
+    |> Pleroma.Repo.chunk_stream(100, :batches)
     |> Stream.each(fn objects ->
       ids =
         objects
@@ -144,34 +128,94 @@ defmodule Mix.Tasks.Pleroma.Database do
     |> Stream.run()
   end
 
-  def run(["vacuum", args]) do
+  def run(["fill_old_hashtags", month_limit]) do
+    import Ecto.Query
+
     start_pleroma()
 
-    case args do
-      "analyze" ->
-        Logger.info("Runnning VACUUM ANALYZE.")
+    month_limit = String.to_integer(month_limit)
 
-        Repo.query!(
-          "vacuum analyze;",
-          [],
-          timeout: :infinity
-        )
+    if month_limit < 1 do
+      shell_error("Invalid `month_limit` argument, needs to be greater than 1")
+    else
+      time_limit = DateTime.utc_now() |> Timex.shift(months: -month_limit)
 
-      "full" ->
-        Logger.info("Runnning VACUUM FULL.")
+      from(
+        o in Object,
+        where: fragment("(?)->>'hashtags' is null", o.data),
+        where: fragment("(?)->>'tag' != '[]'", o.data),
+        where: o.inserted_at < ^time_limit,
+        select: %{id: o.id, tag: fragment("(?)->>'tag'", o.data)}
+      )
+      |> Pleroma.Repo.chunk_stream(200, :batches)
+      |> Stream.each(fn objects ->
+        Repo.transaction(fn ->
+          objects_first = objects |> List.first()
+          objects_last = objects |> List.last()
+
+          Logger.info(
+            "fill_old_hashtags: #{objects_first.id} (#{objects_first.inserted_at}) -- #{
+              objects_last.id
+            } (#{objects_last.inserted_at})"
+          )
+
+          objects
+          |> Enum.map(fn object ->
+            tags =
+              object.tag
+              |> Jason.decode!()
+              |> Enum.filter(&is_bitstring(&1))
+
+            Object
+            |> where([o], o.id == ^object.id)
+            |> update([o],
+              set: [data: fragment("safe_jsonb_set(?, '{hashtags}', ?, true)", o.data, ^tags)]
+            )
+            |> Repo.update_all([], timeout: :infinity)
+          end)
+        end)
+      end)
+      |> Stream.run()
+    end
+  end
 
-        Logger.warn(
-          "Re-packing your entire database may take a while and will consume extra disk space during the process."
-        )
+  def run(["vacuum", args]) do
+    start_pleroma()
 
-        Repo.query!(
-          "vacuum full;",
-          [],
-          timeout: :infinity
-        )
+    Maintenance.vacuum(args)
+  end
 
-      _ ->
-        Logger.error("Error: invalid vacuum argument.")
-    end
+  def run(["ensure_expiration"]) do
+    start_pleroma()
+    days = Pleroma.Config.get([:mrf_activity_expiration, :days], 365)
+
+    Pleroma.Activity
+    |> join(:inner, [a], o in Object,
+      on:
+        fragment(
+          "(?->>'id') = COALESCE((?)->'object'->> 'id', (?)->>'object')",
+          o.data,
+          a.data,
+          a.data
+        )
+    )
+    |> where(local: true)
+    |> where([a], fragment("(? ->> 'type'::text) = 'Create'", a.data))
+    |> where([_a, o], fragment("?->>'type' = 'Note'", o.data))
+    |> Pleroma.Repo.chunk_stream(100, :batches)
+    |> Stream.each(fn activities ->
+      Enum.each(activities, fn activity ->
+        expires_at =
+          activity.inserted_at
+          |> DateTime.from_naive!("Etc/UTC")
+          |> Timex.shift(days: days)
+
+        Pleroma.Workers.PurgeExpiredActivity.enqueue(%{
+          activity_id: activity.id,
+          expires_at: expires_at
+        })
+      end)
+    end)
+    |> Stream.run()
   end
 end