pleroma.database fill_old_hashtags: Add month_limit argument

2 files changed, 45 insertions, 35 deletions

diff --git a/docs/administration/CLI_tasks/database.md b/docs/administration/CLI_tasks/database.md
index a2d2e8cdd..eb85381ff 100644
--- a/docs/administration/CLI_tasks/database.md
+++ b/docs/administration/CLI_tasks/database.md
@@ -93,12 +93,14 @@ Can be safely re-run
 
 ## Fill hashtags for old objects
 
+Migrate hashags fields for old objects, from now to `months_limit` months.
+
 ```sh tab="OTP"
-./bin/pleroma_ctl database fill_old_hashtags
+./bin/pleroma_ctl database fill_old_hashtags <months_limit>
 ```
 
 ```sh tab="From Source"
-mix pleroma.database fill_old_hashtags
+mix pleroma.database fill_old_hashtags <months_limit>
 ```
 
 ## Vacuum the database
diff --git a/lib/mix/tasks/pleroma/database.ex b/lib/mix/tasks/pleroma/database.ex
index 0c1343313..a09852503 100644
--- a/lib/mix/tasks/pleroma/database.ex
+++ b/lib/mix/tasks/pleroma/database.ex
@@ -128,47 +128,55 @@ defmodule Mix.Tasks.Pleroma.Database do
     |> Stream.run()
   end
 
-  def run(["fill_old_hashtags"]) do
+  def run(["fill_old_hashtags", month_limit]) do
     import Ecto.Query
 
     start_pleroma()
 
-    from(
-      o in Object,
-      where: fragment("(?)->>'hashtags' is null", o.data),
-      where: fragment("(?)->>'tag' != '[]'", o.data),
-      select: %{id: o.id, tag: fragment("(?)->>'tag'", o.data)},
-      order_by: [:desc, o.id]
-    )
-    |> 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})"
-        )
+    month_limit = String.to_integer(month_limit)
 
-        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)]
+    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)
+
+      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})"
           )
-          |> Repo.update_all([], timeout: :infinity)
+
+          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)
-    end)
-    |> Stream.run()
+      |> Stream.run()
+    end
   end
 
   def run(["vacuum", args]) do