Merge branch 'dedupe-sharding' into 'develop'

Pleroma.Upload.Filter.Dedupe: sharding directory structure

See merge request pleroma/pleroma!4292

changelog.d/dedupe-sharding.change

@@ -0,0 +1 @@
+Dedupe upload filter now uses a three-level sharding directory structure

lib/pleroma/upload/filter/dedupe.ex

@@ -17,8 +17,16 @@ defmodule Pleroma.Upload.Filter.Dedupe do
       |> Base.encode16(case: :lower)
 
     filename = shasum <> "." <> extension
-    {:ok, :filtered, %Upload{upload | id: shasum, path: filename}}
+
+    {:ok, :filtered, %Upload{upload | id: shasum, path: shard_path(filename)}}
   end
 
   def filter(_), do: {:ok, :noop}
+
+  @spec shard_path(String.t()) :: String.t()
+  def shard_path(
+        <<a::binary-size(2), b::binary-size(2), c::binary-size(2), _::binary>> = filename
+      ) do
+    Path.join([a, b, c, filename])
+  end
 end

test/pleroma/object_test.exs

@@ -174,8 +174,9 @@ defmodule Pleroma.ObjectTest do
 
       filename = Path.basename(href)
 
-      assert {:ok, files} = File.ls(uploads_dir)
+      expected_path = Path.join([uploads_dir, Pleroma.Upload.Filter.Dedupe.shard_path(filename)])
-      assert filename in files
+
+      assert File.exists?(expected_path)
 
       Object.delete(note)
 
@@ -183,8 +184,7 @@ defmodule Pleroma.ObjectTest do
 
       assert Object.get_by_id(note.id).data["deleted"]
       assert Object.get_by_id(attachment.id) == nil
-      assert {:ok, files} = File.ls(uploads_dir)
+      refute File.exists?(expected_path)
-      refute filename in files
     end
 
     test "with objects that have legacy data.url attribute" do

test/pleroma/upload/filter/dedupe_test.exs

@@ -10,6 +10,10 @@ defmodule Pleroma.Upload.Filter.DedupeTest do
 
   @shasum "e30397b58d226d6583ab5b8b3c5defb0c682bda5c31ef07a9f57c1c4986e3781"
 
+  test "generates a shard path for a shasum" do
+    assert "e3/03/97/" <> _path = Dedupe.shard_path(@shasum)
+  end
+
   test "adds shasum" do
     File.cp!(
       "test/fixtures/image.jpg",
@@ -23,10 +27,12 @@ defmodule Pleroma.Upload.Filter.DedupeTest do
       tempfile: Path.absname("test/fixtures/image_tmp.jpg")
     }
 
+    expected_path = Dedupe.shard_path(@shasum <> ".jpg")
+
     assert {
              :ok,
              :filtered,
-             %Pleroma.Upload{id: @shasum, path: @shasum <> ".jpg"}
+             %Pleroma.Upload{id: @shasum, path: ^expected_path}
            } = Dedupe.filter(upload)
   end
 end

test/pleroma/upload_test.exs

@@ -149,6 +149,9 @@ defmodule Pleroma.UploadTest do
 
     test "copies the file to the configured folder with deduping" do
       File.cp!("test/fixtures/image.jpg", "test/fixtures/image_tmp.jpg")
+      expected_filename = "e30397b58d226d6583ab5b8b3c5defb0c682bda5c31ef07a9f57c1c4986e3781.jpg"
+
+      expected_path = Pleroma.Upload.Filter.Dedupe.shard_path(expected_filename)
 
       file = %Plug.Upload{
         content_type: "image/jpeg",
@@ -159,8 +162,7 @@ defmodule Pleroma.UploadTest do
       {:ok, data} = Upload.store(file, filters: [Pleroma.Upload.Filter.Dedupe])
 
       assert List.first(data["url"])["href"] ==
-               Pleroma.Upload.base_url() <>
+               Path.join([Pleroma.Upload.base_url(), expected_path])
-                 "e30397b58d226d6583ab5b8b3c5defb0c682bda5c31ef07a9f57c1c4986e3781.jpg"
     end
 
     test "copies the file to the configured folder without deduping" do