fix(oban): handle orphaned source deletion jobs gracefully

Add rescue block to SourceDeletionWorker to handle cases where the
source no longer exists. This prevents infinite retries when a deletion
job fails and the source has already been removed from the database.

Pinchflat_improvements.md

@@ -129,7 +129,7 @@ sqlite3 /path/to/pinchflat.db "
 docker compose start pinchflat
 ```
 
-## Issue 5: Orphaned deletion jobs for non-existent sources
+## Issue 5: Orphaned deletion jobs for non-existent sources [COMPLETED]
 
 **Problem:** When a source is deleted, a `SourceDeletionWorker` job is created. If this job fails repeatedly (e.g., due to database busy errors or files already deleted), it keeps retrying indefinitely even after the source no longer exists.
 

lib/pinchflat/sources/source_deletion_worker.ex

@@ -34,5 +34,8 @@ defmodule Pinchflat.Sources.SourceDeletionWorker do
     source = Sources.get_source!(source_id)
 
     Sources.delete_source(source, delete_files: delete_files)
+  rescue
+    Ecto.NoResultsError -> Logger.info("#{__MODULE__} discarded: source #{source_id} not found")
+    Ecto.StaleEntryError -> Logger.info("#{__MODULE__} discarded: source #{source_id} stale")
   end
 end

test/pinchflat/sources/source_deletion_worker_test.exs

@@ -48,5 +48,9 @@ defmodule Pinchflat.Sources.SourceDeletionWorkerTest do
       assert_raise Ecto.NoResultsError, fn -> Repo.reload!(media_item) end
       refute File.exists?(media_item.media_filepath)
     end
+
+    test "does not blow up if the record doesn't exist" do
+      assert :ok = perform_job(SourceDeletionWorker, %{id: 0})
+    end
   end
 end