Merge pull request #4114 from nulib/5132-escape-anvl

Escape colons in Ark requests

app/lib/meadow/ark/serializer.ex

@@ -48,6 +48,17 @@ defmodule Meadow.Ark.Serializer do
     |> Enum.join("\n")
   end
 
-  def serialize({key, value}) when is_atom(key),
-    do: Map.get(@datacite_map, key) <> ": " <> String.replace(value, "%", "%25")
+  def serialize({key, value}) when is_atom(key) do
+    escapable =
+      case key do
+        :target -> "%\r\n"
+        _ -> ":%\r\n"
+      end
+
+    [
+      Map.get(@datacite_map, key),
+      URI.encode(value, fn c -> not String.contains?(escapable, <<c>>) end)
+    ]
+    |> Enum.join(": ")
+  end
 end

app/lib/meadow/search/bulk.ex

@@ -40,16 +40,16 @@ defmodule Meadow.Search.Bulk do
   defp upload_batch(docs, index) do
     with_log_metadata module: __MODULE__, index: index do
       bulk_document = docs |> Enum.join("\n")
-      
+
       Logger.info("Uploading batch of #{Enum.count(docs)} documents to #{index}")
-      
+
       case HTTP.post("/#{index}/_bulk", bulk_document <> "\n") do
         {:ok, %{status_code: status} = response} ->
           Logger.info("Bulk upload status: #{status}")
           {:ok, response}
 
         {:retry, response} ->
-          Logger.warn("Bulk upload retrying")
+          Logger.warning("Bulk upload retrying")
           {:retry, response}
 
         {:error, error} ->

app/lib/meadow/search/config.ex

@@ -2,8 +2,6 @@ defmodule Meadow.Search.Config do
   @moduledoc """
   Convenience methods for retrieving search-specific configuration
   """
-  alias Meadow.Search.HTTP
-
   require Logger
 
   def index_configs do

app/lib/meadow/utils/arks.ex

@@ -52,7 +52,7 @@ defmodule Meadow.Arks do
   """
   def mint_ark(%Work{descriptive_metadata: %{ark: ark}} = work)
       when not is_nil(ark) do
-    Logger.warn("Not minting ARK for work #{work.id} because it already has one: #{ark}")
+    Logger.warning("Not minting ARK for work #{work.id} because it already has one: #{ark}")
     {:noop, work}
   end
 

app/lib/meadow_web/resolvers/data.ex

@@ -164,20 +164,6 @@ defmodule MeadowWeb.Resolvers.Data do
     end
   end
 
-  def replace_file_set(_, %{id: id} = params, _) do
-    file_set = FileSets.get_file_set!(id)
-
-    case Pipeline.replace_the_file_set(file_set, Map.delete(params, :id)) do
-      {:error, changeset} ->
-        {:error,
-         message: "Could not replace file set",
-         details: ChangesetErrors.humanize_errors(changeset)}
-
-      {:ok, file_set} ->
-        {:ok, file_set}
-    end
-  end
-
   def update_file_set(_, %{id: id} = params, _) do
     file_set = FileSets.get_file_set!(id)
 

app/test/meadow/ark/serializer_test.exs

@@ -3,37 +3,65 @@ defmodule Meadow.Ark.SerializerTest do
 
   alias Meadow.Ark.Serializer
 
-  @response_body "success: ark:/99999/fk4z90ps4x\n_updated: 1630613597\ndatacite.publisher: Test publisher\n_profile: datacite\ndatacite.title: Test title\n_export: yes\ndatacite.creator: Test creator\n_owner: apitest\n_ownergroup: apitest\n_target: https://test/items/123\n_created: 1630613597\ndatacite.publicationyear: 2021\ndatacite.resourcetype: Image\n_status: public\n"
+  @request_payload """
+  _profile: datacite
+  datacite.creator: Test %25 creator
+  datacite.publicationyear: 2021
+  datacite.publisher: Publisher%3A Test
+  datacite.resourcetype: Image
+  _status: public
+  _target: https://test/items/123
+  datacite.title: 100%25
+  """
 
   describe "serialize/1" do
     test "desconstructs a Meadow.Ark and properly handles ANVL escaping of % characters" do
       ark = %Meadow.Ark{
         ark: "ark:/99999/fk4z90ps4x",
         creator: "Test % creator",
         publication_year: "2021",
-        publisher: "%Test publisher%",
+        publisher: "Publisher: Test",
         resource_type: "Image",
         status: "public",
         target: "https://test/items/123",
         title: "100%"
       }
 
-      assert Serializer.serialize(ark) == "_profile: datacite\ndatacite.creator: Test %25 creator\ndatacite.publicationyear: 2021\ndatacite.publisher: %25Test publisher%25\ndatacite.resourcetype: Image\n_status: public\n_target: https://test/items/123\ndatacite.title: 100%25"
+      assert Serializer.serialize(ark) == String.trim(@request_payload)
     end
   end
 
+  @response_body """
+  success: ark:/99999/fk4z90ps4x
+  _updated: 1630613597
+  datacite.publisher: Test publisher
+  _profile: datacite
+  datacite.title: Test title
+  _export: yes
+  datacite.creator: Test creator
+  _owner: apitest
+  _ownergroup: apitest
+  _target: https://test/items/123
+  _created: 1630613597
+  datacite.publicationyear: 2021
+  datacite.resourcetype: Image
+  _status: public
+  """
+
   describe "deserialize/1" do
     test "builds a Meadow.Ark struct" do
-      assert %Meadow.Ark{
-               ark: "ark:/99999/fk4z90ps4x",
-               creator: "Test creator",
-               publication_year: "2021",
-               publisher: "Test publisher",
-               resource_type: "Image",
-               status: "public",
-               target: "https://test/items/123",
-               title: "Test title"
-             } = Serializer.deserialize(@response_body)
+      expected = %Meadow.Ark{
+        ark: "ark:/99999/fk4z90ps4x",
+        creator: "Test creator",
+        publication_year: "2021",
+        publisher: "Test publisher",
+        resource_type: "Image",
+        status: "public",
+        target: "https://test/items/123",
+        title: "Test title"
+      }
+
+      assert Serializer.deserialize(@response_body) == expected
     end
   end
 end