fix: Better detection IPFS links in NFT metadata fetcher (#10638)

* fix: Better detection IPFS links in NFT metadata fetcher

* Fix tests

* Remove bypass from tests

apps/explorer/lib/explorer/token/metadata_retriever.ex

@@ -13,7 +13,6 @@ defmodule Explorer.Token.MetadataRetriever do
 
   @no_uri_error "no uri"
   @vm_execution_error "VM execution error"
-  @ipfs_protocol "ipfs://"
   @invalid_base64_data "invalid data:application/json;base64"
 
   # https://eips.ethereum.org/EIPS/eip-1155#metadata
@@ -515,31 +514,11 @@ defmodule Explorer.Token.MetadataRetriever do
     end
   end
 
-  # CIDv0 IPFS links # https://docs.ipfs.tech/concepts/content-addressing/#version-0-v0
-  defp fetch_json_from_uri({:ok, ["Qm" <> _ = result]}, _, token_id, hex_token_id, from_base_uri?) do
-    if String.length(result) == 46 do
-      ipfs? = true
-      fetch_json_from_uri({:ok, [ipfs_link(result)]}, ipfs?, token_id, hex_token_id, from_base_uri?)
-    else
-      Logger.warning(["Unknown metadata format result #{inspect(result)}."], fetcher: :token_instances)
-
-      {:error, truncate_error(result)}
-    end
-  end
-
   defp fetch_json_from_uri({:ok, ["'" <> token_uri]}, ipfs?, token_id, hex_token_id, from_base_uri?) do
     token_uri = token_uri |> String.split("'") |> List.first()
     fetch_metadata_inner(token_uri, ipfs?, token_id, hex_token_id, from_base_uri?)
   end
 
-  defp fetch_json_from_uri({:ok, ["http://" <> _ = token_uri]}, ipfs?, token_id, hex_token_id, from_base_uri?) do
-    fetch_metadata_inner(token_uri, ipfs?, token_id, hex_token_id, from_base_uri?)
-  end
-
-  defp fetch_json_from_uri({:ok, ["https://" <> _ = token_uri]}, ipfs?, token_id, hex_token_id, from_base_uri?) do
-    fetch_metadata_inner(token_uri, ipfs?, token_id, hex_token_id, from_base_uri?)
-  end
-
   defp fetch_json_from_uri(
          {:ok, [type = "data:application/json;utf8," <> json]},
          ipfs?,
@@ -587,35 +566,63 @@ defmodule Explorer.Token.MetadataRetriever do
       {:error, @invalid_base64_data}
   end
 
-  defp fetch_json_from_uri({:ok, ["#{@ipfs_protocol}ipfs/" <> right]}, _ipfs?, _token_id, hex_token_id, _from_base_uri?) do
+  defp fetch_json_from_uri({:ok, [token_uri_string]}, ipfs?, token_id, hex_token_id, from_base_uri?) do
-    fetch_from_ipfs(right, hex_token_id)
+    fetch_from_ipfs?(token_uri_string, ipfs?, token_id, hex_token_id, from_base_uri?)
   end
 
-  defp fetch_json_from_uri({:ok, ["ipfs/" <> right]}, _ipfs?, _token_id, hex_token_id, _from_base_uri?) do
+  defp fetch_json_from_uri(uri, _ipfs?, _token_id, _hex_token_id, _from_base_uri?) do
-    fetch_from_ipfs(right, hex_token_id)
+    Logger.warning(["Unknown metadata uri format #{inspect(uri)}."], fetcher: :token_instances)
-  end
 
-  defp fetch_json_from_uri({:ok, [@ipfs_protocol <> right]}, _ipfs?, _token_id, hex_token_id, _from_base_uri?) do
+    {:error, "unknown metadata uri format"}
-    fetch_from_ipfs(right, hex_token_id)
   end
 
-  defp fetch_json_from_uri({:ok, [json]}, _ipfs?, _token_id, hex_token_id, _from_base_uri?) do
+  # credo:disable-for-next-line Credo.Check.Refactor.CyclomaticComplexity
-    json = ExplorerHelper.decode_json(json, true)
+  defp fetch_from_ipfs?(token_uri_string, ipfs?, token_id, hex_token_id, from_base_uri?) do
+    case URI.parse(token_uri_string) do
+      %URI{scheme: "ipfs", path: path} ->
+        resource_id =
+          case path do
+            "/ipfs/" <> resource_id ->
+              resource_id
+
+            "/" <> resource_id ->
+              resource_id
+          end
+
+        fetch_from_ipfs(resource_id, hex_token_id)
+
+      %URI{scheme: _, path: "/ipfs/" <> resource_id} ->
+        fetch_from_ipfs(resource_id, hex_token_id)
+
+      %URI{scheme: _, path: "ipfs/" <> resource_id} ->
+        fetch_from_ipfs(resource_id, hex_token_id)
+
+      %URI{scheme: scheme} when not is_nil(scheme) ->
+        fetch_metadata_inner(token_uri_string, ipfs?, token_id, hex_token_id, from_base_uri?)
+
+      %URI{path: path} ->
+        case path do
+          "Qm" <> <<_::binary-size(44)>> = resource_id ->
+            fetch_from_ipfs(resource_id, hex_token_id)
 
-    check_type(json, hex_token_id)
+          # todo: rewrite for strict CID v1 support
+          "bafybe" <> _ = resource_id ->
+            fetch_from_ipfs(resource_id, hex_token_id)
+
+          _ ->
+            json = ExplorerHelper.decode_json(token_uri_string, true)
+
+            check_type(json, hex_token_id)
+        end
+    end
   rescue
     e ->
-      Logger.warning(["Unknown metadata format #{inspect(json)}.", Exception.format(:error, e, __STACKTRACE__)],
+      Logger.warning(
+        ["Unknown metadata format #{inspect(token_uri_string)}.", Exception.format(:error, e, __STACKTRACE__)],
         fetcher: :token_instances
       )
 
-      {:error, "invalid json"}
+      {:error, "invalid token_uri_string"}
-  end
-
-  defp fetch_json_from_uri(uri, _ipfs?, _token_id, _hex_token_id, _from_base_uri?) do
-    Logger.warning(["Unknown metadata uri format #{inspect(uri)}."], fetcher: :token_instances)
-
-    {:error, "unknown metadata uri format"}
   end
 
   defp fetch_json_from_json_string(json, ipfs?, token_id, hex_token_id, from_base_uri?, type) do

apps/explorer/test/explorer/token/metadata_retriever_test.exs

@@ -885,7 +885,7 @@ defmodule Explorer.Token.MetadataRetrieverTest do
                 }}
     end
 
-    test "fetches image from ipfs link directly", %{bypass: bypass} do
+    test "fetches image from ipfs link directly" do
       path = "/ipfs/bafybeig6nlmyzui7llhauc52j2xo5hoy4lzp6442lkve5wysdvjkizxonu"
 
       json = """
@@ -894,14 +894,19 @@ defmodule Explorer.Token.MetadataRetrieverTest do
       }
       """
 
-      Bypass.expect(bypass, "GET", path, fn conn ->
+      Application.put_env(:explorer, :http_adapter, Explorer.Mox.HTTPoison)
-        Conn.resp(conn, 200, json)
+
+      Explorer.Mox.HTTPoison
+      |> expect(:get, fn "https://ipfs.io/ipfs/bafybeig6nlmyzui7llhauc52j2xo5hoy4lzp6442lkve5wysdvjkizxonu",
+                         _headers,
+                         _options ->
+        {:ok, %HTTPoison.Response{status_code: 200, body: json}}
       end)
 
       data =
         {:ok,
          [
-           "http://localhost:#{bypass.port}#{path}"
+           path
          ]}
 
       assert {:ok,
@@ -912,7 +917,7 @@ defmodule Explorer.Token.MetadataRetrieverTest do
               }} == MetadataRetriever.fetch_json(data)
     end
 
-    test "Fetches metadata from ipfs", %{bypass: bypass} do
+    test "Fetches metadata from ipfs" do
       path = "/ipfs/bafybeid4ed2ua7fwupv4nx2ziczr3edhygl7ws3yx6y2juon7xakgj6cfm/51.json"
 
       json = """
@@ -921,14 +926,19 @@ defmodule Explorer.Token.MetadataRetrieverTest do
       }
       """
 
-      Bypass.expect(bypass, "GET", path, fn conn ->
+      Application.put_env(:explorer, :http_adapter, Explorer.Mox.HTTPoison)
-        Conn.resp(conn, 200, json)
+
+      Explorer.Mox.HTTPoison
+      |> expect(:get, fn "https://ipfs.io/ipfs/bafybeid4ed2ua7fwupv4nx2ziczr3edhygl7ws3yx6y2juon7xakgj6cfm/51.json",
+                         _headers,
+                         _options ->
+        {:ok, %HTTPoison.Response{status_code: 200, body: json}}
       end)
 
       data =
         {:ok,
          [
-           "http://localhost:#{bypass.port}#{path}"
+           path
          ]}
 
       {:ok,
@@ -937,6 +947,7 @@ defmodule Explorer.Token.MetadataRetrieverTest do
        }} = MetadataRetriever.fetch_json(data)
 
       assert "ipfs://bafybeihxuj3gxk7x5p36amzootyukbugmx3pw7dyntsrohg3se64efkuga/51.png" == Map.get(metadata, "image")
+      Application.put_env(:explorer, :http_adapter, HTTPoison)
     end
 
     test "Fetches metadata from '${url}'", %{bypass: bypass} do

cspell.json

@@ -41,6 +41,7 @@
         "badhash",
         "badnumber",
         "badpassword",
+        "bafybe",
         "bafybeid",
         "bafybeig",
         "bafybeihxuj",