dry

Author: ericmj

lib/diff_web/live/diff_live_view.ex

@@ -227,7 +227,7 @@ defmodule DiffWeb.DiffLiveView do
 
             diff_data =
               Jason.encode!(%{
-                "diff" => sanitize_utf8(raw_diff),
+                "diff" => DiffWeb.LiveView.sanitize_utf8(raw_diff),
                 "path_from" => path_from,
                 "path_to" => path_to
               })
@@ -328,50 +328,6 @@ defmodule DiffWeb.DiffLiveView do
   defp parse_version(""), do: {:ok, :latest}
   defp parse_version(input), do: Version.parse(input)
 
-  defp sanitize_utf8(content) when is_binary(content) do
-    case String.valid?(content) do
-      true ->
-        content
-
-      false ->
-        # Multiple fallback strategies for invalid UTF-8
-        content
-        |> sanitize_invalid_bytes()
-    end
-  end
-
-  defp sanitize_utf8(content), do: content
-
-  defp sanitize_invalid_bytes(content) do
-    # Try different encoding conversions and fallbacks
-    cond do
-      # Try converting from common encodings
-      latin1_result = safe_unicode_convert(content, :latin1, :utf8) ->
-        latin1_result
-
-      # Try converting from ISO-8859-1
-      iso_result = safe_unicode_convert(content, {:encoding, :latin1}, :utf8) ->
-        iso_result
-
-      # Last resort: replace invalid bytes with replacement character
-      true ->
-        content
-        |> :binary.bin_to_list()
-        # Replace high bytes with '?'
-        |> Enum.map(fn byte -> if byte > 127, do: 63, else: byte end)
-        |> :binary.list_to_bin()
-    end
-  end
-
-  defp safe_unicode_convert(content, from, to) do
-    case :unicode.characters_to_binary(content, from, to) do
-      result when is_binary(result) -> result
-      _ -> nil
-    end
-  rescue
-    _ -> nil
-  end
-
   defp parse_diff(diff) do
     case String.split(diff, ":", trim: true) do
       [app, from, to] -> {app, from, to, build_url(app, from, to)}

lib/diff_web/views/live_view.ex

@@ -44,7 +44,7 @@ defmodule DiffWeb.LiveView do
     end
   end
 
-  defp sanitize_utf8(content) when is_binary(content) do
+  def sanitize_utf8(content) when is_binary(content) do
     case String.valid?(content) do
       true ->
         content
@@ -55,19 +55,15 @@ defmodule DiffWeb.LiveView do
     end
   end
 
-  defp sanitize_utf8(content), do: content
+  def sanitize_utf8(content), do: content
 
   defp sanitize_invalid_bytes(content) do
     # Try different encoding conversions and fallbacks
     cond do
-      # Try converting from common encodings
+      # Try converting from Latin-1/ISO-8859-1 encoding
       latin1_result = safe_unicode_convert(content, :latin1, :utf8) ->
         latin1_result
 
-      # Try converting from ISO-8859-1
-      iso_result = safe_unicode_convert(content, {:encoding, :latin1}, :utf8) ->
-        iso_result
-
       # Last resort: replace invalid bytes with replacement character
       true ->
         content