Raise when encoding/decoding invalid UTF-8 (#346)

Author: ericmj

.github/workflows/main.yml

@@ -129,7 +129,7 @@ jobs:
 
   conformance-test:
     name: Conformance test (Elixir ${{matrix.elixir}} | Erlang/OTP ${{matrix.otp}})
-    runs-on: ubuntu-18.04
+    runs-on: ubuntu-22.04
     strategy:
       fail-fast: false
       matrix:

lib/protobuf/wire.ex

@@ -71,7 +71,16 @@ defmodule Protobuf.Wire do
   @spec encode(proto_type(), proto_value()) :: iodata()
   def encode(type, value)
 
-  def encode(type, binary) when is_binary(binary) and type in [:string, :bytes],
+  def encode(:string, binary) when is_binary(binary) do
+    unless String.valid?(binary) do
+      raise Protobuf.EncodeError,
+        message: "invalid UTF-8 data for type string: #{inspect(binary)}"
+    end
+
+    encode_from_wire_type(wire_delimited(), binary)
+  end
+
+  def encode(:bytes, binary) when is_binary(binary),
     do: encode_from_wire_type(wire_delimited(), binary)
 
   def encode(:int32, n) when n in @sint32_range, do: encode_from_wire_type(wire_varint(), n)
@@ -133,7 +142,16 @@ defmodule Protobuf.Wire do
   @spec decode(proto_type(), binary() | integer()) :: proto_value()
   def decode(type, value)
 
-  def decode(type, val) when type in [:string, :bytes], do: val
+  def decode(:string, binary) when is_binary(binary) do
+    unless String.valid?(binary) do
+      raise Protobuf.DecodeError,
+        message: "invalid UTF-8 data for type string: #{inspect(binary)}"
+    end
+
+    binary
+  end
+
+  def decode(:bytes, binary) when is_binary(binary), do: binary
 
   def decode(:int32, val) do
     <<n::signed-integer-32>> = <<val::32>>

test/protobuf/decoder_test.exs

@@ -198,6 +198,12 @@ defmodule Protobuf.DecoderTest do
              %TestMsg.Oneof{first: {:b, ""}, second: {:c, 0}}
   end
 
+  test "raises on invalid UTF-8" do
+    assert_raise Protobuf.DecodeError, "invalid UTF-8 data for type string: <<255>>", fn ->
+      Decoder.decode(<<10, 1, 255>>, TestMsg.Scalars)
+    end
+  end
+
   test "decodes with transformer module" do
     assert TestMsg.WithTransformModule.decode(<<8, 42>>) == 42
 

test/protobuf/encoder_test.exs

@@ -254,6 +254,16 @@ defmodule Protobuf.EncoderTest do
                255, 1, 202, 62, 3, 102, 111, 111>>
   end
 
+  test "raises on invalid UTF-8" do
+    message =
+      "Got error when encoding TestMsg.Scalars#string: " <>
+        "** (Protobuf.EncodeError) invalid UTF-8 data for type string: <<255>>"
+
+    assert_raise Protobuf.EncodeError, message, fn ->
+      Encoder.encode(%TestMsg.Scalars{string: <<255>>})
+    end
+  end
+
   test "encodes with transformer module" do
     msg = %TestMsg.ContainsTransformModule{field: 0}
     assert Encoder.encode(msg) == <<10, 0>>

test/protobuf/wire_test.exs

@@ -99,6 +99,8 @@ defmodule Protobuf.WireTest do
 
     test "string" do
       assert encode(:string, "testing") == <<7, 116, 101, 115, 116, 105, 110, 103>>
+
+      assert_raise Protobuf.EncodeError, fn -> encode(:string, <<255>>) end
     end
 
     test "bytes" do
@@ -340,6 +342,8 @@ defmodule Protobuf.WireTest do
 
     test "string" do
       assert "testing" == Wire.decode(:string, <<116, 101, 115, 116, 105, 110, 103>>)
+
+      assert_raise Protobuf.DecodeError, fn -> Wire.decode(:string, <<255>>) end
     end
 
     test "bytes" do