Unable to decode BigDecimal via ByteBuffer

[sergiimk]

When upgrading avro4s from version 2.0.4 to latest I get following exception when reading file containing DECIMAL type:

com.sksamuel.avro4s.Avro4sDecodingException: Unable to decode '[1, 31, 113, -5, 4, -53]' to BigDecimal via ByteBuffer

Sample file attached: example.zip

parquet-tools utility seems to have no problem reading this file.

I'm using the following DTO when reading:

case class Foo(
  // These columns are present in the file but we can skip them when reading
  // system_time: Timestamp,
  // event_time: Timestamp,
  decimal_13_4: Option[BigDecimal],
  decimal_38_18: Option[BigDecimal]
)

Read using following bit of code:

def read[T: Encoder: Decoder](path: Path)(
    implicit schemaFor: SchemaFor[T]
  ): List[T] = {
    val format = RecordFormat[T]

    val reader =
      AvroParquetReader
        .builder[GenericRecord](new org.apache.hadoop.fs.Path(path.toUri))
        .withDataModel(GenericData.get)
        .build()

    val records = Stream
      .continually(reader.read)
      .takeWhile(_ != null)
      .map(format.from)
      .toList

    reader.close()
    records
  }

Reproducible in hello-world project using the following dependencies:

scalaVersion := "2.12.13"
libraryDependencies += "com.sksamuel.avro4s" %% "avro4s-core" % "4.0.9"
libraryDependencies += "org.apache.parquet" % "parquet-avro" % "1.12.0"
libraryDependencies += "org.apache.hadoop" % "hadoop-client" % "3.3.0"