fix comments

Author: nullccxsy

src/iceberg/avro/avro_stream_internal.cc

@@ -145,7 +145,8 @@ void AvroOutputStream::flush() {
   }
 }
 
-std::shared_ptr<::arrow::io::OutputStream> AvroOutputStream::get_output_stream() const {
+const std::shared_ptr<::arrow::io::OutputStream>& AvroOutputStream::arrow_output_stream()
+    const {
   return output_stream_;
 }
 

src/iceberg/avro/avro_stream_internal.h

@@ -87,7 +87,7 @@ class AvroOutputStream : public ::avro::OutputStream {
   /// store, if any.
   void flush() override;
 
-  std::shared_ptr<::arrow::io::OutputStream> get_output_stream() const;
+  const std::shared_ptr<::arrow::io::OutputStream>& arrow_output_stream() const;
 
  private:
   std::shared_ptr<::arrow::io::OutputStream> output_stream_;

src/iceberg/avro/avro_writer.cc

@@ -53,9 +53,6 @@ Result<std::unique_ptr<AvroOutputStream>> CreateOutputStream(const WriterOptions
 
 }  // namespace
 
-// A stateful context to keep track of the writing progress.
-struct WriteContext {};
-
 class AvroWriter::Impl {
  public:
   Status Open(const WriterOptions& options) {
@@ -70,7 +67,7 @@ class AvroWriter::Impl {
     constexpr int64_t kDefaultBufferSize = 1024 * 1024;
     ICEBERG_ASSIGN_OR_RAISE(auto output_stream,
                             CreateOutputStream(options, kDefaultBufferSize));
-    arrow_output_stream_ = output_stream->get_output_stream();
+    arrow_output_stream_ = output_stream->arrow_output_stream();
     writer_ = std::make_unique<::avro::DataFileWriter<::avro::GenericDatum>>(
         std::move(output_stream), *avro_schema_);
     datum_ = std::make_unique<::avro::GenericDatum>(*avro_schema_);
@@ -105,19 +102,19 @@ class AvroWriter::Impl {
   int64_t length() { return total_bytes_; }
 
  private:
-  int64_t total_bytes_ = 0;
   // The schema to write.
   std::shared_ptr<::iceberg::Schema> write_schema_;
   // The avro schema to write.
   std::shared_ptr<::avro::ValidSchema> avro_schema_;
+  // Arrow output stream of the Avro file to write
+  std::shared_ptr<::arrow::io::OutputStream> arrow_output_stream_;
   // The avro writer to write the data into a datum.
   std::unique_ptr<::avro::DataFileWriter<::avro::GenericDatum>> writer_;
-  // Arrow schema for data conversion (C API format)
-  ArrowSchema arrow_schema_;
-  // Reusable Avro datum for writing individual records
+  // Reusable Avro datum for writing individual records.
   std::unique_ptr<::avro::GenericDatum> datum_;
-  // Arrow output stream for writing data to filesystem
-  std::shared_ptr<::arrow::io::OutputStream> arrow_output_stream_;
+  // Arrow schema to write data.
+  ArrowSchema arrow_schema_;
+  int64_t total_bytes_ = 0;
 };
 
 AvroWriter::~AvroWriter() = default;

test/avro_test.cc

@@ -135,9 +135,11 @@ class AvroReaderTest : public TempFileTestBase {
     ASSERT_TRUE(file_info_result.ok());
     ASSERT_EQ(file_info_result->size(), writer->length().value());
 
-    auto reader_result = ReaderFactoryRegistry::Open(
-        FileFormatType::kAvro,
-        {.path = temp_avro_file_, .io = file_io_, .projection = schema});
+    auto reader_result = ReaderFactoryRegistry::Open(FileFormatType::kAvro,
+                                                     {.path = temp_avro_file_,
+                                                      .length = file_info_result->size(),
+                                                      .io = file_io_,
+                                                      .projection = schema});
     ASSERT_THAT(reader_result, IsOk());
     auto reader = std::move(reader_result.value());
     ASSERT_NO_FATAL_FAILURE(VerifyNextBatch(*reader, expected_string));