From 8657822f49c1df46a68343ed21bbf1e4ee0f00a2 Mon Sep 17 00:00:00 2001
From: Adrien Aury <44274230+adrienaury@users.noreply.github.com>
Date: Fri, 29 Sep 2023 22:56:01 +0000
Subject: [PATCH] perf: use buffered output

---
 internal/infra/datarowreader_jsonline.go | 24 ++++++++++++++++++++++--
 pkg/mimo/benchmark_test.go               | 14 +++-----------
 2 files changed, 25 insertions(+), 13 deletions(-)

diff --git a/internal/infra/datarowreader_jsonline.go b/internal/infra/datarowreader_jsonline.go
index 05d0b5a..d1800ad 100644
--- a/internal/infra/datarowreader_jsonline.go
+++ b/internal/infra/datarowreader_jsonline.go
@@ -29,6 +29,8 @@ import (
 	"github.com/cgi-fr/mimo/pkg/mimo"
 )
 
+const linebreak byte = 10
+
 type DataRowReaderJSONLine struct {
 	input  *bufio.Scanner
 	output *bufio.Writer
@@ -72,8 +74,10 @@ func (drr *DataRowReaderJSONLine) ReadDataRowAndWrite() (mimo.DataRow, error) {
 	var data mimo.DataRow
 
 	if drr.input.Scan() {
-		if _, err := drr.output.Write(append(drr.input.Bytes(), '\n')); err != nil {
-			return nil, fmt.Errorf("%w", err)
+		if drr.output != nil {
+			if err := drr.writeLine(); err != nil {
+				return nil, err
+			}
 		}
 
 		data = mimo.DataRow{}
@@ -93,6 +97,22 @@ func (drr *DataRowReaderJSONLine) ReadDataRowAndWrite() (mimo.DataRow, error) {
 	return data, nil
 }
 
+func (drr *DataRowReaderJSONLine) writeLine() error {
+	if _, err := drr.output.Write(drr.input.Bytes()); err != nil {
+		return fmt.Errorf("%w", err)
+	}
+
+	if err := drr.output.WriteByte(linebreak); err != nil {
+		return fmt.Errorf("%w", err)
+	}
+
+	return nil
+}
+
 func (drr *DataRowReaderJSONLine) Flush() error {
+	if drr.output == nil {
+		return nil
+	}
+
 	return fmt.Errorf("%w", drr.output.Flush())
 }
diff --git a/pkg/mimo/benchmark_test.go b/pkg/mimo/benchmark_test.go
index 6b08fa9..9facbc1 100644
--- a/pkg/mimo/benchmark_test.go
+++ b/pkg/mimo/benchmark_test.go
@@ -18,8 +18,6 @@
 package mimo_test
 
 import (
-	"io"
-	"os"
 	"testing"
 
 	"github.com/cgi-fr/mimo/internal/infra"
@@ -35,13 +33,11 @@ func BenchmarkInMemory(b *testing.B) {
 		b.FailNow()
 	}
 
-	file, err := os.Open("testdata/masked.jsonl")
+	maskedReader, err := infra.NewDataRowReaderJSONLineFromFile("testdata/masked.jsonl")
 	if err != nil {
 		b.FailNow()
 	}
 
-	maskedReader := infra.NewDataRowReaderJSONLine(file, io.Discard)
-
 	driver := mimo.NewDriver(
 		realReader,
 		maskedReader,
@@ -76,13 +72,11 @@ func BenchmarkOnDisk(b *testing.B) {
 		b.FailNow()
 	}
 
-	file, err := os.Open("testdata/masked.jsonl")
+	maskedReader, err := infra.NewDataRowReaderJSONLineFromFile("testdata/masked.jsonl")
 	if err != nil {
 		b.FailNow()
 	}
 
-	maskedReader := infra.NewDataRowReaderJSONLine(file, io.Discard)
-
 	driver := mimo.NewDriver(
 		realReader,
 		maskedReader,
@@ -123,13 +117,11 @@ func BenchmarkAllOptions(b *testing.B) {
 		b.FailNow()
 	}
 
-	file, err := os.Open("testdata/single-100-2.jsonl")
+	maskedReader, err := infra.NewDataRowReaderJSONLineFromFile("testdata/single-100-2.jsonl")
 	if err != nil {
 		b.FailNow()
 	}
 
-	maskedReader := infra.NewDataRowReaderJSONLine(file, io.Discard)
-
 	driver := mimo.NewDriver(
 		realReader,
 		maskedReader,