mydump,restore: reduce parser memory allocation

lightning/mydump/parser.go

@@ -1,6 +1,7 @@
 package mydump
 
 import (
+	"bytes"
 	"io"
 
 	"github.com/pkg/errors"
@@ -12,7 +13,7 @@ type ChunkParser struct {
 	// states for the lexer
 	reader      io.Reader
 	buf         []byte
-	bufSize     int
+	blockBuf    []byte
 	isLastChunk bool
 
 	lastRow Row
@@ -24,6 +25,10 @@ type ChunkParser struct {
 	// The list of columns in the form `(a, b, c)` in the last INSERT statement.
 	// Assumed to be constant throughout the entire file.
 	Columns []byte
+
+	// cache
+	remainBuf *bytes.Buffer
+	appendBuf *bytes.Buffer
 }
 
 // Chunk represents a portion of the data file.
@@ -43,8 +48,10 @@ type Row struct {
 // NewChunkParser creates a new parser which can read chunks out of a file.
 func NewChunkParser(reader io.Reader) *ChunkParser {
 	return &ChunkParser{
-		reader:  reader,
-		bufSize: 8192,
+		reader:    reader,
+		blockBuf:  make([]byte, 8192),
+		remainBuf: &bytes.Buffer{},
+		appendBuf: &bytes.Buffer{},
 	}
 }
 
@@ -85,15 +92,18 @@ func tryAppendTo(out *[]byte, tail []byte) {
 }
 
 func (parser *ChunkParser) readBlock() error {
-	block := make([]byte, parser.bufSize)
-
-	n, err := io.ReadFull(parser.reader, block)
+	n, err := io.ReadFull(parser.reader, parser.blockBuf)
 	switch err {
 	case io.ErrUnexpectedEOF, io.EOF:
 		parser.isLastChunk = true
 		fallthrough
 	case nil:
-		tryAppendTo(&parser.buf, block[:n])
+		parser.remainBuf.Reset()
+		parser.remainBuf.Write(parser.buf)
+		parser.appendBuf.Reset()
+		parser.appendBuf.Write(parser.remainBuf.Bytes())
+		parser.appendBuf.Write(parser.blockBuf[:n])
+		parser.buf = parser.appendBuf.Bytes()
 		return nil
 	default:
 		return errors.Trace(err)

lightning/restore/restore.go

@@ -26,6 +26,7 @@ import (
 	verify "github.com/pingcap/tidb-lightning/lightning/verification"
 	tidbcfg "github.com/pingcap/tidb/config"
 	"github.com/pingcap/tidb/meta/autoid"
+	"github.com/pingcap/tidb/util/hack"
 	"github.com/pingcap/tidb/util/kvencoder"
 	"github.com/pkg/errors"
 )
@@ -1386,7 +1387,7 @@ func (cr *chunkRestore) restore(
 
 		// sql -> kv
 		start = time.Now()
-		kvs, _, err := kvEncoder.SQL2KV(buffer.String())
+		kvs, _, err := kvEncoder.SQL2KV(hack.String(buffer.Bytes()))
 		encodeDur := time.Since(start)
 		encodeTotalDur += encodeDur
 		metric.BlockEncodeSecondsHistogram.Observe(encodeDur.Seconds())