Reducing memory allocation events

Former-commit-id: c94e79ba116464504580fc397270ead154063971

pkg/obiformats/fastseq_json_header.go

@@ -1,8 +1,10 @@
 package obiformats
 
 import (
+	"bytes"
 	"math"
 	"strings"
+	"unsafe"
 
 	log "github.com/sirupsen/logrus"
 
@@ -85,17 +87,26 @@ func ParseFastSeqJsonHeader(sequence *obiseq.BioSequence) {
 	}
 }
 
-func FormatFastSeqJsonHeader(sequence *obiseq.BioSequence) string {
+func WriteFastSeqJsonHeader(buffer *bytes.Buffer, sequence *obiseq.BioSequence) {
+
 	annotations := sequence.Annotations()
 
 	if len(annotations) > 0 {
-		text, err := obiutils.JsonMarshal(sequence.Annotations())
+		err := obiutils.JsonMarshalByteBuffer(buffer, sequence.Annotations())
 
 		if err != nil {
 			log.Fatal(err)
 		}
+	}
+}
 
-		return string(text)
+func FormatFastSeqJsonHeader(sequence *obiseq.BioSequence) string {
+	annotations := sequence.Annotations()
+	buffer := bytes.Buffer{}
+
+	if len(annotations) > 0 {
+		obiutils.JsonMarshalByteBuffer(&buffer, sequence.Annotations())
+		return unsafe.String(unsafe.SliceData(buffer.Bytes()), len(buffer.Bytes()))
 	}
 
 	return ""

pkg/obiformats/fastseq_obi_header.go

@@ -7,6 +7,7 @@ import (
 	"regexp"
 	"strconv"
 	"strings"
+	"unsafe"
 
 	log "github.com/sirupsen/logrus"
 
@@ -298,17 +299,18 @@ func ParseFastSeqOBIHeader(sequence *obiseq.BioSequence) {
 	}
 }
 
-func FormatFastSeqOBIHeader(sequence *obiseq.BioSequence) string {
+func WriteFastSeqOBIHeade(buffer *bytes.Buffer, sequence *obiseq.BioSequence) {
+
 	annotations := sequence.Annotations()
 
-	if annotations != nil {
-		var text strings.Builder
+	if len(annotations) > 0 {
 
 		for key, value := range annotations {
 			if key != "definition" {
+
 				switch t := value.(type) {
 				case string:
-					text.WriteString(fmt.Sprintf("%s=%s; ", key, t))
+					buffer.WriteString(fmt.Sprintf("%s=%s; ", key, t))
 				case map[string]int,
 					map[string]string,
 					map[string]interface{},
@@ -318,16 +320,30 @@ func FormatFastSeqOBIHeader(sequence *obiseq.BioSequence) string {
 						log.Fatalf("Cannot convert %v value", value)
 					}
 					tv = bytes.ReplaceAll(tv, []byte(`"`), []byte("'"))
-					text.WriteString(fmt.Sprintf("%s=", key))
-					text.Write(tv)
-					text.WriteString("; ")
+					buffer.WriteString(fmt.Sprintf("%s=", key))
+					buffer.Write(tv)
+					buffer.WriteString("; ")
 				default:
-					text.WriteString(fmt.Sprintf("%s=%v; ", key, value))
+					buffer.WriteString(fmt.Sprintf("%s=%v; ", key, value))
 				}
 			}
 		}
 
-		return text.String() + " " + sequence.Definition()
+		if sequence.HasDefinition() {
+			buffer.WriteByte(' ')
+			buffer.WriteString(sequence.Definition())
+		}
+	}
+
+}
+
+func FormatFastSeqOBIHeader(sequence *obiseq.BioSequence) string {
+	annotations := sequence.Annotations()
+
+	if annotations != nil {
+		var text bytes.Buffer
+		WriteFastSeqOBIHeade(&text, sequence)
+		return unsafe.String(unsafe.SliceData(text.Bytes()), len(text.String()))
 	}
 
 	return ""

pkg/obiformats/fastseq_write_fasta.go

@@ -81,12 +81,16 @@ func FormatFastaBatch(batch obiiter.BioSequenceBatch, formater FormatHeader, ski
 	var bs bytes.Buffer
 
 	// Iterate over each sequence in the batch
-	for _, seq := range batch.Slice() {
+	for i, seq := range batch.Slice() {
 		// Check if the sequence is empty
 		if seq.Len() > 0 {
 			// Format the sequence using the provided formater function
 			formattedSeq := FormatFasta(seq, formater)
 
+			if i == 0 {
+				bs.Grow(len(formattedSeq) * len(batch.Slice()) * 5 / 4)
+			}
+
 			// Append the formatted sequence to the buffer
 			bs.WriteString(formattedSeq)
 			bs.WriteByte('\n')

pkg/obiformats/fastseq_write_fastq.go

@@ -2,7 +2,6 @@ package obiformats
 
 import (
 	"bytes"
-	"fmt"
 	"io"
 	"os"
 	"sync"
@@ -15,39 +14,52 @@ import (
 	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obiutils"
 )
 
-// The function FormatFastq takes a BioSequence object, a quality shift value, and a header formatter
-// function as input, and returns a formatted string in FASTQ format.
-func FormatFastq(seq *obiseq.BioSequence, formater FormatHeader) string {
-
-	q := seq.QualitiesString()
+func _formatFastq(buff *bytes.Buffer, seq *obiseq.BioSequence, formater FormatHeader) {
 
 	info := ""
 	if formater != nil {
 		info = formater(seq)
 	}
 
-	f := fmt.Sprintf("@%s %s\n%s\n+\n%s",
-		seq.Id(), info,
-		seq.String(),
-		q,
-	)
+	buff.WriteByte('@')
+	buff.WriteString(seq.Id())
+	buff.WriteByte(' ')
 
-	if f[0] != '@' {
-		log.Panicln("FormatFastq: FASTQ format error")
-	}
+	buff.WriteString(info)
+	buff.WriteByte('\n')
 
-	return f
+	buff.Write(seq.Sequence())
+	buff.WriteString("\n+\n")
+
+	q := seq.QualitiesString()
+	buff.WriteString(q)
+	buff.WriteByte('\n')
+
+}
+
+// The function FormatFastq takes a BioSequence object, a quality shift value, and a header formatter
+// function as input, and returns a formatted string in FASTQ format.
+func FormatFastq(seq *obiseq.BioSequence, formater FormatHeader) string {
+
+	var buff bytes.Buffer
+
+	_formatFastq(&buff, seq, formater)
+
+	return buff.String()
 }
 
 func FormatFastqBatch(batch obiiter.BioSequenceBatch,
 	formater FormatHeader, skipEmpty bool) []byte {
 	var bs bytes.Buffer
 
-	for _, seq := range batch.Slice() {
+	for i, seq := range batch.Slice() {
 		if seq.Len() > 0 {
-			fs := FormatFastq(seq, formater)
-			bs.WriteString(fs)
-			bs.WriteString("\n")
+			_formatFastq(&bs, seq, formater)
+
+			if i == 0 {
+
+				bs.Grow(len(bs.Bytes()) * len(batch.Slice()) * 5 / 4)
+			}
 		} else {
 			if skipEmpty {
 				log.Warnf("Sequence %s is empty and skiped in output", seq.Id())