From bc82422bc5362407a3560bf7f392e29e4d2bf9bb Mon Sep 17 00:00:00 2001
From: Eric Coissac <eric.coissac@metabarcoding.org>
Date: Tue, 28 Mar 2023 11:43:04 +0700
Subject: [PATCH] Reduce redundante call to bytes.ToLower and substitute the
 last call by an home made version doing the conversion in place

Former-commit-id: d9ea22f649d97be352f8dbb37acc1495df830118
---
 pkg/obiformats/ecopcr_read.go  |  3 +--
 pkg/obiformats/embl_read.go    |  4 ++--
 pkg/obiformats/fastseq_read.go |  3 +--
 pkg/obiformats/genbank_read.go |  2 +-
 pkg/obiseq/biosequence.go      |  5 ++---
 pkg/obiutils/bytes.go          | 11 +++++++++++
 6 files changed, 18 insertions(+), 10 deletions(-)
 create mode 100644 pkg/obiutils/bytes.go

diff --git a/pkg/obiformats/ecopcr_read.go b/pkg/obiformats/ecopcr_read.go
index cf5be7d..96b68b7 100644
--- a/pkg/obiformats/ecopcr_read.go
+++ b/pkg/obiformats/ecopcr_read.go
@@ -1,7 +1,6 @@
 package obiformats
 
 import (
-	"bytes"
 	"compress/gzip"
 	"encoding/csv"
 	"fmt"
@@ -70,7 +69,7 @@ func __read_ecopcr_bioseq__(file *__ecopcr_file__) (*obiseq.BioSequence, error)
 		comment = strings.TrimSpace(record[19])
 	}
 
-	bseq := obiseq.NewBioSequence(name, bytes.ToLower(sequence), comment)
+	bseq := obiseq.NewBioSequence(name, sequence, comment)
 	annotation := bseq.Annotations()
 
 	annotation["ac"] = name
diff --git a/pkg/obiformats/embl_read.go b/pkg/obiformats/embl_read.go
index a0f81ec..0ff9a6a 100644
--- a/pkg/obiformats/embl_read.go
+++ b/pkg/obiformats/embl_read.go
@@ -141,10 +141,10 @@ func _ParseEmblFile(source string, input <-chan _FileChunk, out obiiter.IBioSequ
 				}
 			case line == "//":
 				sequence := obiseq.NewBioSequence(id,
-					bytes.ToLower(seqBytes.Bytes()),
+					seqBytes.Bytes(),
 					defBytes.String())
 				sequence.SetSource(source)
-				
+
 				sequence.SetFeatures(featBytes.Bytes())
 
 				annot := sequence.Annotations()
diff --git a/pkg/obiformats/fastseq_read.go b/pkg/obiformats/fastseq_read.go
index e73db30..e18a3b5 100644
--- a/pkg/obiformats/fastseq_read.go
+++ b/pkg/obiformats/fastseq_read.go
@@ -7,7 +7,6 @@ package obiformats
 import "C"
 
 import (
-	"bytes"
 	"fmt"
 	"os"
 	"path"
@@ -42,7 +41,7 @@ func _FastseqReader(source string,
 			comment = ""
 		}
 
-		rep := obiseq.NewBioSequence(name, bytes.ToLower(sequence), comment)
+		rep := obiseq.NewBioSequence(name, sequence, comment)
 		rep.SetSource(source)
 		if s.qual.l > C.ulong(0) {
 			cquality := unsafe.Slice(s.qual.s, C.int(s.qual.l))
diff --git a/pkg/obiformats/genbank_read.go b/pkg/obiformats/genbank_read.go
index 1ca1f59..195c974 100644
--- a/pkg/obiformats/genbank_read.go
+++ b/pkg/obiformats/genbank_read.go
@@ -69,7 +69,7 @@ func _ParseGenbankFile(source string,
 			case line == "//":
 				log.Debugln("Total lines := ", nl)
 				sequence := obiseq.NewBioSequence(id,
-					bytes.ToLower(seqBytes.Bytes()),
+					seqBytes.Bytes(),
 					defBytes.String())
 				sequence.SetSource(source)
 				state = inHeader
diff --git a/pkg/obiseq/biosequence.go b/pkg/obiseq/biosequence.go
index 99b1e4c..e38b685 100644
--- a/pkg/obiseq/biosequence.go
+++ b/pkg/obiseq/biosequence.go
@@ -11,10 +11,10 @@
 package obiseq
 
 import (
-	"bytes"
 	"crypto/md5"
 	"sync/atomic"
 
+	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiutils"
 	log "github.com/sirupsen/logrus"
 )
 
@@ -210,7 +210,6 @@ func (s *BioSequence) Source() string {
 	return s.source
 }
 
-
 // Returning the MD5 hash of the sequence.
 func (s *BioSequence) MD5() [16]byte {
 	return md5.Sum(s.sequence)
@@ -244,7 +243,7 @@ func (s *BioSequence) SetSequence(sequence []byte) {
 	if s.sequence != nil {
 		RecycleSlice(&s.sequence)
 	}
-	s.sequence = bytes.ToLower(sequence)
+	s.sequence = obiutils.InPlaceToLower(sequence)
 }
 
 // Setting the qualities of the BioSequence.
diff --git a/pkg/obiutils/bytes.go b/pkg/obiutils/bytes.go
new file mode 100644
index 0000000..b17d226
--- /dev/null
+++ b/pkg/obiutils/bytes.go
@@ -0,0 +1,11 @@
+package obiutils
+
+func InPlaceToLower(data []byte) []byte {
+	for i,l := range data {
+		if l >= 'A' && l <='Z' {
+			data[i]|=32
+		}
+	}
+
+	return data
+}
\ No newline at end of file