Patch header parting and formatiing

pkg/obichunk/chunk_on_disk.go

@@ -81,13 +81,7 @@ func ISequenceChunkOnDisk(iterator obiiter.IBioSequenceBatch,
 				panic(err)
 			}
 
-			//chunck := make(obiseq.BioSequenceSlice, 0, 10000)
-			chunck := obiseq.MakeBioSequenceSlice()
-			for iseq.Next() {
-				b := iseq.Get()
-				chunck = append(chunck, b.Slice()...)
-				b.Recycle()
-			}
+			chunck := iseq.Load()
 
 			newIter.Push(obiiter.MakeBioSequenceBatch(order, chunck))
 			log.Infof("Start processing of batch %d/%d : %d sequences",

pkg/obiformats/fastseq_json_header.go

@@ -1,9 +1,11 @@
 package obiformats
 
 import (
-	log "github.com/sirupsen/logrus"
+	"math"
 	"strings"
 
+	log "github.com/sirupsen/logrus"
+
 	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
 	"github.com/goccy/go-json"
 )
@@ -42,6 +44,16 @@ func _parse_json_header_(header string, annotations obiseq.Annotation) string {
 	stop++
 
 	err := json.Unmarshal([]byte(header)[start:stop], &annotations)
+
+	for k, v := range annotations {
+		switch vt := v.(type) {
+		case float64 :
+			if vt == math.Floor(vt) {
+				annotations[k] = int(vt)
+			}
+		}
+	}
+	
 	if err != nil {
 		log.Fatalf("annotation parsing error on %s : %v\n", header, err)
 	}

pkg/obiformats/fastseq_obi_header.go

@@ -3,6 +3,8 @@ package obiformats
 import (
 	"bytes"
 	"fmt"
+	"log"
+	"math"
 	"regexp"
 	"strconv"
 	"strings"
@@ -251,9 +253,20 @@ func ParseOBIFeatures(text string, annotations obiseq.Annotation) string {
 			} // End of not string
 		} // End of not numeric
 
-		annotations[key] = value
+		switch vt := value.(type) {
+		case float64:
+			if vt == math.Floor(vt) {
+				annotations[key] = int(vt)
+			}
+		default:
+			annotations[key] = value
+		}
 
-		d = part[stop:]
+		if stop < len(part) {
+			d = part[stop:]
+		} else {
+			d = []byte{}
+		}
 		//m = __obi_header_key_pattern__.FindIndex(d)
 		m = __match__key__(d)
 	}
@@ -280,6 +293,16 @@ func FormatFastSeqOBIHeader(sequence *obiseq.BioSequence) string {
 			switch t := value.(type) {
 			case string:
 				text.WriteString(fmt.Sprintf("%s=%s; ", key, t))
+			case map[string]int,
+				map[string]interface{}:
+				tv, err := json.Marshal(t)
+				if err != nil {
+					log.Fatalf("Cannot convert %v value", value)
+				}
+				tv = bytes.ReplaceAll(tv, []byte(`"`), []byte("'"))
+				text.WriteString(fmt.Sprintf("%s=", key))
+				text.Write(tv)
+				text.WriteString("; ")
 			default:
 				text.WriteString(fmt.Sprintf("%s=%v; ", key, value))
 			}

pkg/obiiter/batchiterator.go

@@ -612,3 +612,15 @@ func (iterator IBioSequenceBatch) FilterOn(predicate obiseq.SequencePredicate,
 
 	return trueIter.Rebatch(size)
 }
+
+func (iterator IBioSequenceBatch) Load() obiseq.BioSequenceSlice {
+
+	chunck := obiseq.MakeBioSequenceSlice()
+	for iterator.Next() {
+		b := iterator.Get()
+		chunck = append(chunck, b.Slice()...)
+		b.Recycle()
+	}
+
+	return chunck
+}

pkg/obiseq/biosequence.go

@@ -257,3 +257,4 @@ func (s *BioSequence) WriteByte(data byte) error {
 	s.sequence = append(s.sequence, data)
 	return nil
 }
+

pkg/obiseq/merge.go

@@ -2,8 +2,10 @@ package obiseq
 
 import (
 	"fmt"
-	log "github.com/sirupsen/logrus"
 	"strings"
+
+	"git.metabarcoding.org/lecasofts/go/obitools/pkg/goutils"
+	log "github.com/sirupsen/logrus"
 )
 
 type StatsOnValues map[string]int
@@ -33,6 +35,16 @@ func (sequence *BioSequence) StatsOn(key string, na string) StatsOnValues {
 		case StatsOnValues:
 			stats = istat
 			newstat = false
+		case map[string]interface{}:
+			stats = make(StatsOnValues, len(istat))
+			var err error
+			for k, v := range istat {
+				stats[k], err = goutils.InterfaceToInt(v)
+				if err != nil {
+					log.Panicf("In sequence %s : %s stat tag not only containing integer values %s",
+						sequence.Id(), mkey, istat)
+				}
+			}
 		default:
 			stats = make(StatsOnValues, 100)
 			annotations[mkey] = stats