obicsv debug

pkg/obioptions/version.go

@@ -7,7 +7,7 @@ import (
 // TODO: The version number is extracted from git. This induces that the version
 // corresponds to the last commit, and not the one when the file will be
 // commited
-var _Commit = "36327c7"
+var _Commit = "3d06978"
 var _Version = "Release 4.2.0"
 
 // Version returns the version of the obitools package.

pkg/obitax/taxonomy.go

@@ -141,6 +141,9 @@ func (taxonomy *Taxonomy) TaxidSting(id string) (string, error) {
 //   - If the taxid is unknown, the method will log a fatal error.
 func (taxonomy *Taxonomy) Taxon(taxid string) *Taxon {
 	taxonomy = taxonomy.OrDefault(false)
+	if taxonomy == nil {
+		return nil
+	}
 
 	id, err := taxonomy.Id(taxid)
 

pkg/obitools/obicsv/iter.go

@@ -2,6 +2,7 @@ package obicsv
 
 import (
 	"fmt"
+	"slices"
 	"sync"
 	"sync/atomic"
 	"time"
@@ -222,7 +223,7 @@ func (iterator *ICSVRecord) SetHeader(header CSVHeader) {
 }
 
 func (iterator *ICSVRecord) AppendField(field string) {
-	iterator.header = append(iterator.header, field)
+	iterator.header.AppendField(field)
 }
 
 func (iterator *ICSVRecord) Next() bool {
@@ -339,3 +340,9 @@ func (iterator *ICSVRecord) Consume() {
 		iterator.Get()
 	}
 }
+
+func (head *CSVHeader) AppendField(field string) {
+	if !slices.Contains(*head, field) {
+		*head = append(*head, field)
+	}
+}

pkg/obitools/obicsv/options.go

@@ -64,6 +64,7 @@ func CSVOptionSet(options *getoptions.GetOpt) {
 }
 
 func OptionSet(options *getoptions.GetOpt) {
+	obiconvert.InputOptionSet(options)
 	obiconvert.OutputModeOptionSet(options)
 	CSVOptionSet(options)
 }

pkg/obitools/obicsv/sequence.go

@@ -1,40 +1,45 @@
 package obicsv
 
 import (
+	"log"
+	"slices"
+
 	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obiiter"
 	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obioptions"
-
-	log "github.com/sirupsen/logrus"
 )
 
 func CSVSequenceHeader(opt Options) CSVHeader {
 	keys := opt.CSVKeys()
-	record := make([]string, 0, len(keys)+4)
+	record := make(CSVHeader, 0, len(keys)+4)
 
 	if opt.CSVId() {
-		record = append(record, "id")
+		record.AppendField("id")
 	}
 
 	if opt.CSVCount() {
-		record = append(record, "count")
+		record.AppendField("count")
 	}
 
 	if opt.CSVTaxon() {
-		record = append(record, "taxid")
+		record.AppendField("taxid")
 	}
 
 	if opt.CSVDefinition() {
-		record = append(record, "definition")
+		record.AppendField("definition")
 	}
 
-	record = append(record, opt.CSVKeys()...)
+	for _, field := range opt.CSVKeys() {
+		if field != "definition" {
+			record.AppendField(field)
+		}
+	}
 
 	if opt.CSVSequence() {
-		record = append(record, "sequence")
+		record.AppendField("sequence")
 	}
 
 	if opt.CSVQuality() {
-		record = append(record, "quality")
+		record.AppendField("quality")
 	}
 
 	return record
@@ -110,11 +115,22 @@ func NewCSVSequenceIterator(iter obiiter.IBioSequence, options ...WithOption) *I
 
 	opt := MakeOptions(options)
 
+	if opt.CSVAutoColumn() {
+		if iter.Next() {
+			batch := iter.Get()
+			if len(batch.Slice()) == 0 {
+				log.Panicf("first batch should not be empty")
+			}
+			auto_slot := batch.Slice().AttributeKeys(true).Members()
+			slices.Sort(auto_slot)
+			CSVKeys(auto_slot)(opt)
+			iter.PushBack()
+		}
+	}
+
 	newIter := NewICSVRecord()
 	newIter.SetHeader(CSVSequenceHeader(opt))
 
-	log.Warnf("", newIter.Header())
-
 	nwriters := opt.ParallelWorkers()
 	newIter.Add(nwriters)
 

pkg/obitools/obicsv/writer.go

@@ -16,8 +16,6 @@ func FormatCVSBatch(batch CSVRecordBatch, header CSVHeader, navalue string) *byt
 	buff := new(bytes.Buffer)
 	csv := csv.NewWriter(buff)
 
-	log.Warn("Header:", header)
-
 	if batch.Order() == 0 {
 		csv.Write(header)
 	}