Refactoring of some code

pkg/obitax/issuubcladeof.go

@@ -1,11 +1,5 @@
 package obitax
 
-import (
-	log "github.com/sirupsen/logrus"
-
-	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
-)
-
 func (taxon *TaxNode) IsSubCladeOf(parent *TaxNode) bool {
 
 	for taxon.taxid != parent.taxid && taxon.parent != taxon.taxid {
@@ -25,18 +19,3 @@ func (taxon *TaxNode) IsBelongingSubclades(clades *TaxonSet) bool {
 
 	return ok
 }
-
-func IsSubCladeOf(taxonomy Taxonomy, taxid int) obiseq.SequencePredicate {
-	parent, err := taxonomy.Taxon(taxid)
-
-	if err != nil {
-		log.Fatalf("Cannot find taxon : %d (%v)", taxid, err)
-	}
-
-	f := func(sequence *obiseq.BioSequence) bool {
-		taxon, err := taxonomy.Taxon(sequence.Taxid())
-		return err == nil && taxon.IsSubCladeOf(parent)
-	}
-
-	return f
-}

pkg/obitax/sequence_predicate.go

@@ -0,0 +1,71 @@
+package obitax
+
+import (
+	log "github.com/sirupsen/logrus"
+
+	"git.metabarcoding.org/lecasofts/go/obitools/pkg/goutils"
+	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
+)
+
+func (taxonomy *Taxonomy) IsAValidTaxon(withAutoCorrection ...bool) obiseq.SequencePredicate {
+	deprecatedTaxidsWarning := make(map[int]bool)
+
+	autocorrection := false
+	if len(withAutoCorrection) > 0 {
+		autocorrection = withAutoCorrection[0]
+	}
+
+	f := func(sequence *obiseq.BioSequence) bool {
+		taxid := sequence.Taxid()
+		taxon, err := taxonomy.Taxon(taxid)
+
+		if err == nil && taxon.taxid != taxid {
+			if autocorrection {
+				sequence.SetTaxid(taxon.taxid)
+				log.Printf("Sequence %s : Taxid %d updated with %d", taxid, taxon.taxid)
+			} else {
+				if _, ok := deprecatedTaxidsWarning[taxid]; !ok {
+					deprecatedTaxidsWarning[taxid] = true
+					log.Printf("Taxid %d is deprecated and must be replaced by %d", taxid, taxon.taxid)
+				}
+			}
+		}
+
+		return err == nil
+	}
+
+	return f
+}
+
+// A function that takes a taxonomy and a taxid as arguments and returns a function that takes a
+// pointer to a BioSequence as an argument and returns a boolean.
+func (taxonomy *Taxonomy) IsSubCladeOf(taxid int) obiseq.SequencePredicate {
+	parent, err := taxonomy.Taxon(taxid)
+
+	if err != nil {
+		log.Fatalf("Cannot find taxon : %d (%v)", taxid, err)
+	}
+
+	f := func(sequence *obiseq.BioSequence) bool {
+		taxon, err := taxonomy.Taxon(sequence.Taxid())
+		return err == nil && taxon.IsSubCladeOf(parent)
+	}
+
+	return f
+}
+
+func (taxonomy *Taxonomy) HasRequiredRank(rank string) obiseq.SequencePredicate {
+
+	if !goutils.Contains(taxonomy.RankList(), rank) {
+		log.Fatalf("%s is not a valid rank (allowed ranks are %v)",
+			rank,
+			taxonomy.RankList())
+	}
+
+	f := func(sequence *obiseq.BioSequence) bool {
+		taxon, err := taxonomy.Taxon(sequence.Taxid())
+		return err == nil && taxon.HasRankDefined(rank)
+	}
+
+	return f
+}

pkg/obitax/taxon.go

@@ -2,8 +2,6 @@ package obitax
 
 import (
 	"regexp"
-
-	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
 )
 
 type TaxNode struct {
@@ -72,17 +70,6 @@ func (node *TaxNode) HasRankDefined(rank string) bool {
 	for node.rank != rank && node.parent != node.taxid {
 		node = node.pparent
 	}
-
 	return node.rank == rank
 
 }
-
-func HasRankDefined(taxonomy Taxonomy, rank string) obiseq.SequencePredicate {
-
-	f := func(sequence *obiseq.BioSequence) bool {
-		taxon, err := taxonomy.Taxon(sequence.Taxid())
-		return err == nil && taxon.HasRankDefined(rank)
-	}
-
-	return f
-}

pkg/obitax/taxonomy.go

@@ -2,7 +2,6 @@ package obitax
 
 import (
 	"fmt"
-	log "github.com/sirupsen/logrus"
 )
 
 type TaxName struct {
@@ -63,7 +62,6 @@ func (taxonomy *Taxonomy) Taxon(taxid int) (*TaxNode, error) {
 		if !aok {
 			return nil, fmt.Errorf("Taxid %d is not part of the taxonomy", taxid)
 		}
-		log.Printf("Taxid %d is deprecated and must be replaced by %d", taxid, a.taxid)
 		t = a
 	}
 	return t, nil