Add managment of the taxonomy alias politic

cmd/obitools/obitaxonomy/main.go

@@ -3,6 +3,7 @@ package main
 import (
 	"os"
 
+	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obidefault"
 	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obioptions"
 	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obitax"
 	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obitools/obiconvert"
@@ -52,13 +53,19 @@ func main() {
 
 	case obitaxonomy.CLIRequestsPathForTaxid() != "NA":
 
-		taxon, err := obitax.DefaultTaxonomy().Taxon(obitaxonomy.CLIRequestsPathForTaxid())
+		taxon, isAlias, err := obitax.DefaultTaxonomy().Taxon(obitaxonomy.CLIRequestsPathForTaxid())
 
 		if err != nil {
 			log.Fatalf("Cannot identify the requested taxon: %s (%v)",
 				obitaxonomy.CLIRequestsPathForTaxid(), err)
 		}
 
+		if isAlias {
+			if obidefault.FailOnTaxonomy() {
+				log.Fatalf("Taxon %s is an alias for %s", taxon.String(), taxon.Parent().String())
+			}
+		}
+
 		s := taxon.Path()
 
 		if s == nil {

pkg/obidefault/taxonomy.go

@@ -2,6 +2,8 @@ package obidefault
 
 var __taxonomy__ = ""
 var __alternative_name__ = false
+var __fail_on_taxonomy__ = false
+var __update_taxid__ = false
 
 func SelectedTaxonomy() string {
 	return __taxonomy__
@@ -30,3 +32,27 @@ func SetSelectedTaxonomy(taxonomy string) {
 func SetAlternativeNamesSelected(alt bool) {
 	__alternative_name__ = alt
 }
+
+func SetFailOnTaxonomy(fail bool) {
+	__fail_on_taxonomy__ = fail
+}
+
+func SetUpdateTaxid(update bool) {
+	__update_taxid__ = update
+}
+
+func FailOnTaxonomyPtr() *bool {
+	return &__fail_on_taxonomy__
+}
+
+func UpdateTaxidPtr() *bool {
+	return &__update_taxid__
+}
+
+func FailOnTaxonomy() bool {
+	return __fail_on_taxonomy__
+}
+
+func UpdateTaxid() bool {
+	return __update_taxid__
+}

pkg/obiformats/fastseq_json_header.go

@@ -9,7 +9,6 @@ import (
 	log "github.com/sirupsen/logrus"
 
 	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obiseq"
-	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obitax"
 	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obiutils"
 	"github.com/buger/jsonparser"
 )
@@ -201,8 +200,6 @@ func _parse_json_array_interface(str []byte, sequence *obiseq.BioSequence) ([]in
 }
 
 func _parse_json_header_(header string, sequence *obiseq.BioSequence) string {
-	taxonomy := obitax.DefaultTaxonomy()
-
 	annotations := sequence.Annotations()
 	start := -1
 	stop := -1
@@ -291,13 +288,8 @@ func _parse_json_header_(header string, sequence *obiseq.BioSequence) string {
 
 			case skey == "taxid":
 				if dataType == jsonparser.Number || dataType == jsonparser.String {
-					taxid := obiutils.UnsafeString(value)
+					taxid := string(value)
-					taxon, err := taxonomy.Taxon(taxid)
+					sequence.SetTaxid(taxid)
-					if err == nil {
-						sequence.SetTaxon(taxon)
-					} else {
-						sequence.SetTaxid(string(value))
-					}
 				} else {
 					log.Fatalf("%s: Cannot parse taxid %s", sequence.Id(), string(value))
 				}
@@ -306,15 +298,7 @@ func _parse_json_header_(header string, sequence *obiseq.BioSequence) string {
 				if dataType == jsonparser.Number || dataType == jsonparser.String {
 					rank, _ := obiutils.SplitInTwo(skey, '_')
 
-					taxid := obiutils.UnsafeString(value)
+					taxid := string(value)
-					taxon, err := taxonomy.Taxon(taxid)
-
-					if err == nil {
-						taxid = taxon.String()
-					} else {
-						taxid = string(value)
-					}
-
 					sequence.SetTaxid(taxid, rank)
 				} else {
 					log.Fatalf("%s: Cannot parse taxid %s", sequence.Id(), string(value))

pkg/obilua/obitaxonomy.go

@@ -1,6 +1,7 @@
 package obilua
 
 import (
+	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obidefault"
 	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obitax"
 	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obiutils"
 	lua "github.com/yuin/gopher-lua"
@@ -98,13 +99,18 @@ func taxonomyGetCode(luaState *lua.LState) int {
 func taxonomyGetTaxon(luaState *lua.LState) int {
 	taxo := checkTaxonomy(luaState)
 	taxid := luaState.CheckString(2)
-	taxon, err := taxo.Taxon(taxid)
+	taxon, isAlias, err := taxo.Taxon(taxid)
 
 	if err != nil {
 		luaState.RaiseError("%s : Error on taxon taxon: %v", taxid, err)
 		return 0
 	}
 
+	if isAlias && obidefault.FailOnTaxonomy() {
+		luaState.RaiseError("%s : Taxon is an alias of %s", taxid, taxon.String())
+		return 0
+	}
+
 	luaState.Push(taxon2Lua(luaState, taxon))
 	return 1
 }

pkg/obioptions/options.go

@@ -177,6 +177,15 @@ func LoadTaxonomyOptionSet(options *getoptions.GetOpt, required, alternatiive bo
 			options.Alias("a"),
 			options.Description("Enable the search on all alternative names and not only scientific names."))
 	}
+
+	options.BoolVar(obidefault.FailOnTaxonomyPtr(), "fail-on-taxonomy",
+		obidefault.FailOnTaxonomy(),
+		options.Description("Make obitools failing on error if a used taxid is not a currently valid one"),
+	)
+
+	options.BoolVar(obidefault.UpdateTaxidPtr(), "update-taxid", obidefault.UpdateTaxid(),
+		options.Description("Make obitools automatically updating the taxid that are declared merged to a newest one."),
+	)
 }
 
 // CLIIsDebugMode returns whether the CLI is in debug mode.

pkg/obioptions/version.go

@@ -8,7 +8,7 @@ import (
 // corresponds to the last commit, and not the one when the file will be
 // commited
 
-var _Commit = "f2e81ad"
+var _Commit = "e2563cd"
 var _Version = "Release 4.2.0"
 
 // Version returns the version of the obitools package.

pkg/obiseq/taxonomy_lca.go

@@ -4,6 +4,7 @@ import (
 	"math"
 	"strings"
 
+	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obidefault"
 	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obitax"
 	log "github.com/sirupsen/logrus"
 )
@@ -15,7 +16,7 @@ func (sequence *BioSequence) TaxonomicDistribution(taxonomy *obitax.Taxonomy) ma
 	taxonomy = taxonomy.OrDefault(true)
 
 	for taxid, v := range taxids {
-		t, err := taxonomy.Taxon(taxid)
+		t, isAlias, err := taxonomy.Taxon(taxid)
 		if err != nil {
 			log.Fatalf(
 				"On sequence %s taxid %s is not defined in taxonomy: %s (%v)",
@@ -25,6 +26,11 @@ func (sequence *BioSequence) TaxonomicDistribution(taxonomy *obitax.Taxonomy) ma
 				err,
 			)
 		}
+
+		if isAlias && obidefault.FailOnTaxonomy() {
+			log.Fatalf("On sequence %s taxid %s is an alias on %s",
+				sequence.Id(), taxid, t.String())
+		}
 		taxons[t.Node] = v
 	}
 	return taxons

pkg/obiseq/taxonomy_methods.go

@@ -5,6 +5,7 @@ import (
 
 	log "github.com/sirupsen/logrus"
 
+	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obidefault"
 	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obitax"
 	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obiutils"
 )
@@ -16,7 +17,7 @@ func (s *BioSequence) Taxon(taxonomy *obitax.Taxonomy) *obitax.Taxon {
 		return nil
 	}
 
-	taxon, _ := taxonomy.Taxon(taxid)
+	taxon, _, _ := taxonomy.Taxon(taxid)
 
 	return taxon
 }
@@ -28,6 +29,8 @@ func (s *BioSequence) Taxon(taxonomy *obitax.Taxonomy) *obitax.Taxon {
 //	taxid - the taxid to set.
 func (s *BioSequence) SetTaxid(taxid string, rank ...string) {
 	var err error
+	var isAlias bool
+
 	if taxid == "" {
 		taxid = "NA"
 	} else {
@@ -35,16 +38,39 @@ func (s *BioSequence) SetTaxid(taxid string, rank ...string) {
 		taxon := (*obitax.Taxon)(nil)
 
 		if taxonomy != nil {
-			taxon, err = taxonomy.Taxon(taxid)
+			taxon, isAlias, err = taxonomy.Taxon(taxid)
 
 			if err != nil {
-				log.Warnf("%s: Taxid: %v is unknown from taxonomy (%v)",
+				if obidefault.FailOnTaxonomy() {
-					s.Id(), taxid, err)
+					log.Fatalf("%s: Taxid: %v is unknown from taxonomy (%v)",
+						s.Id(), taxid, err)
+				} else {
+					log.Warnf("%s: Taxid: %v is unknown from taxonomy (%v)",
+						s.Id(), taxid, err)
+				}
+			}
+
+			if isAlias {
+				if obidefault.FailOnTaxonomy() {
+					log.Fatalf("%s: Taxid: %v is an alias from taxonomy (%v) to %s",
+						s.Id(), taxid, taxonomy.Name(), taxon.String())
+				} else {
+					if obidefault.UpdateTaxid() {
+						log.Warnf("%s: Taxid: %v is updated to %s",
+							s.Id(), taxid, taxon.String())
+						taxid = taxon.String()
+					} else {
+						log.Warnf("%s: Taxid %v has to be updated to %s",
+							s.Id(), taxid, taxon.String())
+					}
+				}
+
+			} else {
+				if taxon != nil {
+					taxid = taxon.String()
+				}
 			}
-		}
 
-		if taxon != nil {
-			taxid = taxon.String()
 		}
 	}
 

pkg/obiseq/taxonomy_predicate.go

@@ -63,7 +63,7 @@ func IsSubCladeOfSlot(taxonomy *obitax.Taxonomy, key string) SequencePredicate {
 		val, ok := sequence.GetStringAttribute(key)
 
 		if ok {
-			parent, err := taxonomy.Taxon(val)
+			parent, _, err := taxonomy.Taxon(val)
 
 			if err != nil {
 				log.Warnf("%s: %s is unkown from the taxonomy (%v)", sequence.Id(), val, err)

pkg/obitax/iterator.go

@@ -216,7 +216,7 @@ func (taxon *Taxon) ISubTaxonomy() *ITaxon {
 }
 
 func (taxonomy *Taxonomy) ISubTaxonomy(taxid string) *ITaxon {
-	taxon, err := taxonomy.Taxon(taxid)
+	taxon, _, err := taxonomy.Taxon(taxid)
 
 	if err != nil {
 		return nil

pkg/obitax/ncbitaxdump_read.go

@@ -91,7 +91,7 @@ func loadNameTable(reader io.Reader, taxonomy *Taxonomy, onlysn bool) int {
 
 		if !onlysn || classname == "scientific name" {
 			n++
-			taxon, err := taxonomy.Taxon(taxid)
+			taxon, _, err := taxonomy.Taxon(taxid)
 
 			if err != nil {
 				log.Fatalf("%s: is unknown from the taxonomy", taxid)
@@ -202,7 +202,7 @@ func LoadNCBITaxDump(directory string, onlysn bool) (*Taxonomy, error) {
 	n = loadMergedTable(buffered, taxonomy)
 	log.Printf("%d merged taxa read\n", n)
 
-	root, err := taxonomy.Taxon("1")
+	root, _, err := taxonomy.Taxon("1")
 
 	if err != nil {
 		log.Fatal("cannot find the root taxon (1) in the NCBI tax dump")

pkg/obitax/ncbitaxdump_readtar.go

@@ -134,7 +134,7 @@ func LoadNCBITarTaxDump(path string, onlysn bool) (*Taxonomy, error) {
 	n = loadMergedTable(buffered, taxonomy)
 	log.Printf("%d merged taxa read\n", n)
 
-	root, err := taxonomy.Taxon("1")
+	root, _, err := taxonomy.Taxon("1")
 
 	if err != nil {
 		log.Fatal("cannot find the root taxon (1) in the NCBI tax dump")

pkg/obitax/taxonomy.go

@@ -129,28 +129,30 @@ func (taxonomy *Taxonomy) TaxidString(id string) (string, error) {
 // Returns:
 //   - A pointer to the Taxon instance associated with the provided taxid.
 //   - If the taxid is unknown, the method will log a fatal error.
-func (taxonomy *Taxonomy) Taxon(taxid string) (*Taxon, error) {
+func (taxonomy *Taxonomy) Taxon(taxid string) (*Taxon, bool, error) {
 	taxonomy = taxonomy.OrDefault(false)
 	if taxonomy == nil {
-		return nil, errors.New("cannot extract taxon from nil taxonomy")
+		return nil, false, errors.New("cannot extract taxon from nil taxonomy")
 	}
 
 	id, err := taxonomy.Id(taxid)
 
 	if err != nil {
-		return nil, fmt.Errorf("Taxid %s: %v", taxid, err)
+		return nil, false, fmt.Errorf("Taxid %s: %v", taxid, err)
 	}
 
 	taxon := taxonomy.nodes.Get(id)
+	isAlias := taxon.Node.id != id
 
 	if taxon == nil {
 		return nil,
+			false,
 			fmt.Errorf("Taxid %s is not part of the taxonomy %s",
 				taxid,
 				taxonomy.name)
 	}
 
-	return taxon, nil
+	return taxon, isAlias, nil
 }
 
 // AsTaxonSet returns the set of taxon nodes contained within the Taxonomy.
@@ -385,7 +387,7 @@ func (taxonomy *Taxonomy) InsertPathString(path []string) (*Taxonomy, error) {
 	}
 
 	var current *Taxon
-	current, err = taxonomy.Taxon(taxid)
+	current, _, err = taxonomy.Taxon(taxid)
 
 	if err != nil {
 		return nil, err
@@ -396,7 +398,7 @@ func (taxonomy *Taxonomy) InsertPathString(path []string) (*Taxonomy, error) {
 	}
 
 	for _, id := range path[1:] {
-		taxon, err := taxonomy.Taxon(id)
+		taxon, _, err := taxonomy.Taxon(id)
 		if err == nil {
 			if !current.SameAs(taxon.Parent()) {
 				return nil, errors.New("path is not consistent with the taxonomy, parent mismatch")

pkg/obitools/obigrep/options.go

@@ -248,14 +248,14 @@ func CLIRestrictTaxonomyPredicate() obiseq.SequencePredicate {
 	if len(_BelongTaxa) > 0 {
 		taxonomy := obitax.DefaultTaxonomy()
 
-		taxon, err := taxonomy.Taxon(_BelongTaxa[0])
+		taxon, _, err := taxonomy.Taxon(_BelongTaxa[0])
 		if err != nil {
 			p = obiseq.IsSubCladeOfSlot(taxonomy, _BelongTaxa[0])
 		} else {
 			p = obiseq.IsSubCladeOf(taxonomy, taxon)
 		}
 		for _, staxid := range _BelongTaxa[1:] {
-			taxon, err := taxonomy.Taxon(staxid)
+			taxon, _, err := taxonomy.Taxon(staxid)
 			if err != nil {
 				p2 = obiseq.IsSubCladeOfSlot(taxonomy, staxid)
 			} else {
@@ -278,7 +278,7 @@ func CLIAvoidTaxonomyPredicate() obiseq.SequencePredicate {
 	if len(_NotBelongTaxa) > 0 {
 		taxonomy := obitax.DefaultTaxonomy()
 
-		taxon, err := taxonomy.Taxon(_NotBelongTaxa[0])
+		taxon, _, err := taxonomy.Taxon(_NotBelongTaxa[0])
 		if err != nil {
 			p = obiseq.IsSubCladeOfSlot(taxonomy, _NotBelongTaxa[0])
 		} else {
@@ -286,7 +286,7 @@ func CLIAvoidTaxonomyPredicate() obiseq.SequencePredicate {
 		}
 
 		for _, taxid := range _NotBelongTaxa[1:] {
-			taxon, err := taxonomy.Taxon(taxid)
+			taxon, _, err := taxonomy.Taxon(taxid)
 			if err != nil {
 				p2 = obiseq.IsSubCladeOfSlot(taxonomy, taxid)
 			} else {

pkg/obitools/obitag/obitag.go

@@ -43,7 +43,7 @@ func MatchDistanceIndex(taxonomy *obitax.Taxonomy, distance int, distanceIdx map
 		taxon = taxonomy.Root()
 	} else {
 		var err error
-		taxon, err = taxonomy.Taxon(distanceIdx[keys[i]])
+		taxon, _, err = taxonomy.Taxon(distanceIdx[keys[i]])
 		if err != nil {
 			log.Panicf("Cannot identify taxon %s in %s (%v)", distanceIdx[keys[i]], taxonomy.Name(), err)
 		}
@@ -197,7 +197,7 @@ func Identify(sequence *obiseq.BioSequence,
 				log.Panic("Problem in identification line : ", best.Id(), "idx:", idx, "distance:", d)
 			}
 
-			match_taxon, err := taxo.Taxon(identification)
+			match_taxon, _, err := taxo.Taxon(identification)
 
 			if err == nil {
 				taxon, _ = taxon.LCA(match_taxon)

pkg/obitools/obitaxonomy/options.go

@@ -91,7 +91,7 @@ func CLITaxonomicalRestrictions() (*obitax.TaxonSet, error) {
 
 	ts := taxonomy.NewTaxonSet()
 	for _, taxid := range __taxonomical_restriction__ {
-		tx, err := taxonomy.Taxon(taxid)
+		tx, _, err := taxonomy.Taxon(taxid)
 
 		if err != nil {
 			return nil, fmt.Errorf(