Adds possibility to extract a taxonomy from taxonomic path included in sequence files

pkg/obitax/iterator.go

@@ -218,9 +218,9 @@ func (taxon *Taxon) ISubTaxonomy() *ITaxon {
 }
 
 func (taxonomy *Taxonomy) ISubTaxonomy(taxid string) *ITaxon {
-	taxon := taxonomy.Taxon(taxid)
+	taxon, err := taxonomy.Taxon(taxid)
 
-	if taxon == nil {
+	if err != nil {
 		return nil
 	}
 

pkg/obitax/ncbitaxdump_read.go

@@ -91,7 +91,13 @@ func loadNameTable(reader io.Reader, taxonomy *Taxonomy, onlysn bool) int {
 
 		if !onlysn || classname == "scientific name" {
 			n++
-			taxonomy.Taxon(taxid).SetName(name, classname)
+			taxon, err := taxonomy.Taxon(taxid)
+
+			if err != nil {
+				log.Fatalf("%s: is unknown from the taxonomy", taxid)
+			}
+
+			taxon.SetName(name, classname)
 		}
 	}
 
@@ -196,7 +202,11 @@ func LoadNCBITaxDump(directory string, onlysn bool) (*Taxonomy, error) {
 	n = loadMergedTable(buffered, taxonomy)
 	log.Printf("%d merged taxa read\n", n)
 
-	root := taxonomy.Taxon("1")
+	root, err := taxonomy.Taxon("1")
+
+	if err != nil {
+		log.Fatal("cannot find the root taxon (1) in the NCBI tax dump")
+	}
 	taxonomy.SetRoot(root)
 
 	return taxonomy, nil

pkg/obitax/ncbitaxdump_readtar.go

@@ -134,7 +134,12 @@ func LoadNCBITarTaxDump(path string, onlysn bool) (*Taxonomy, error) {
 	n = loadMergedTable(buffered, taxonomy)
 	log.Printf("%d merged taxa read\n", n)
 
-	root := taxonomy.Taxon("1")
+	root, err := taxonomy.Taxon("1")
+
+	if err != nil {
+		log.Fatal("cannot find the root taxon (1) in the NCBI tax dump")
+	}
+
 	taxonomy.SetRoot(root)
 
 	return taxonomy, nil

pkg/obitax/string_parser.go

@@ -0,0 +1,64 @@
+package obitax
+
+import (
+	"errors"
+	"strings"
+)
+
+// ParseTaxonString parses a string in the format "code:taxid [scientific name]@rank"
+// and returns the individual components. It handles extra whitespace around components.
+//
+// Parameters:
+//   - taxonStr: The string to parse in the format "code:taxid [scientific name]@rank"
+//
+// Returns:
+//   - code: The taxonomy code
+//   - taxid: The taxon identifier
+//   - scientificName: The scientific name (without brackets)
+//   - rank: The rank
+//   - error: An error if the string format is invalid
+func ParseTaxonString(taxonStr string) (code, taxid, scientificName, rank string, err error) {
+	// Trim any leading/trailing whitespace from the entire string
+	taxonStr = strings.TrimSpace(taxonStr)
+
+	// Split by '@' to separate rank
+	parts := strings.Split(taxonStr, "@")
+	if len(parts) > 2 {
+		return "", "", "", "", errors.New("invalid format: multiple '@' characters found")
+	}
+
+	mainPart := strings.TrimSpace(parts[0])
+	if len(parts) == 2 {
+		rank = strings.TrimSpace(parts[1])
+	} else {
+		rank = "no rank"
+	}
+
+	// Find scientific name part (enclosed in square brackets)
+	startBracket := strings.Index(mainPart, "[")
+	endBracket := strings.LastIndex(mainPart, "]")
+
+	if startBracket == -1 || endBracket == -1 || startBracket > endBracket {
+		return "", "", "", "", errors.New("invalid format: scientific name must be enclosed in square brackets")
+	}
+
+	// Extract and clean scientific name
+	scientificName = strings.TrimSpace(mainPart[startBracket+1 : endBracket])
+
+	// Process code:taxid part
+	idPart := strings.TrimSpace(mainPart[:startBracket])
+	idComponents := strings.Split(idPart, ":")
+
+	if len(idComponents) != 2 {
+		return "", "", "", "", errors.New("invalid format: missing taxonomy code separator ':'")
+	}
+
+	code = strings.TrimSpace(idComponents[0])
+	taxid = strings.TrimSpace(idComponents[1])
+
+	if code == "" || taxid == "" || scientificName == "" {
+		return "", "", "", "", errors.New("invalid format: code, taxid and scientific name cannot be empty")
+	}
+
+	return code, taxid, scientificName, rank, nil
+}

pkg/obitax/taxon.go

@@ -1,6 +1,7 @@
 package obitax
 
 import (
+	"errors"
 	"iter"
 	"regexp"
 
@@ -379,3 +380,29 @@ func (taxon *Taxon) SameAs(other *Taxon) bool {
 
 	return taxon.Taxonomy == other.Taxonomy && taxon.Node.id == other.Node.id
 }
+
+func (taxon *Taxon) AddChild(child string, replace bool) (*Taxon, error) {
+	if taxon == nil {
+		return nil, errors.New("nil taxon")
+	}
+
+	code, taxid, scientific_name, rank, err := ParseTaxonString(child)
+
+	if err != nil {
+		return nil, err
+	}
+
+	if taxon.Taxonomy.code != code {
+		return nil, errors.New("taxonomy code mismatch")
+	}
+
+	newTaxon, err := taxon.Taxonomy.AddTaxon(taxid, *taxon.Node.id, rank, false, replace)
+
+	if err != nil {
+		return nil, err
+	}
+
+	newTaxon.SetName(scientific_name, "scientific name")
+
+	return newTaxon, nil
+}

pkg/obitax/taxonomy.go

@@ -12,7 +12,6 @@ import (
 	"fmt"
 
 	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obiutils"
-	log "github.com/sirupsen/logrus"
 )
 
 // Taxonomy represents a hierarchical classification of taxa.
@@ -130,27 +129,28 @@ 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 {
+func (taxonomy *Taxonomy) Taxon(taxid string) (*Taxon, error) {
 	taxonomy = taxonomy.OrDefault(false)
 	if taxonomy == nil {
-		return nil
+		return nil, errors.New("cannot extract taxon from nil taxonomy")
 	}
 
 	id, err := taxonomy.Id(taxid)
 
 	if err != nil {
-		log.Fatalf("Taxid %s: %v", taxid, err)
+		return nil, fmt.Errorf("Taxid %s: %v", taxid, err)
 	}
 
 	taxon := taxonomy.nodes.Get(id)
 
 	if taxon == nil {
-		log.Fatalf("Taxid %s is not part of the taxonomy %s",
-			taxid,
-			taxonomy.name)
+		return nil,
+			fmt.Errorf("Taxid %s is not part of the taxonomy %s",
+				taxid,
+				taxonomy.name)
 	}
 
-	return taxon
+	return taxon, nil
 }
 
 // AsTaxonSet returns the set of taxon nodes contained within the Taxonomy.
@@ -353,3 +353,63 @@ func (taxonomy *Taxonomy) HasRoot() bool {
 	taxonomy = taxonomy.OrDefault(false)
 	return taxonomy != nil && taxonomy.root != nil
 }
+
+func (taxonomy *Taxonomy) InsertPathString(path []string) (*Taxonomy, error) {
+	if len(path) == 0 {
+		return nil, errors.New("path is empty")
+	}
+
+	code, taxid, scientific_name, rank, err := ParseTaxonString(path[0])
+
+	if taxonomy == nil {
+		taxonomy = NewTaxonomy(code, code, obiutils.AsciiAlphaNumSet)
+	}
+
+	if err != nil {
+		return nil, err
+	}
+
+	if taxonomy.Len() == 0 {
+
+		if code != taxonomy.code {
+			return nil, fmt.Errorf("cannot insert taxon %s into taxonomy %s with code %s",
+				path[0], taxonomy.name, taxonomy.code)
+		}
+
+		root, err := taxonomy.AddTaxon(taxid, taxid, rank, true, true)
+
+		if err != nil {
+			return nil, err
+		}
+		root.SetName(scientific_name, "scientificName")
+	}
+
+	var current *Taxon
+	current, err = taxonomy.Taxon(taxid)
+
+	if err != nil {
+		return nil, err
+	}
+
+	if !current.IsRoot() {
+		return nil, errors.New("path does not start with a root node")
+	}
+
+	for _, id := range path[1:] {
+		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")
+			}
+			current = taxon
+		} else {
+			current, err = current.AddChild(id, false)
+
+			if err != nil {
+				return nil, err
+			}
+		}
+	}
+
+	return taxonomy, nil
+}

pkg/obitax/taxonset.go

@@ -212,7 +212,8 @@ func (set *TaxonSet) Sort() *TaxonSlice {
 		pushed = false
 		for _, node := range set.set {
 			if !parent[node] && (parent[set.Get(node.parent).Node] ||
-				!set.Contains(node.parent)) {
+				!set.Contains(node.parent) ||
+				node == taxonomy.Root().Node) {
 				pushed = true
 				taxa.slice = append(taxa.slice, node)
 				parent[node] = true