add nomenclatural code flag (close #265)

config.go

@@ -13,6 +13,10 @@ type Config struct {
 	// BatchSize sets the maximum number of elements in names-strings slice.
 	BatchSize int
 
+	// Code defines nomenclatural code that might modify parsing process in
+	// some cases.
+	nomcode.Code
+
 	// Debug sets a "debug" state for parsing. The debug state forces output
 	// format to showing parsed ast tree.
 	Debug bool
@@ -36,17 +40,15 @@ type Config struct {
 	// a stream of name-strings.
 	JobsNum int
 
+	nomcode.NomCode
+
 	// Port to run wer-service.
 	Port int
 
 	// WithCapitalization flag, when true, the first letter of a name-string
 	// is capitalized, if appropriate.
 	WithCapitalization bool
 
-	// WithCultivars flag, when true, cultivar names will be parsed and
-	// modify cardinality, normalized and canonical output.
-	WithCultivars bool
-
 	// WithDetails can be set to true when a simplified output is not sufficient
 	// for obtaining a required information.
 	WithDetails bool
@@ -145,10 +147,10 @@ func OptWithCapitaliation(b bool) Option {
 	}
 }
 
-// OptWithCultivars sets the EnableCultivars field.
-func OptWithCultivars(b bool) Option {
+// OptCode sets nomenclatural code to modify parsing process.
+func OptWithCultivars(c nomcode.Code) Option {
 	return func(cfg *Config) {
-		cfg.WithCultivars = b
+		cfg.Code = c
 	}
 }
 

ent/nomcode/nomcode.go

@@ -0,0 +1,52 @@
+package nomcode
+
+import (
+	"log/slog"
+	"strings"
+)
+
+// Code represents a nomenclatural code.
+type Code int
+
+// Constants for different nomenclatural codes.
+const (
+	Unknown    Code = iota // Unknown code
+	Zoological             // Zoological code
+	Botanical              // Botanical code
+	Cultivar               // Cultivar code
+)
+
+// New creates a new Code from a string representation.
+// It accepts short codes ('b', 'z', 'c') and full names
+// ('botanical', 'zoological', 'cultivar') as well as
+// official abbreviations ('icn', 'iczn', 'icncp').
+// The input string is case-insensitive.
+func New(s string) Code {
+	sOrig := s
+	s = strings.ToLower(s)
+	switch s {
+	case "b", "bot", "botanical", "icn":
+		return Botanical
+	case "z", "zoo", "zoological", "iczn":
+		return Zoological
+	case "c", "cult", "cultivar", "icncp":
+		return Cultivar
+	default:
+		slog.Warn("Cannot determine nomenclatural code", "input", sOrig)
+		return Unknown
+	}
+}
+
+// String returns the official abbreviation of the nomenclatural code.
+func (c Code) String() string {
+	switch c {
+	case Zoological:
+		return "ICZN"
+	case Botanical:
+		return "ICN"
+	case Cultivar:
+		return "ICNCP"
+	default:
+		return ""
+	}
+}

gnparser/cmd/root.go

@@ -80,6 +80,7 @@ gnparser -j 5 -p 8080
 		withPreserveDiaeresesFlag(cmd)
 		batchSizeFlag(cmd)
 		spGrCutFlag(cmd)
+
 		port := portFlag(cmd)
 		cfg := gnparser.NewConfig(opts...)
 		batchSize = cfg.BatchSize
@@ -139,7 +140,18 @@ func init() {
 		"maximum number of names in a batch send for processing.")
 
 	rootCmd.Flags().BoolP("cultivar", "C", false,
-		"include cultivar epithets and graft-chimeras in normalized and canonical outputs")
+		"parse according to  cultivar code ICNCP (DEPRECATED, use nomenclatural-code instead)")
+
+	codeHelp := `Modifies the parser's behavior in ambiguous cases, sometimes 
+introducing additional parsing rules.
+
+Accepted values are:
+  - 'bot', 'icn', 'botanical' for botanical code
+  - 'cult', 'icncp', 'cultivar' for cultivar code
+  - 'zoo', 'iczn', 'zoological' for zoological code
+
+If not set, the parser will attempt to determine the appropriate code/s.`
+	rootCmd.Flags().StringP("nomenclatural-code", "n", "", codeHelp)
 
 	rootCmd.Flags().BoolP("capitalize", "c", false,
 		"capitalize the first letter of input name-strings")
@@ -149,9 +161,15 @@ func init() {
 
 	rootCmd.Flags().BoolP("details", "d", false, "provides more details")
 
-	formatHelp := "sets output format. Can be one of:\n" +
-		"'csv', 'tsv', 'compact', 'pretty'\n" +
-		"default is 'csv'"
+	formatHelp := `Sets the output format.
+
+Accepted values are:
+  - 'csv': Comma-separated values
+  - 'tsv': Tab-separated values
+  - 'compact': Compact JSON format
+  - 'pretty': Human-readable JSON format
+
+If not set, the output format defaults to 'csv'.`
 	rootCmd.Flags().StringP("format", "f", "", formatHelp)
 
 	rootCmd.Flags().BoolP("ignore_tags", "i", false,