extend cf to infraspecies (close #245)

CHANGELOG.md

@@ -4,7 +4,7 @@
 
 ## [v1.7.5] - 2023-09-12 Tue
 
-- Fix [#249]: ignore `cf` at the end of the strings.
+- Fix [#249]: allow `cf` at the end of the strings, cf for infraspecies.
 - Fix [#248]: do not escape double quotes for TSV output.
 - Fix [#246]: ignore `ms` at the end of the strings.
 

ent/parsed/details.go

@@ -50,14 +50,15 @@ type InfraspeciesElem struct {
 
 // Comparison are details for a surrogate comparison name.
 type Comparison struct {
-	// Genus is the genus of a name.
-	Genus string `json:"genus"`
-	// Species is a specific epithet of a name.
-	Species string `json:"species,omitempty"`
-	// Cultivar is a value of a cultivar of a binomial.
-	Cultivar string `json:"cultivar,omitempty"`
-	// SpeciesAuthorship the authorship of Species.
-	SpeciesAuthorship *Authorship `json:"authorship,omitempty"`
+	// Genus is used if no species information is given
+	Genus string `json:"genus,omitempty"`
+
+	// Species are details for the binomial part of a name.
+	*Species
+
+	// InfraSpecies is an infraspecific epthet of a name.
+	InfraSpecies *InfraspeciesElem `json:"infraspecies,omitempty"`
+
 	// CompMarker, usually "cf.".
 	CompMarker string `json:"comparisonMarker"`
 }
@@ -91,7 +92,6 @@ type DetailsGraftChimeraFormula struct {
 // isDetails implements Details interface.
 func (DetailsHybridFormula) isDetails() {}
 
-
 // isDetails implements Details interface.
 func (DetailsGraftChimeraFormula) isDetails() {}
 

ent/parser/ast.go

@@ -533,16 +533,60 @@ func (p *Engine) newApproxNode(n *node32) *approxNode {
 }
 
 type comparisonNode struct {
-	Genus      *parsed.Word
-	SpEpithet  *spEpithetNode
-	Comparison *parsed.Word
+	Genus          *parsed.Word
+	SpEpithet      *spEpithetNode
+	InfraSpEpithet *infraspEpithetNode
+	Comparison     *parsed.Word
+	Cardinality    int
 }
 
 func (p *Engine) newComparisonNode(n *node32) *comparisonNode {
+	n = n.up
+	switch n.pegRule {
+	case ruleNameCompIsp:
+		return p.newCompIspNode(n)
+	case ruleNameCompSp:
+		return p.newCompSpNode(n)
+	default:
+		var res *comparisonNode
+		return res
+	}
+}
+
+func (p *Engine) newCompIspNode(n *node32) *comparisonNode {
 	var cn *comparisonNode
-	if n.pegRule != ruleNameComp {
-		return cn
+	n = n.up
+	var gen, comp *parsed.Word
+	var spEp *spEpithetNode
+	var ispEp *infraspEpithetNode
+	for n != nil {
+		switch n.pegRule {
+		case ruleGenusWord:
+			gen = p.newWordNode(n, parsed.GenusType)
+		case ruleComparison:
+			comp = p.newWordNode(n, parsed.ComparisonMarkerType)
+		case ruleSpeciesEpithet:
+			spEp = p.newSpeciesEpithetNode(n)
+			p.cardinality = 2
+		case ruleInfraspEpithet:
+			ispEp = p.newInfraspEpithetNode(n)
+			p.cardinality = 3
+		}
+		n = n.next
+	}
+	cn = &comparisonNode{
+		Genus:          gen,
+		Comparison:     comp,
+		SpEpithet:      spEp,
+		InfraSpEpithet: ispEp,
+		Cardinality:    3,
 	}
+	return cn
+
+}
+
+func (p *Engine) newCompSpNode(n *node32) *comparisonNode {
+	var cn *comparisonNode
 	n = n.up
 	var gen, comp *parsed.Word
 	var spEp *spEpithetNode
@@ -560,9 +604,10 @@ func (p *Engine) newComparisonNode(n *node32) *comparisonNode {
 		n = n.next
 	}
 	cn = &comparisonNode{
-		Genus:      gen,
-		Comparison: comp,
-		SpEpithet:  spEp,
+		Genus:       gen,
+		Comparison:  comp,
+		SpEpithet:   spEp,
+		Cardinality: 2,
 	}
 	return cn
 }
@@ -1137,9 +1182,13 @@ func (n *node32) flatChildren() []*node32 {
 func (p *Engine) newWordNode(n *node32, wt parsed.WordType) *parsed.Word {
 	t := n.token32
 	val := p.nodeValue(n)
+	norm := val
+	if n.pegRule == ruleComparison {
+		norm = "cf."
+	}
 	wrd := parsed.Word{
 		Verbatim:   val,
-		Normalized: val,
+		Normalized: norm,
 		Type:       wt,
 		Start:      int(t.begin),
 		End:        int(t.end),

ent/parser/engine.go

@@ -210,6 +210,8 @@ var nodeRules = map[pegRule]struct{}{
 	ruleName:                            {},
 	ruleNameApprox:                      {},
 	ruleNameComp:                        {},
+	ruleNameCompSp:                      {},
+	ruleNameCompIsp:                     {},
 	ruleNameSpecies:                     {},
 	ruleNamedGenusGraftChimera:          {},
 	ruleNamedGenusHybrid:                {},

ent/parser/grammar.peg

@@ -41,7 +41,11 @@ NameUninomial <- (UninomialCombo / Uninomial) (_ CultivarWordGroup)?
 
 NameApprox <- GenusWord (_ SpeciesEpithet)? _ Approximation ApproxNameIgnored
 
-NameComp <- GenusWord _ Comparison (_ SpeciesEpithet)?
+NameComp <- NameCompIsp / NameCompSp
+
+NameCompSp <- GenusWord _ Comparison (_ SpeciesEpithet)?
+
+NameCompIsp <- GenusWord _ SpeciesEpithet _ Comparison (_ InfraspEpithet)? 
 
 NameSpecies <- GenusWord (_? ( Subgenus / SubgenusOrSuperspecies))?
                _ SpeciesEpithet (_ InfraspGroup)? (_ CultivarWordGroup)?