Merge feature/gen-3c: unknown allele displays '-' (stores '?'), fix Unbekannter Farbschlag (wildcard + E-locus family fallback) [god-QA pending]

gerbil-manager-web/src/genetics/__tests__/genetics.test.ts

@@ -173,8 +173,8 @@ describe('Farbschlag catalog', () => {
   })
 
   it('falls back to Unbekannter Farbschlag for uncatalogued genotypes', () => {
-    // Colourpoint marked + dilute + grey combo not in the catalog.
-    const match = farbschlagFor(fromDisplayString('aa cchmcchm dd ee gg PP spsp rere'))
+    // E-dominant (no Fuchs/Schimmel family) + an uncatalogued cchm/dd/gg/pp combo.
+    const match = farbschlagFor(fromDisplayString('aa cchmcchm dd EE gg pp spsp rere'))
     expect(match.unknown).toBe(true)
     expect(match.name).toBe('Unbekannter Farbschlag')
   })
@@ -222,7 +222,7 @@ describe('Farbschlag catalog', () => {
   it('genotypeToFarbschlag (DATA-1 denormalization contract) returns the plain name', () => {
     expect(genotypeToFarbschlag(wildType())).toBe('Agouti')
     expect(genotypeToFarbschlag(fromDisplayString('aa CC DD EE GG pp spsp rere'))).toBe('Platin')
-    expect(genotypeToFarbschlag(fromDisplayString('aa cchmcchm dd ee gg PP spsp rere'))).toBe(
+    expect(genotypeToFarbschlag(fromDisplayString('aa cchmcchm dd EE gg pp spsp rere'))).toBe(
       'Unbekannter Farbschlag',
     )
   })
@@ -350,3 +350,32 @@ describe('GEN-3a: Schimmel catalog fix (breeder C5)', () => {
     expect(BASE_COLORS.some((e) => e.name === 'Schwarzschimmel')).toBe(false)
   })
 })
+
+describe("GEN-3c: unknown allele displays as '-' (stored as '?')", () => {
+  it("accepts '-' input, stores '?', displays '-'", () => {
+    const g = fromDisplayString('Aa C- DD EE GG Pp spsp rere')
+    expect(g.C).toEqual(['C', '?']) // stored internal contract stays '?'
+    expect(toDisplayString(g)).toBe('Aa C- DD EE GG Pp spsp rere') // displayed as '-'
+  })
+  it("'?' and '-' inputs are equivalent", () => {
+    expect(toDisplayString(fromDisplayString('Aa C? DD EE GG Pp spsp rere'))).toBe(
+      'Aa C- DD EE GG Pp spsp rere',
+    )
+  })
+})
+
+describe('GEN-3c: no Unbekannt when the E locus is known (family fallback)', () => {
+  it('eef with unknown other loci -> Fuchsschimmel (the reported bug case)', () => {
+    expect(genotypeToFarbschlag(fromDisplayString('aa C- D- eef Gg Pp spsp --'))).toBe(
+      'Fuchsschimmel',
+    )
+  })
+  it('ee -> Fuchs family, efef -> a Schimmel (never Unbekannt) even with unknowns', () => {
+    expect(farbschlagFor(fromDisplayString('a- C- D- ee G- P-')).unknown).toBe(false)
+    expect(farbschlagFor(fromDisplayString('A- C- D- efef G- P-')).unknown).toBe(false)
+  })
+  it('unknown allele no longer blocks a match (wildcard, never Unbekannt)', () => {
+    // C unknown must not force Unbekannt — it resolves to SOME named variety.
+    expect(farbschlagFor(fromDisplayString('aa C- DD EE GG PP spsp rere')).unknown).toBe(false)
+  })
+})

gerbil-manager-web/src/genetics/catalog.ts

@@ -20,9 +20,15 @@
  *    meta rows dropped, 17 matched the frozen names). Genotypes normalized from
  *    portal notation (c[chm]->cchm, c[h]->ch, e[f]->ef, '-'/'--' = unknown).
  */
-import { LOCUS_ORDER, type LocusKey } from './loci'
-import { makeGenotype, toDisplayString, wildType, type AllelePair, type Genotype } from './genotype'
-import { phenotypeTokens, type PhenotypeTokens } from './phenotype'
+import { LOCUS_ORDER, dominantAllele, type LocusKey } from './loci'
+import {
+  makeGenotype,
+  toDisplayString,
+  wildType,
+  WILDCARD,
+  type AllelePair,
+  type Genotype,
+} from './genotype'
 
 export interface FarbschlagEntry {
   /** German variety name. FROZEN for the original 18 (DB keys). */
@@ -138,25 +144,64 @@ export interface FarbschlagMatch {
   readonly unknown: boolean
 }
 
-function matches(tokens: PhenotypeTokens, entry: FarbschlagEntry): boolean {
-  return (Object.keys(entry.tokens) as LocusKey[]).every(
-    (locus) => tokens[locus] === entry.tokens[locus],
-  )
+/**
+ * Expressed token at a locus, or null when UNKNOWN (a '?' allele) — null acts as
+ * a wildcard in matching. The E locus is PAIR-aware so the Fuchs/Schimmel family
+ * is distinguishable: ee->'e', e/ef->'eef' (Fuchsschimmel), ef/ef->'ef' (Schimmel).
+ */
+function locusToken(g: Genotype, locus: LocusKey): string | null {
+  const [x, y] = g[locus]
+  if (x === WILDCARD || y === WILDCARD) return null
+  if (locus === 'E') {
+    if (x === y) return x // ee->'e', efef->'ef', EE->'E'
+    if ((x === 'e' && y === 'ef') || (x === 'ef' && y === 'e')) return 'eef'
+    return dominantAllele('E', x, y) // E/ef, E/e -> 'E'
+  }
+  return dominantAllele(locus, x, y)
+}
+
+function matches(g: Genotype, entry: FarbschlagEntry): boolean {
+  // Unknown loci (null token) match anything (treated as wildcard, GEN-3c).
+  return (Object.keys(entry.tokens) as LocusKey[]).every((locus) => {
+    const t = locusToken(g, locus)
+    return t === null || t === entry.tokens[locus]
+  })
+}
+
+/**
+ * GEN-3c family fallback: the E locus alone names the Fuchs/Schimmel family even
+ * when other loci are unknown (so genotypes never fall through to "Unbekannt").
+ *   ee -> Fuchs | e/ef -> Fuchsschimmel | ef/ef -> Schimmel | e/? -> Fuchs (for now)
+ * Returns null when E is dominant (full colour) or fully unknown.
+ */
+function eFamily(g: Genotype): string | null {
+  const [x, y] = g.E
+  if (x === 'e' && y === 'e') return 'Fuchs'
+  if ((x === 'e' && y === 'ef') || (x === 'ef' && y === 'e')) return 'Fuchsschimmel'
+  if (x === 'ef' && y === 'ef') return 'Schimmel'
+  if ((x === 'e' || y === 'e') && (x === WILDCARD || y === WILDCARD)) return 'Fuchs'
+  return null
 }
 
 /** Resolve a genotype to its German Farbschlag (with Schecke/Rex modifiers). */
 export function farbschlagFor(g: Genotype): FarbschlagMatch {
-  const tokens = phenotypeTokens(g)
-  const base = BASE_COLORS.find((e) => matches(tokens, e)) ?? null
+  const family = eFamily(g)
+  // When the animal is a Fuchs/Schimmel-family genotype, only entries that
+  // explicitly pin the E locus may name it (so e.g. eef can't masquerade as a
+  // C-series white via wildcard C); otherwise fall back to the family name.
+  const base = family
+    ? (BASE_COLORS.find((e) => e.tokens.E !== undefined && matches(g, e)) ?? null)
+    : (BASE_COLORS.find((e) => matches(g, e)) ?? null)
 
   const modifiers: string[] = []
-  if (tokens.Sp === 'Sp') modifiers.push('Schecke')
-  if (tokens.Re === 'Re') modifiers.push('Rex')
+  if (locusToken(g, 'Sp') === 'Sp') modifiers.push('Schecke')
+  if (locusToken(g, 'Re') === 'Re') modifiers.push('Rex')
 
-  if (!base) {
+  const baseName = base?.name ?? family
+  if (!baseName) {
     return { name: UNKNOWN_FARBSCHLAG, base: null, unknown: true }
   }
-  const name = [base.name, ...modifiers].join(' ')
+  const name = [baseName, ...modifiers].join(' ')
   return { name, base, unknown: false }
 }
 

gerbil-manager-web/src/genetics/genotype.ts

@@ -79,12 +79,14 @@ export function wildType(): Genotype {
  * The Sls locus is OMITTED when wild-type (sl/sl) so legacy 8-locus strings and
  * the colour catalog stay byte-identical; it only appears for WP/Sls carriers
  * (e.g. "… spsp rere Slsl"). Round-trips: a missing Sls re-parses to sl/sl.
+ * GEN-3c: unknown alleles are STORED as '?' but DISPLAYED as '-' (breeder
+ * convention) — e.g. ['C','?'] renders "C-".
  */
 export function toDisplayString(g: Genotype): string {
   return LOCUS_ORDER.filter(
     (locus) => locus !== 'Sls' || !(g.Sls[0] === 'sl' && g.Sls[1] === 'sl'),
   )
-    .map((locus) => g[locus][0] + g[locus][1])
+    .map((locus) => (g[locus][0] + g[locus][1]).replace(/\?/g, '-'))
     .join(' ')
 }
 
@@ -144,6 +146,9 @@ function normalizeToken(tok: string): string | null {
   if (t === 'WP') t = 'Slsl'
   t = t.replace(/S\(l\)/g, 'Sl').replace(/s\(l\)/g, 'sl')
   t = t.replace(/Uw/g, 'G').replace(/uw/g, 'g')
+  // GEN-3c: '-' is the breeder's UNKNOWN marker on input; store internally as '?'
+  // (the frozen storage contract keeps '?'; only DISPLAY renders '-').
+  t = t.replace(/-/g, '?')
   return t
 }
 
@@ -193,7 +198,9 @@ export function fromDisplayString(input: string): Genotype {
     const [a, b] = splitToken(token)
     const refAllele = a === WILDCARD ? b : a
     if (refAllele === WILDCARD) {
-      throw new Error(`Token "${token}" is fully unknown; cannot infer its locus`)
+      // Fully-unknown token ("--"/"??", e.g. a positional placeholder): the locus
+      // can't be inferred — skip it (defaults to wild-type), don't throw.
+      continue
     }
     const locus = ALLELE_TO_LOCUS[refAllele]
     if (!locus) throw new Error(`Unknown allele "${refAllele}" in token "${token}"`)