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

/**
 * Farbschlag catalog: genotype -> German variety name + the full named catalog.
 *
 * A catalog entry matches when every locus it SPECIFIES equals the genotype's
 * phenotype token (see phenotype.ts). Unspecified loci are wildcards. The first
 * matching entry wins; Schecke/Rex are appended as modifiers.
 *
 * Two roles, one list:
 *  - REVERSE MATCH (genotypeToFarbschlag): first match wins, so the frozen/base
 *    names are listed FIRST and win their token-set. Genotype-identical synonyms
 *    (e.g. "-hell" intensity variants) are listed after and simply won't be the
 *    reverse-match result — that's expected (shade isn't encoded in the 6 loci).
 *  - FULL NAMED LIST (CATALOG seed + dropdowns + images): every entry is a
 *    distinct user-pickable variety with its own image.
 *
 * Sources:
 *  - First 18: de.wikibooks Rennmaus-Farbgenetik, rennmaus-info, clan-of-topolino,
 *    rennmauswelten. These names are a FROZEN contract (DB keys) — never rename.
 *  - Remaining 55 (GEN-2): baseportal.de "Farben" DB (78 rows; 6 "zur Scheckung"
 *    meta rows dropped, 17 matched the frozen names). Genotypes normalized from
 *    portal notation (c[chm]->cchm, c[h]->ch, e[f]->ef, '-'/'--' = unknown).
 */
import { LOCI, 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). */
  readonly name: string
  /** Common English name (reference only; UI is German). */
  readonly english?: string
  /** Required phenotype tokens; omitted loci are wildcards. */
  readonly tokens: Partial<Record<LocusKey, string>>
  /** Image filename under /farbschlaege/ (public), if available. */
  readonly image?: string
}

/**
 * Ordered so the earliest match wins. The frozen 18 come first (they win their
 * token-sets); the baseportal varieties follow, base before "-hell" within each
 * portal collision group.
 */
export const BASE_COLORS: readonly FarbschlagEntry[] = [
  // ── Frozen names (DB-key contract; do not rename) ──
  // GEN-3a: 'Schwarzschimmel' REMOVED (breeder C5: no such variety; efef base is
  // Orangeschimmel — see the GEN-2 block below). This was an authorized exception
  // to the frozen-name rule; the ColorVariety seed drops it too.
  { name: 'Pink Eyed White (PEW)', english: 'Pink Eyed White', tokens: { C: 'ch', P: 'p' }, image: 'rotaugen-weiss-pew-d-sep-e-sep.jpg' },
  { name: 'Hermelin', english: 'Dark Tailed White', tokens: { A: 'a', C: 'ch', D: 'D', P: 'P' }, image: 'hermelin.jpeg' },
  { name: 'Himalaya', english: 'Himalayan', tokens: { A: 'A', C: 'ch', D: 'D', P: 'P' }, image: 'himalaya.jpg' },
  { name: 'Zobel', english: 'Sable', tokens: { A: 'a', C: 'cchm', D: 'D', E: 'E', G: 'g', P: 'P' }, image: 'zobel.jpeg' },
  { name: 'Rotaugenschimmel', english: 'Red-Eyed Roan', tokens: { C: 'C', D: 'D', E: 'ef', G: 'G', P: 'p' }, image: 'rotaugen-schimmel.jpg' },
  { name: 'Agouti', english: 'Golden Agouti', tokens: { A: 'A', C: 'C', D: 'D', E: 'E', G: 'G', P: 'P' }, image: 'agouti-mit-erklaerung-der-genloci.JPG' },
  { name: 'Schwarz', english: 'Black', tokens: { A: 'a', C: 'C', D: 'D', E: 'E', G: 'G', P: 'P' }, image: 'schwarz.jpg' },
  { name: 'Silberagouti', english: 'Grey Agouti', tokens: { A: 'A', C: 'C', D: 'D', E: 'E', G: 'g', P: 'P' }, image: 'silberagouti.jpg' },
  { name: 'Anthrazit', english: 'Slate', tokens: { A: 'a', C: 'C', D: 'D', E: 'E', G: 'g', P: 'P' }, image: 'anthrazit.jpg' },
  { name: 'Algierfuchs', english: 'Dark-Eyed Honey', tokens: { A: 'A', C: 'C', D: 'D', E: 'e', G: 'G', P: 'P' }, image: 'algierfuchs.jpg' },
  { name: 'Blau', english: 'Blue', tokens: { A: 'a', C: 'C', D: 'd', E: 'E', G: 'G', P: 'P' }, image: 'blau-schwarz-dd.JPG' },
  { name: 'Gold', english: 'Argente Golden', tokens: { A: 'A', C: 'C', D: 'D', E: 'E', G: 'G', P: 'p' }, image: 'gold.jpg' },
  { name: 'Platin', english: 'Lilac', tokens: { A: 'a', C: 'C', D: 'D', E: 'E', G: 'G', P: 'p' }, image: 'platin.JPG' },
  { name: 'Goldfuchs', english: 'Yellow Fox', tokens: { A: 'A', C: 'C', D: 'D', E: 'e', G: 'G', P: 'p' }, image: 'goldfuchs.jpg' },
  { name: 'Rotfuchs', english: 'Argente Nutmeg', tokens: { A: 'a', C: 'C', D: 'D', E: 'e', G: 'G', P: 'p' }, image: 'rotfuchs.JPG' },
  { name: 'dd Gold', english: 'dd Argente Golden', tokens: { A: 'A', C: 'C', D: 'd', E: 'E', G: 'G', P: 'p' }, image: 'gold-dd.jpg' },
  { name: 'dd Platin', english: 'dd Lilac', tokens: { A: 'a', C: 'C', D: 'd', E: 'E', G: 'G', P: 'p' }, image: 'platin-dd.jpg' },

  // ── baseportal.de varieties (GEN-2), normalized to the frozen allele table ──
  { name: 'Altweiss (REW)', tokens: { A: 'a', C: 'C', D: 'D', E: 'E', G: 'g', P: 'p' }, image: 'altweiss-rew.jpeg' },
  { name: 'Apricot (Blassfuchs)', tokens: { A: 'A', C: 'C', D: 'D', E: 'e', G: 'g', P: 'p' }, image: 'apricot-blassfuchs.jpg' },
  { name: 'Blaufuchs', tokens: { A: 'a', C: 'C', D: 'D', E: 'e', G: 'g', P: 'P' }, image: 'blaufuchs.jpg' },
  { name: 'C-Separator', tokens: { A: 'a', C: 'C', D: 'D', E: 'e', G: 'g', P: 'p' }, image: 'c-separator.jpg' },
  { name: 'Elfenbein', tokens: { A: 'A', C: 'C', D: 'D', E: 'E', G: 'g', P: 'p' }, image: 'elfenbein.jpg' },
  { name: 'Kohlfuchs', tokens: { A: 'a', C: 'C', D: 'D', E: 'e', G: 'G', P: 'P' }, image: 'kohlfuchs.jpg' },
  { name: 'Polarfuchs', tokens: { A: 'A', C: 'C', D: 'D', E: 'e', G: 'g', P: 'P' }, image: 'polarfuchs.jpg' },
  { name: 'Saphir', tokens: { A: 'a', C: 'C', D: 'D', E: 'E', G: 'G', P: 'p' }, image: 'saphir.jpg' },
  // GEN-3a: efef base (otherwise wild C/D/G/P) = Orangeschimmel (breeder C5).
  { name: 'Orangeschimmel', tokens: { C: 'C', D: 'D', E: 'ef', G: 'G', P: 'P' }, image: 'schimmel-orangeschimmel.jpg' },
  { name: 'Topas', tokens: { A: 'A', C: 'C', D: 'D', E: 'E', G: 'G', P: 'p' }, image: 'topas.jpg' },
  { name: 'Platin-Hell', tokens: { A: 'a', C: 'C', D: 'D', E: 'E', G: 'G', P: 'p' }, image: 'platin-hell.jpg' },
  { name: 'Agouti dd', tokens: { A: 'A', C: 'C', D: 'd', E: 'E', G: 'G', P: 'P' }, image: 'agouti-dd.jpg' },
  { name: 'Silberagouti dd', tokens: { A: 'A', C: 'C', D: 'd', E: 'E', G: 'g', P: 'P' }, image: 'silberagouti-dd.jpg' },
  { name: 'Kohlfuchs dd', tokens: { A: 'a', C: 'C', D: 'd', E: 'e', G: 'G', P: 'P' }, image: 'kohlfuchs-dd.jpg' },
  { name: 'Anthrazit dd', tokens: { A: 'a', C: 'C', D: 'd', E: 'E', G: 'g', P: 'P' }, image: 'anthrazit-dd.jpg' },
  // GEN-3a: efef gg base = Silberschimmel (breeder C5) — listed before the
  // A-specific Polarfuchsschimmel so the canonical efef-gg reverse-matches here.
  { name: 'Silberschimmel', tokens: { C: 'C', D: 'D', E: 'ef', G: 'g', P: 'P' }, image: 'silberschimmel.jpg' },
  { name: 'Polarfuchsschimmel', tokens: { A: 'A', C: 'C', D: 'D', E: 'ef', G: 'g', P: 'P' }, image: 'polarfuchsschimmel.jpg' },
  { name: 'Algierfuchsschimmel', tokens: { A: 'A', C: 'C', D: 'D', E: 'ef', G: 'G', P: 'P' }, image: 'algierfuchsschimmel.jpg' },
  { name: 'Kohlfuchsschimmel', tokens: { A: 'a', C: 'C', D: 'D', E: 'ef', G: 'G', P: 'P' }, image: 'kohlfuchsschimmel.jpg' },
  { name: 'Blaufuchsschimmel', tokens: { A: 'a', C: 'C', D: 'D', E: 'ef', G: 'g', P: 'P' }, image: 'blaufuchsschimmel.jpg' },
  { name: 'Kohlfuchs, hell', tokens: { A: 'a', C: 'C', D: 'D', E: 'e', G: 'G', P: 'P' }, image: 'kohlfuchs-hell.jpg' },
  { name: 'Goldfuchs, hell', tokens: { A: 'A', C: 'C', D: 'D', E: 'e', G: 'G', P: 'p' }, image: 'goldfuchs-hell.jpg' },
  { name: 'Goldfuchsschimmel', tokens: { A: 'A', C: 'C', D: 'D', E: 'ef', G: 'G', P: 'p' }, image: 'goldfuchsschimmel.jpg' },
  { name: 'Gold-Hell', tokens: { A: 'A', C: 'C', D: 'D', E: 'E', G: 'G', P: 'p' }, image: 'gold-hell.jpg' },
  { name: 'Blaufuchs, hell', tokens: { A: 'a', C: 'C', D: 'D', E: 'e', G: 'g', P: 'P' }, image: 'blaufuchs-hell.jpeg' },
  { name: 'Rotfuchsschimmel', tokens: { A: 'a', C: 'C', D: 'D', E: 'ef', G: 'G', P: 'p' }, image: 'rotfuchsschimmel.jpg' },
  { name: 'Polarfuchs, hell', tokens: { A: 'A', C: 'C', D: 'D', E: 'e', G: 'g', P: 'P' }, image: 'polarfuchs-hell.jpeg' },
  { name: 'Kohlfuchsschimmel, hell', tokens: { A: 'a', C: 'C', D: 'D', E: 'ef', G: 'G', P: 'P' }, image: 'kohlfuchsschimmel-hell.jpg' },
  { name: 'Rotfuchs, hell', tokens: { A: 'a', C: 'C', D: 'D', E: 'e', G: 'G', P: 'p' }, image: 'rotfuchs-hell.jpg' },
  { name: 'Kohlfuchs-Hell', tokens: { A: 'a', C: 'C', D: 'D', E: 'e', G: 'G', P: 'P' }, image: 'kohlfuchs-hell-2.jpg' },
  { name: 'Algierfuchs, hell', tokens: { A: 'A', C: 'C', D: 'D', E: 'e', G: 'G', P: 'P' }, image: 'algierfuchs-hell.JPG' },
  { name: 'Topas dd', tokens: { A: 'A', C: 'C', D: 'd', E: 'E', G: 'G', P: 'p' }, image: 'topas-dd.jpg' },
  { name: 'Blaufuchs dd', tokens: { A: 'a', C: 'C', D: 'd', E: 'e', G: 'g', P: 'p' }, image: 'blaufuchs-dd.jpg' },

  // ── GEN-3f: c^chm colourpoint varieties, reconciled to the breeder's CP- naming ──
  // The merged GEN-3e colourpointName() rule is authoritative: aa points are the
  // marten/sable group (Marder/Siam, +gg Zobel/Zobel-Hell — E and D irrelevant);
  // A- points take the 'CP-<base colour>' prefix and the '-Hell' shade variants
  // collapse (other loci irrelevant for the CP prefix). These names == the strings
  // in her live data (god: extract animals.json) so the re-import name-matches and
  // the Farbschlag mismatch hint stops. The het cchm/ch points (Siam, Zobel-Hell,
  // CP-Fuchs-Hell) use the 'cchm/ch' pair token. The agouti fox/dilute points
  // (CP-Fuchs/CP-Blaufuchs) resolve through the engine's E-family fallback to
  // 'CP-Fuchs'; their distinct dropdown names remain for hand-pick + import match.
  { name: 'Marder', tokens: { A: 'a', C: 'cchm', D: 'D', E: 'E', G: 'G', P: 'P' }, image: 'marder.JPG' },
  { name: 'Siam', tokens: { A: 'a', C: 'cchm/ch', D: 'D', E: 'E', G: 'G', P: 'P' }, image: 'siam-marder-hell.JPG' },
  { name: 'Zobel-Hell', tokens: { A: 'a', C: 'cchm/ch', D: 'D', E: 'E', G: 'g', P: 'P' }, image: 'zobel-hell.jpg' },
  { name: 'CP-Agouti', tokens: { A: 'A', C: 'cchm', D: 'D', E: 'E', G: 'G', P: 'P' }, image: 'agouti-cp.jpg' },
  { name: 'CP-Silberagouti', tokens: { A: 'A', C: 'cchm', D: 'D', E: 'E', G: 'g', P: 'P' }, image: 'silberagouti-cp.JPG' },
  { name: 'CP-Algierfuchs', tokens: { A: 'A', C: 'cchm', D: 'D', E: 'e', G: 'G', P: 'P' }, image: 'algierfuchs-cp.jpg' },
  { name: 'CP-Polarfuchs', tokens: { A: 'A', C: 'cchm', D: 'D', E: 'e', G: 'g', P: 'P' }, image: 'polarfuchs-cp.jpg' },
  { name: 'CP-Fuchs', tokens: { A: 'A', C: 'cchm', D: 'd', E: 'e', G: 'G', P: 'P' } },
  { name: 'CP-Fuchs-Hell', tokens: { A: 'A', C: 'cchm/ch', D: 'd', E: 'e', G: 'G', P: 'P' } },
  { name: 'CP-Blaufuchs', tokens: { A: 'A', C: 'cchm', D: 'd', E: 'e', G: 'g', P: 'P' } },
  { name: 'CP-Orangeschimmel', tokens: { C: 'cchm', D: 'D', E: 'ef', G: 'G', P: 'P' } },
]

export const UNKNOWN_FARBSCHLAG = 'Unbekannter Farbschlag'

export interface FarbschlagMatch {
  /** German variety name (with Schecke/Rex modifiers appended), or the fallback. */
  readonly name: string
  /** The matched base entry, if any. */
  readonly base: FarbschlagEntry | null
  /** True when no base colour matched (name is the fallback). */
  readonly unknown: boolean
}

/**
 * Expressed token at a locus. GEN-3d: an UNKNOWN allele ('?') is resolved to the
 * MOST-DOMINANT allele of the locus (the safer default) rather than acting as a
 * match-anything wildcard — so an unknown-C animal reads as full-colour 'C', not
 * a c^h/c^chm colourpoint white. The E locus stays PAIR-aware so the Fuchs/
 * Schimmel family is distinguishable: ee->'e', e/ef->'eef', ef/ef->'ef'.
 * (The Fuchs/Schimmel FAMILY for unknown-E is still handled by eFamily on the
 * raw pair, which runs before this.)
 */
function locusToken(g: Genotype, locus: LocusKey): string {
  // Default an unknown allele to the WILD-TYPE reading: most-dominant for the
  // colour loci (unknown-C => full-colour 'C', not a white), but the recessive
  // UNMARKED allele for the spotting/rex markers (unknown-Sp must NOT imply Schecke).
  const alleles = LOCI[locus].alleles
  const isMarker = locus === 'Sp' || locus === 'Re' || locus === 'Sls'
  const fallback = isMarker ? alleles[alleles.length - 1] : alleles[0]
  const [x, y] = g[locus].map((a) => (a === WILDCARD ? fallback : a))
  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 {
  return (Object.keys(entry.tokens) as LocusKey[]).every(
    (locus) => locusToken(g, locus) === 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). */
/** Resolve an allele pair to concrete alleles, defaulting unknown to wild-type. */
function resolvedPair(g: Genotype, locus: LocusKey): [string, string] {
  const alleles = LOCI[locus].alleles
  const isMarker = locus === 'Sp' || locus === 'Re' || locus === 'Sls'
  const fallback = isMarker ? alleles[alleles.length - 1] : alleles[0]
  const [x, y] = g[locus].map((a) => (a === WILDCARD ? fallback : a))
  return [x, y]
}

/** Base colour name (no modifiers, no colourpoint prefix), via E-family + matches. */
function baseColourFor(g: Genotype): string | null {
  const family = eFamily(g)
  const base = family
    ? (BASE_COLORS.find((e) => e.tokens.E !== undefined && matches(g, e)) ?? null)
    : (BASE_COLORS.find((e) => matches(g, e)) ?? null)
  return base?.name ?? family
}

/**
 * GEN-3e: the C-locus colourpoint NAMING transform (breeder-authoritative).
 * Returns the colourpoint name, or null when it doesn't apply (full C present,
 * or chch — which the base matcher names Hermelin/Himalaya, preserving both).
 *   aa cchm/cchm        -> Marder      | aa cchm/ch        -> Siam
 *   aa cchm/cchm gg     -> Zobel       | aa cchm/ch gg     -> Zobel-Hell
 *   A- cchm/cchm | cchm/ch -> CP-<base colour> (base computed as if C were full)
 */
function colourpointName(g: Genotype): string | null {
  const c = resolvedPair(g, 'C')
  if (c.includes('C')) return null // a full C allele => full colour, no CP
  if (c[0] === 'ch' && c[1] === 'ch') return null // chch -> base matcher (Hermelin/Himalaya)
  // Remaining: cchm/cchm or cchm/ch (colourpoint, no full C, not chch).
  const bothCchm = c[0] === 'cchm' && c[1] === 'cchm'
  const agouti = resolvedPair(g, 'A').includes('A')
  if (!agouti) {
    const [g1, g2] = resolvedPair(g, 'G')
    const grey = g1 === 'g' && g2 === 'g'
    if (grey) return bothCchm ? 'Zobel' : 'Zobel-Hell'
    return bothCchm ? 'Marder' : 'Siam'
  }
  // A- colourpoint -> CP-<base colour>, base as if C were full.
  const base = baseColourFor(makeGenotype({ ...g, C: ['C', 'C'] }))
  return base ? `CP-${base}` : null
}

export function farbschlagFor(g: Genotype): FarbschlagMatch {
  const modifiers: string[] = []
  if (locusToken(g, 'Sp') === 'Sp') modifiers.push('Schecke')
  if (locusToken(g, 'Re') === 'Re') modifiers.push('Rex')

  const baseName = colourpointName(g) ?? baseColourFor(g)
  if (!baseName) {
    return { name: UNKNOWN_FARBSCHLAG, base: null, unknown: true }
  }
  const name = [baseName, ...modifiers].join(' ')
  return { name, base: null, unknown: false }
}

/**
 * Convenience: genotype -> German Farbschlag name (with Schecke/Rex modifiers),
 * or "Unbekannter Farbschlag". This is the DENORMALIZATION contract for DATA-1:
 * the backend mirrors this mapping into a stored Farbschlag column (recomputed
 * whenever the genotype changes) so Gridify list filtering/sorting never has to
 * parse genotype strings in SQL. The catalog (BASE_COLORS) is self-contained in
 * this file so it can be ported/kept in sync with the backend.
 */
export function genotypeToFarbschlag(g: Genotype): string {
  return farbschlagFor(g).name
}

/**
 * A representative full genotype for a catalog entry: each specified locus is
 * homozygous for its token allele; unspecified loci take the wild-type allele.
 * This is the entry's CanonicalGenotype for DB seeding.
 */
export function representativeGenotype(entry: FarbschlagEntry): Genotype {
  const base = wildType()
  const out = {} as Record<LocusKey, AllelePair>
  for (const locus of LOCUS_ORDER) {
    const token = entry.tokens[locus]
    if (!token) {
      out[locus] = base[locus]
      continue
    }
    // GEN-3f: a token may encode a HETEROZYGOUS pair as "x/y" (e.g. the het
    // colourpoints Siam/Zobel-Hell use C: 'cchm/ch'); otherwise it's homozygous.
    const [a, b] = token.includes('/') ? (token.split('/') as [string, string]) : [token, token]
    out[locus] = [a, b]
  }
  return makeGenotype(out)
}

/**
 * DB seed view for DATA-2's ColorVariety table. Each row mirrors the table
 * shape (Name, CanonicalGenotype, SortOrder, ImageFile). SortOrder = position.
 *
 * IMPORTANT: `name` values become DB keys the UI filters on — renames are
 * BREAKING changes and must be routed through god. Adding new varieties is safe.
 */
export interface ColorVarietySeed {
  readonly name: string
  readonly english?: string
  readonly canonicalGenotype: string
  readonly sortOrder: number
  readonly image?: string
}

export const CATALOG: readonly ColorVarietySeed[] = BASE_COLORS.map((entry, i) => ({
  name: entry.name,
  english: entry.english,
  canonicalGenotype: toDisplayString(representativeGenotype(entry)),
  sortOrder: i,
  image: entry.image,
}))

/** Number of catalogued varieties. */
export const CATALOG_SIZE = BASE_COLORS.length

/** Public path to a Farbschlag image (served from gerbil-manager-web/public). */
export function farbschlagImageUrl(image: string | undefined): string | null {
  return image ? `/farbschlaege/${image}` : null
}

/** Look up a catalog entry by exact variety name. */
export function findVariety(name: string): FarbschlagEntry | undefined {
  return BASE_COLORS.find((e) => e.name === name)
}