FEAT: genotype parsing resilience and displayGenotypeSafe format helper

This commit is contained in:
2026-06-08 21:19:11 +02:00
parent 7d8166c555
commit 04d189bd2f
5 changed files with 83 additions and 7 deletions

View File

@@ -16,6 +16,7 @@ import {
fromJSON,
wildType,
extractGenotypeFlags,
displayGenotypeSafe,
} from '../genotype'
import { combineLocus } from '../punnett'
import { LOCI, type LocusKey } from '../loci'
@@ -686,4 +687,48 @@ describe('GEN-3h: breeder bracket-notation display + E-locus e-before-ef order',
expect(g.Sp).toEqual(['Sp', 'sp'])
expect(toDisplayString(g)).toBe('aa c[chm]c[chm] Dd e- Gg Pp Spsp')
})
// ── GENOTYPE-PARSE-CRASH / displayGenotypeSafe resilience ───────────────
it('handles blank c locus and c[-] resiliently without crashing', () => {
// Vandana's genotype
const rawVandana = 'Aa Cc Dd eef Gg P? spsp ??'
const g = fromDisplayString(rawVandana)
expect(g.C).toEqual(['C', '?'])
expect(toDisplayString(g)).toBe('Aa C- Dd ee[f] Gg P- spsp')
// cchmc allele
const g2 = fromDisplayString('Aa cchmc Dd ee Gg Pp spsp')
expect(g2.C).toEqual(['cchm', '?'])
expect(toDisplayString(g2)).toBe('Aa c[chm]- Dd ee Gg Pp spsp')
// c[-] and blank c standalone
const g3 = fromDisplayString('Aa c[-] Dd ee Gg Pp spsp')
expect(g3.C).toEqual(['C', 'C']) // default to wild-type since c[-] was skipped
expect(toDisplayString(g3)).toBe('Aa CC Dd ee Gg Pp spsp')
const g4 = fromDisplayString('Aa c Dd ee Gg Pp spsp')
expect(g4.C).toEqual(['C', 'C']) // default to wild-type since c was skipped
expect(toDisplayString(g4)).toBe('Aa CC Dd ee Gg Pp spsp')
})
it('displayGenotypeSafe formats parsed genotypes and handles fallback safely', () => {
// 1. Parseable with unknown alleles, replacing ? with -
expect(displayGenotypeSafe('Aa C? Dd')).toBe('Aa C- Dd EE GG PP spsp')
// 2. Parseable with trailing ?? (stripped completely)
expect(displayGenotypeSafe('Aa CC Dd EE GG Pp spsp ??')).toBe('Aa CC Dd EE GG Pp spsp')
expect(displayGenotypeSafe('Aa Cc Dd eef Gg P? spsp ??')).toBe('Aa C- Dd ee[f] Gg P- spsp')
expect(displayGenotypeSafe('Aa CC Dd EE GG Pp spsp Slsl ??')).toBe('Aa CC Dd EE GG Pp spsp Slsl')
// 3. Unparseable fallback: removes ?? / -- / [-], replaces remaining ? with -
expect(displayGenotypeSafe('Aa Cc XX ??')).toBe('Aa Cc XX')
expect(displayGenotypeSafe('Aa Cc XX --')).toBe('Aa Cc XX')
expect(displayGenotypeSafe('Aa Cc XX [-]')).toBe('Aa Cc XX')
expect(displayGenotypeSafe('Aa Cc? XX')).toBe('Aa Cc- XX')
// 4. Handles null/undefined/empty
expect(displayGenotypeSafe(null)).toBe('')
expect(displayGenotypeSafe(undefined)).toBe('')
expect(displayGenotypeSafe('')).toBe('')
})
})

View File

@@ -192,6 +192,11 @@ function normalizeToken(tok: string): string | null {
// the generic [-]→? rule below (which makes the bracket-dash a wildcard,
// leaving the leading allele intact for splitToken).
t = t.replace(/(?<=[A-Za-z])e\[-\]/g, '?')
t = t.replace(/(?<=[A-Za-z])c\[-\]/g, '?')
t = t.replace(/(?<=[A-Za-z])c$/g, '?')
t = t.replace(/^c\[-\]$/g, '??')
t = t.replace(/^c\?$/g, '??')
t = t.replace(/^c$/g, '??')
t = t.replace(/\[-\]/g, '?') // bare/standalone bracket-unknown → wildcard
// GEN-3c: plain dash is the breeder's UNKNOWN marker on input; store internally as '?'.
t = t.replace(/-/g, '?')
@@ -257,6 +262,25 @@ export function fromDisplayString(input: string): Genotype {
return makeGenotype({ ...base, ...acc })
}
export function displayGenotypeSafe(raw: string | null | undefined): string {
if (!raw) return ''
try {
const parsed = fromDisplayString(raw)
return toDisplayString(parsed)
} catch {
// Fallback path:
return raw
.trim()
.split(/\s+/)
.filter((tok) => {
const norm = normalizeToken(tok)
return norm !== null && norm !== '?' && norm !== '??'
})
.map((tok) => tok.replace(/\?/g, '-'))
.join(' ')
}
}
export function hasUnknown(g: Genotype): boolean {
return LOCUS_ORDER.some((l) => g[l][0] === WILDCARD || g[l][1] === WILDCARD)
}

View File

@@ -20,6 +20,7 @@ export {
toJSON,
fromJSON,
hasUnknown,
displayGenotypeSafe,
WILDCARD,
} from './genotype'
export type { Genotype, AllelePair } from './genotype'