49 lines
2.2 KiB
TypeScript
49 lines
2.2 KiB
TypeScript
import { describe, expect, it } from 'vitest'
|
|
import { evaluateEvidenceRules, normalizeEvidenceText, scoreEvidenceAnchor } from './evidenceMatching.js'
|
|
|
|
const barricelliRule = {
|
|
id: 'rule-barricelli',
|
|
name: 'Contemporary Barricelli fire report',
|
|
flagKey: 'barricelli.child-rescue-source',
|
|
minimumAnchorMatches: 1,
|
|
anchors: [{
|
|
id: 'parents',
|
|
phrase: 'den italienske maler og opfinder Barricelli og frue, født Aall',
|
|
minimumSimilarity: 0.72,
|
|
}, {
|
|
id: 'drink',
|
|
phrase: 'Han vækkede nemlig sin mor for at faa noget at drikke',
|
|
minimumSimilarity: 0.72,
|
|
}],
|
|
}
|
|
|
|
describe('evidence text matching', () => {
|
|
it('normalizes historical Norwegian characters and page layout noise', () => {
|
|
expect(normalizeEvidenceText('Født Aall — 2½ aar\n gammel')).toBe('fodt aall 2 1 2 aar gammel')
|
|
})
|
|
|
|
it('tolerates plausible OCR substitutions in a distinctive passage', () => {
|
|
const result = scoreEvidenceAnchor(
|
|
normalizeEvidenceText('I kvistleiligheden boede den italienske maler og opfinder Barrioelli og frue, født Aall, med sin lille søn.'),
|
|
barricelliRule.anchors[0].phrase,
|
|
)
|
|
expect(result.similarity).toBeGreaterThan(0.9)
|
|
})
|
|
|
|
it('awards the data-defined flag when one configured anchor is present', () => {
|
|
const evaluations = evaluateEvidenceRules('Han vækkede nemlig sin mor for at faa noget at drikke, og da ser hun huset brænder.', [barricelliRule])
|
|
expect(evaluations[0]).toMatchObject({ matched: true, matchedAnchorCount: 1, flagKey: 'barricelli.child-rescue-source' })
|
|
})
|
|
|
|
it('does not match generic words from an unrelated fire report', () => {
|
|
const evaluations = evaluateEvidenceRules('A family escaped from a boarding-house fire during the night.', [barricelliRule])
|
|
expect(evaluations[0]).toMatchObject({ matched: false, matchedAnchorCount: 0 })
|
|
})
|
|
|
|
it('can require multiple anchors for a stricter level rule', () => {
|
|
const rule = { ...barricelliRule, minimumAnchorMatches: 2 }
|
|
expect(evaluateEvidenceRules(barricelliRule.anchors[0].phrase, [rule])[0].matched).toBe(false)
|
|
expect(evaluateEvidenceRules(barricelliRule.anchors.map(anchor => anchor.phrase).join(' '), [rule])[0].matched).toBe(true)
|
|
})
|
|
})
|