gupi-osint-board/server/evidenceMatching.test.ts

import { describe, expect, it } from 'vitest'
import { readFileSync } from 'node:fs'
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,
  }],
}

const sceneSevenFixture = (name: string) => readFileSync(
  new URL(`../mysteries/barricelli-scene-7/fixtures/${name}`, import.meta.url),
  'utf8',
)
const sceneSevenManifest = JSON.parse(readFileSync(
  new URL('../mysteries/barricelli-scene-7/mystery.json', import.meta.url),
  'utf8',
)) as {
  documents: unknown[]
  goals: { key:string;requiredFlags:string[] }[]
  evidenceMatchRules: { name:string;flagKey:string;minimumAnchorMatches:number;anchors:{ phrase:string;minimumSimilarity:number }[] }[]
}
const authoredPatentRule = sceneSevenManifest.evidenceMatchRules[0]
const patentRule = { id:'rule-patent',...authoredPatentRule,
  anchors:authoredPatentRule.anchors.map((anchor,index) => ({ id:`anchor-${index}`,...anchor })) }

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)
  })

  it('accepts the Scene 7 Google Patents OCR fixture', () => {
    const evaluation = evaluateEvidenceRules(sceneSevenFixture('google-patents-target-ocr.txt'), [patentRule])[0]
    expect(evaluation).toMatchObject({ matched: true, flagKey: 'scene7.nils_inventor_proved' })
    expect(evaluation.matchedAnchorCount).toBeGreaterThanOrEqual(2)
  })

  it('keeps the Scene 7 assignment empty and ties its goal to the source flag', () => {
    expect(sceneSevenManifest.documents).toEqual([])
    expect(sceneSevenManifest.goals).toEqual([
      expect.objectContaining({ key:'barricelli.inventor-proof',requiredFlags:['scene7.nils_inventor_proved'] }),
    ])
    expect(authoredPatentRule.flagKey).toBe('scene7.nils_inventor_proved')
  })

  it('does not confuse the father-only source with proof about Nils', () => {
    expect(evaluateEvidenceRules(sceneSevenFixture('father-only-negative-ocr.txt'), [patentRule])[0])
      .toMatchObject({ matched: false })
  })

  it('tolerates a cropped, line-broken patent result without accepting unrelated patents or empty OCR', () => {
    const cropped = 'GB 695913 A — Improved chest of drawers\nInventor: Nils Aall Barri-\ncelli'
    expect(evaluateEvidenceRules(cropped, [patentRule])[0].matched).toBe(true)
    expect(evaluateEvidenceRules('US123456A Improved umbrella stand — Inventor Ada Example', [patentRule])[0].matched).toBe(false)
    expect(evaluateEvidenceRules('', [patentRule])[0].matched).toBe(false)
  })
})