グカネータ / 質問パターン見直し (#41) (#365)

Reviewed-on: #365
Co-authored-by: miteruzo <miteruzo@naver.com>
Co-committed-by: miteruzo <miteruzo@naver.com>

frontend/src/lib/gekanator.test.ts

@@ -2,6 +2,7 @@ import { beforeEach, describe, expect, it, vi } from 'vitest'
 
 import { apiPost } from '@/lib/api'
 import {
+  buildGekanatorQuestions,
   expectedAnswerForQuestion,
   restoreGekanatorQuestion,
   saveGekanatorExtraQuestionAnswers,
@@ -146,6 +147,23 @@ describe('expectedAnswerForQuestion', () => {
 
     expect(expectedAnswerForQuestion(question, post({ tags: [] }))).toBe('no')
   })
+
+  it('ignores example answers for direct title facts', () => {
+    const question: StoredGekanatorQuestion = {
+      id: 'title:length-at-least:20',
+      text: 'タイトルは 20 文字以上?',
+      kind: 'title',
+      condition: {
+        type: 'title-length-at-least',
+        length: 20,
+      },
+      exampleAnswers: {
+        1: 'yes',
+      },
+    }
+
+    expect(expectedAnswerForQuestion(question, post({ id: 1, title: 'short' }))).toBe('no')
+  })
 })
 
 describe('restoreGekanatorQuestion', () => {
@@ -187,6 +205,65 @@ describe('restoreGekanatorQuestion', () => {
     expect(question.test(post({ id: 2 }))).toBe(false)
     expect(question.test(post({ id: 3 }))).toBe(false)
   })
+
+  it('tests a post_similarity question against its configured partial answer', () => {
+    const question = restoreGekanatorQuestion({
+      id: 'post-similarity:10',
+      text: '喜多ちゃんが泣いてる?',
+      kind: 'post_similarity',
+      source: 'user_suggested',
+      priorityWeight: 1.2,
+      condition: {
+        type: 'post-similarity',
+        postId: 999,
+        answer: 'partial',
+        threshold: 0.65,
+      },
+      exampleAnswers: {
+        1: 'partial',
+        2: 'yes',
+      },
+    })
+
+    expect(question.test(post({ id: 1 }))).toBe(true)
+    expect(question.test(post({ id: 2 }))).toBe(false)
+  })
+
+  it('normalizes legacy title-length-greater-than questions', () => {
+    const question = restoreGekanatorQuestion({
+      id: 'title:length-greater-than:20',
+      text: '題名が長めの投稿?',
+      kind: 'title',
+      condition: {
+        type: 'title-length-greater-than',
+        length: 20,
+      },
+    })
+
+    expect(question.id).toBe('title:length-at-least:21')
+    expect(question.condition).toEqual({
+      type: 'title-length-at-least',
+      length: 21,
+    })
+    expect(question.test(post({ title: 'x'.repeat(20) }))).toBe(false)
+    expect(question.test(post({ title: 'x'.repeat(21) }))).toBe(true)
+  })
+})
+
+describe('buildGekanatorQuestions', () => {
+  it('builds quantitative title length questions', () => {
+    const questions = buildGekanatorQuestions([
+      post({ id: 1, title: 'a' }),
+      post({ id: 2, title: 'bb' }),
+      post({ id: 3, title: 'ccc' }),
+      post({ id: 4, title: 'dddd' }),
+    ])
+    const titleQuestion = questions.find(question =>
+      question.condition.type === 'title-length-at-least')
+
+    expect(titleQuestion?.text).toMatch(/^タイトルは \d+ 文字以上\?$/)
+    expect(titleQuestion?.id).toMatch(/^title:length-at-least:\d+$/)
+  })
 })
 
 describe('Gekanator API writers', () => {

frontend/src/lib/gekanator.ts

@@ -372,10 +372,12 @@ export const fetchGekanatorQuestions = async (): Promise<StoredGekanatorQuestion
 
 
 export const fetchGekanatorExtraQuestions = async (
-    gameId: number,
+    gameId:  number,
+    nonce?: string,
 ): Promise<GekanatorExtraQuestion[]> => {
   const data = await apiGet<{ questions: GekanatorExtraQuestion[] }> (
-      `/gekanator/games/${ gameId }/extra_questions`)
+      `/gekanator/games/${ gameId }/extra_questions`,
+      { params: nonce ? { nonce } : undefined })
   return data.questions
 }
 

frontend/src/lib/gekanatorCandidateRecovery.test.ts

@@ -0,0 +1,151 @@
+import { describe, expect, it } from 'vitest'
+
+import {
+    candidatePostsFor,
+    hardFilteredPostsForAnswer,
+    recoverCandidatePosts,
+} from '@/lib/gekanatorCandidateRecovery'
+
+import type {
+    GekanatorAnswerLog,
+    GekanatorAnswerValue,
+    GekanatorQuestion,
+} from '@/lib/gekanator'
+import type { Post } from '@/types'
+
+
+const post = (id: number): Post => ({
+  id,
+  versionNo: 1,
+  url: `https://example.com/posts/${ id }`,
+  title: `post ${ id }`,
+  thumbnail: null,
+  thumbnailBase: null,
+  tags: [],
+  viewed: false,
+  related: [],
+  originalCreatedFrom: null,
+  originalCreatedBefore: null,
+  createdAt: '2026-06-10T00:00:00.000Z',
+  updatedAt: '2026-06-10T00:00:00.000Z',
+  uploadedUser: null,
+})
+
+
+const postSimilarityQuestion = (
+    id:       string,
+    answers:  Record<`${ number }`, GekanatorAnswerValue>,
+): GekanatorQuestion => ({
+  id,
+  text: `${ id }?`,
+  kind: 'post_similarity',
+  condition: {
+      type: 'post-similarity',
+      postId: 9999,
+      answer: 'yes',
+      threshold: 0.65 },
+  source: 'user_suggested',
+  priorityWeight: 1,
+  exampleAnswers: answers,
+  test: candidate => answers[String (candidate.id) as `${ number }`] === 'yes',
+})
+
+
+const answer = (
+    question:  GekanatorQuestion,
+    value:     GekanatorAnswerValue,
+): GekanatorAnswerLog => ({
+  questionId: question.id,
+  questionText: question.text,
+  questionCondition: question.condition,
+  answer: value,
+  originalAnswer: value,
+})
+
+
+describe('candidatePostsFor', () => {
+  it('lets recovered candidates ignore old answers but not later answers', () => {
+    const posts = [post (1), post (2), post (3)]
+    const oldQuestion = postSimilarityQuestion ('old', {
+	1: 'no',
+	2: 'yes',
+	3: 'yes',
+    })
+    const laterQuestion = postSimilarityQuestion ('later', {
+	1: 'no',
+	2: 'no',
+	3: 'yes',
+    })
+
+    const candidates = candidatePostsFor ({
+	posts,
+	questions: [oldQuestion, laterQuestion],
+	answers: [answer (oldQuestion, 'yes'), answer (laterQuestion, 'yes')],
+	softenedQuestionIds: new Set (),
+	rejectedPostIds: new Set (),
+	recoveredCandidatePosts: new Map ([
+	    [1, 1],
+	    [3, 1],
+	]) })
+
+    expect(candidates.map (candidate => candidate.id)).toEqual ([3])
+  })
+
+  it('does not let recovered candidates bypass explicit rejected posts', () => {
+    const posts = [post (1), post (2)]
+    const question = postSimilarityQuestion ('question', {
+	1: 'yes',
+	2: 'yes',
+    })
+
+    const candidates = candidatePostsFor ({
+	posts,
+	questions: [question],
+	answers: [answer (question, 'yes')],
+	softenedQuestionIds: new Set (),
+	rejectedPostIds: new Set ([1]),
+	recoveredCandidatePosts: new Map ([[1, 1]]) })
+
+    expect(candidates.map (candidate => candidate.id)).toEqual ([2])
+  })
+})
+
+
+describe('hardFilteredPostsForAnswer', () => {
+  it('returns zero candidates without falling back to the original pool', () => {
+    const posts = [post (1), post (2)]
+    const question = postSimilarityQuestion ('question', {
+	1: 'yes',
+	2: 'yes',
+    })
+
+    expect(hardFilteredPostsForAnswer ({
+	posts,
+	question,
+	answer: 'no',
+    })).toEqual ([])
+  })
+})
+
+
+describe('recoverCandidatePosts', () => {
+  it('recovers high-score non-rejected, non-eligible candidates in staged batches', () => {
+    const posts = Array.from ({ length: 10 }, (_value, index) => post (index + 1))
+    const scores = new Map (posts.map (candidate => [candidate.id, candidate.id]))
+
+    const recovered = recoverCandidatePosts ({
+	posts,
+	scores,
+	rejectedPostIds: new Set ([10]),
+	recoveredCandidatePosts: new Map ([[8, 1]]),
+	eligiblePostIds: new Set ([9]),
+	answerCountAtRecovery: 2,
+	recoveryStepCount: 0,
+    })
+
+    expect(recovered?.recoveryStepCount).toBe (1)
+    expect([...(recovered?.recoveredCandidatePosts.keys () ?? [])])
+      .toEqual ([8, 7, 6, 5, 4, 3, 2])
+    expect(recovered?.recoveredCandidatePosts.get (7)).toBe (2)
+  })
+})

frontend/src/lib/gekanatorCandidateRecovery.ts

@@ -0,0 +1,146 @@
+import { expectedAnswerForQuestion } from '@/lib/gekanator'
+
+import type {
+    GekanatorAnswerLog,
+    GekanatorAnswerValue,
+    GekanatorQuestion,
+} from '@/lib/gekanator'
+import type { Post } from '@/types'
+
+
+export type RecoveredCandidatePost = {
+    postId:                number
+    answerCountAtRecovery: number }
+
+
+export const candidatePostsFor = ({
+    posts,
+    questions,
+    answers,
+    softenedQuestionIds,
+    rejectedPostIds,
+    recoveredCandidatePosts,
+}: {
+    posts:                   Post[]
+    questions:               GekanatorQuestion[]
+    answers:                 GekanatorAnswerLog[]
+    softenedQuestionIds:     Set<string>
+    rejectedPostIds:         Set<number>
+    recoveredCandidatePosts: Map<number, number>
+}): Post[] => {
+  const questionById = new Map (questions.map (question => [question.id, question]))
+
+  return posts.filter (post => {
+    if (rejectedPostIds.has (post.id))
+      return false
+
+    const answerCountAtRecovery = recoveredCandidatePosts.get (post.id)
+
+    return answers.every ((answer, index) => {
+      if (answerCountAtRecovery !== undefined && index < answerCountAtRecovery)
+	return true
+
+      if (softenedQuestionIds.has (answer.questionId))
+	return true
+
+      const question = questionById.get (answer.questionId)
+      if (!(question))
+	return true
+
+      switch (answer.answer)
+      {
+      case 'yes':
+      case 'no': {
+	const expected = expectedAnswerForQuestion (question, post)
+	return expected === null || expected === 'unknown' || expected === answer.answer
+      }
+      default:
+	return true
+      }
+    })
+  })
+}
+
+
+export const hardFilteredPostsForAnswer = ({
+    posts,
+    question,
+    answer,
+}: {
+    posts:     Post[]
+    question:  GekanatorQuestion
+    answer:    GekanatorAnswerValue
+}): Post[] => {
+  if (answer === 'unknown')
+    return posts
+
+  return posts.filter (post => {
+    const expected = expectedAnswerForQuestion (question, post)
+    return expected === null || expected === 'unknown' || expected === answer
+  })
+}
+
+
+const concreteAnswerOptions: GekanatorAnswerValue[] = [
+    'yes',
+    'no',
+    'partial',
+    'probably_no']
+
+
+export const allConcreteAnswerOptionsExhausted = (
+    posts:      Post[],
+    question:   GekanatorQuestion | null,
+): boolean => {
+  if (!(question))
+    return false
+
+  return concreteAnswerOptions.every (answer =>
+      hardFilteredPostsForAnswer ({ posts, question, answer }).length === 0)
+}
+
+
+const nextRecoveryBatchSize = (recoveryStepCount: number): number =>
+    6 * (2 ** recoveryStepCount)
+
+
+export const recoverCandidatePosts = ({
+    posts,
+    scores,
+    rejectedPostIds,
+    recoveredCandidatePosts,
+    eligiblePostIds,
+    answerCountAtRecovery,
+    recoveryStepCount,
+}: {
+    posts:                   Post[]
+    scores:                  Map<number, number>
+    rejectedPostIds:         Set<number>
+    recoveredCandidatePosts: Map<number, number>
+    eligiblePostIds:         Set<number>
+    answerCountAtRecovery:   number
+    recoveryStepCount:       number
+}): {
+    recoveredCandidatePosts: Map<number, number>
+    recoveryStepCount:       number
+} | null => {
+  const recovered = new Map (recoveredCandidatePosts)
+  const candidates = posts
+      .filter (post =>
+	  !(rejectedPostIds.has (post.id))
+	  && !(eligiblePostIds.has (post.id))
+	  && !(recovered.has (post.id)))
+      .sort ((a, b) =>
+	  (scores.get (b.id) ?? Number.NEGATIVE_INFINITY)
+	      - (scores.get (a.id) ?? Number.NEGATIVE_INFINITY))
+      .slice (0, nextRecoveryBatchSize (recoveryStepCount))
+
+  if (candidates.length === 0)
+    return null
+
+  candidates.forEach (post => recovered.set (post.id, answerCountAtRecovery))
+
+  return {
+      recoveredCandidatePosts: recovered,
+      recoveryStepCount: recoveryStepCount + 1 }
+}

frontend/src/lib/gekanatorQuestionFilters.ts

@@ -0,0 +1,167 @@
+import { titleLengthMinimumForCondition } from '@/lib/gekanator'
+
+import type {
+    GekanatorAnswerLog,
+    GekanatorAnswerValue,
+    GekanatorQuestion,
+} from '@/lib/gekanator'
+
+
+export const monthForCondition = (
+    condition:  GekanatorQuestion['condition'],
+): number | null => {
+  if (condition.type === 'original-month')
+    return condition.month
+
+  if (condition.type !== 'original-month-day')
+    return null
+
+  const month = Number (condition.monthDay.split ('-')[0])
+  return Number.isInteger (month) ? month : null
+}
+
+
+const isTitleLengthContradiction = (
+    candidate:  GekanatorQuestion['condition'],
+    previous:   GekanatorQuestion['condition'],
+    answer:     GekanatorAnswerValue,
+): boolean => {
+  const candidateLength = titleLengthMinimumForCondition (candidate)
+  const previousLength = titleLengthMinimumForCondition (previous)
+  if (candidateLength === null || previousLength === null)
+    return false
+
+  switch (answer)
+  {
+  case 'yes':
+    return candidateLength <= previousLength
+  case 'no':
+    return candidateLength >= previousLength
+  default:
+    return false
+  }
+}
+
+
+const isQuestionRedundantAfterAnswers = (
+    question:  GekanatorQuestion,
+    answers:   GekanatorAnswerLog[],
+): boolean => answers.some (answer => {
+  const previous = answer.questionCondition
+  return previous !== undefined
+      && isTitleLengthContradiction (question.condition, previous, answer.answer)
+})
+
+
+const isSourceFactBlocked = (
+    candidate:  GekanatorQuestion['condition'],
+    previous:   GekanatorQuestion['condition'],
+    answer:     GekanatorAnswerValue,
+): boolean => {
+  if (candidate.type !== 'source' || previous.type !== 'source')
+    return false
+
+  switch (answer)
+  {
+  case 'yes':
+    return true
+  case 'no':
+    return candidate.host === previous.host
+  default:
+    return false
+  }
+}
+
+
+const isOriginalYearFactBlocked = (
+    candidate:  GekanatorQuestion['condition'],
+    previous:   GekanatorQuestion['condition'],
+    answer:     GekanatorAnswerValue,
+): boolean => {
+  if (candidate.type !== 'original-year' || previous.type !== 'original-year')
+    return false
+
+  switch (answer)
+  {
+  case 'yes':
+    return true
+  case 'no':
+    return candidate.year === previous.year
+  default:
+    return false
+  }
+}
+
+
+const isOriginalMonthFactBlocked = (
+    candidate:  GekanatorQuestion['condition'],
+    previous:   GekanatorQuestion['condition'],
+    answer:     GekanatorAnswerValue,
+): boolean => {
+  switch (answer)
+  {
+  case 'yes':
+    if (previous.type === 'original-month')
+      {
+	if (candidate.type === 'original-month')
+	  return true
+
+	if (candidate.type === 'original-month-day')
+	  return monthForCondition (candidate) !== previous.month
+
+	return false
+      }
+
+    if (previous.type === 'original-month-day')
+      return candidate.type === 'original-month'
+	  || candidate.type === 'original-month-day'
+
+    return false
+  case 'no':
+    if (previous.type === 'original-month')
+      {
+	if (candidate.type === 'original-month')
+	  return candidate.month === previous.month
+
+	if (candidate.type === 'original-month-day')
+	  return monthForCondition (candidate) === previous.month
+
+	return false
+      }
+
+    if (previous.type === 'original-month-day')
+      return candidate.type === 'original-month-day'
+	  && candidate.monthDay === previous.monthDay
+
+    return false
+  default:
+    return false
+  }
+}
+
+
+const isFactQuestionBlocked = (
+    candidate:  GekanatorQuestion['condition'],
+    previous:   GekanatorQuestion['condition'],
+    answer:     GekanatorAnswerValue,
+): boolean => {
+  if (!(answer === 'yes' || answer === 'no'))
+    return false
+
+  return isSourceFactBlocked (candidate, previous, answer)
+      || isOriginalYearFactBlocked (candidate, previous, answer)
+      || isOriginalMonthFactBlocked (candidate, previous, answer)
+}
+
+
+export const isQuestionHardFilteredAfterAnswers = (
+    question:  GekanatorQuestion,
+    answers:   GekanatorAnswerLog[],
+): boolean => answers.some (answer => {
+  const previous = answer.questionCondition
+  if (previous === undefined)
+    return false
+
+  return isQuestionRedundantAfterAnswers (question, [answer])
+      || isFactQuestionBlocked (question.condition, previous, answer.answer)
+})

frontend/src/lib/queryKeys.ts

@@ -12,8 +12,8 @@ export const gekanatorKeys = {
     root:      ['gekanator'] as const,
     posts:     () => ['gekanator', 'posts'] as const,
     questions: () => ['gekanator', 'questions'] as const,
-    extraQuestions: (gameId: number) =>
-		       ['gekanator', 'games', gameId, 'extra-questions'] as const }
+    extraQuestions: (gameId: number, nonce: string) =>
+		       ['gekanator', 'games', gameId, 'extra-questions', nonce] as const }
 
 export const tagsKeys = {
     root:         ['tags'] as const,