btrc-hub/frontend/src/lib/gekanator.ts

import { apiGet, apiPost } from '@/lib/api'

import type { Post } from '@/types'

export type GekanatorAnswerValue =
    | 'yes'
    | 'no'
    | 'partial'
    | 'probably_no'
    | 'unknown'

export type GekanatorAnswerLog = {
    questionId:       string
    questionText:     string
    questionCondition?:  GekanatorQuestionCondition
    answer:           GekanatorAnswerValue
    originalAnswer:   GekanatorAnswerValue }

export type GekanatorQuestionKind =
    | 'tag'
    | 'source'
    | 'title'
    | 'original_date'

export type GekanatorQuestionSource =
    | 'default'
    | 'user_suggested'
    | 'ai_generated'
    | 'admin_curated'

export type GekanatorQuestionCondition =
    | { type: 'tag'; key: string }
    | { type: 'source'; host: string }
    | { type: 'original-year'; year: number }
    | { type: 'original-month'; month: number }
    | { type: 'original-month-day'; monthDay: string }
    | { type: 'title-length-greater-than'; length: number }
    | { type: 'title-has-ascii' }

export type StoredGekanatorQuestion = {
    id:              string
    text:            string
    kind:            GekanatorQuestionKind
    condition:       GekanatorQuestionCondition
    source?:         GekanatorQuestionSource
    priorityWeight?: number }

export type GekanatorQuestion = {
    id:              string
    text:            string
    kind:            GekanatorQuestionKind
    condition:       GekanatorQuestionCondition
    source:          GekanatorQuestionSource
    priorityWeight:  number
    test:            (post: Post) => boolean }

const countBy = <T extends string | number> (values: T[]): Map<T, number> => {
  const counts = new Map<T, number> ()
  values.forEach (value => counts.set (value, (counts.get (value) ?? 0) + 1))
  return counts
}


const median = (values: number[]): number => {
  const sorted = [...values].sort ((a, b) => a - b)
  return sorted[Math.floor (sorted.length / 2)] ?? 0
}


const hostOf = (post: Post): string | null => {
  try
  {
    return new URL (post.url).hostname.replace (/^www\./, '')
  }
  catch
  {
    return null
  }
}


const originalYearOf = (post: Post): number | null => {
  const value = post.originalCreatedFrom || post.originalCreatedBefore
  if (!(value))
    return null

  const date = new Date (value)
  if (Number.isNaN (date.getTime ()))
    return null

  return date.getFullYear ()
}


const originalDateOf = (post: Post): Date | null => {
  const value = post.originalCreatedFrom || post.originalCreatedBefore
  if (!(value))
    return null

  const date = new Date (value)
  if (Number.isNaN (date.getTime ()))
    return null

  return date
}


const originalMonthOf = (post: Post): number | null => {
  const date = originalDateOf (post)
  if (!(date))
    return null

  return date.getMonth () + 1
}


const originalMonthDayOf = (post: Post): string | null => {
  const date = originalDateOf (post)
  if (!(date))
    return null

  return `${ date.getMonth () + 1 }-${ date.getDate () }`
}


const tagQuestionKey = ({ category, name }: { category: string; name: string }): string =>
    `${ category }:${ name }`


const tagFromQuestionKey = (key: string): { category: string; name: string } => {
  const [category, ...rest] = key.split (':')
  return { category: category ?? '', name: rest.join (':') }
}


const nicoTagLabel = (name: string): string => name.replace (/^nico:/, '')


const tagQuestionText = (category: string, label: string): string => {
  switch (category)
  {
  case 'deerjikist':
    return `作者・ニジラーとして「${ label }」に関係してゐる?`
  case 'meme':
    return `元ネタ・ミームとして「${ label }」に関係しさう?`
  case 'character':
    return `「${ label }」といふキャラクターが関係してゐる?`
  case 'material':
    return `素材として「${ label }」に関係してゐる?`
  case 'nico':
    return `ニコニコに「${ label }」といふタグが付いてゐる?`
  case 'general':
  case 'meta':
  default:
    return `内容として「${ label }」に関係しさう?`
  }
}


const questionableTag = (post: Post, key: string): boolean => {
  const { category, name } = tagFromQuestionKey (key)

  return (
    post.tags.some (tag =>
	tag.name === name
	&& tag.category === category
	&& !(tag.category === 'meta')
	&& !(tag.name.includes ('タグ希望'))
	&& !(tag.name.includes ('bot操作'))))
}


const questionMatches = (
    post:       Post,
    condition:  GekanatorQuestionCondition,
): boolean => {
  switch (condition.type)
  {
  case 'tag':
    return questionableTag (post, condition.key)
  case 'source':
    return hostOf (post) === condition.host
  case 'original-year':
    return originalYearOf (post) === condition.year
  case 'original-month':
    return originalMonthOf (post) === condition.month
  case 'original-month-day':
    return originalMonthDayOf (post) === condition.monthDay
  case 'title-length-greater-than':
    return (post.title?.length ?? 0) > condition.length
  case 'title-has-ascii':
    return /[A-Za-z0-9]/.test (post.title ?? '')
  }
}


export const restoreGekanatorQuestion = (
    question:  StoredGekanatorQuestion,
): GekanatorQuestion => ({
  ...question,
  source: question.source ?? 'default',
  priorityWeight: question.priorityWeight ?? 1,
  test: (post: Post) => questionMatches (post, question.condition) })


export const storeGekanatorQuestion = (
    question:  GekanatorQuestion,
): StoredGekanatorQuestion => ({
  id: question.id,
  text: question.text,
  kind: question.kind,
  condition: question.condition,
  source: question.source,
  priorityWeight: question.priorityWeight })


export const fetchGekanatorPosts = async (): Promise<Post[]> => {
  const data = await apiGet<{ posts: Post[] }> ('/gekanator/posts')
  return data.posts
}


export const fetchGekanatorQuestions = async (): Promise<StoredGekanatorQuestion[]> => {
  const data = await apiGet<{ questions: StoredGekanatorQuestion[] }> ('/gekanator/questions')
  return data.questions
}


export const buildGekanatorQuestions = (posts: Post[]): GekanatorQuestion[] => {
  const tagCounts = countBy (posts.flatMap (post =>
      post.tags
      .filter (tag =>
	  !(tag.category === 'meta')
	  && !(tag.name.includes ('タグ希望'))
	  && !(tag.name.includes ('bot操作')))
      .map (tag => tagQuestionKey (tag))))
  const hosts = countBy (posts.map (hostOf).filter ((host): host is string => Boolean (host)))
  const originalYears = countBy (
      posts
      .map (originalYearOf)
      .filter ((year): year is number => year !== null))
  const originalMonths = countBy (
      posts
      .map (originalMonthOf)
      .filter ((month): month is number => month !== null))
  const originalMonthDays = countBy (
      posts
      .map (originalMonthDayOf)
      .filter ((monthDay): monthDay is string => monthDay !== null))
  const titleLengthMedian = median (posts.map (post => post.title?.length ?? 0))

  const usefulEntries = <T extends string | number> (counts: Map<T, number>) =>
      [...counts.entries ()]
      .filter (([, count]) => count > 0 && count < posts.length)
      .sort ((a, b) => Math.abs (posts.length / 2 - a[1])
	       - Math.abs (posts.length / 2 - b[1]))
      .slice (0, 80)

  const tagQuestions = usefulEntries (tagCounts)
      .filter (([, count]) => count >= 2 && count <= Math.max (2, posts.length * .7))
      .slice (0, 80)
      .map (([key]) => {
	const { category, name } = tagFromQuestionKey (String (key))
	const label = category === 'nico' ? nicoTagLabel (name) : name

	return {
	    id:   `tag:${ key }`,
	    text: tagQuestionText (category, label),
	    kind: 'tag' as const,
	    condition: { type: 'tag' as const, key: String (key) },
	    source: 'default' as const,
	    priorityWeight: 1,
	    test: (post: Post) => questionableTag (post, String (key)) }
      })

  const sourceQuestions = usefulEntries (hosts)
      .filter (([, count]) => count >= 2 && count <= Math.max (2, posts.length * .7))
      .slice (0, 20)
      .map (([host]) => ({
	  id:   `source:${ host }`,
	  text: `${ host } の投稿を思ひ浮かべてゐる?`,
	  kind: 'source' as const,
	  condition: { type: 'source' as const, host },
	  source: 'default' as const,
	  priorityWeight: 1,
	  test: (post: Post) => hostOf (post) === host }))

  const originalYearQuestions = usefulEntries (originalYears)
      .filter (([, count]) => count >= 2 && count <= Math.max (2, posts.length * .7))
      .slice (0, 20)
      .map (([year]) => ({
	  id:   `original-year:${ year }`,
	  text: `オリジナルの投稿年は ${ year } 年?`,
	  kind: 'original_date' as const,
	  condition: { type: 'original-year' as const, year },
	  source: 'default' as const,
	  priorityWeight: 1,
	  test: (post: Post) => originalYearOf (post) === year }))

  const originalMonthQuestions = usefulEntries (originalMonths)
      .filter (([, count]) => count >= 2 && count <= Math.max (2, posts.length * .7))
      .slice (0, 20)
      .map (([month]) => ({
	  id:   `original-month:${ month }`,
	  text: `オリジナルの投稿月は ${ month } 月?`,
	  kind: 'original_date' as const,
	  condition: { type: 'original-month' as const, month },
	  source: 'default' as const,
	  priorityWeight: 1,
	  test: (post: Post) => originalMonthOf (post) === month }))

  const originalMonthDayQuestions = usefulEntries (originalMonthDays)
      .filter (([, count]) => count >= 2 && count <= Math.max (2, posts.length * .7))
      .slice (0, 20)
      .map (([monthDay]) => {
	const [month, day] = String (monthDay).split ('-')

	return {
	    id:   `original-month-day:${ monthDay }`,
	    text: `オリジナルの投稿日は ${ month } 月 ${ day } 日?`,
	    kind: 'original_date' as const,
	    condition: { type: 'original-month-day' as const, monthDay: String (monthDay) },
	    source: 'default' as const,
	    priorityWeight: 1,
	    test: (post: Post) => originalMonthDayOf (post) === monthDay }
      })

  const titleQuestions = [
      {
	  id:   'title:long',
	  text: '題名が長めの投稿?',
	  kind: 'title' as const,
	  condition: {
	      type: 'title-length-greater-than' as const,
	      length: titleLengthMedian },
	  source: 'default' as const,
	  priorityWeight: 1,
	  test: (post: Post) => (post.title?.length ?? 0) > titleLengthMedian },
      {
	  id:   'title:ascii',
	  text: '題名に英数字が混じってゐる?',
	  kind: 'title' as const,
	  condition: { type: 'title-has-ascii' as const },
	  source: 'default' as const,
	  priorityWeight: 1,
	  test: (post: Post) => /[A-Za-z0-9]/.test (post.title ?? '') }]
      .filter (question => {
	const yes = posts.filter (post => question.test (post)).length
	const no = posts.length - yes
	return yes >= 2 && no >= 2 && yes <= posts.length * .7 && no <= posts.length * .7
      })

  return [
      ...sourceQuestions,
      ...originalYearQuestions,
      ...originalMonthQuestions,
      ...originalMonthDayQuestions,
      ...titleQuestions,
      ...tagQuestions]
}


export const saveGekanatorGame = async ({
    guessedPostId,
    correctPostId,
    answers,
}: {
    guessedPostId:  number
    correctPostId:  number
    answers:        GekanatorAnswerLog[]
}): Promise<{ id: number }> =>
    await apiPost ('/gekanator/games', {
	guessed_post_id: guessedPostId,
	correct_post_id: correctPostId,
	answers: answers.map (answer => ({
	    question_id: answer.questionId,
	    question_text: answer.questionText,
	    question_condition: answer.questionCondition ?? null,
	    answer: answer.answer,
	    original_answer: answer.originalAnswer })) })


export const saveGekanatorQuestionSuggestion = async ({
    gekanatorGameId,
    questionText,
    answer,
}: {
    gekanatorGameId:  number
    questionText:     string
    answer:           GekanatorAnswerValue
}): Promise<{ id: number; count: number }> =>
    await apiPost ('/gekanator/question_suggestions', {
	gekanator_game_id: gekanatorGameId,
	question_text: questionText,
	answer })