#361

frontend/src/lib/gekanator.ts

@@ -13,6 +13,7 @@ export type GekanatorAnswerLog = {
     questionId:       string
     questionText:     string
     questionCondition?:  GekanatorQuestionCondition
+    questionMode?:    'normal' | 'winning_run'
     answer:           GekanatorAnswerValue
     originalAnswer:   GekanatorAnswerValue }
 
@@ -29,6 +30,8 @@ export type GekanatorQuestionSource =
     | 'ai_generated'
     | 'admin_curated'
 
+export type GekanatorPerformanceMode = 'lite' | 'normal'
+
 export type GekanatorQuestionCondition =
     | { type: 'tag'; key: string }
     | { type: 'source'; host: string }
@@ -38,6 +41,7 @@ export type GekanatorQuestionCondition =
     | { type: 'title-length-at-least'; length: number }
     | { type: 'title-length-greater-than'; length: number }
     | { type: 'title-has-ascii' }
+    | { type: 'title-contains'; text: string }
     | {
 	  type: 'post-similarity'
 	  postId: number
@@ -76,6 +80,13 @@ export type GekanatorQuestion = {
     exampleAnswers?: Record<`${ number }`, GekanatorAnswerValue>
     test:            (post: Post) => boolean }
 
+export type BuildGekanatorQuestionsOptions = {
+    includeTitleContains?: boolean
+    tagQuestionCap?: number
+    titleContainsCap?: number
+    totalQuestionCap?: number
+}
+
 
 export const normalizeTitleLengthCondition = (
     condition:  GekanatorQuestionCondition,
@@ -127,6 +138,8 @@ export const questionIdForCondition = (
     return `title:length-at-least:${ titleLengthMinimumForCondition (condition) }`
   case 'title-has-ascii':
     return 'title:ascii'
+  case 'title-contains':
+    return `title:contains:${ condition.text }`
   }
 }
 
@@ -292,6 +305,8 @@ const questionMatches = (
     return (post.title?.length ?? 0) > question.condition.length
   case 'title-has-ascii':
     return /[A-Za-z0-9]/.test (post.title ?? '')
+  case 'title-contains':
+    return (post.title ?? '').includes (question.condition.text)
   case 'post-similarity':
     return false
   }
@@ -319,6 +334,7 @@ export const expectedAnswerForQuestion = (
   case 'title-length-at-least':
   case 'title-length-greater-than':
   case 'title-has-ascii':
+  case 'title-contains':
     return questionMatches (post, question) ? 'yes' : 'no'
   case 'post-similarity':
     return null
@@ -382,7 +398,16 @@ export const fetchGekanatorExtraQuestions = async (
 }
 
 
-export const buildGekanatorQuestions = (posts: Post[]): GekanatorQuestion[] => {
+export const buildGekanatorQuestions = (
+    posts:    Post[],
+    options:  BuildGekanatorQuestionsOptions = { },
+): GekanatorQuestion[] => {
+  const {
+      includeTitleContains = true,
+      tagQuestionCap = 192,
+      titleContainsCap = 24,
+      totalQuestionCap = Number.POSITIVE_INFINITY,
+  } = options
   const tagCounts = countBy (posts.flatMap (post =>
       post.tags
       .filter (tag =>
@@ -404,17 +429,31 @@ export const buildGekanatorQuestions = (posts: Post[]): GekanatorQuestion[] => {
       .map (originalMonthDayOf)
       .filter ((monthDay): monthDay is string => monthDay !== null))
   const titleLengthMedian = median (posts.map (post => post.title?.length ?? 0))
+  const titleWordCounts =
+      includeTitleContains
+	? countBy (
+	    posts.flatMap (post =>
+		Array.from (
+		    new Set (
+			(post.title ?? '')
+			.match (
+			    /[\p{Script=Han}\p{Script=Hiragana}\p{Script=Katakana}A-Za-z0-9]{2,}/gu)
+			?? []))))
+	: new Map<string, number> ()
 
-  const usefulEntries = <T extends string | number> (counts: Map<T, number>) =>
+  const usefulEntries = <T extends string | number> (
+      counts:  Map<T, number>,
+      cap:     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)
+      .slice (0, cap)
 
-  const tagQuestions = usefulEntries (tagCounts)
+  const tagQuestions = usefulEntries (tagCounts, Math.max (tagQuestionCap, 80))
       .filter (([, count]) => count >= 2 && count <= Math.max (2, posts.length * .7))
-      .slice (0, 80)
+      .slice (0, tagQuestionCap)
       .map (([key]) => {
 	const { category, name } = tagFromQuestionKey (String (key))
 	const label = category === 'nico' ? nicoTagLabel (name) : name
@@ -429,7 +468,7 @@ export const buildGekanatorQuestions = (posts: Post[]): GekanatorQuestion[] => {
 	    test: (post: Post) => questionableTag (post, String (key)) }
       })
 
-  const sourceQuestions = usefulEntries (hosts)
+  const sourceQuestions = usefulEntries (hosts, 20)
       .filter (([, count]) => count >= 2 && count <= Math.max (2, posts.length * .7))
       .slice (0, 20)
       .map (([host]) => ({
@@ -441,7 +480,7 @@ export const buildGekanatorQuestions = (posts: Post[]): GekanatorQuestion[] => {
 	  priorityWeight: 1,
 	  test: (post: Post) => hostOf (post) === host }))
 
-  const originalYearQuestions = usefulEntries (originalYears)
+  const originalYearQuestions = usefulEntries (originalYears, 20)
       .filter (([, count]) => count >= 2 && count <= Math.max (2, posts.length * .7))
       .slice (0, 20)
       .map (([year]) => ({
@@ -453,7 +492,7 @@ export const buildGekanatorQuestions = (posts: Post[]): GekanatorQuestion[] => {
 	  priorityWeight: 1,
 	  test: (post: Post) => originalYearOf (post) === year }))
 
-  const originalMonthQuestions = usefulEntries (originalMonths)
+  const originalMonthQuestions = usefulEntries (originalMonths, 20)
       .filter (([, count]) => count >= 2 && count <= Math.max (2, posts.length * .7))
       .slice (0, 20)
       .map (([month]) => ({
@@ -465,7 +504,7 @@ export const buildGekanatorQuestions = (posts: Post[]): GekanatorQuestion[] => {
 	  priorityWeight: 1,
 	  test: (post: Post) => originalMonthOf (post) === month }))
 
-  const originalMonthDayQuestions = usefulEntries (originalMonthDays)
+  const originalMonthDayQuestions = usefulEntries (originalMonthDays, 20)
       .filter (([, count]) => count >= 2 && count <= Math.max (2, posts.length * .7))
       .slice (0, 20)
       .map (([monthDay]) => {
@@ -505,6 +544,23 @@ export const buildGekanatorQuestions = (posts: Post[]): GekanatorQuestion[] => {
 	const no = posts.length - yes
 	return yes >= 2 && no >= 2 && yes <= posts.length * .7 && no <= posts.length * .7
       })
+  const titleContainsQuestions =
+      includeTitleContains
+	? usefulEntries (titleWordCounts, titleContainsCap)
+	    .filter (([word, count]) =>
+		String (word).length <= 24
+		&& count >= 2
+		&& count <= Math.max (2, posts.length * .7))
+	    .slice (0, titleContainsCap)
+	    .map (([word]) => ({
+		id:   `title:contains:${ word }`,
+		text: `題名に「${ word }」が含まれる?`,
+		kind: 'title' as const,
+		condition: { type: 'title-contains' as const, text: String (word) },
+		source: 'default' as const,
+		priorityWeight: .96,
+		test: (post: Post) => (post.title ?? '').includes (String (word)) }))
+	: []
 
   return [
       ...sourceQuestions,
@@ -512,7 +568,8 @@ export const buildGekanatorQuestions = (posts: Post[]): GekanatorQuestion[] => {
       ...originalMonthQuestions,
       ...originalMonthDayQuestions,
       ...titleQuestions,
-      ...tagQuestions]
+      ...titleContainsQuestions,
+      ...tagQuestions].slice (0, totalQuestionCap)
 }
 
 

frontend/src/lib/gekanatorCandidateRecovery.ts

@@ -100,7 +100,7 @@ export const allConcreteAnswerOptionsExhausted = (
 }
 
 
-const nextRecoveryBatchSize = (recoveryStepCount: number): number =>
+const nextRecoveryTargetSize = (recoveryStepCount: number): number =>
     6 * (2 ** recoveryStepCount)
 
 
@@ -125,6 +125,16 @@ export const recoverCandidatePosts = ({
     recoveryStepCount:       number
 } | null => {
   const recovered = new Map (recoveredCandidatePosts)
+  const targetSize = nextRecoveryTargetSize (recoveryStepCount)
+  const countedPostIds = new Set ([
+      ...eligiblePostIds,
+      ...recovered.keys ()])
+  const addCount = targetSize - countedPostIds.size
+  if (addCount <= 0)
+    return {
+	recoveredCandidatePosts: recovered,
+	recoveryStepCount: recoveryStepCount + 1 }
+
   const candidates = posts
       .filter (post =>
 	  !(rejectedPostIds.has (post.id))
@@ -133,7 +143,7 @@ export const recoverCandidatePosts = ({
       .sort ((a, b) =>
 	  (scores.get (b.id) ?? Number.NEGATIVE_INFINITY)
 	      - (scores.get (a.id) ?? Number.NEGATIVE_INFINITY))
-      .slice (0, nextRecoveryBatchSize (recoveryStepCount))
+      .slice (0, addCount)
 
   if (candidates.length === 0)
     return null