@k2works/claude-code-booster 3.4.1 → 3.6.0

package/lib/assets/docs/article/grokkingfp/typescript/part-2.md

@@ -1,1129 +1,1129 @@
-# Part II: 関数型スタイルのプログラミング
-
-本章では、関数型プログラミングの核心となるテクニックを学びます。イミュータブルなデータ操作、高階関数、そして `chain`（flatMap）による複雑なデータ変換を習得します。
-
----
-
-## 第3章: イミュータブルなデータ操作
-
-### 3.1 イミュータブルとは
-
-イミュータブル（不変）とは、一度作成されたデータが変更されないことを意味します。データを「変更」する代わりに、新しいデータを「作成」します。
-
-```plantuml
-@startuml
-!theme plain
-
-rectangle "ミュータブル（可変）" as mut {
-  card "list = [A, B]" as m1
-  card "list.push(C)" as m2
-  card "list = [A, B, C]" as m3
-  m1 --> m2 : 変更
-  m2 --> m3 : 同じ配列が変わる
-}
-
-rectangle "イミュータブル（不変）" as immut {
-  card "list1 = [A, B]" as i1
-  card "list2 = [...list1, C]" as i2
-  card "list1 = [A, B]\nlist2 = [A, B, C]" as i3
-  i1 --> i2 : 新規作成
-  i2 --> i3 : 両方が存在
-}
-
-@enduml
-```
-
-TypeScript では `readonly` 修飾子と fp-ts の `ReadonlyArray` モジュールを使ってイミュータブルなデータを扱います。
-
-### 3.2 配列の基本操作
-
-**ソースファイル**: `app/typescript/src/ch03_immutable_data.ts`
-
-#### appended - 要素の追加
-
-```typescript
-// スプレッド演算子を使ったイミュータブルな追加
-const appended = <T>(lst: readonly T[], element: T): readonly T[] =>
-  [...lst, element]
-
-const appleBook = ['Apple', 'Book'] as const
-const appleBookMango = appended(appleBook, 'Mango')
-
-// appleBook.length === 2       // 元のリストは変わらない
-// appleBookMango.length === 3  // 新しいリストが作成される
-// appleBookMango === ['Apple', 'Book', 'Mango']
-```
-
-#### slice - リストの切り出し
-
-```typescript
-const firstTwo = <T>(lst: readonly T[]): readonly T[] =>
-  lst.slice(0, 2)
-
-const lastTwo = <T>(lst: readonly T[]): readonly T[] =>
-  lst.slice(Math.max(0, lst.length - 2))
-
-// firstTwo(['a', 'b', 'c']) === ['a', 'b']
-// lastTwo(['a', 'b', 'c']) === ['b', 'c']
-```
-
-```plantuml
-@startuml
-!theme plain
-
-rectangle "slice(start, end)" {
-  card "['a', 'b', 'c', 'd']" as orig
-  card "slice(1, 3)" as op
-  card "['b', 'c']" as result
-
-  orig --> op
-  op --> result
-}
-
-note bottom of result
-  インデックス 1 から 3 の手前まで
-  (end は含まない)
-end note
-
-@enduml
-```
-
-### 3.3 リストの変換例
-
-**ソースファイル**: `app/typescript/src/ch03_immutable_data.ts`
-
-```typescript
-// 最初の2要素を末尾に移動
-const moveFirstTwoToEnd = <T>(lst: readonly T[]): readonly T[] => {
-  const firstTwo = lst.slice(0, 2)
-  const withoutFirstTwo = lst.slice(2)
-  return [...withoutFirstTwo, ...firstTwo]
-}
-
-// moveFirstTwoToEnd(['a', 'b', 'c']) === ['c', 'a', 'b']
-
-// 最後の要素の前に挿入
-const insertBeforeLast = <T>(lst: readonly T[], element: T): readonly T[] => {
-  if (lst.length === 0) return [element]
-  const last = lst.slice(lst.length - 1)
-  const withoutLast = lst.slice(0, lst.length - 1)
-  return [...withoutLast, element, ...last]
-}
-
-// insertBeforeLast(['a', 'b'], 'c') === ['a', 'c', 'b']
-```
-
-### 3.4 旅程の再計画
-
-**ソースファイル**: `app/typescript/src/ch03_immutable_data.ts`
-
-旅行の計画変更をイミュータブルに行う例です。
-
-```typescript
-const replan = (
-  plan: readonly string[],
-  newCity: string,
-  beforeCity: string
-): readonly string[] => {
-  const index = plan.indexOf(beforeCity)
-  if (index === -1) return [...plan, newCity]
-  return [...plan.slice(0, index), newCity, ...plan.slice(index)]
-}
-
-const planA = ['Paris', 'Berlin', 'Kraków']
-const planB = replan(planA, 'Vienna', 'Kraków')
-
-// planB === ['Paris', 'Berlin', 'Vienna', 'Kraków']
-// planA === ['Paris', 'Berlin', 'Kraków']  // 元の計画は変わらない!
-```
-
-```plantuml
-@startuml
-!theme plain
-
-rectangle "旅程の再計画" {
-  card "Plan A:\nParis → Berlin → Kraków" as planA
-  card "replan(planA, 'Vienna', 'Kraków')" as op
-  card "Plan B:\nParis → Berlin → Vienna → Kraków" as planB
-
-  planA --> op
-  op --> planB
-}
-
-note right of planA
-  Plan A は変更されない
-  (イミュータブル)
-end note
-
-@enduml
-```
-
-### 3.5 イミュータブルなオブジェクト
-
-TypeScript では `readonly` 修飾子を使ってイミュータブルな型を定義します。
-
-```typescript
-interface Point {
-  readonly x: number
-  readonly y: number
-}
-
-const createPoint = (x: number, y: number): Point => ({ x, y })
-
-// スプレッド演算子でコピーしながら更新
-const withX = (point: Point, newX: number): Point => ({ ...point, x: newX })
-const withY = (point: Point, newY: number): Point => ({ ...point, y: newY })
-
-const p1 = createPoint(1, 2)
-const p2 = withX(p1, 10)
-
-// p1.x === 1   // 元のPointは変わらない
-// p2.x === 10
-// p2.y === 2
-```
-
-### 3.6 fp-ts を使ったイミュータブル操作
-
-fp-ts の `ReadonlyArray` モジュールを使うと、より型安全にイミュータブル操作ができます。
-
-```typescript
-import { pipe } from 'fp-ts/function'
-import * as RA from 'fp-ts/ReadonlyArray'
-
-// append - 要素を追加
-const appendWithFpts = <T>(lst: readonly T[], element: T): readonly T[] =>
-  pipe(lst, RA.append(element))
-
-// prepend - 先頭に追加
-const prependWithFpts = <T>(lst: readonly T[], element: T): readonly T[] =>
-  pipe(lst, RA.prepend(element))
-
-// concat - 連結
-const concatWithFpts = <T>(
-  lst1: readonly T[],
-  lst2: readonly T[]
-): readonly T[] =>
-  pipe(lst1, RA.concat(lst2))
-```
-
----
-
-## 第4章: 関数を値として扱う
-
-### 4.1 高階関数とは
-
-高階関数（Higher-Order Function）とは、以下のいずれかを満たす関数です:
-
-1. 関数を引数として受け取る
-2. 関数を戻り値として返す
-
-```plantuml
-@startuml
-!theme plain
-
-rectangle "高階関数" {
-  rectangle "関数を引数に取る" as arg {
-    card "sortBy(wordScore)" as sortBy
-    card "map(transform)" as map
-    card "filter(predicate)" as filter
-  }
-
-  rectangle "関数を返す" as ret {
-    card "largerThan(n)" as larger
-    card "divisibleBy(n)" as divisible
-  }
-}
-
-@enduml
-```
-
-### 4.2 関数を引数として渡す
-
-**ソースファイル**: `app/typescript/src/ch04_higher_order.ts`
-
-#### sortBy - ソート基準を関数で指定
-
-fp-ts の `Ord` を使って型安全にソートできます。
-
-```typescript
-import { pipe } from 'fp-ts/function'
-import * as RA from 'fp-ts/ReadonlyArray'
-import * as Ord from 'fp-ts/Ord'
-import * as N from 'fp-ts/number'
-
-const score = (word: string): number => word.replaceAll('a', '').length
-
-// 任意のスコア関数でソート（降順）
-const rankedWords = (
-  words: readonly string[],
-  wordScore: (word: string) => number
-): readonly string[] =>
-  pipe(
-    words,
-    RA.sortBy([Ord.reverse(Ord.contramap(wordScore)(N.Ord))])
-  )
-
-const words = ['rust', 'java']
-const sortedWords = rankedWords(words, score)
-// sortedWords === ['rust', 'java']
-// rust: 4文字 (r, u, s, t), java: 2文字 (j, v)
-```
-
-```plantuml
-@startuml
-!theme plain
-
-rectangle "sortBy の動作" {
-  card "['rust', 'java']" as input
-  card "score('rust') = 4\nscore('java') = 2" as scores
-  card "['rust', 'java']" as output
-
-  input --> scores : score関数を適用
-  scores --> output : スコア順にソート
-}
-
-@enduml
-```
-
-#### map - 各要素を変換
-
-```typescript
-const getLengths = (words: readonly string[]): readonly number[] =>
-  pipe(
-    words,
-    RA.map((w) => w.length)
-  )
-
-// getLengths(['scala', 'rust', 'ada']) === [5, 4, 3]
-
-const doubleAll = (numbers: readonly number[]): readonly number[] =>
-  pipe(
-    numbers,
-    RA.map((n) => n * 2)
-  )
-
-// doubleAll([5, 1, 2, 4, 0]) === [10, 2, 4, 8, 0]
-```
-
-```plantuml
-@startuml
-!theme plain
-
-rectangle "map の動作" {
-  card "['scala', 'rust', 'ada']" as input
-  card "length関数を各要素に適用" as op
-  card "[5, 4, 3]" as output
-
-  input --> op
-  op --> output
-}
-
-@enduml
-```
-
-#### filter - 条件に合う要素を抽出
-
-```typescript
-const filterOdd = (numbers: readonly number[]): readonly number[] =>
-  pipe(
-    numbers,
-    RA.filter((n) => n % 2 === 1)
-  )
-
-// filterOdd([5, 1, 2, 4, 0]) === [5, 1]
-
-const filterLargerThan = (
-  numbers: readonly number[],
-  threshold: number
-): readonly number[] =>
-  pipe(
-    numbers,
-    RA.filter((n) => n > threshold)
-  )
-
-// filterLargerThan([5, 1, 2, 4, 0], 4) === [5]
-```
-
-#### reduce - 畳み込み
-
-```typescript
-const sumAll = (numbers: readonly number[]): number =>
-  pipe(
-    numbers,
-    RA.reduce(0, (acc, n) => acc + n)
-  )
-
-// sumAll([5, 1, 2, 4, 100]) === 112
-
-const findMax = (numbers: readonly number[]): number | undefined =>
-  pipe(
-    numbers,
-    RA.reduce(undefined as number | undefined, (max, n) =>
-      max === undefined || n > max ? n : max
-    )
-  )
-
-// findMax([5, 1, 2, 4, 15]) === 15
-```
-
-```plantuml
-@startuml
-!theme plain
-
-rectangle "reduce の動作" {
-  card "初期値: 0" as init
-  card "0 + 5 = 5" as s1
-  card "5 + 1 = 6" as s2
-  card "6 + 2 = 8" as s3
-  card "8 + 4 = 12" as s4
-  card "12 + 100 = 112" as s5
-
-  init --> s1
-  s1 --> s2
-  s2 --> s3
-  s3 --> s4
-  s4 --> s5
-}
-
-@enduml
-```
-
-### 4.3 型とインターフェース
-
-**ソースファイル**: `app/typescript/src/ch04_higher_order.ts`
-
-```typescript
-interface ProgrammingLanguage {
-  readonly name: string
-  readonly year: number
-}
-
-const createLanguage = (name: string, year: number): ProgrammingLanguage =>
-  ({ name, year })
-
-const java = createLanguage('Java', 1995)
-const scala = createLanguage('Scala', 2004)
-const languages = [java, scala]
-
-// フィールドにアクセス
-const getNames = (langs: readonly ProgrammingLanguage[]): readonly string[] =>
-  pipe(
-    langs,
-    RA.map((lang) => lang.name)
-  )
-
-// getNames(languages) === ['Java', 'Scala']
-
-// 条件でフィルタ
-const filterYoungLanguages = (
-  langs: readonly ProgrammingLanguage[],
-  afterYear: number
-): readonly ProgrammingLanguage[] =>
-  pipe(
-    langs,
-    RA.filter((lang) => lang.year > afterYear)
-  )
-
-// filterYoungLanguages(languages, 2000) === [scala]
-```
-
-### 4.4 関数を返す関数
-
-**ソースファイル**: `app/typescript/src/ch04_higher_order.ts`
-
-```typescript
-const largerThan = (n: number): ((i: number) => boolean) => (i) => i > n
-
-// 使用例
-pipe([5, 1, 2, 4, 0], RA.filter(largerThan(4)))
-// => [5]
-
-pipe([5, 1, 2, 4, 0], RA.filter(largerThan(1)))
-// => [5, 2, 4]
-```
-
-```plantuml
-@startuml
-!theme plain
-
-rectangle "関数を返す関数" {
-  card "largerThan(4)" as call
-  card "(i) => i > 4" as fn
-  card "filter に渡す" as filter
-
-  call --> fn : 関数を生成
-  fn --> filter : 生成された関数を使用
-}
-
-note bottom
-  largerThan(4) は
-  「4より大きいか判定する関数」を返す
-end note
-
-@enduml
-```
-
-### 4.5 カリー化
-
-カリー化（Currying）は、複数の引数を取る関数を、引数を1つずつ取る関数のチェーンに変換する技法です。
-
-```typescript
-// 通常の関数
-const largerThanNormal = (n: number, i: number): boolean => i > n
-
-// カリー化された関数
-const largerThan = (n: number) => (i: number): boolean => i > n
-
-// 使用例
-const isLargerThan4 = largerThan(4)  // 部分適用
-isLargerThan4(5)  // => true
-isLargerThan4(3)  // => false
-
-// filter に直接渡す
-pipe([5, 1, 2, 4, 0], RA.filter(largerThan(4)))
-// => [5]
-```
-
-```plantuml
-@startuml
-!theme plain
-
-rectangle "カリー化" {
-  rectangle "通常の関数" as normal {
-    card "f(a, b, c)" as n1
-  }
-
-  rectangle "カリー化された関数" as curried {
-    card "f(a)(b)(c)" as c1
-  }
-
-  normal --> curried : カリー化
-}
-
-note bottom of curried
-  部分適用が可能
-  f(a) → 関数を返す
-  f(a)(b) → 関数を返す
-  f(a)(b)(c) → 結果を返す
-end note
-
-@enduml
-```
-
-### 4.6 ワードスコアリングの例
-
-**ソースファイル**: `app/typescript/src/ch04_higher_order.ts`
-
-複数のスコアリングロジックを組み合わせる例です。
-
-```typescript
-const score = (word: string): number => word.replaceAll('a', '').length
-const bonus = (word: string): number => (word.includes('c') ? 5 : 0)
-const penalty = (word: string): number => (word.includes('s') ? 7 : 0)
-
-// カリー化された高スコア単語フィルタ
-const highScoringWordsWithThreshold =
-  (wordScoreFn: (word: string) => number) =>
-  (higherThan: number) =>
-  (words: readonly string[]): readonly string[] =>
-    pipe(
-      words,
-      RA.filter((word) => wordScoreFn(word) > higherThan)
-    )
-
-const words = ['ada', 'haskell', 'scala', 'java', 'rust']
-const scorer = highScoringWordsWithThreshold(wordScoreWithBonusAndPenalty)
-
-scorer(1)(words)
-// => ['java']
-
-scorer(0)(words)
-// => ['ada', 'scala', 'java']
-```
-
----
-
-## 第5章: chain（flatMap）とネスト構造
-
-### 5.1 flatten と flatMap
-
-**ソースファイル**: `app/typescript/src/ch05_flatmap.ts`
-
-#### flatten - ネストしたリストを平坦化
-
-```typescript
-import { pipe } from 'fp-ts/function'
-import * as RA from 'fp-ts/ReadonlyArray'
-
-interface Book {
-  readonly title: string
-  readonly authors: readonly string[]
-}
-
-const books: readonly Book[] = [
-  { title: 'FP in Scala', authors: ['Chiusano', 'Bjarnason'] },
-  { title: 'The Hobbit', authors: ['Tolkien'] },
-]
-
-// map だけだとネストする
-const authorLists = pipe(books, RA.map((book) => book.authors))
-// [['Chiusano', 'Bjarnason'], ['Tolkien']]
-
-// flatten で平坦化
-const authors = pipe(authorLists, RA.flatten)
-// ['Chiusano', 'Bjarnason', 'Tolkien']
-```
-
-```plantuml
-@startuml
-!theme plain
-
-rectangle "flatten の動作" {
-  card "[['A', 'B'], ['C']]" as nested
-  card "flatten" as op
-  card "['A', 'B', 'C']" as flat
-
-  nested --> op
-  op --> flat
-}
-
-@enduml
-```
-
-#### chain = map + flatten
-
-fp-ts では `flatMap` は `chain` という名前で提供されています。
-
-```typescript
-// map して flatten
-const authors1 = pipe(
-  books,
-  RA.map((book) => book.authors),
-  RA.flatten
-)
-
-// chain で同じことを1行で
-const authors2 = pipe(
-  books,
-  RA.chain((book) => book.authors)
-)
-
-// authors1 === authors2
-// ['Chiusano', 'Bjarnason', 'Tolkien']
-```
-
-### 5.2 chain によるリストサイズの変化
-
-```typescript
-// 要素数が増える
-pipe([1, 2, 3], RA.chain((i) => [i, i + 10]))
-// [1, 11, 2, 12, 3, 13] - 6要素
-
-// 要素数が同じ
-pipe([1, 2, 3], RA.chain((i) => [i * 2]))
-// [2, 4, 6] - 3要素
-
-// 要素数が減る（フィルタリング効果）
-pipe([1, 2, 3], RA.chain((i) => (i % 2 === 0 ? [i] : [])))
-// [2] - 1要素
-```
-
-```plantuml
-@startuml
-!theme plain
-
-rectangle "chain とサイズの変化" {
-  rectangle "増加" as inc {
-    card "[1, 2, 3]" as i1
-    card "i => [i, i+10]" as f1
-    card "[1, 11, 2, 12, 3, 13]" as r1
-    i1 --> f1
-    f1 --> r1
-  }
-
-  rectangle "減少（フィルタ）" as dec {
-    card "[1, 2, 3]" as i2
-    card "i => if even [i] else []" as f2
-    card "[2]" as r2
-    i2 --> f2
-    f2 --> r2
-  }
-}
-
-@enduml
-```
-
-### 5.3 ネストした chain
-
-複数のリストを組み合わせる場合、chain をネストします。
-
-```typescript
-interface Book {
-  readonly title: string
-  readonly authors: readonly string[]
-}
-
-interface Movie {
-  readonly title: string
-}
-
-const bookAdaptations = (author: string): readonly Movie[] => {
-  if (author === 'Tolkien') {
-    return [
-      { title: 'An Unexpected Journey' },
-      { title: 'The Desolation of Smaug' },
-    ]
-  }
-  return []
-}
-
-const books: readonly Book[] = [
-  { title: 'FP in Scala', authors: ['Chiusano', 'Bjarnason'] },
-  { title: 'The Hobbit', authors: ['Tolkien'] },
-]
-
-// ネストした chain
-const recommendations = pipe(
-  books,
-  RA.chain((book) =>
-    pipe(
-      book.authors,
-      RA.chain((author) =>
-        pipe(
-          bookAdaptations(author),
-          RA.map(
-            (movie) =>
-              `You may like ${movie.title}, because you liked ${author}'s ${book.title}`
-          )
-        )
-      )
-    )
-  )
-)
-
-// 結果:
-// [
-//   "You may like An Unexpected Journey, because you liked Tolkien's The Hobbit",
-//   "You may like The Desolation of Smaug, because you liked Tolkien's The Hobbit"
-// ]
-```
-
-```plantuml
-@startuml
-!theme plain
-
-rectangle "ネストした chain のフロー" {
-  card "books" as books
-  card "book.authors" as authors
-  card "bookAdaptations(author)" as movies
-  card "recommendation string" as result
-
-  books --> authors : chain
-  authors --> movies : chain
-  movies --> result : map
-}
-
-note bottom
-  最後だけ map
-  途中は chain
-end note
-
-@enduml
-```
-
-### 5.4 Point の生成例
-
-**ソースファイル**: `app/typescript/src/ch05_flatmap.ts`
-
-```typescript
-interface Point {
-  readonly x: number
-  readonly y: number
-}
-
-const createPoint = (x: number, y: number): Point => ({ x, y })
-
-// 全ての (x, y) の組み合わせから Point を生成
-const generatePoints = (
-  xs: readonly number[],
-  ys: readonly number[]
-): readonly Point[] =>
-  pipe(
-    xs,
-    RA.chain((x) =>
-      pipe(
-        ys,
-        RA.map((y) => createPoint(x, y))
-      )
-    )
-  )
-
-generatePoints([1], [-2, 7])
-// [{ x: 1, y: -2 }, { x: 1, y: 7 }]
-
-generatePoints([1, 2], [-2, 7])
-// [{ x: 1, y: -2 }, { x: 1, y: 7 }, { x: 2, y: -2 }, { x: 2, y: 7 }]
-```
-
-### 5.5 Option との組み合わせ
-
-**ソースファイル**: `app/typescript/src/ch05_flatmap.ts`
-
-fp-ts の `Option` 型と組み合わせることで、安全なデータ変換ができます。
-
-```typescript
-import * as O from 'fp-ts/Option'
-
-// Option のリストから Some の値だけを取り出す
-const compactOptions = <T>(options: readonly O.Option<T>[]): readonly T[] =>
-  pipe(options, RA.compact)
-
-compactOptions([O.some(1), O.none, O.some(2), O.none, O.some(3)])
-// [1, 2, 3]
-
-// filterMap - map と filter を同時に
-const filterMapExample = <T, U>(
-  func: (item: T) => O.Option<U>,
-  lst: readonly T[]
-): readonly U[] =>
-  pipe(lst, RA.filterMap(func))
-
-filterMapExample(
-  (s: string) => (s.length > 2 ? O.some(s.toUpperCase()) : O.none),
-  ['a', 'abc', 'de', 'xyz']
-)
-// ['ABC', 'XYZ']
-```
-
-### 5.6 traverse - 効果の反転
-
-**ソースファイル**: `app/typescript/src/ch05_flatmap.ts`
-
-`traverse` は、リストの各要素に `Option` を返す関数を適用し、全てが `Some` なら `Some<配列>` を返し、一つでも `None` なら `None` を返します。
-
-```typescript
-const traverseOption = <T, U>(
-  func: (item: T) => O.Option<U>,
-  lst: readonly T[]
-): O.Option<readonly U[]> =>
-  pipe(lst, RA.traverse(O.Applicative)(func))
-
-const safeDivide = (n: number): O.Option<number> =>
-  n === 0 ? O.none : O.some(10 / n)
-
-traverseOption(safeDivide, [1, 2, 5])
-// some([10, 5, 2])
-
-traverseOption(safeDivide, [1, 0, 5])
-// none - 0で割る要素があるため全体が失敗
-```
-
-### 5.7 円内の点の判定
-
-**ソースファイル**: `app/typescript/src/ch05_flatmap.ts`
-
-chain でフィルタリングも行う例です。
-
-```typescript
-interface Point {
-  readonly x: number
-  readonly y: number
-}
-
-const points: readonly Point[] = [
-  { x: 5, y: 2 },
-  { x: 1, y: 1 },
-]
-const radiuses = [2, 1]
-
-const isInside = (point: Point, radius: number): boolean =>
-  radius * radius >= point.x * point.x + point.y * point.y
-
-// 全組み合わせを生成
-const allCombinations = (
-  points: readonly Point[],
-  radiuses: readonly number[]
-): readonly string[] =>
-  pipe(
-    radiuses,
-    RA.chain((r) =>
-      pipe(
-        points,
-        RA.map(
-          (point) =>
-            `Point(${point.x},${point.y}) is within a radius of ${r}: ${isInside(point, r)}`
-        )
-      )
-    )
-  )
-
-// 結果:
-// [
-//   "Point(5,2) is within a radius of 2: false",
-//   "Point(1,1) is within a radius of 2: true",
-//   "Point(5,2) is within a radius of 1: false",
-//   "Point(1,1) is within a radius of 1: false"
-// ]
-```
-
-#### chain によるフィルタリング
-
-```typescript
-// 円内の点のみを抽出
-const insidePoints = (
-  points: readonly Point[],
-  radiuses: readonly number[]
-): readonly string[] =>
-  pipe(
-    radiuses,
-    RA.chain((r) =>
-      pipe(
-        points,
-        RA.filter((point) => isInside(point, r)),
-        RA.map(
-          (point) => `Point(${point.x},${point.y}) is within a radius of ${r}`
-        )
-      )
-    )
-  )
-
-// 結果: ["Point(1,1) is within a radius of 2"]
-```
-
-```plantuml
-@startuml
-!theme plain
-
-rectangle "フィルタリングの3つの方法" {
-  rectangle "1. filter を使う" as m1 {
-    card "RA.filter((p) => isInside(p, r))"
-  }
-
-  rectangle "2. chain でフィルタ" as m2 {
-    card "RA.chain((p) => isInside(p, r) ? [p] : [])"
-  }
-
-  rectangle "3. filterMap を使う" as m3 {
-    card "RA.filterMap((p) =>\n  isInside(p, r) ? O.some(p) : O.none)"
-  }
-}
-
-@enduml
-```
-
----
-
-## まとめ
-
-### Part II で学んだこと
-
-```plantuml
-@startuml
-!theme plain
-
-rectangle "Part II: 関数型スタイルのプログラミング" {
-  rectangle "第3章" as ch3 {
-    card "イミュータブルデータ"
-    card "slice / spread"
-    card "ReadonlyArray"
-  }
-
-  rectangle "第4章" as ch4 {
-    card "高階関数"
-    card "map / filter / reduce"
-    card "関数を返す関数"
-    card "カリー化"
-    card "Ord によるソート"
-  }
-
-  rectangle "第5章" as ch5 {
-    card "flatten / chain"
-    card "ネストした変換"
-    card "Option との組み合わせ"
-    card "traverse"
-  }
-}
-
-ch3 --> ch4
-ch4 --> ch5
-
-@enduml
-```
-
-### キーポイント
-
-| 章 | 主要概念 | キー操作 |
-|----|----------|----------|
-| 第3章 | イミュータブル | `slice`, スプレッド演算子, `readonly` |
-| 第4章 | 高階関数 | `RA.map`, `RA.filter`, `RA.reduce`, `RA.sortBy` |
-| 第5章 | 平坦化 | `RA.flatten`, `RA.chain`, `RA.traverse` |
-
-### Scala vs TypeScript fp-ts 比較
-
-| Scala | TypeScript (fp-ts) | 説明 |
-|-------|-------------------|------|
-| `list.appended(x)` | `[...list, x]` | 末尾に追加 |
-| `list.slice(a, b)` | `list.slice(a, b)` | スライス |
-| `list.map(f)` | `pipe(list, RA.map(f))` | map |
-| `list.filter(p)` | `pipe(list, RA.filter(p))` | filter |
-| `list.foldLeft(z)(f)` | `pipe(list, RA.reduce(z, f))` | reduce |
-| `list.flatMap(f)` | `pipe(list, RA.chain(f))` | flatMap |
-| `list.flatten` | `pipe(list, RA.flatten)` | flatten |
-| `list.sortBy(f)` | `pipe(list, RA.sortBy([ord]))` | sortBy |
-
-### 重要な法則
-
-1. **イミュータブルデータ**: 元のデータは変更せず、新しいデータを作成する
-2. **関数は値**: 関数を引数として渡したり、戻り値として返したりできる
-3. **chain パターン**: ネストした構造を平坦化しながら変換する
-4. **traverse パターン**: 効果（Option等）を反転させてリスト全体を扱う
-
-### 次のステップ
-
-Part III では、以下のトピックを学びます:
-
-- `Option` 型による安全なエラーハンドリング
-- `Either` 型と複合的なエラー処理
-- エラー処理のパターン
-
----
-
-## 演習問題
-
-### 問題 1: イミュータブルな操作
-
-以下の関数を実装してください。リストの中央に要素を挿入する関数です。
-
-```typescript
-const insertAtMiddle = <T>(list: readonly T[], element: T): readonly T[] => ???
-
-// 期待される動作
-// insertAtMiddle(['a', 'b', 'c', 'd'], 'X') === ['a', 'b', 'X', 'c', 'd']
-// insertAtMiddle(['a', 'b'], 'X') === ['a', 'X', 'b']
-```
-
-<details>
-<summary>解答</summary>
-
-```typescript
-const insertAtMiddle = <T>(list: readonly T[], element: T): readonly T[] => {
-  const middle = Math.floor(list.length / 2)
-  const before = list.slice(0, middle)
-  const after = list.slice(middle)
-  return [...before, element, ...after]
-}
-```
-
-</details>
-
-### 問題 2: 高階関数
-
-以下の関数を実装してください。条件を満たす要素の数をカウントする関数です。
-
-```typescript
-const countWhere = <T>(
-  list: readonly T[],
-  predicate: (item: T) => boolean
-): number => ???
-
-// 期待される動作
-// countWhere([1, 2, 3, 4, 5], (i) => i > 3) === 2
-// countWhere(['a', 'bb', 'ccc'], (s) => s.length > 1) === 2
-```
-
-<details>
-<summary>解答</summary>
-
-```typescript
-import { pipe } from 'fp-ts/function'
-import * as RA from 'fp-ts/ReadonlyArray'
-
-const countWhere = <T>(
-  list: readonly T[],
-  predicate: (item: T) => boolean
-): number =>
-  pipe(
-    list,
-    RA.filter(predicate),
-    (filtered) => filtered.length
-  )
-
-// または reduce を使って
-const countWhere2 = <T>(
-  list: readonly T[],
-  predicate: (item: T) => boolean
-): number =>
-  pipe(
-    list,
-    RA.reduce(0, (count, elem) => (predicate(elem) ? count + 1 : count))
-  )
-```
-
-</details>
-
-### 問題 3: chain の活用
-
-以下のネストした配列を chain で平坦化してください。
-
-```typescript
-const nested = [[1, 2], [10, 20], [100, 200]]
-// それぞれの配列の各要素を足し合わせた全ての組み合わせを生成
-// 期待される結果: [111, 211, 121, 221, 112, 212, 122, 222]
-```
-
-<details>
-<summary>解答</summary>
-
-```typescript
-import { pipe } from 'fp-ts/function'
-import * as RA from 'fp-ts/ReadonlyArray'
-
-const result = pipe(
-  [1, 2],
-  RA.chain((x) =>
-    pipe(
-      [10, 20],
-      RA.chain((y) =>
-        pipe(
-          [100, 200],
-          RA.map((z) => x + y + z)
-        )
-      )
-    )
-  )
-)
-
-// 結果: [111, 211, 121, 221, 112, 212, 122, 222]
-```
-
-</details>
-
-### 問題 4: chain によるフィルタリング
-
-filter を使わずに、chain だけで偶数のみを抽出するコードを書いてください。
-
-```typescript
-const numbers = [1, 2, 3, 4, 5, 6]
-// 偶数のみを抽出: [2, 4, 6]
-```
-
-<details>
-<summary>解答</summary>
-
-```typescript
-import { pipe } from 'fp-ts/function'
-import * as RA from 'fp-ts/ReadonlyArray'
-
-const evenNumbers = pipe(
-  numbers,
-  RA.chain((n) => (n % 2 === 0 ? [n] : []))
-)
-
-// または関数に分離
-const evenFilter = (n: number): readonly number[] =>
-  n % 2 === 0 ? [n] : []
-
-const evenNumbers2 = pipe(numbers, RA.chain(evenFilter))
-```
-
-</details>
+# Part II: 関数型スタイルのプログラミング
+
+本章では、関数型プログラミングの核心となるテクニックを学びます。イミュータブルなデータ操作、高階関数、そして `chain`（flatMap）による複雑なデータ変換を習得します。
+
+---
+
+## 第3章: イミュータブルなデータ操作
+
+### 3.1 イミュータブルとは
+
+イミュータブル（不変）とは、一度作成されたデータが変更されないことを意味します。データを「変更」する代わりに、新しいデータを「作成」します。
+
+```plantuml
+@startuml
+!theme plain
+
+rectangle "ミュータブル（可変）" as mut {
+  card "list = [A, B]" as m1
+  card "list.push(C)" as m2
+  card "list = [A, B, C]" as m3
+  m1 --> m2 : 変更
+  m2 --> m3 : 同じ配列が変わる
+}
+
+rectangle "イミュータブル（不変）" as immut {
+  card "list1 = [A, B]" as i1
+  card "list2 = [...list1, C]" as i2
+  card "list1 = [A, B]\nlist2 = [A, B, C]" as i3
+  i1 --> i2 : 新規作成
+  i2 --> i3 : 両方が存在
+}
+
+@enduml
+```
+
+TypeScript では `readonly` 修飾子と fp-ts の `ReadonlyArray` モジュールを使ってイミュータブルなデータを扱います。
+
+### 3.2 配列の基本操作
+
+**ソースファイル**: `app/typescript/src/ch03_immutable_data.ts`
+
+#### appended - 要素の追加
+
+```typescript
+// スプレッド演算子を使ったイミュータブルな追加
+const appended = <T>(lst: readonly T[], element: T): readonly T[] =>
+  [...lst, element]
+
+const appleBook = ['Apple', 'Book'] as const
+const appleBookMango = appended(appleBook, 'Mango')
+
+// appleBook.length === 2       // 元のリストは変わらない
+// appleBookMango.length === 3  // 新しいリストが作成される
+// appleBookMango === ['Apple', 'Book', 'Mango']
+```
+
+#### slice - リストの切り出し
+
+```typescript
+const firstTwo = <T>(lst: readonly T[]): readonly T[] =>
+  lst.slice(0, 2)
+
+const lastTwo = <T>(lst: readonly T[]): readonly T[] =>
+  lst.slice(Math.max(0, lst.length - 2))
+
+// firstTwo(['a', 'b', 'c']) === ['a', 'b']
+// lastTwo(['a', 'b', 'c']) === ['b', 'c']
+```
+
+```plantuml
+@startuml
+!theme plain
+
+rectangle "slice(start, end)" {
+  card "['a', 'b', 'c', 'd']" as orig
+  card "slice(1, 3)" as op
+  card "['b', 'c']" as result
+
+  orig --> op
+  op --> result
+}
+
+note bottom of result
+  インデックス 1 から 3 の手前まで
+  (end は含まない)
+end note
+
+@enduml
+```
+
+### 3.3 リストの変換例
+
+**ソースファイル**: `app/typescript/src/ch03_immutable_data.ts`
+
+```typescript
+// 最初の2要素を末尾に移動
+const moveFirstTwoToEnd = <T>(lst: readonly T[]): readonly T[] => {
+  const firstTwo = lst.slice(0, 2)
+  const withoutFirstTwo = lst.slice(2)
+  return [...withoutFirstTwo, ...firstTwo]
+}
+
+// moveFirstTwoToEnd(['a', 'b', 'c']) === ['c', 'a', 'b']
+
+// 最後の要素の前に挿入
+const insertBeforeLast = <T>(lst: readonly T[], element: T): readonly T[] => {
+  if (lst.length === 0) return [element]
+  const last = lst.slice(lst.length - 1)
+  const withoutLast = lst.slice(0, lst.length - 1)
+  return [...withoutLast, element, ...last]
+}
+
+// insertBeforeLast(['a', 'b'], 'c') === ['a', 'c', 'b']
+```
+
+### 3.4 旅程の再計画
+
+**ソースファイル**: `app/typescript/src/ch03_immutable_data.ts`
+
+旅行の計画変更をイミュータブルに行う例です。
+
+```typescript
+const replan = (
+  plan: readonly string[],
+  newCity: string,
+  beforeCity: string
+): readonly string[] => {
+  const index = plan.indexOf(beforeCity)
+  if (index === -1) return [...plan, newCity]
+  return [...plan.slice(0, index), newCity, ...plan.slice(index)]
+}
+
+const planA = ['Paris', 'Berlin', 'Kraków']
+const planB = replan(planA, 'Vienna', 'Kraków')
+
+// planB === ['Paris', 'Berlin', 'Vienna', 'Kraków']
+// planA === ['Paris', 'Berlin', 'Kraków']  // 元の計画は変わらない!
+```
+
+```plantuml
+@startuml
+!theme plain
+
+rectangle "旅程の再計画" {
+  card "Plan A:\nParis → Berlin → Kraków" as planA
+  card "replan(planA, 'Vienna', 'Kraków')" as op
+  card "Plan B:\nParis → Berlin → Vienna → Kraków" as planB
+
+  planA --> op
+  op --> planB
+}
+
+note right of planA
+  Plan A は変更されない
+  (イミュータブル)
+end note
+
+@enduml
+```
+
+### 3.5 イミュータブルなオブジェクト
+
+TypeScript では `readonly` 修飾子を使ってイミュータブルな型を定義します。
+
+```typescript
+interface Point {
+  readonly x: number
+  readonly y: number
+}
+
+const createPoint = (x: number, y: number): Point => ({ x, y })
+
+// スプレッド演算子でコピーしながら更新
+const withX = (point: Point, newX: number): Point => ({ ...point, x: newX })
+const withY = (point: Point, newY: number): Point => ({ ...point, y: newY })
+
+const p1 = createPoint(1, 2)
+const p2 = withX(p1, 10)
+
+// p1.x === 1   // 元のPointは変わらない
+// p2.x === 10
+// p2.y === 2
+```
+
+### 3.6 fp-ts を使ったイミュータブル操作
+
+fp-ts の `ReadonlyArray` モジュールを使うと、より型安全にイミュータブル操作ができます。
+
+```typescript
+import { pipe } from 'fp-ts/function'
+import * as RA from 'fp-ts/ReadonlyArray'
+
+// append - 要素を追加
+const appendWithFpts = <T>(lst: readonly T[], element: T): readonly T[] =>
+  pipe(lst, RA.append(element))
+
+// prepend - 先頭に追加
+const prependWithFpts = <T>(lst: readonly T[], element: T): readonly T[] =>
+  pipe(lst, RA.prepend(element))
+
+// concat - 連結
+const concatWithFpts = <T>(
+  lst1: readonly T[],
+  lst2: readonly T[]
+): readonly T[] =>
+  pipe(lst1, RA.concat(lst2))
+```
+
+---
+
+## 第4章: 関数を値として扱う
+
+### 4.1 高階関数とは
+
+高階関数（Higher-Order Function）とは、以下のいずれかを満たす関数です:
+
+1. 関数を引数として受け取る
+2. 関数を戻り値として返す
+
+```plantuml
+@startuml
+!theme plain
+
+rectangle "高階関数" {
+  rectangle "関数を引数に取る" as arg {
+    card "sortBy(wordScore)" as sortBy
+    card "map(transform)" as map
+    card "filter(predicate)" as filter
+  }
+
+  rectangle "関数を返す" as ret {
+    card "largerThan(n)" as larger
+    card "divisibleBy(n)" as divisible
+  }
+}
+
+@enduml
+```
+
+### 4.2 関数を引数として渡す
+
+**ソースファイル**: `app/typescript/src/ch04_higher_order.ts`
+
+#### sortBy - ソート基準を関数で指定
+
+fp-ts の `Ord` を使って型安全にソートできます。
+
+```typescript
+import { pipe } from 'fp-ts/function'
+import * as RA from 'fp-ts/ReadonlyArray'
+import * as Ord from 'fp-ts/Ord'
+import * as N from 'fp-ts/number'
+
+const score = (word: string): number => word.replaceAll('a', '').length
+
+// 任意のスコア関数でソート（降順）
+const rankedWords = (
+  words: readonly string[],
+  wordScore: (word: string) => number
+): readonly string[] =>
+  pipe(
+    words,
+    RA.sortBy([Ord.reverse(Ord.contramap(wordScore)(N.Ord))])
+  )
+
+const words = ['rust', 'java']
+const sortedWords = rankedWords(words, score)
+// sortedWords === ['rust', 'java']
+// rust: 4文字 (r, u, s, t), java: 2文字 (j, v)
+```
+
+```plantuml
+@startuml
+!theme plain
+
+rectangle "sortBy の動作" {
+  card "['rust', 'java']" as input
+  card "score('rust') = 4\nscore('java') = 2" as scores
+  card "['rust', 'java']" as output
+
+  input --> scores : score関数を適用
+  scores --> output : スコア順にソート
+}
+
+@enduml
+```
+
+#### map - 各要素を変換
+
+```typescript
+const getLengths = (words: readonly string[]): readonly number[] =>
+  pipe(
+    words,
+    RA.map((w) => w.length)
+  )
+
+// getLengths(['scala', 'rust', 'ada']) === [5, 4, 3]
+
+const doubleAll = (numbers: readonly number[]): readonly number[] =>
+  pipe(
+    numbers,
+    RA.map((n) => n * 2)
+  )
+
+// doubleAll([5, 1, 2, 4, 0]) === [10, 2, 4, 8, 0]
+```
+
+```plantuml
+@startuml
+!theme plain
+
+rectangle "map の動作" {
+  card "['scala', 'rust', 'ada']" as input
+  card "length関数を各要素に適用" as op
+  card "[5, 4, 3]" as output
+
+  input --> op
+  op --> output
+}
+
+@enduml
+```
+
+#### filter - 条件に合う要素を抽出
+
+```typescript
+const filterOdd = (numbers: readonly number[]): readonly number[] =>
+  pipe(
+    numbers,
+    RA.filter((n) => n % 2 === 1)
+  )
+
+// filterOdd([5, 1, 2, 4, 0]) === [5, 1]
+
+const filterLargerThan = (
+  numbers: readonly number[],
+  threshold: number
+): readonly number[] =>
+  pipe(
+    numbers,
+    RA.filter((n) => n > threshold)
+  )
+
+// filterLargerThan([5, 1, 2, 4, 0], 4) === [5]
+```
+
+#### reduce - 畳み込み
+
+```typescript
+const sumAll = (numbers: readonly number[]): number =>
+  pipe(
+    numbers,
+    RA.reduce(0, (acc, n) => acc + n)
+  )
+
+// sumAll([5, 1, 2, 4, 100]) === 112
+
+const findMax = (numbers: readonly number[]): number | undefined =>
+  pipe(
+    numbers,
+    RA.reduce(undefined as number | undefined, (max, n) =>
+      max === undefined || n > max ? n : max
+    )
+  )
+
+// findMax([5, 1, 2, 4, 15]) === 15
+```
+
+```plantuml
+@startuml
+!theme plain
+
+rectangle "reduce の動作" {
+  card "初期値: 0" as init
+  card "0 + 5 = 5" as s1
+  card "5 + 1 = 6" as s2
+  card "6 + 2 = 8" as s3
+  card "8 + 4 = 12" as s4
+  card "12 + 100 = 112" as s5
+
+  init --> s1
+  s1 --> s2
+  s2 --> s3
+  s3 --> s4
+  s4 --> s5
+}
+
+@enduml
+```
+
+### 4.3 型とインターフェース
+
+**ソースファイル**: `app/typescript/src/ch04_higher_order.ts`
+
+```typescript
+interface ProgrammingLanguage {
+  readonly name: string
+  readonly year: number
+}
+
+const createLanguage = (name: string, year: number): ProgrammingLanguage =>
+  ({ name, year })
+
+const java = createLanguage('Java', 1995)
+const scala = createLanguage('Scala', 2004)
+const languages = [java, scala]
+
+// フィールドにアクセス
+const getNames = (langs: readonly ProgrammingLanguage[]): readonly string[] =>
+  pipe(
+    langs,
+    RA.map((lang) => lang.name)
+  )
+
+// getNames(languages) === ['Java', 'Scala']
+
+// 条件でフィルタ
+const filterYoungLanguages = (
+  langs: readonly ProgrammingLanguage[],
+  afterYear: number
+): readonly ProgrammingLanguage[] =>
+  pipe(
+    langs,
+    RA.filter((lang) => lang.year > afterYear)
+  )
+
+// filterYoungLanguages(languages, 2000) === [scala]
+```
+
+### 4.4 関数を返す関数
+
+**ソースファイル**: `app/typescript/src/ch04_higher_order.ts`
+
+```typescript
+const largerThan = (n: number): ((i: number) => boolean) => (i) => i > n
+
+// 使用例
+pipe([5, 1, 2, 4, 0], RA.filter(largerThan(4)))
+// => [5]
+
+pipe([5, 1, 2, 4, 0], RA.filter(largerThan(1)))
+// => [5, 2, 4]
+```
+
+```plantuml
+@startuml
+!theme plain
+
+rectangle "関数を返す関数" {
+  card "largerThan(4)" as call
+  card "(i) => i > 4" as fn
+  card "filter に渡す" as filter
+
+  call --> fn : 関数を生成
+  fn --> filter : 生成された関数を使用
+}
+
+note bottom
+  largerThan(4) は
+  「4より大きいか判定する関数」を返す
+end note
+
+@enduml
+```
+
+### 4.5 カリー化
+
+カリー化（Currying）は、複数の引数を取る関数を、引数を1つずつ取る関数のチェーンに変換する技法です。
+
+```typescript
+// 通常の関数
+const largerThanNormal = (n: number, i: number): boolean => i > n
+
+// カリー化された関数
+const largerThan = (n: number) => (i: number): boolean => i > n
+
+// 使用例
+const isLargerThan4 = largerThan(4)  // 部分適用
+isLargerThan4(5)  // => true
+isLargerThan4(3)  // => false
+
+// filter に直接渡す
+pipe([5, 1, 2, 4, 0], RA.filter(largerThan(4)))
+// => [5]
+```
+
+```plantuml
+@startuml
+!theme plain
+
+rectangle "カリー化" {
+  rectangle "通常の関数" as normal {
+    card "f(a, b, c)" as n1
+  }
+
+  rectangle "カリー化された関数" as curried {
+    card "f(a)(b)(c)" as c1
+  }
+
+  normal --> curried : カリー化
+}
+
+note bottom of curried
+  部分適用が可能
+  f(a) → 関数を返す
+  f(a)(b) → 関数を返す
+  f(a)(b)(c) → 結果を返す
+end note
+
+@enduml
+```
+
+### 4.6 ワードスコアリングの例
+
+**ソースファイル**: `app/typescript/src/ch04_higher_order.ts`
+
+複数のスコアリングロジックを組み合わせる例です。
+
+```typescript
+const score = (word: string): number => word.replaceAll('a', '').length
+const bonus = (word: string): number => (word.includes('c') ? 5 : 0)
+const penalty = (word: string): number => (word.includes('s') ? 7 : 0)
+
+// カリー化された高スコア単語フィルタ
+const highScoringWordsWithThreshold =
+  (wordScoreFn: (word: string) => number) =>
+  (higherThan: number) =>
+  (words: readonly string[]): readonly string[] =>
+    pipe(
+      words,
+      RA.filter((word) => wordScoreFn(word) > higherThan)
+    )
+
+const words = ['ada', 'haskell', 'scala', 'java', 'rust']
+const scorer = highScoringWordsWithThreshold(wordScoreWithBonusAndPenalty)
+
+scorer(1)(words)
+// => ['java']
+
+scorer(0)(words)
+// => ['ada', 'scala', 'java']
+```
+
+---
+
+## 第5章: chain（flatMap）とネスト構造
+
+### 5.1 flatten と flatMap
+
+**ソースファイル**: `app/typescript/src/ch05_flatmap.ts`
+
+#### flatten - ネストしたリストを平坦化
+
+```typescript
+import { pipe } from 'fp-ts/function'
+import * as RA from 'fp-ts/ReadonlyArray'
+
+interface Book {
+  readonly title: string
+  readonly authors: readonly string[]
+}
+
+const books: readonly Book[] = [
+  { title: 'FP in Scala', authors: ['Chiusano', 'Bjarnason'] },
+  { title: 'The Hobbit', authors: ['Tolkien'] },
+]
+
+// map だけだとネストする
+const authorLists = pipe(books, RA.map((book) => book.authors))
+// [['Chiusano', 'Bjarnason'], ['Tolkien']]
+
+// flatten で平坦化
+const authors = pipe(authorLists, RA.flatten)
+// ['Chiusano', 'Bjarnason', 'Tolkien']
+```
+
+```plantuml
+@startuml
+!theme plain
+
+rectangle "flatten の動作" {
+  card "[['A', 'B'], ['C']]" as nested
+  card "flatten" as op
+  card "['A', 'B', 'C']" as flat
+
+  nested --> op
+  op --> flat
+}
+
+@enduml
+```
+
+#### chain = map + flatten
+
+fp-ts では `flatMap` は `chain` という名前で提供されています。
+
+```typescript
+// map して flatten
+const authors1 = pipe(
+  books,
+  RA.map((book) => book.authors),
+  RA.flatten
+)
+
+// chain で同じことを1行で
+const authors2 = pipe(
+  books,
+  RA.chain((book) => book.authors)
+)
+
+// authors1 === authors2
+// ['Chiusano', 'Bjarnason', 'Tolkien']
+```
+
+### 5.2 chain によるリストサイズの変化
+
+```typescript
+// 要素数が増える
+pipe([1, 2, 3], RA.chain((i) => [i, i + 10]))
+// [1, 11, 2, 12, 3, 13] - 6要素
+
+// 要素数が同じ
+pipe([1, 2, 3], RA.chain((i) => [i * 2]))
+// [2, 4, 6] - 3要素
+
+// 要素数が減る（フィルタリング効果）
+pipe([1, 2, 3], RA.chain((i) => (i % 2 === 0 ? [i] : [])))
+// [2] - 1要素
+```
+
+```plantuml
+@startuml
+!theme plain
+
+rectangle "chain とサイズの変化" {
+  rectangle "増加" as inc {
+    card "[1, 2, 3]" as i1
+    card "i => [i, i+10]" as f1
+    card "[1, 11, 2, 12, 3, 13]" as r1
+    i1 --> f1
+    f1 --> r1
+  }
+
+  rectangle "減少（フィルタ）" as dec {
+    card "[1, 2, 3]" as i2
+    card "i => if even [i] else []" as f2
+    card "[2]" as r2
+    i2 --> f2
+    f2 --> r2
+  }
+}
+
+@enduml
+```
+
+### 5.3 ネストした chain
+
+複数のリストを組み合わせる場合、chain をネストします。
+
+```typescript
+interface Book {
+  readonly title: string
+  readonly authors: readonly string[]
+}
+
+interface Movie {
+  readonly title: string
+}
+
+const bookAdaptations = (author: string): readonly Movie[] => {
+  if (author === 'Tolkien') {
+    return [
+      { title: 'An Unexpected Journey' },
+      { title: 'The Desolation of Smaug' },
+    ]
+  }
+  return []
+}
+
+const books: readonly Book[] = [
+  { title: 'FP in Scala', authors: ['Chiusano', 'Bjarnason'] },
+  { title: 'The Hobbit', authors: ['Tolkien'] },
+]
+
+// ネストした chain
+const recommendations = pipe(
+  books,
+  RA.chain((book) =>
+    pipe(
+      book.authors,
+      RA.chain((author) =>
+        pipe(
+          bookAdaptations(author),
+          RA.map(
+            (movie) =>
+              `You may like ${movie.title}, because you liked ${author}'s ${book.title}`
+          )
+        )
+      )
+    )
+  )
+)
+
+// 結果:
+// [
+//   "You may like An Unexpected Journey, because you liked Tolkien's The Hobbit",
+//   "You may like The Desolation of Smaug, because you liked Tolkien's The Hobbit"
+// ]
+```
+
+```plantuml
+@startuml
+!theme plain
+
+rectangle "ネストした chain のフロー" {
+  card "books" as books
+  card "book.authors" as authors
+  card "bookAdaptations(author)" as movies
+  card "recommendation string" as result
+
+  books --> authors : chain
+  authors --> movies : chain
+  movies --> result : map
+}
+
+note bottom
+  最後だけ map
+  途中は chain
+end note
+
+@enduml
+```
+
+### 5.4 Point の生成例
+
+**ソースファイル**: `app/typescript/src/ch05_flatmap.ts`
+
+```typescript
+interface Point {
+  readonly x: number
+  readonly y: number
+}
+
+const createPoint = (x: number, y: number): Point => ({ x, y })
+
+// 全ての (x, y) の組み合わせから Point を生成
+const generatePoints = (
+  xs: readonly number[],
+  ys: readonly number[]
+): readonly Point[] =>
+  pipe(
+    xs,
+    RA.chain((x) =>
+      pipe(
+        ys,
+        RA.map((y) => createPoint(x, y))
+      )
+    )
+  )
+
+generatePoints([1], [-2, 7])
+// [{ x: 1, y: -2 }, { x: 1, y: 7 }]
+
+generatePoints([1, 2], [-2, 7])
+// [{ x: 1, y: -2 }, { x: 1, y: 7 }, { x: 2, y: -2 }, { x: 2, y: 7 }]
+```
+
+### 5.5 Option との組み合わせ
+
+**ソースファイル**: `app/typescript/src/ch05_flatmap.ts`
+
+fp-ts の `Option` 型と組み合わせることで、安全なデータ変換ができます。
+
+```typescript
+import * as O from 'fp-ts/Option'
+
+// Option のリストから Some の値だけを取り出す
+const compactOptions = <T>(options: readonly O.Option<T>[]): readonly T[] =>
+  pipe(options, RA.compact)
+
+compactOptions([O.some(1), O.none, O.some(2), O.none, O.some(3)])
+// [1, 2, 3]
+
+// filterMap - map と filter を同時に
+const filterMapExample = <T, U>(
+  func: (item: T) => O.Option<U>,
+  lst: readonly T[]
+): readonly U[] =>
+  pipe(lst, RA.filterMap(func))
+
+filterMapExample(
+  (s: string) => (s.length > 2 ? O.some(s.toUpperCase()) : O.none),
+  ['a', 'abc', 'de', 'xyz']
+)
+// ['ABC', 'XYZ']
+```
+
+### 5.6 traverse - 効果の反転
+
+**ソースファイル**: `app/typescript/src/ch05_flatmap.ts`
+
+`traverse` は、リストの各要素に `Option` を返す関数を適用し、全てが `Some` なら `Some<配列>` を返し、一つでも `None` なら `None` を返します。
+
+```typescript
+const traverseOption = <T, U>(
+  func: (item: T) => O.Option<U>,
+  lst: readonly T[]
+): O.Option<readonly U[]> =>
+  pipe(lst, RA.traverse(O.Applicative)(func))
+
+const safeDivide = (n: number): O.Option<number> =>
+  n === 0 ? O.none : O.some(10 / n)
+
+traverseOption(safeDivide, [1, 2, 5])
+// some([10, 5, 2])
+
+traverseOption(safeDivide, [1, 0, 5])
+// none - 0で割る要素があるため全体が失敗
+```
+
+### 5.7 円内の点の判定
+
+**ソースファイル**: `app/typescript/src/ch05_flatmap.ts`
+
+chain でフィルタリングも行う例です。
+
+```typescript
+interface Point {
+  readonly x: number
+  readonly y: number
+}
+
+const points: readonly Point[] = [
+  { x: 5, y: 2 },
+  { x: 1, y: 1 },
+]
+const radiuses = [2, 1]
+
+const isInside = (point: Point, radius: number): boolean =>
+  radius * radius >= point.x * point.x + point.y * point.y
+
+// 全組み合わせを生成
+const allCombinations = (
+  points: readonly Point[],
+  radiuses: readonly number[]
+): readonly string[] =>
+  pipe(
+    radiuses,
+    RA.chain((r) =>
+      pipe(
+        points,
+        RA.map(
+          (point) =>
+            `Point(${point.x},${point.y}) is within a radius of ${r}: ${isInside(point, r)}`
+        )
+      )
+    )
+  )
+
+// 結果:
+// [
+//   "Point(5,2) is within a radius of 2: false",
+//   "Point(1,1) is within a radius of 2: true",
+//   "Point(5,2) is within a radius of 1: false",
+//   "Point(1,1) is within a radius of 1: false"
+// ]
+```
+
+#### chain によるフィルタリング
+
+```typescript
+// 円内の点のみを抽出
+const insidePoints = (
+  points: readonly Point[],
+  radiuses: readonly number[]
+): readonly string[] =>
+  pipe(
+    radiuses,
+    RA.chain((r) =>
+      pipe(
+        points,
+        RA.filter((point) => isInside(point, r)),
+        RA.map(
+          (point) => `Point(${point.x},${point.y}) is within a radius of ${r}`
+        )
+      )
+    )
+  )
+
+// 結果: ["Point(1,1) is within a radius of 2"]
+```
+
+```plantuml
+@startuml
+!theme plain
+
+rectangle "フィルタリングの3つの方法" {
+  rectangle "1. filter を使う" as m1 {
+    card "RA.filter((p) => isInside(p, r))"
+  }
+
+  rectangle "2. chain でフィルタ" as m2 {
+    card "RA.chain((p) => isInside(p, r) ? [p] : [])"
+  }
+
+  rectangle "3. filterMap を使う" as m3 {
+    card "RA.filterMap((p) =>\n  isInside(p, r) ? O.some(p) : O.none)"
+  }
+}
+
+@enduml
+```
+
+---
+
+## まとめ
+
+### Part II で学んだこと
+
+```plantuml
+@startuml
+!theme plain
+
+rectangle "Part II: 関数型スタイルのプログラミング" {
+  rectangle "第3章" as ch3 {
+    card "イミュータブルデータ"
+    card "slice / spread"
+    card "ReadonlyArray"
+  }
+
+  rectangle "第4章" as ch4 {
+    card "高階関数"
+    card "map / filter / reduce"
+    card "関数を返す関数"
+    card "カリー化"
+    card "Ord によるソート"
+  }
+
+  rectangle "第5章" as ch5 {
+    card "flatten / chain"
+    card "ネストした変換"
+    card "Option との組み合わせ"
+    card "traverse"
+  }
+}
+
+ch3 --> ch4
+ch4 --> ch5
+
+@enduml
+```
+
+### キーポイント
+
+| 章 | 主要概念 | キー操作 |
+|----|----------|----------|
+| 第3章 | イミュータブル | `slice`, スプレッド演算子, `readonly` |
+| 第4章 | 高階関数 | `RA.map`, `RA.filter`, `RA.reduce`, `RA.sortBy` |
+| 第5章 | 平坦化 | `RA.flatten`, `RA.chain`, `RA.traverse` |
+
+### Scala vs TypeScript fp-ts 比較
+
+| Scala | TypeScript (fp-ts) | 説明 |
+|-------|-------------------|------|
+| `list.appended(x)` | `[...list, x]` | 末尾に追加 |
+| `list.slice(a, b)` | `list.slice(a, b)` | スライス |
+| `list.map(f)` | `pipe(list, RA.map(f))` | map |
+| `list.filter(p)` | `pipe(list, RA.filter(p))` | filter |
+| `list.foldLeft(z)(f)` | `pipe(list, RA.reduce(z, f))` | reduce |
+| `list.flatMap(f)` | `pipe(list, RA.chain(f))` | flatMap |
+| `list.flatten` | `pipe(list, RA.flatten)` | flatten |
+| `list.sortBy(f)` | `pipe(list, RA.sortBy([ord]))` | sortBy |
+
+### 重要な法則
+
+1. **イミュータブルデータ**: 元のデータは変更せず、新しいデータを作成する
+2. **関数は値**: 関数を引数として渡したり、戻り値として返したりできる
+3. **chain パターン**: ネストした構造を平坦化しながら変換する
+4. **traverse パターン**: 効果（Option等）を反転させてリスト全体を扱う
+
+### 次のステップ
+
+Part III では、以下のトピックを学びます:
+
+- `Option` 型による安全なエラーハンドリング
+- `Either` 型と複合的なエラー処理
+- エラー処理のパターン
+
+---
+
+## 演習問題
+
+### 問題 1: イミュータブルな操作
+
+以下の関数を実装してください。リストの中央に要素を挿入する関数です。
+
+```typescript
+const insertAtMiddle = <T>(list: readonly T[], element: T): readonly T[] => ???
+
+// 期待される動作
+// insertAtMiddle(['a', 'b', 'c', 'd'], 'X') === ['a', 'b', 'X', 'c', 'd']
+// insertAtMiddle(['a', 'b'], 'X') === ['a', 'X', 'b']
+```
+
+<details>
+<summary>解答</summary>
+
+```typescript
+const insertAtMiddle = <T>(list: readonly T[], element: T): readonly T[] => {
+  const middle = Math.floor(list.length / 2)
+  const before = list.slice(0, middle)
+  const after = list.slice(middle)
+  return [...before, element, ...after]
+}
+```
+
+</details>
+
+### 問題 2: 高階関数
+
+以下の関数を実装してください。条件を満たす要素の数をカウントする関数です。
+
+```typescript
+const countWhere = <T>(
+  list: readonly T[],
+  predicate: (item: T) => boolean
+): number => ???
+
+// 期待される動作
+// countWhere([1, 2, 3, 4, 5], (i) => i > 3) === 2
+// countWhere(['a', 'bb', 'ccc'], (s) => s.length > 1) === 2
+```
+
+<details>
+<summary>解答</summary>
+
+```typescript
+import { pipe } from 'fp-ts/function'
+import * as RA from 'fp-ts/ReadonlyArray'
+
+const countWhere = <T>(
+  list: readonly T[],
+  predicate: (item: T) => boolean
+): number =>
+  pipe(
+    list,
+    RA.filter(predicate),
+    (filtered) => filtered.length
+  )
+
+// または reduce を使って
+const countWhere2 = <T>(
+  list: readonly T[],
+  predicate: (item: T) => boolean
+): number =>
+  pipe(
+    list,
+    RA.reduce(0, (count, elem) => (predicate(elem) ? count + 1 : count))
+  )
+```
+
+</details>
+
+### 問題 3: chain の活用
+
+以下のネストした配列を chain で平坦化してください。
+
+```typescript
+const nested = [[1, 2], [10, 20], [100, 200]]
+// それぞれの配列の各要素を足し合わせた全ての組み合わせを生成
+// 期待される結果: [111, 211, 121, 221, 112, 212, 122, 222]
+```
+
+<details>
+<summary>解答</summary>
+
+```typescript
+import { pipe } from 'fp-ts/function'
+import * as RA from 'fp-ts/ReadonlyArray'
+
+const result = pipe(
+  [1, 2],
+  RA.chain((x) =>
+    pipe(
+      [10, 20],
+      RA.chain((y) =>
+        pipe(
+          [100, 200],
+          RA.map((z) => x + y + z)
+        )
+      )
+    )
+  )
+)
+
+// 結果: [111, 211, 121, 221, 112, 212, 122, 222]
+```
+
+</details>
+
+### 問題 4: chain によるフィルタリング
+
+filter を使わずに、chain だけで偶数のみを抽出するコードを書いてください。
+
+```typescript
+const numbers = [1, 2, 3, 4, 5, 6]
+// 偶数のみを抽出: [2, 4, 6]
+```
+
+<details>
+<summary>解答</summary>
+
+```typescript
+import { pipe } from 'fp-ts/function'
+import * as RA from 'fp-ts/ReadonlyArray'
+
+const evenNumbers = pipe(
+  numbers,
+  RA.chain((n) => (n % 2 === 0 ? [n] : []))
+)
+
+// または関数に分離
+const evenFilter = (n: number): readonly number[] =>
+  n % 2 === 0 ? [n] : []
+
+const evenNumbers2 = pipe(numbers, RA.chain(evenFilter))
+```
+
+</details>