refactor: split TypoCorrection-related API for separate namaespace

2025-12-03 02:58:27 +00:00 · 2025-06-28 00:27:30 +09:00
parent 74d4d412c3
commit 854874fdc2
3 changed files with 35 additions and 43 deletions
--- a/Sources/KanaKanjiConverterModule/DicdataStore/DicdataStore.swift
+++ b/Sources/KanaKanjiConverterModule/DicdataStore/DicdataStore.swift
@@ -318,7 +318,7 @@ public final class DicdataStore {
            segments.append((segments.last ?? "") + String(inputData.input[rightIndex].character.toKatakana()))
        }
        // MARK: 誤り訂正の対象を列挙する。非常に重い処理。
-        var stringToInfo = inputData.getRangesWithTypos(fromIndex, rightIndexRange: toIndexLeft ..< toIndexRight)
+        var stringToInfo = TypoCorrection.getRangesWithTypos(inputs: inputData.input, leftIndex: fromIndex, rightIndexRange: toIndexLeft ..< toIndexRight)
        // MARK: 検索対象を列挙していく。
        let stringSet: [([Character], [UInt8])] = stringToInfo.keys.map {($0, $0.map(self.character2charId))}
        let (minCharIDsCount, maxCharIDsCount) = stringSet.lazy.map {$0.1.count}.minAndMax() ?? (0, -1)
@@ -425,7 +425,7 @@ public final class DicdataStore {
        }

        // MARK: 誤り訂正なし
-        let stringToEndIndex = inputData.getRangesWithoutTypos(fromIndex, rightIndexRange: toIndexLeft ..< toIndexRight)
+        let stringToEndIndex = TypoCorrection.getRangesWithoutTypos(inputs: inputData.input, leftIndex: fromIndex, rightIndexRange: toIndexLeft ..< toIndexRight)
        // MARK: 検索対象を列挙していく。
        guard let (minString, maxString) = stringToEndIndex.keys.minAndMax(by: {$0.count < $1.count}) else {
            debug(#function, "minString/maxString is nil", stringToEndIndex)
@@ -485,7 +485,7 @@ public final class DicdataStore {
        let segment = inputData.input[fromIndex...toIndex].reduce(into: "") {$0.append($1.character)}.toKatakana()

        // TODO: 最適化の余地あり
-        let string2penalty = inputData.getRangeWithTypos(fromIndex, toIndex).filter {
+        let string2penalty = TypoCorrection.getRangeWithTypos(inputs: inputData.input, leftIndex: fromIndex, rightIndex: toIndex).filter {
            needTypoCorrection || $0.value == 0.0
        }

--- a/Sources/KanaKanjiConverterModule/DicdataStore/TypoCorrection.swift
+++ b/Sources/KanaKanjiConverterModule/DicdataStore/TypoCorrection.swift
@@ -1,15 +1,8 @@
-//
-//  TypoCorrection.swift
-//  Keyboard
-//
-//  Created by ensan on 2022/12/18.
-//  Copyright © 2022 ensan. All rights reserved.
-//
 import SwiftUtils

 // MARK: 誤り訂正用のAPI
-extension ComposingText {
-    private func shouldBeRemovedForDicdataStore(components: [ConvertTargetElement]) -> Bool {
+enum TypoCorrection {
+    private static func shouldBeRemovedForDicdataStore(components: [ComposingText.ConvertTargetElement]) -> Bool {
        // 判定に使うのは最初の1エレメントの最初の文字で十分
        guard let first = components.first?.string.first?.toKatakana() else {
            return false
@@ -21,7 +14,7 @@ extension ComposingText {
    /// getRangeWithTyposの複数版にあたる。`result`の計算が一回で済む分、高速になる。
    /// 例えば`left=4, rightIndexRange=6..<10`の場合、`4...6, 4...7, 4...8, 4...9`の範囲で計算する
    /// `left <= rightIndexRange.startIndex`が常に成り立つ
-    func getRangesWithTypos(_ left: Int, rightIndexRange: Range<Int>) -> [[Character]: (endIndex: Int, penalty: PValue)] {
+    static func getRangesWithTypos(inputs: [ComposingText.InputElement], leftIndex left: Int, rightIndexRange: Range<Int>) -> [[Character]: (endIndex: Int, penalty: PValue)] {
        let count = rightIndexRange.endIndex - left
        debug(#function, left, rightIndexRange, count)
        let nodes = (0..<count).map {(i: Int) in
@@ -30,7 +23,7 @@ extension ComposingText {
                if count <= j {
                    return []
                }
-                return Self.getTypo(self.input[left + i ... left + j])
+                return Self.getTypo(inputs[left + i ... left + j])
            }
        }

@@ -39,12 +32,12 @@ extension ComposingText {
        var stringToInfo: [([Character], (endIndex: Int, penalty: PValue))] = []

        // 深さ優先で列挙する
-        var stack: [(convertTargetElements: [ConvertTargetElement], lastElement: InputElement, count: Int, penalty: PValue)] = nodes[0].compactMap { typoCandidate in
+        var stack: [(convertTargetElements: [ComposingText.ConvertTargetElement], lastElement: ComposingText.InputElement, count: Int, penalty: PValue)] = nodes[0].compactMap { typoCandidate in
            guard let firstElement = typoCandidate.inputElements.first else {
                return nil
            }
-            if Self.isLeftSideValid(first: firstElement, of: self.input, from: left) {
-                var convertTargetElements = [ConvertTargetElement]()
+            if ComposingText.isLeftSideValid(first: firstElement, of: inputs, from: left) {
+                var convertTargetElements = [ComposingText.ConvertTargetElement]()
                for element in typoCandidate.inputElements {
                    ComposingText.updateConvertTargetElements(currentElements: &convertTargetElements, newElement: element)
                }
@@ -54,7 +47,7 @@ extension ComposingText {
        }
        while let (convertTargetElements, lastElement, count, penalty) = stack.popLast() {
            if rightIndexRange.contains(count + left - 1) {
-                if let convertTarget = ComposingText.getConvertTargetIfRightSideIsValid(lastElement: lastElement, of: self.input, to: count + left, convertTargetElements: convertTargetElements)?.map({$0.toKatakana()}) {
+                if let convertTarget = ComposingText.getConvertTargetIfRightSideIsValid(lastElement: lastElement, of: inputs, to: count + left, convertTargetElements: convertTargetElements)?.map({$0.toKatakana()}) {
                    stringToInfo.append((convertTarget, (count + left - 1, penalty)))
                }
            }
@@ -65,7 +58,7 @@ extension ComposingText {
            // 訂正数上限(3個)
            if penalty >= maxPenalty {
                var convertTargetElements = convertTargetElements
-                let correct = [self.input[left + count]].map {InputElement(character: $0.character.toKatakana(), inputStyle: $0.inputStyle)}
+                let correct = [inputs[left + count]].map {ComposingText.InputElement(character: $0.character.toKatakana(), inputStyle: $0.inputStyle)}
                if count + correct.count > nodes.endIndex {
                    continue
                }
@@ -100,7 +93,7 @@ extension ComposingText {
    /// closedRangeでもらう
    /// 例えば`left=4, rightIndexRange=6..<10`の場合、`4...6, 4...7, 4...8, 4...9`の範囲で計算する
    /// `left <= rightIndexRange.startIndex`が常に成り立つ
-    func getRangesWithoutTypos(_ left: Int, rightIndexRange: Range<Int>) -> [[Character]: Int] {
+    static func getRangesWithoutTypos(inputs: [ComposingText.InputElement], leftIndex left: Int, rightIndexRange: Range<Int>) -> [[Character]: Int] {
        let count = rightIndexRange.endIndex - left
        debug(#function, left, rightIndexRange, count)
        let nodes = (0..<count).map {(i: Int) in
@@ -110,7 +103,7 @@ extension ComposingText {
                    return []
                }
                // frozen: trueとしているため、typo候補は含まれない
-                return Self.getTypo(self.input[left + i ... left + j], frozen: true)
+                return Self.getTypo(inputs[left + i ... left + j], frozen: true)
            }
        }

@@ -118,12 +111,12 @@ extension ComposingText {
        var stringToInfo: [([Character], Int)] = []

        // 深さ優先で列挙する
-        var stack: [(convertTargetElements: [ConvertTargetElement], lastElement: InputElement, count: Int)] = nodes[0].compactMap { typoCandidate in
+        var stack: [(convertTargetElements: [ComposingText.ConvertTargetElement], lastElement: ComposingText.InputElement, count: Int)] = nodes[0].compactMap { typoCandidate in
            guard let firstElement = typoCandidate.inputElements.first else {
                return nil
            }
-            if Self.isLeftSideValid(first: firstElement, of: self.input, from: left) {
-                var convertTargetElements = [ConvertTargetElement]()
+            if ComposingText.isLeftSideValid(first: firstElement, of: inputs, from: left) {
+                var convertTargetElements = [ComposingText.ConvertTargetElement]()
                for element in typoCandidate.inputElements {
                    ComposingText.updateConvertTargetElements(currentElements: &convertTargetElements, newElement: element)
                }
@@ -133,7 +126,7 @@ extension ComposingText {
        }
        while case .some((var convertTargetElements, let lastElement, let count)) = stack.popLast() {
            if rightIndexRange.contains(count + left - 1) {
-                if let convertTarget = ComposingText.getConvertTargetIfRightSideIsValid(lastElement: lastElement, of: self.input, to: count + left, convertTargetElements: convertTargetElements)?.map({$0.toKatakana()}) {
+                if let convertTarget = ComposingText.getConvertTargetIfRightSideIsValid(lastElement: lastElement, of: inputs, to: count + left, convertTargetElements: convertTargetElements)?.map({$0.toKatakana()}) {
                    stringToInfo.append((convertTarget, (count + left - 1)))
                }
            }
@@ -148,7 +141,7 @@ extension ComposingText {
                for element in $0.inputElements {
                    ComposingText.updateConvertTargetElements(currentElements: &convertTargetElements, newElement: element)
                }
-                if shouldBeRemovedForDicdataStore(components: convertTargetElements) {
+                if Self.shouldBeRemovedForDicdataStore(components: convertTargetElements) {
                    return nil
                }
                return (
@@ -162,7 +155,7 @@ extension ComposingText {
    }


-    func getRangeWithTypos(_ left: Int, _ right: Int) -> [[Character]: PValue] {
+    static func getRangeWithTypos(inputs: [ComposingText.InputElement], leftIndex left: Int, rightIndex right: Int) -> [[Character]: PValue] {
        // 各iから始まる候補を列挙する
        // 例えばinput = [d(あ), r(s), r(i), r(t), r(s), d(は), d(は), d(れ)]の場合
        // nodes =      [[d(あ)], [r(s)], [r(i)], [r(t), [r(t), r(a)]], [r(s)], [d(は), d(ば), d(ぱ)], [d(れ)]]
@@ -174,19 +167,19 @@ extension ComposingText {
                if count <= j {
                    return []
                }
-                return Self.getTypo(self.input[left + i ... left + j])
+                return Self.getTypo(inputs[left + i ... left + j])
            }
        }

        let maxPenalty: PValue = 3.5 * 3

        // 深さ優先で列挙する
-        var stack: [(convertTargetElements: [ConvertTargetElement], lastElement: InputElement, count: Int, penalty: PValue)] = nodes[0].compactMap { typoCandidate in
+        var stack: [(convertTargetElements: [ComposingText.ConvertTargetElement], lastElement: ComposingText.InputElement, count: Int, penalty: PValue)] = nodes[0].compactMap { typoCandidate in
            guard let firstElement = typoCandidate.inputElements.first else {
                return nil
            }
-            if Self.isLeftSideValid(first: firstElement, of: self.input, from: left) {
-                var convertTargetElements = [ConvertTargetElement]()
+            if ComposingText.isLeftSideValid(first: firstElement, of: inputs, from: left) {
+                var convertTargetElements = [ComposingText.ConvertTargetElement]()
                for element in typoCandidate.inputElements {
                    ComposingText.updateConvertTargetElements(currentElements: &convertTargetElements, newElement: element)
                }
@@ -199,7 +192,7 @@ extension ComposingText {

        while let (convertTargetElements, lastElement, count, penalty) = stack.popLast() {
            if count + left - 1 == right {
-                if let convertTarget = ComposingText.getConvertTargetIfRightSideIsValid(lastElement: lastElement, of: self.input, to: count + left, convertTargetElements: convertTargetElements)?.map({$0.toKatakana()}) {
+                if let convertTarget = ComposingText.getConvertTargetIfRightSideIsValid(lastElement: lastElement, of: inputs, to: count + left, convertTargetElements: convertTargetElements)?.map({$0.toKatakana()}) {
                    stringToPenalty.append((convertTarget, penalty))
                }
                continue
@@ -211,7 +204,7 @@ extension ComposingText {
            // 訂正数上限(3個)
            if penalty >= maxPenalty {
                var convertTargetElements = convertTargetElements
-                let correct = [self.input[left + count]].map {InputElement(character: $0.character.toKatakana(), inputStyle: $0.inputStyle)}
+                let correct = [inputs[left + count]].map {ComposingText.InputElement(character: $0.character.toKatakana(), inputStyle: $0.inputStyle)}
                if count + correct.count > nodes.endIndex {
                    continue
                }
@@ -228,7 +221,7 @@ extension ComposingText {
                    for element in $0.inputElements {
                        ComposingText.updateConvertTargetElements(currentElements: &convertTargetElements, newElement: element)
                    }
-                    if shouldBeRemovedForDicdataStore(components: convertTargetElements) {
+                    if Self.shouldBeRemovedForDicdataStore(components: convertTargetElements) {
                        return nil
                    }
                    return (
@@ -243,7 +236,7 @@ extension ComposingText {
        return Dictionary(stringToPenalty, uniquingKeysWith: max)
    }

-    private static func getTypo(_ elements: some Collection<InputElement>, frozen: Bool = false) -> [TypoCandidate] {
+    private static func getTypo(_ elements: some Collection<ComposingText.InputElement>, frozen: Bool = false) -> [TypoCandidate] {
        let key = elements.reduce(into: "") {$0.append($1.character)}.toKatakana()

        if (elements.allSatisfy {$0.inputStyle == .direct}) {
@@ -251,19 +244,19 @@ extension ComposingText {
            if key.count > 1 {
                return dictionary[key, default: []].map {
                    TypoCandidate(
-                        inputElements: $0.value.map {InputElement(character: $0, inputStyle: .direct)},
+                        inputElements: $0.value.map {ComposingText.InputElement(character: $0, inputStyle: .direct)},
                        weight: $0.weight
                    )
                }
            } else if key.count == 1 {
                var result = dictionary[key, default: []].map {
                    TypoCandidate(
-                        inputElements: $0.value.map {InputElement(character: $0, inputStyle: .direct)},
+                        inputElements: $0.value.map {ComposingText.InputElement(character: $0, inputStyle: .direct)},
                        weight: $0.weight
                    )
                }
                // そのまま
-                result.append(TypoCandidate(inputElements: key.map {InputElement(character: $0, inputStyle: .direct)}, weight: 0))
+                result.append(TypoCandidate(inputElements: key.map {ComposingText.InputElement(character: $0, inputStyle: .direct)}, weight: 0))
                return result
            }
        }
@@ -272,20 +265,20 @@ extension ComposingText {
            if key.count > 1 {
                return dictionary[key, default: []].map {
                    TypoCandidate(
-                        inputElements: $0.map {InputElement(character: $0, inputStyle: .roman2kana)},
+                        inputElements: $0.map {ComposingText.InputElement(character: $0, inputStyle: .roman2kana)},
                        weight: 3.5
                    )
                }
            } else if key.count == 1 {
                var result = dictionary[key, default: []].map {
                    TypoCandidate(
-                        inputElements: $0.map {InputElement(character: $0, inputStyle: .roman2kana)},
+                        inputElements: $0.map {ComposingText.InputElement(character: $0, inputStyle: .roman2kana)},
                        weight: 3.5
                    )
                }
                // そのまま
                result.append(
-                    TypoCandidate(inputElements: key.map {InputElement(character: $0, inputStyle: .roman2kana)}, weight: 0)
+                    TypoCandidate(inputElements: key.map {ComposingText.InputElement(character: $0, inputStyle: .roman2kana)}, weight: 0)
                )
                return result
            }
@@ -306,7 +299,7 @@ extension ComposingText {
    }

    struct TypoCandidate: Equatable {
-        var inputElements: [InputElement]
+        var inputElements: [ComposingText.InputElement]
        var weight: PValue
    }

--- a/Sources/KanaKanjiConverterModule/LOUDS/LOUDS.swift
+++ b/Sources/KanaKanjiConverterModule/LOUDS/LOUDS.swift
@@ -238,7 +238,6 @@ package struct LOUDS: Sendable {
    /// - Note: より適切な名前に変更したい
    @inlinable func byfixNodeIndices(targets: [[UInt8]], depth: Range<Int>) -> [Int] {
        // 辞書順でソートする
-//        let targets = targets.sorted(by: Self.lexLessThan)
        var targets = targets
        targets.sort(by: Self.lexLessThan)
        // 最終出力となる