feat: 特殊な変換のプロトコルを定め、外から挿入できるようにした

2025-12-03 02:58:27 +00:00 · 2025-05-25 18:20:08 +09:00
parent 9c6e30e8b0
commit fcc6219484
4 changed files with 141 additions and 27 deletions
--- a/Sources/CliTool/Subcommands/DictCommands/ReadCommand.swift
+++ b/Sources/CliTool/Subcommands/DictCommands/ReadCommand.swift
@@ -106,8 +106,6 @@ extension Subcommands.Dict {
                requireJapanesePrediction: false,
                requireEnglishPrediction: false,
                keyboardLanguage: .ja_JP,
-                typographyLetterCandidate: false,
-                unicodeCandidate: true,
                englishCandidateInRoman2KanaInput: true,
                fullWidthRomanCandidate: false,
                halfWidthKanaCandidate: false,
@@ -117,6 +115,7 @@ extension Subcommands.Dict {
                memoryDirectoryURL: URL(fileURLWithPath: self.dictionaryDirectory),
                sharedContainerURL: URL(fileURLWithPath: self.dictionaryDirectory),
                textReplacer: .empty,
+                specialCandidateProviders: nil,
                metadata: .init(versionString: "anco for debugging")
            )
        }
--- a/Sources/KanaKanjiConverterModule/Converter/ConvertRequestOptions.swift
+++ b/Sources/KanaKanjiConverterModule/Converter/ConvertRequestOptions.swift
@@ -9,6 +9,46 @@
 import Foundation

 public struct ConvertRequestOptions: Sendable {
+    /// 変換リクエストに必要な設定データ
+    ///
+    /// - parameters:
+    ///   - N_best: 変換候補の数。上位`N`件までの言語モデル上の妥当性を保証します。大きくすると計算量が増加します。
+    ///   - requireJapanesePrediction: 日本語の予測変換候補の必要性。`false`にすると、日本語の予測変換候補を出力しなくなります。
+    ///   - requireEnglishPrediction: 英語の予測変換候補の必要性。`false`にすると、英語の予測変換候補を出力しなくなります。ローマ字入力を用いた日本語入力では`false`にした方が良いでしょう。
+    ///   - keyboardLanguage: キーボードの言語を指定します。
+    ///   - englishCandidateInRoman2KanaInput: `true`の場合、日本語ローマ字入力時に英語変換候補を出力します。`false`の場合、ローマ字入力時に英語変換候補を出力しません。
+    ///   - fullWidthRomanCandidate: `true`の場合、全角英数字の変換候補が出力に含まれるようになります。
+    ///   - halfWidthKanaCandidate: `true`の場合、半角カナの変換候補が出力に含まれるようになります。
+    ///   - learningType: 学習モードを指定します。詳しくは`LearningType`を参照してください。
+    ///   - maxMemoryCount: 学習が有効な場合に保持するデータの最大数を指定します。`0`の場合`learningType`を`nothing`に指定する方が適切です。
+    ///   - shouldResetMemory: `true`の場合、変換を開始する前に学習データをリセットします。
+    ///   - dictionaryResourceURL: 内蔵辞書データの読み出し先を指定します。
+    ///   - memoryDirectoryURL: 学習データの保存先を指定します。書き込み可能なディレクトリを指定してください。
+    ///   - sharedContainerURL: ユーザ辞書など、キーボード外で書き込んだ設定データの保存されているディレクトリを指定します。
+    ///   - textReplacer: 予測変換のための置換機を指定します。
+    ///   - specialCandidateProviders: 特殊変換を実施する変換関数を挿入します
+    ///   - metadata: メタデータを指定します。詳しくは`ConvertRequestOptions.Metadata`を参照してください。
+    public init(N_best: Int = 10, requireJapanesePrediction: Bool, requireEnglishPrediction: Bool, keyboardLanguage: KeyboardLanguage, englishCandidateInRoman2KanaInput: Bool = false, fullWidthRomanCandidate: Bool = false, halfWidthKanaCandidate: Bool = false, learningType: LearningType, maxMemoryCount: Int = 65536, shouldResetMemory: Bool = false, dictionaryResourceURL: URL, memoryDirectoryURL: URL, sharedContainerURL: URL, textReplacer: TextReplacer, specialCandidateProviders: [any SpecialCandidateProvider]?, zenzaiMode: ZenzaiMode = .off, preloadDictionary: Bool = false, metadata: ConvertRequestOptions.Metadata?) {
+        self.N_best = N_best
+        self.requireJapanesePrediction = requireJapanesePrediction
+        self.requireEnglishPrediction = requireEnglishPrediction
+        self.keyboardLanguage = keyboardLanguage
+        self.englishCandidateInRoman2KanaInput = englishCandidateInRoman2KanaInput
+        self.fullWidthRomanCandidate = fullWidthRomanCandidate
+        self.halfWidthKanaCandidate = halfWidthKanaCandidate
+        self.learningType = learningType
+        self.maxMemoryCount = maxMemoryCount
+        self.shouldResetMemory = shouldResetMemory
+        self.memoryDirectoryURL = memoryDirectoryURL
+        self.sharedContainerURL = sharedContainerURL
+        self.metadata = metadata
+        self.textReplacer = textReplacer
+        self.specialCandidateProviders = specialCandidateProviders ?? KanaKanjiConverter.defaultSpecialCandidateProviders
+        self.zenzaiMode = zenzaiMode
+        self.preloadDictionary = preloadDictionary
+        self.dictionaryResourceURL = dictionaryResourceURL
+    }
+
    /// 変換リクエストに必要な設定データ
    ///
    /// - parameters:
@@ -29,13 +69,26 @@ public struct ConvertRequestOptions: Sendable {
    ///   - sharedContainerURL: ユーザ辞書など、キーボード外で書き込んだ設定データの保存されているディレクトリを指定します。
    ///   - textReplacer: 予測変換のための置換機を指定します。
    ///   - metadata: メタデータを指定します。詳しくは`ConvertRequestOptions.Metadata`を参照してください。
+    @available(*, deprecated, message: "it be removed in AzooKeyKanaKanjiConverter v1.0")
    public init(N_best: Int = 10, requireJapanesePrediction: Bool, requireEnglishPrediction: Bool, keyboardLanguage: KeyboardLanguage, typographyLetterCandidate: Bool = false, unicodeCandidate: Bool = true, englishCandidateInRoman2KanaInput: Bool = false, fullWidthRomanCandidate: Bool = false, halfWidthKanaCandidate: Bool = false, learningType: LearningType, maxMemoryCount: Int = 65536, shouldResetMemory: Bool = false, dictionaryResourceURL: URL, memoryDirectoryURL: URL, sharedContainerURL: URL, textReplacer: TextReplacer, zenzaiMode: ZenzaiMode = .off, preloadDictionary: Bool = false, metadata: ConvertRequestOptions.Metadata?) {
+
+        var specialCandidateProviders = [any SpecialCandidateProvider]()
+        if typographyLetterCandidate {
+            specialCandidateProviders.append(.typography)
+        }
+        if unicodeCandidate {
+            specialCandidateProviders.append(.unicode)
+        }
+        specialCandidateProviders.append(.emailAddress)
+        specialCandidateProviders.append(.timeExpression)
+        specialCandidateProviders.append(.calendar)
+        specialCandidateProviders.append(.version)
+
        self.N_best = N_best
        self.requireJapanesePrediction = requireJapanesePrediction
        self.requireEnglishPrediction = requireEnglishPrediction
        self.keyboardLanguage = keyboardLanguage
-        self.typographyLetterCandidate = typographyLetterCandidate
-        self.unicodeCandidate = unicodeCandidate
+        self.englishCandidateInRoman2KanaInput = englishCandidateInRoman2KanaInput
        self.englishCandidateInRoman2KanaInput = englishCandidateInRoman2KanaInput
        self.fullWidthRomanCandidate = fullWidthRomanCandidate
        self.halfWidthKanaCandidate = halfWidthKanaCandidate
@@ -46,6 +99,7 @@ public struct ConvertRequestOptions: Sendable {
        self.sharedContainerURL = sharedContainerURL
        self.metadata = metadata
        self.textReplacer = textReplacer
+        self.specialCandidateProviders = specialCandidateProviders
        self.zenzaiMode = zenzaiMode
        self.preloadDictionary = preloadDictionary
        self.dictionaryResourceURL = dictionaryResourceURL
@@ -103,8 +157,6 @@ public struct ConvertRequestOptions: Sendable {
    public var requireEnglishPrediction: Bool
    public var keyboardLanguage: KeyboardLanguage
    // KeyboardSettingのinjection用途
-    public var typographyLetterCandidate: Bool
-    public var unicodeCandidate: Bool
    public var englishCandidateInRoman2KanaInput: Bool
    public var fullWidthRomanCandidate: Bool
    public var halfWidthKanaCandidate: Bool
@@ -117,6 +169,8 @@ public struct ConvertRequestOptions: Sendable {
    public var memoryDirectoryURL: URL
    public var sharedContainerURL: URL
    public var dictionaryResourceURL: URL
+    /// providers to generate "special" candidates such as Unicode conversion.
+    public var specialCandidateProviders: [any SpecialCandidateProvider]
    public var zenzaiMode: ZenzaiMode
    public var preloadDictionary: Bool
    // メタデータ
@@ -131,11 +185,7 @@ public struct ConvertRequestOptions: Sendable {
            requireJapanesePrediction: true,
            requireEnglishPrediction: true,
            keyboardLanguage: .ja_JP,
-            typographyLetterCandidate: false,
-            unicodeCandidate: true,
            englishCandidateInRoman2KanaInput: true,
-            fullWidthRomanCandidate: true,
-            halfWidthKanaCandidate: false,
            learningType: .inputAndOutput,
            maxMemoryCount: 65536,
            shouldResetMemory: false,
@@ -146,6 +196,7 @@ public struct ConvertRequestOptions: Sendable {
            // dummy data, won't work
            sharedContainerURL: Bundle.main.bundleURL,
            textReplacer: .empty,
+            specialCandidateProviders: nil,
            preloadDictionary: false,
            metadata: nil
        )
--- a/Sources/KanaKanjiConverterModule/Converter/KanaKanjiConverter.swift
+++ b/Sources/KanaKanjiConverterModule/Converter/KanaKanjiConverter.swift
@@ -18,6 +18,13 @@ import EfficientNGram
    }

    private var converter = Kana2Kanji()
+    nonisolated(unsafe) public static let defaultSpecialCandidateProviders: [any SpecialCandidateProvider] = [
+        CalendarSpecialCandidateProvider(),
+        EmailAddressSpecialCandidateProvider(),
+        UnicodeSpecialCandidateProvider(),
+        VersionSpecialCandidateProvider(),
+        TimeExpressionSpecialCandidateProvider()
+    ]
    @MainActor private var checker = SpellChecker()
    private var checkerInitialized: [KeyboardLanguage: Bool] = [.none: true, .ja_JP: true]

@@ -141,23 +148,10 @@ import EfficientNGram
    ///   - string: 入力されたString
    /// - Returns:
    ///   `賢い変換候補
-    private func getWiseCandidate(_ inputData: ComposingText, options: ConvertRequestOptions) -> [Candidate] {
-        var result = [Candidate]()
-
-        // toWarekiCandidates/toSeirekiCandidatesは以前は設定可能にしていたが、特にoffにする需要がなさそうなので常時有効化した
-        result.append(contentsOf: self.toWarekiCandidates(inputData))
-        result.append(contentsOf: self.toSeirekiCandidates(inputData))
-        result.append(contentsOf: self.toEmailAddressCandidates(inputData))
-
-        if options.typographyLetterCandidate {
-            result.append(contentsOf: self.typographicalCandidates(inputData))
+    private func getSpecialCandidate(_ inputData: ComposingText, options: ConvertRequestOptions) -> [Candidate] {
+        options.specialCandidateProviders.flatMap { provider in
+            provider.provideCandidates(converter: self, inputData: inputData, options: options)
        }
-        if options.unicodeCandidate {
-            result.append(contentsOf: self.unicodeCandidates(inputData))
-        }
-        result.append(contentsOf: self.toVersionCandidate(inputData, options: options))
-        result.append(contentsOf: self.convertToTimeExpression(inputData))
-        return result
    }

    /// 変換候補の重複を除去する関数。
@@ -563,7 +557,7 @@ import EfficientNGram
        seenCandidate.formUnion(word_candidates.map {$0.text})

        // 賢く変換するパターン（任意件数）
-        let wise_candidates: [Candidate] = self.getUniqueCandidate(self.getWiseCandidate(inputData, options: options), seenCandidates: seenCandidate)
+        let wise_candidates: [Candidate] = self.getUniqueCandidate(self.getSpecialCandidate(inputData, options: options), seenCandidates: seenCandidate)
        // 途中でwise_candidatesを挟む
        word_candidates.insert(contentsOf: wise_candidates, at: min(5, word_candidates.endIndex))

--- a/Sources/KanaKanjiConverterModule/Converter/SpecialCandidateProvider.swift
+++ b/Sources/KanaKanjiConverterModule/Converter/SpecialCandidateProvider.swift
@@ -0,0 +1,70 @@
+public protocol SpecialCandidateProvider: Sendable {
+    @MainActor
+    func provideCandidates(converter: KanaKanjiConverter, inputData: ComposingText, options: ConvertRequestOptions) -> [Candidate]
+}
+
+public struct CalendarSpecialCandidateProvider: SpecialCandidateProvider {
+    public init() {}
+    @MainActor public func provideCandidates(converter: KanaKanjiConverter, inputData: ComposingText, options _: ConvertRequestOptions) -> [Candidate] {
+        converter.toWarekiCandidates(inputData) + converter.toSeirekiCandidates(inputData)
+    }
+}
+
+public struct EmailAddressSpecialCandidateProvider: SpecialCandidateProvider {
+    public init() {}
+    @MainActor public func provideCandidates(converter: KanaKanjiConverter, inputData: ComposingText, options _: ConvertRequestOptions) -> [Candidate] {
+        converter.toEmailAddressCandidates(inputData)
+    }
+}
+
+public struct TypographySpecialCandidateProvider: SpecialCandidateProvider {
+    public init() {}
+    @MainActor public func provideCandidates(converter: KanaKanjiConverter, inputData: ComposingText, options _: ConvertRequestOptions) -> [Candidate] {
+        converter.typographicalCandidates(inputData)
+    }
+}
+
+public struct UnicodeSpecialCandidateProvider: SpecialCandidateProvider {
+    public init() {}
+    @MainActor public func provideCandidates(converter: KanaKanjiConverter, inputData: ComposingText, options _: ConvertRequestOptions) -> [Candidate] {
+        converter.unicodeCandidates(inputData)
+    }
+}
+
+public struct VersionSpecialCandidateProvider: SpecialCandidateProvider {
+    public init() {}
+    @MainActor public func provideCandidates(converter: KanaKanjiConverter, inputData: ComposingText, options: ConvertRequestOptions) -> [Candidate] {
+        converter.toVersionCandidate(inputData, options: options)
+    }
+}
+
+public struct TimeExpressionSpecialCandidateProvider: SpecialCandidateProvider {
+    public init() {}
+    @MainActor public func provideCandidates(converter: KanaKanjiConverter, inputData: ComposingText, options _: ConvertRequestOptions) -> [Candidate] {
+        converter.convertToTimeExpression(inputData)
+    }
+}
+
+public extension SpecialCandidateProvider where Self == CalendarSpecialCandidateProvider {
+    static var calendar: Self { .init() }
+}
+
+public extension SpecialCandidateProvider where Self == EmailAddressSpecialCandidateProvider {
+    static var emailAddress: Self { .init() }
+}
+
+public extension SpecialCandidateProvider where Self == TypographySpecialCandidateProvider {
+    static var typography: Self { .init() }
+}
+
+public extension SpecialCandidateProvider where Self == UnicodeSpecialCandidateProvider {
+    static var unicode: Self { .init() }
+}
+
+public extension SpecialCandidateProvider where Self == VersionSpecialCandidateProvider {
+    static var version: Self { .init() }
+}
+
+public extension SpecialCandidateProvider where Self == TimeExpressionSpecialCandidateProvider {
+    static var timeExpression: Self { .init() }
+}