[cli] 完全一致変換のみを出力するモードを追加 (#88)

* cli: add option to report score

* cli: 完全一致変換をサポート

Sources/CliTool/Subcommands/RunCommand.swift

@@ -15,6 +15,9 @@ extension Subcommands {
         @Flag(name: [.customLong("disable_prediction")], help: "Disable producing prediction candidates.")
         var disablePrediction = false
 
+        @Flag(name: [.customLong("only_whole_conversion")], help: "Show only whole conversion (完全一致変換).")
+        var onlyWholeConversion = false
+
         @Flag(name: [.customLong("report_score")], help: "Show internal score for the candidate.")
         var reportScore = false
 
@@ -25,19 +28,31 @@ extension Subcommands {
             var composingText = ComposingText()
             composingText.insertAtCursorPosition(input, inputStyle: .direct)
             let result = converter.requestCandidates(composingText, options: requestOptions())
-            for candidate in result.mainResults.prefix(self.displayTopN) {
+            let mainResults = result.mainResults.filter {
+                !self.onlyWholeConversion || $0.data.reduce(into: "", {$0.append(contentsOf: $1.ruby)}) == input.toKatakana()
+            }
+            for candidate in mainResults.prefix(self.displayTopN) {
                 if self.reportScore {
                     print("\(candidate.text) \(bold: "score:") \(candidate.value)")
                 } else {
                     print(candidate.text)
                 }
             }
+            if self.onlyWholeConversion {
+                // entropyを示す
+                let expValues = mainResults.map { exp(Double($0.value)) }
+                let sumOfExpValues = expValues.reduce(into: 0, +=)
+                // 確率値に補正
+                let probs = expValues.map { $0 / sumOfExpValues }
+                let entropy = -probs.reduce(into: 0) { $0 += $1 * log($1) }
+                print("\(bold: "Entropy:") \(entropy)")
+            }
         }
 
         func requestOptions() -> ConvertRequestOptions {
-            .withDefaultDictionary(
-                N_best: configNBest,
-                requireJapanesePrediction: !disablePrediction,
+            var option: ConvertRequestOptions = .withDefaultDictionary(
+                N_best: self.onlyWholeConversion ? max(self.configNBest, self.displayTopN) : self.configNBest,
+                requireJapanesePrediction: !self.onlyWholeConversion && !self.disablePrediction,
                 requireEnglishPrediction: false,
                 keyboardLanguage: .ja_JP,
                 typographyLetterCandidate: false,
@@ -52,6 +67,10 @@ extension Subcommands {
                 sharedContainerURL: URL(fileURLWithPath: ""),
                 metadata: .init(appVersionString: "anco")
             )
+            if self.onlyWholeConversion {
+                option.requestQuery = .完全一致
+            }
+            return option
         }
     }
 }

Sources/KanaKanjiConverterModule/Converter/ConvertRequestOptions.swift

@@ -49,6 +49,26 @@ public struct ConvertRequestOptions: Sendable {
         self.dictionaryResourceURL = dictionaryResourceURL
     }
 
+    package 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 = TextReplacer(), metadata: ConvertRequestOptions.Metadata, requestQuery: RequestQuery) {
+        self.N_best = N_best
+        self.requireJapanesePrediction = requireJapanesePrediction
+        self.requireEnglishPrediction = requireEnglishPrediction
+        self.keyboardLanguage = keyboardLanguage
+        self.typographyLetterCandidate = typographyLetterCandidate
+        self.unicodeCandidate = unicodeCandidate
+        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.dictionaryResourceURL = dictionaryResourceURL
+    }
+
     public var N_best: Int
     public var requireJapanesePrediction: Bool
     public var requireEnglishPrediction: Bool
@@ -71,6 +91,9 @@ public struct ConvertRequestOptions: Sendable {
     // メタデータ
     public var metadata: Metadata
 
+    // MARK: プライベートAPI
+    package var requestQuery: RequestQuery = .default
+
     static var `default`: Self {
         Self(
             N_best: 10,
@@ -103,4 +126,9 @@ public struct ConvertRequestOptions: Sendable {
         }
         var appVersionString: String
     }
+
+    package enum RequestQuery: Sendable {
+        case `default`
+        case 完全一致
+    }
 }

Sources/KanaKanjiConverterModule/Converter/KanaKanjiConverter.swift

@@ -428,6 +428,10 @@ import SwiftUtils
         let sums: [(CandidateData, Candidate)] = clauseResult.map {($0, converter.processClauseCandidate($0))}
         // 文章全体を変換した場合の候補上位5件を作る
         let whole_sentence_unique_candidates = self.getUniqueCandidate(sums.map {$0.1})
+        if case .完全一致 = options.requestQuery {
+            // 完全一致候補のみが要求されている場合、ここで全てのデータを返してreturnする
+            return ConversionResult(mainResults: whole_sentence_unique_candidates.sorted(by: {$0.value > $1.value}), firstClauseResults: [])
+        }
         let sentence_candidates = whole_sentence_unique_candidates.min(count: 5, sortedBy: {$0.value > $1.value})
         // 予測変換を最大3件作成する
         let prediction_candidates: [Candidate] = options.requireJapanesePrediction ? Array(self.getUniqueCandidate(self.getPredictionCandidate(sums, composingText: inputData, options: options)).min(count: 3, sortedBy: {$0.value > $1.value})) : []