refactor: Zenz関連の処理の変更を容易にするためリファクタリングを実施 (#325)

* refactor: 内部的にリファクタの概念を導入

* feat: split file

* refactor: remove predictNextCharacter-related features as it is no longer used

* refactor: remove zenz-v1 support

* refactor: remove unused api

* feat: refactor to make ZenzContext more simple

Author: ensan-hcl

Sources/CliTool/Anco.swift

@@ -11,7 +11,6 @@ struct Anco: AsyncParsableCommand {
             Subcommands.Evaluate.self,
             Subcommands.ZenzEvaluate.self,
             Subcommands.Session.self,
-            Subcommands.ExperimentalPredict.self,
             Subcommands.NGram.self
         ],
         defaultSubcommand: Subcommands.Run.self

Sources/CliTool/Subcommands/ExperimentalPredict.swift

@@ -42,8 +42,6 @@ extension Subcommands {
         var configZenzaiProfile: String?
         @Option(name: [.customLong("config_topic")], help: "enable topic prompting for zenz-v3 and later.")
         var configZenzaiTopic: String?
-        @Flag(name: [.customLong("zenz_v1")], help: "Use zenz_v1 model.")
-        var zenzV1 = false
         @Flag(name: [.customLong("zenz_v2")], help: "Use zenz_v2 model.")
         var zenzV2 = false
         @Flag(name: [.customLong("zenz_v3")], help: "Use zenz_v3 model.")
@@ -86,13 +84,13 @@ extension Subcommands {
         }
 
         @MainActor mutating func run() async {
-            if self.zenzV1 || self.zenzV2 {
+            if self.zenzV2 {
                 print("\(bold: "We strongly recommend to use zenz-v3 models")")
             }
-            if (self.zenzV1 || self.zenzV2 || self.zenzV3) && self.zenzWeightPath.isEmpty {
+            if (self.zenzV2 || self.zenzV3) && self.zenzWeightPath.isEmpty {
                 preconditionFailure("\(bold: "zenz version is specified but --zenz weight is not specified")")
             }
-            if !self.zenzWeightPath.isEmpty && (!self.zenzV1 && !self.zenzV2 && !self.zenzV3) {
+            if !self.zenzWeightPath.isEmpty && (!self.zenzV2 && !self.zenzV3) {
                 print("zenz version is not specified. By default, zenz-v3 will be used.")
             }
 
@@ -193,17 +191,6 @@ extension Subcommands {
                         print("anything should not be saved because the learning type is not for update memory")
                     }
                     continue
-                case ":p", ":pred":
-                    // 次の文字の予測を取得する
-                    let results = converter.predictNextCharacter(
-                        leftSideContext: leftSideContext,
-                        count: 10,
-                        options: requestOptions(learningType: learningType, memoryDirectory: memoryDirectory, leftSideContext: leftSideContext)
-                    )
-                    if let firstCandidate = results.first {
-                        leftSideContext.append(firstCandidate.character)
-                    }
-                    continue
                 case let command where command == ":ip" || command.hasPrefix(":ip "):
                     // 入力中の次の文字の予測を取得する (zenz-v3)
                     let parts = command.split(separator: " ")
@@ -263,7 +250,6 @@ extension Subcommands {
                     \(bold: ":d, :del") - delete one character
                     \(bold: ":n, :next") - see more candidates
                     \(bold: ":s, :save") - save memory to temporary directory
-                    \(bold: ":p, :pred") - predict next one character
                     \(bold: ":ip [n] [max_entropy=F] [min_length=N]") - predict next input character(s) (zenz-v3)
                     \(bold: ":%d") - select candidate at that index (like :3 to select 3rd candidate)
                     \(bold: ":ctx %s") - set the string as context
@@ -350,9 +336,7 @@ extension Subcommands {
         }
 
         func requestOptions(learningType: LearningType, memoryDirectory: URL, leftSideContext: String?) -> ConvertRequestOptions {
-            let zenzaiVersionDependentMode: ConvertRequestOptions.ZenzaiVersionDependentMode = if self.zenzV1 {
-                .v1
-            } else if self.zenzV2 {
+            let zenzaiVersionDependentMode: ConvertRequestOptions.ZenzaiVersionDependentMode = if self.zenzV2 {
                 .v2(.init(profile: self.configZenzaiProfile, leftSideContext: leftSideContext))
             } else {
                 .v3(.init(profile: self.configZenzaiProfile, topic: self.configZenzaiTopic, leftSideContext: leftSideContext))

Sources/CliTool/Subcommands/SessionCommand.swift

@@ -0,0 +1,49 @@
+import Foundation
+import HeapModule
+
+struct FixedSizeHeap<Element: Comparable> {
+    private var size: Int
+    private var heap: Heap<Element>
+
+    init(size: Int) {
+        self.size = size
+        self.heap = []
+    }
+
+    mutating func removeMax() {
+        self.heap.removeMax()
+    }
+
+    mutating func removeMin() {
+        self.heap.removeMin()
+    }
+
+    @discardableResult
+    mutating func insertIfPossible(_ element: Element) -> Bool {
+        if self.heap.count < self.size {
+            self.heap.insert(element)
+            return true
+        } else if let min = self.heap.min, element > min {
+            self.heap.replaceMin(with: element)
+            return true
+        } else {
+            return false
+        }
+    }
+
+    var unordered: [Element] {
+        self.heap.unordered
+    }
+
+    var max: Element? {
+        self.heap.max
+    }
+
+    var min: Element? {
+        self.heap.min
+    }
+
+    var isEmpty: Bool {
+        self.heap.isEmpty
+    }
+}

Sources/KanaKanjiConverterModule/ConversionAlgorithms/Zenzai/Zenz/FixedSizeHeap.swift

@@ -26,7 +26,7 @@ package final class Zenz {
     }
 
     package func endSession() {
-        try? self.zenzContext?.reset_context()
+        try? self.zenzContext?.resetContext()
     }
 
     func candidateEvaluate(
@@ -36,12 +36,13 @@ package final class Zenz {
         prefixConstraint: Kana2Kanji.PrefixConstraint,
         personalizationMode: (mode: ConvertRequestOptions.ZenzaiMode.PersonalizationMode, base: EfficientNGram, personal: EfficientNGram)?,
         versionDependentConfig: ConvertRequestOptions.ZenzaiVersionDependentMode
-    ) -> ZenzContext.CandidateEvaluationResult {
+    ) -> CandidateEvaluationResult {
         guard let zenzContext else {
             return .error
         }
         for candidate in candidates {
-            return zenzContext.evaluate_candidate(
+            return ZenzCandidateEvaluator.evaluate(
+                context: zenzContext,
                 input: convertTarget.toKatakana(),
                 candidate: candidate,
                 requestRichCandidates: requestRichCandidates,
@@ -53,20 +54,6 @@ package final class Zenz {
         return .error
     }
 
-    func predictNextCharacter(leftSideContext: String, count: Int) -> [(character: Character, value: Float)] {
-        guard let zenzContext else {
-            return []
-        }
-        return zenzContext.predict_next_character(leftSideContext: leftSideContext, count: count)
-    }
-
-    func predictNextInputCharacter(leftSideContext: String, composingText: String, count: Int, versionDependentConfig: ConvertRequestOptions.ZenzaiVersionDependentMode) -> [(character: Character, value: Float)] {
-        guard let zenzContext else {
-            return []
-        }
-        return zenzContext.predict_next_input_character(leftSideContext: leftSideContext, composingText: composingText, count: count, versionDependentConfig: versionDependentConfig)
-    }
-
     func predictNextInputText(
         leftSideContext: String,
         composingText: String,
@@ -79,7 +66,8 @@ package final class Zenz {
         guard let zenzContext else {
             return ""
         }
-        return zenzContext.predict_next_input_text(
+        return ZenzInputTextGenerator.generate(
+            context: zenzContext,
             leftSideContext: leftSideContext,
             composingText: composingText,
             count: count,
@@ -91,6 +79,9 @@ package final class Zenz {
     }
 
     package func pureGreedyDecoding(pureInput: String, maxCount: Int = .max) -> String {
-        self.zenzContext?.pure_greedy_decoding(leftSideContext: pureInput, maxCount: maxCount) ?? ""
+        guard let zenzContext else {
+            return ""
+        }
+        return ZenzPureGreedyDecoder.decode(context: zenzContext, leftSideContext: pureInput, maxCount: maxCount)
     }
 }