add prediction candidate

2025-12-03 02:58:27 +00:00 · 2023-09-19 22:06:48 +09:00
parent f163350c0b
commit abd896f649
4 changed files with 126 additions and 19 deletions
--- a/Sources/KanaKanjiConverterModule/Candidate.swift
+++ b/Sources/KanaKanjiConverterModule/Candidate.swift
@@ -72,13 +72,13 @@ public struct Candidate: Sendable {
    /// 入力となるテキスト
    public var text: String
    /// 評価値
-    public let value: PValue
+    public var value: PValue
    /// composingText.inputにおいて対応する文字数。
    public var correspondingCount: Int
    /// 最後のmid(予測変換に利用)
-    public let lastMid: Int
+    public var lastMid: Int
    /// DicdataElement列
-    public let data: [DicdataElement]
+    public var data: [DicdataElement]
    /// 変換として選択した際に実行する`action`。
    /// - note: 括弧を入力した際にカーソルを移動するために追加した変数
    public var actions: [CompleteAction]
--- a/Sources/KanaKanjiConverterModule/Converter/KanaKanjiConverter.swift
+++ b/Sources/KanaKanjiConverterModule/Converter/KanaKanjiConverter.swift
@@ -595,15 +595,15 @@ import SwiftUtils
    }
    
    /// 変換確定後の予測変換候補を要求する関数
-    public func requestPredictionCandidates(leftSideCandidate: Candidate, options: ConvertRequestOptions) -> [Candidate] {
+    public func requestPredictionCandidates(leftSideCandidate: Candidate, options: ConvertRequestOptions) -> [PredictionCandidate] {
        // ゼロヒント予測変換に基づく候補を列挙
        let zeroHintResults = self.converter.getZeroHintPredictionCandidates(preparts: [leftSideCandidate], N_best: 10)
        // 予測変換に基づく候補を列挙
-        // TODO: implement
+        let predictionResults = self.converter.getPredictionCandidates(prepart: leftSideCandidate, N_best: 10)
        // 学習・ユーザ辞書に基づく候補を列挙
        // TODO: implement
        // 絵文字、記号類を列挙
        // TODO: implement
-        return zeroHintResults
+        return zeroHintResults.chained(predictionResults).max(count: 10, sortedBy: {$0.value < $1.value})
    }
 }
--- a/Sources/KanaKanjiConverterModule/Kana2Kanji/Kana2Kanji.swift
+++ b/Sources/KanaKanjiConverterModule/Kana2Kanji/Kana2Kanji.swift
@@ -67,6 +67,63 @@ struct Kana2Kanji {
        )
    }

+    func getPredictionCandidates(prepart: Candidate, N_best: Int) -> [PredictionCandidate] {
+        var result: [PredictionCandidate] = []
+        var count = 1
+        var prefixCandidate = prepart
+        prefixCandidate.actions = []
+        var prefixCandidateData = prepart.data
+        var totalWord = ""
+        var totalRuby = ""
+        var totalData: [DicdataElement] = []
+        while count <= min(prepart.data.count, 3), let element = prefixCandidateData.popLast() {
+            defer {
+                count += 1
+            }
+            // prefixCandidateを更新する
+            do {
+                prefixCandidate.value -= element.value()
+                prefixCandidate.value -= self.dicdataStore.getCCValue(prefixCandidateData.last?.rcid ?? CIDData.BOS.cid, element.lcid)
+                if DicdataStore.includeMMValueCalculation(element) {
+                    let previousMid = prefixCandidateData.last(where: DicdataStore.includeMMValueCalculation)?.mid ?? MIDData.BOS.mid
+                    prefixCandidate.lastMid = previousMid
+                    prefixCandidate.value -= self.dicdataStore.getMMValue(previousMid, element.mid)
+                }
+                prefixCandidate.data = prefixCandidateData
+                
+                prefixCandidate.text = prefixCandidateData.reduce(into: "") { $0 += $1.word }
+                prefixCandidate.correspondingCount = prefixCandidateData.reduce(into: 0) { $0 += $1.ruby.count }
+            }
+
+            
+            totalWord.insert(contentsOf: element.word, at: totalWord.startIndex)
+            totalRuby.insert(contentsOf: element.ruby, at: totalRuby.startIndex)
+            totalData.insert(element, at: 0)
+            let dicdata = self.dicdataStore.getPredictionLOUDSDicdata(key: totalRuby).filter {$0.word.hasPrefix(totalWord)}
+            
+            for data in dicdata {
+                let ccValue = self.dicdataStore.getCCValue(prefixCandidateData.last?.rcid ?? CIDData.BOS.cid, data.lcid)
+                let includeMMValueCalculation = DicdataStore.includeMMValueCalculation(data)
+                let mmValue = includeMMValueCalculation ? self.dicdataStore.getMMValue(prefixCandidate.lastMid, data.mid):.zero
+                let wValue = data.value()
+                let newValue = prefixCandidate.value + mmValue + ccValue + wValue
+                // 追加すべきindexを取得する
+                let lastindex: Int = (result.lastIndex(where: {$0.value >= newValue}) ?? -1) + 1
+                if lastindex == N_best {
+                    continue
+                }
+                // カウントがオーバーしている場合は除去する
+                if result.count >= N_best {
+                    result.removeLast()
+                }
+                // 共通接頭辞を切り落とす
+                let text = String(data.word.dropFirst(totalWord.count))
+                result.insert(.replacement(.init(text: text, targetData: totalData, replacementData: [data], value: newValue)), at: lastindex)
+            }
+        }
+        return result
+    }
+
    /// 入力がない状態から、妥当な候補を探す
    /// - parameters:
    ///   - preparts: Candidate列。以前確定した候補など
@@ -75,8 +132,8 @@ struct Kana2Kanji {
    ///   ゼロヒント予測変換の結果
    /// - note:
    ///   「食べちゃ-てる」「食べちゃ-いる」などの間抜けな候補を返すことが多いため、学習によるもの以外を無効化している。
-    func getZeroHintPredictionCandidates(preparts: some Collection<Candidate>, N_best: Int) -> [Candidate] {
-        var result: [Candidate] = []
+    func getZeroHintPredictionCandidates(preparts: some Collection<Candidate>, N_best: Int) -> [PredictionCandidate] {
+        var result: [PredictionCandidate] = []
        for candidate in preparts {
            if let last = candidate.data.last {
                let dicdata = self.dicdataStore.getZeroHintPredictionDicdata(lastRcid: last.rcid)
@@ -92,21 +149,11 @@ struct Kana2Kanji {
                    if lastindex == N_best {
                        continue
                    }
-                    // データを作成する
-                    var nodedata = candidate.data
-                    nodedata.append(data)
-                    let candidate = Candidate(
-                        text: data.word,
-                        value: data.value(),
-                        correspondingCount: data.ruby.count,
-                        lastMid: data.mid,
-                        data: [data]
-                    )
                    // カウントがオーバーしている場合は除去する
                    if result.count >= N_best {
                        result.removeLast()
                    }
-                    result.insert(candidate, at: lastindex)
+                    result.insert(.additional(.init(text: data.word, data: [data], value: newValue)), at: lastindex)
                }
            }
        }
--- a/Sources/KanaKanjiConverterModule/PredictionCandidate.swift
+++ b/Sources/KanaKanjiConverterModule/PredictionCandidate.swift
@@ -0,0 +1,60 @@
+//
+//  PredictionCandidate.swift
+//
+//
+//  Created by miwa on 2023/09/19.
+//
+
+import Foundation
+
+public enum PredictionCandidate: Sendable {
+    case additional(AdditionalPredictionCandidate)
+    case replacement(ReplacementPredictionCandidate)
+    
+    public struct AdditionalPredictionCandidate: Sendable {
+        public var text: String
+        public var data: [DicdataElement]
+        public var value: PValue
+    }
+    public struct ReplacementPredictionCandidate: Sendable {
+        /// 予測変換として表示するデータ
+        public var text: String
+        /// 置換対象のデータ
+        public var targetData: [DicdataElement]
+        /// 置換後のデータ
+        public var replacementData: [DicdataElement]
+        /// 重み
+        public var value: PValue
+    }
+
+    public var value: PValue {
+        switch self {
+        case .additional(let c):
+            c.value
+        case .replacement(let c):
+            c.value
+        }
+    }
+    
+    public func join(to candidate: consuming Candidate) -> Candidate {
+        switch self {
+        case .additional(let c):
+            for data in c.data {
+                candidate.text.append(contentsOf: data.word)
+                candidate.data.append(data)
+            }
+            candidate.value = c.value
+            candidate.correspondingCount = candidate.data.reduce(into: 0) { $0 += $1.ruby.count }
+            candidate.lastMid = c.data.last(where: DicdataStore.includeMMValueCalculation)?.mid ?? candidate.lastMid
+            return candidate
+        case .replacement(let c):
+            candidate.data.removeLast(c.targetData.count)
+            candidate.data.append(contentsOf: c.replacementData)
+            candidate.text = candidate.data.reduce(into: "") {$0 += $1.word}
+            candidate.value = c.value
+            candidate.lastMid = candidate.data.last(where: DicdataStore.includeMMValueCalculation)?.mid ?? MIDData.BOS.mid
+            candidate.correspondingCount = candidate.data.reduce(into: 0) { $0 += $1.ruby.count }
+            return candidate
+        }
+    }
+}