LMAssembly // Integrate EtenDOS SCPC data into the codebase.

Packages/vChewing_LangModelAssembly/Package.swift

@@ -26,10 +26,6 @@ let package = Package(
         .product(name: "Megrez", package: "vChewing_Megrez"),
         .product(name: "Shared", package: "vChewing_Shared"),
         .product(name: "PinyinPhonaConverter", package: "vChewing_PinyinPhonaConverter"),
-      ],
-      resources: [
-        .process("Resources/sequenceDataFromEtenDOS-chs.json"),
-        .process("Resources/sequenceDataFromEtenDOS-cht.json"),
       ]
     ),
     .testTarget(

Packages/vChewing_LangModelAssembly/Sources/LangModelAssembly/LMInstantiator.swift

@@ -64,8 +64,9 @@ public extension LMAssembly {
       lmUserOverride = .init(dataURL: uomDataURL)
     }
 
-    public func setOptions(handler: (inout Config) -> Void) {
+    @discardableResult public func setOptions(handler: (inout Config) -> Void) -> LMInstantiator {
       handler(&config)
+      return self
     }
 
     @discardableResult public static func connectSQLDB(dbPath: String, dropPreviousConnection: Bool = true) -> Bool {
@@ -97,6 +98,7 @@ public extension LMAssembly {
 
     // 磁帶資料模組。「currentCassette」對外唯讀，僅用來讀取磁帶本身的中繼資料（Metadata）。
     static var lmCassette = LMCassette()
+    static var lmPlainBopomofo = LMPlainBopomofo()
 
     // 聲明使用者語言模組。
     // 使用者語言模組使用多執行緒的話，可能會導致一些問題。有時間再仔細排查看看。
@@ -111,7 +113,6 @@ public extension LMAssembly {
     )
     var lmReplacements = LMReplacements()
     var lmAssociates = LMAssociates()
-    var lmPlainBopomofo = LMPlainBopomofo()
 
     // 半衰记忆模组
     var lmUserOverride: LMUserOverride
@@ -189,23 +190,6 @@ public extension LMAssembly {
       }
     }
 
-    public func loadSCPCSequencesData() {
-      let fileName = !isCHS ? "sequenceDataFromEtenDOS-cht" : "sequenceDataFromEtenDOS-chs"
-      guard let path = Bundle.module.path(forResource: fileName, ofType: "json") else {
-        vCLog("lmPlainBopomofo: File name access failure: \(fileName)")
-        return
-      }
-      DispatchQueue.main.async {
-        if FileManager.default.isReadableFile(atPath: path) {
-          self.lmPlainBopomofo.clear()
-          self.lmPlainBopomofo.open(path)
-          vCLog("lmPlainBopomofo: \(self.lmPlainBopomofo.count) entries of data loaded from: \(path)")
-        } else {
-          vCLog("lmPlainBopomofo: File access failure: \(path)")
-        }
-      }
-    }
-
     public var isCassetteDataLoaded: Bool { Self.lmCassette.isLoaded }
     public static func loadCassetteData(path: String) {
       DispatchQueue.main.async {
@@ -340,7 +324,9 @@ public extension LMAssembly {
 
       // 如果有檢測到使用者自訂逐字選字語料庫內的相關資料的話，在這裡先插入。
       if config.isSCPCEnabled {
-        rawAllUnigrams += lmPlainBopomofo.valuesFor(key: keyChain).map { Megrez.Unigram(value: $0, score: 0) }
+        rawAllUnigrams += Self.lmPlainBopomofo.valuesFor(key: keyChain, isCHS: isCHS).map {
+          Megrez.Unigram(value: $0, score: 0)
+        }
       }
 
       // 用 reversed 指令讓使用者語彙檔案內的詞條優先順序隨著行數增加而逐漸增高。

Packages/vChewing_LangModelAssembly/Sources/LangModelAssembly/SubLMs/lmPlainBopomofo.swift

@@ -9,60 +9,35 @@
 import Foundation
 import Shared
 
-public extension LMAssembly {
-  @frozen struct LMPlainBopomofo {
-    public private(set) var filePath: String?
-    var dataMap: [String: String] = [:]
+extension LMAssembly {
+  struct LMPlainBopomofo {
+    @usableFromInline typealias DataMap = [String: [String: String]]
+    let dataMap: DataMap
 
     public var count: Int { dataMap.count }
 
     public init() {
-      dataMap = [:]
-    }
-
-    public var isLoaded: Bool { !dataMap.isEmpty }
-
-    @discardableResult public mutating func open(_ path: String) -> Bool {
-      if isLoaded { return false }
-      let oldPath = filePath
-      filePath = nil
-
       do {
-        let rawData = try Data(contentsOf: URL(fileURLWithPath: path))
-        let rawJSON = try JSONDecoder().decode([String: String].self, from: rawData)
+        let rawData = jsnEtenDosSequence.data(using: .utf8) ?? .init([])
+        let rawJSON = try JSONDecoder().decode([String: [String: String]].self, from: rawData)
         dataMap = rawJSON
       } catch {
-        filePath = oldPath
         vCLog("\(error)")
-        vCLog("↑ Exception happened when reading JSON file at: \(path).")
-        return false
+        vCLog("↑ Exception happened when parsing raw JSON sequence data from vChewing LMAssembly.")
+        dataMap = [:]
       }
-
-      filePath = path
-      return true
     }
 
-    public mutating func clear() {
-      filePath = nil
-      dataMap.removeAll()
-    }
-
-    public func saveData() {
-      guard let filePath = filePath, let plistURL = URL(string: filePath) else { return }
-      do {
-        let plistData = try PropertyListSerialization.data(fromPropertyList: dataMap, format: .binary, options: 0)
-        try plistData.write(to: plistURL)
-      } catch {
-        vCLog("Failed to save current database to: \(filePath)")
-      }
-    }
+    public var isLoaded: Bool { !dataMap.isEmpty }
 
-    public func valuesFor(key: String) -> [String] {
+    public func valuesFor(key: String, isCHS: Bool) -> [String] {
       var pairs: [String] = []
-      if let arrRangeRecords: String = dataMap[key]?.trimmingCharacters(in: .newlines) {
+      let subKey = isCHS ? "S" : "T"
+      if let arrRangeRecords: String = dataMap[key]?[subKey] {
         pairs.append(contentsOf: arrRangeRecords.map(\.description))
       }
-      return pairs.deduplicated
+      // 這裡不做去重複處理，因為倚天中文系統注音排序適應者們已經形成了肌肉記憶。
+      return pairs
     }
 
     public func hasValuesFor(key: String) -> Bool { dataMap.keys.contains(key) }