LMAssembly // Pack LMUserOverride inside LMInstantiator, etc.

Packages/vChewing_LangModelAssembly/README.md

@@ -1,17 +1,17 @@
 # LangModelAssembly
 
-威注音輸入法的語言模組總成套裝。
+威注音輸入法的語言模組總成套裝，以 LMAssembly 命名空間承載下述唯二對外物件：
 
-- vChewingLM：總命名空間，也承載一些在套裝內共用的工具函式。
 - LMConsolidator：自動格式整理模組。
-- LMInstantiator：語言模組副本化模組。另有其日期時間擴充模組可用（對 CIN 磁帶模式無效）。
+- LMInstantiator：語言模組副本化模組，亦集成一些自身功能擴展。
+
+LMAssembly 總命名空間也承載一些在套裝內共用的工具函式。
 
 以下是子模組：
 
-- lmCassette：專門用來處理 CIN 磁帶檔案的模組，命名為「遠野」引擎。
 - LMAssociates：關聯詞語模組。
+- lmCassette：專門用來處理 CIN 磁帶檔案的模組，命名為「遠野」引擎。
 - LMCoreEX：可以直接讀取 TXT 格式的帶有權重資料的語彙檔案的模組。
-- LMCoreJSON：專門用來讀取原廠 JSON 檔案的模組。
 - lmPlainBopomofo：專門用來讀取使用者自訂ㄅ半候選字順序覆蓋定義檔案（plist）的模組。
 - lmReplacements：專門用來讀取使用者語彙置換模式的辭典資料的模組。
 - lmUserOverride：半衰記憶模組。

Packages/vChewing_LangModelAssembly/Sources/LangModelAssembly/InputToken.swift

@@ -11,29 +11,31 @@ import Foundation
 /// 工作原理：先用 InputToken.parse 分析原始字串，給出準確的 Token。
 /// 然後再讓這個 Token 用 .translated() 自我表述出轉換結果。
 
-public enum InputToken {
-  case timeZone(shortened: Bool)
-  case timeNow(shortened: Bool)
-  case date(dayDelta: Int = 0, yearDelta: Int = 0, shortened: Bool = true, luna: Bool = false)
-  case week(dayDelta: Int = 0, shortened: Bool = true)
-  case year(yearDelta: Int = 0)
-  case yearGanzhi(yearDelta: Int = 0)
-  case yearZodiac(yearDelta: Int = 0)
+extension LMAssembly {
+  enum InputToken {
+    case timeZone(shortened: Bool)
+    case timeNow(shortened: Bool)
+    case date(dayDelta: Int = 0, yearDelta: Int = 0, shortened: Bool = true, luna: Bool = false)
+    case week(dayDelta: Int = 0, shortened: Bool = true)
+    case year(yearDelta: Int = 0)
+    case yearGanzhi(yearDelta: Int = 0)
+    case yearZodiac(yearDelta: Int = 0)
+  }
 }
 
 // MARK: - 正式對外投入使用的 API。
 
 public extension String {
   func parseAsInputToken(isCHS: Bool) -> [String] {
-    InputToken.parse(from: self).map { $0.translated(isCHS: isCHS) }.flatMap { $0 }.deduplicated
+    LMAssembly.InputToken.parse(from: self).map { $0.translated(isCHS: isCHS) }.flatMap { $0 }.deduplicated
   }
 }
 
 // MARK: - Parser parsing raw token value to construct token.
 
-public extension InputToken {
-  static func parse(from rawToken: String) -> [InputToken] {
-    var result: [InputToken] = []
+extension LMAssembly.InputToken {
+  static func parse(from rawToken: String) -> [LMAssembly.InputToken] {
+    var result: [LMAssembly.InputToken] = []
     guard rawToken.prefix(6) == "MACRO@" else { return result }
     var mapParams: [String: Int] = [:]
     let tokenComponents = rawToken.dropFirst(6).split(separator: "_").map { param in
@@ -69,7 +71,7 @@ public extension InputToken {
 
 // MARK: - Parser parsing token itself.
 
-public extension InputToken {
+extension LMAssembly.InputToken {
   func translated(isCHS: Bool) -> [String] {
     let locale = Locale(identifier: isCHS ? "zh-Hans" : "zh-Hant-TW")
     let formatter = DateFormatter()

Packages/vChewing_LangModelAssembly/Sources/LangModelAssembly/LMConsolidator.swift

@@ -10,7 +10,7 @@ import Foundation
 import LineReader
 import Shared
 
-public extension vChewingLM {
+public extension LMAssembly {
   enum LMConsolidator {
     public static let kPragmaHeader = "# 𝙵𝙾𝚁𝙼𝙰𝚃 𝚘𝚛𝚐.𝚊𝚝𝚎𝚕𝚒𝚎𝚛𝙸𝚗𝚖𝚞.𝚟𝚌𝚑𝚎𝚠𝚒𝚗𝚐.𝚞𝚜𝚎𝚛𝙻𝚊𝚗𝚐𝚞𝚊𝚐𝚎𝙼𝚘𝚍𝚎𝚕𝙳𝚊𝚝𝚊.𝚏𝚘𝚛𝚖𝚊𝚝𝚝𝚎𝚍"
 

Packages/vChewing_LangModelAssembly/Sources/LangModelAssembly/LMInstantiator.swift

@@ -11,7 +11,7 @@ import Megrez
 import Shared
 import SQLite3
 
-public extension vChewingLM {
+public extension LMAssembly {
   /// 語言模組副本化模組（LMInstantiator，下稱「LMI」）自身為符合天權星組字引擎內
   /// 的 LangModelProtocol 協定的模組、統籌且整理來自其它子模組的資料（包括使
   /// 用者語彙、繪文字模組、語彙濾除表、原廠語言模組等）。
@@ -56,8 +56,12 @@ public extension vChewingLM {
     public var config = Config()
 
     // 這句需要留著，不然無法被 package 外界存取。
-    public init(isCHS: Bool = false) {
+    public init(
+      isCHS: Bool = false,
+      uomDataURL: URL? = nil
+    ) {
       self.isCHS = isCHS
+      lmUserOverride = .init(dataURL: uomDataURL)
     }
 
     public func setOptions(handler: (inout Config) -> Void) {
@@ -109,6 +113,9 @@ public extension vChewingLM {
     var lmAssociates = LMAssociates()
     var lmPlainBopomofo = LMPlainBopomofo()
 
+    // 半衰记忆模组
+    var lmUserOverride: LMUserOverride
+
     // MARK: - 工具函式
 
     public func resetFactoryJSONModels() {}

Packages/vChewing_LangModelAssembly/Sources/LangModelAssembly/LMInstantiator_CassetteExtension.swift

@@ -10,7 +10,7 @@ import Foundation
 import Megrez
 import Shared
 
-public extension vChewingLM.LMInstantiator {
+public extension LMAssembly.LMInstantiator {
   /// 磁帶模式專用：當前磁帶所規定的花牌鍵。
   var cassetteWildcardKey: String { Self.lmCassette.wildcardKey }
   /// 磁帶模式專用：當前磁帶規定的最大碼長。

Packages/vChewing_LangModelAssembly/Sources/LangModelAssembly/LMInstantiator_DateTimeExtension.swift

@@ -11,7 +11,7 @@ import Megrez
 
 // MARK: - 日期時間便捷輸入功能
 
-extension vChewingLM.LMInstantiator {
+extension LMAssembly.LMInstantiator {
   func queryDateTimeUnigrams(with key: String = "") -> [Megrez.Unigram] {
     guard let tokenTrigger = TokenTrigger(rawValue: key) else { return [] }
     var results = [Megrez.Unigram]()

Packages/vChewing_LangModelAssembly/Sources/LangModelAssembly/LMInstantiator_NumPadExtension.swift

@@ -9,7 +9,7 @@
 import Foundation
 import Megrez
 
-public extension vChewingLM.LMInstantiator {
+public extension LMAssembly.LMInstantiator {
   func supplyNumPadUnigrams(key: String) -> [Megrez.Unigram] {
     guard let status = config.numPadFWHWStatus else { return [] }
     let initials = "_NumPad_"

Packages/vChewing_LangModelAssembly/Sources/LangModelAssembly/LMInstantiator_SQLExtension.swift

@@ -31,30 +31,32 @@ import SQLite3
    ) WITHOUT ROWID;
  */
 
-enum CoreColumn: Int32 {
-  case theDataCHS = 1 // 簡體中文
-  case theDataCHT = 2 // 繁體中文
-  case theDataCNS = 3 // 全字庫
-  case theDataMISC = 4 // 待辦
-  case theDataSYMB = 5 // 符號圖
-  case theDataCHEW = 6 // 注音文
+extension LMAssembly.LMInstantiator {
+  enum CoreColumn: Int32 {
+    case theDataCHS = 1 // 簡體中文
+    case theDataCHT = 2 // 繁體中文
+    case theDataCNS = 3 // 全字庫
+    case theDataMISC = 4 // 待辦
+    case theDataSYMB = 5 // 符號圖
+    case theDataCHEW = 6 // 注音文
 
-  var name: String { String(describing: self) }
+    var name: String { String(describing: self) }
 
-  var id: Int32 { rawValue }
+    var id: Int32 { rawValue }
 
-  var defaultScore: Double {
-    switch self {
-    case .theDataCHEW: return -1
-    case .theDataCNS: return -11
-    case .theDataSYMB: return -13
-    case .theDataMISC: return -10
-    default: return -9.9
+    var defaultScore: Double {
+      switch self {
+      case .theDataCHEW: return -1
+      case .theDataCNS: return -11
+      case .theDataSYMB: return -13
+      case .theDataMISC: return -10
+      default: return -9.9
+      }
     }
   }
 }
 
-extension vChewingLM.LMInstantiator {
+extension LMAssembly.LMInstantiator {
   fileprivate static func querySQL(strStmt sqlQuery: String, coreColumn column: CoreColumn, handler: (String) -> Void) {
     guard Self.ptrSQL != nil else { return }
     performStatementSansResult { ptrStatement in
@@ -134,7 +136,9 @@ extension vChewingLM.LMInstantiator {
   /// - parameters:
   ///   - key: 讀音索引鍵。
   ///   - column: 資料欄位。
-  func factoryUnigramsFor(key: String, column: CoreColumn) -> [Megrez.Unigram] {
+  func factoryUnigramsFor(
+    key: String, column: LMAssembly.LMInstantiator.CoreColumn
+  ) -> [Megrez.Unigram] {
     if key == "_punctuation_list" { return [] }
     var grams: [Megrez.Unigram] = []
     var gramsHW: [Megrez.Unigram] = []
@@ -210,7 +214,7 @@ extension vChewingLM.LMInstantiator {
   }
 }
 
-private extension vChewingLM.LMInstantiator {
+private extension LMAssembly.LMInstantiator {
   /// 內部函式，用以將注音讀音索引鍵進行加密。
   ///
   /// 使用這種加密字串作為索引鍵，可以增加對 json 資料庫的存取速度。
@@ -258,7 +262,7 @@ private extension vChewingLM.LMInstantiator {
   ]
 }
 
-public extension vChewingLM.LMInstantiator {
+public extension LMAssembly.LMInstantiator {
   @discardableResult static func connectToTestSQLDB() -> Bool {
     Self.connectSQLDB(dbPath: #":memory:"#) && sqlTestCoreLMData.runAsSQLExec(dbPointer: &ptrSQL)
   }

Packages/vChewing_LangModelAssembly/Sources/LangModelAssembly/LMInstantiator_UOMRepresentable.swift

@@ -0,0 +1,60 @@
+// (c) 2021 and onwards The vChewing Project (MIT-NTL License).
+// ====================
+// This code is released under the MIT license (SPDX-License-Identifier: MIT)
+// ... with NTL restriction stating that:
+// No trademark license is granted to use the trade names, trademarks, service
+// marks, or product names of Contributor, except as required to fulfill notice
+// requirements defined in MIT License.
+
+import Foundation
+import Megrez
+
+public extension LMAssembly.LMInstantiator {
+  func performUOMObservation(
+    walkedBefore: [Megrez.Node],
+    walkedAfter: [Megrez.Node],
+    cursor: Int,
+    timestamp: Double,
+    saveCallback: (() -> Void)? = nil
+  ) {
+    lmUserOverride.performObservation(
+      walkedBefore: walkedBefore,
+      walkedAfter: walkedAfter,
+      cursor: cursor,
+      timestamp: timestamp,
+      saveCallback: saveCallback
+    )
+  }
+
+  func fetchUOMSuggestion(
+    currentWalk: [Megrez.Node],
+    cursor: Int,
+    timestamp: Double
+  ) -> LMAssembly.OverrideSuggestion {
+    lmUserOverride.fetchSuggestion(
+      currentWalk: currentWalk,
+      cursor: cursor,
+      timestamp: timestamp
+    )
+  }
+
+  func loadUOMData(fromURL fileURL: URL? = nil) {
+    lmUserOverride.loadData(fromURL: fileURL)
+  }
+
+  func saveUOMData(toURL fileURL: URL? = nil) {
+    lmUserOverride.saveData(toURL: fileURL)
+  }
+
+  func clearUOMData(withURL fileURL: URL? = nil) {
+    lmUserOverride.clearData(withURL: fileURL)
+  }
+
+  func bleachSpecifiedUOMSuggestions(targets: [String], saveCallback: (() -> Void)? = nil) {
+    lmUserOverride.bleachSpecifiedSuggestions(targets: targets, saveCallback: saveCallback)
+  }
+
+  func bleachUOMUnigrams(saveCallback: (() -> Void)? = nil) {
+    lmUserOverride.bleachUnigrams(saveCallback: saveCallback)
+  }
+}

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

@@ -10,8 +10,8 @@ import Megrez
 import PinyinPhonaConverter
 import Shared
 
-public extension vChewingLM {
-  @frozen struct LMAssociates {
+extension LMAssembly {
+  struct LMAssociates {
     public private(set) var filePath: String?
     var rangeMap: [String: [(Range<String.Index>, Int)]] = [:]
     var strData: String = ""

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

@@ -12,9 +12,9 @@ import LineReader
 import Megrez
 import Shared
 
-public extension vChewingLM {
+extension LMAssembly {
   /// 磁帶模組，用來方便使用者自行擴充字根輸入法。
-  @frozen struct LMCassette {
+  struct LMCassette {
     public private(set) var filePath: String?
     public private(set) var nameShort: String = ""
     public private(set) var nameENG: String = ""
@@ -45,7 +45,7 @@ public extension vChewingLM {
   }
 }
 
-public extension vChewingLM.LMCassette {
+extension LMAssembly.LMCassette {
   /// 計算頻率時要用到的東西 - fscale
   private static let fscale = 2.7
   /// 萬用花牌字符，哪怕花牌鍵仍不可用。
@@ -86,7 +86,7 @@ public extension vChewingLM.LMCassette {
     if FileManager.default.fileExists(atPath: path) {
       do {
         guard let fileHandle = FileHandle(forReadingAtPath: path) else {
-          throw vChewingLM.FileErrors.fileHandleError("")
+          throw LMAssembly.FileErrors.fileHandleError("")
         }
         let lineReader = try LineReader(file: fileHandle)
         var theMaxKeyLength = 1

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

@@ -10,12 +10,12 @@ import Megrez
 import PinyinPhonaConverter
 import Shared
 
-public extension vChewingLM {
+extension LMAssembly {
   /// 與之前的 LMCore 不同，LMCoreEX 不在辭典內記錄實體，而是記錄 range 範圍。
   /// 需要資料的時候，直接拿 range 去 strData 取資料。
   /// 資料記錄原理與上游 C++ 的 ParselessLM 差不多，但用的是 Swift 原生手段。
   /// 主要時間消耗仍在 For 迴圈，但這個算法可以顯著減少記憶體佔用。
-  @frozen struct LMCoreEX {
+  struct LMCoreEX {
     public private(set) var filePath: String?
     /// 資料庫辭典。索引內容為注音字串，資料內容則為字串首尾範圍、方便自 strData 取資料。
     var rangeMap: [String: [Range<String.Index>]] = [:]

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

@@ -9,7 +9,7 @@
 import Foundation
 import Shared
 
-public extension vChewingLM {
+public extension LMAssembly {
   @frozen struct LMPlainBopomofo {
     public private(set) var filePath: String?
     var dataMap: [String: String] = [:]
@@ -29,13 +29,8 @@ public extension vChewingLM {
 
       do {
         let rawData = try Data(contentsOf: URL(fileURLWithPath: path))
-        if let rawJSON = try? JSONSerialization.jsonObject(with: rawData) as? [String: String] {
-          dataMap = rawJSON
-        } else {
-          filePath = oldPath
-          vCLog("↑ Exception happened when reading JSON file at: \(path).")
-          return false
-        }
+        let rawJSON = try JSONDecoder().decode([String: String].self, from: rawData)
+        dataMap = rawJSON
       } catch {
         filePath = oldPath
         vCLog("\(error)")

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

@@ -8,8 +8,8 @@
 
 import Shared
 
-public extension vChewingLM {
-  @frozen struct LMReplacements {
+extension LMAssembly {
+  struct LMReplacements {
     public private(set) var filePath: String?
     var rangeMap: [String: Range<String.Index>] = [:]
     var strData: String = ""