Megrez // Allow resynchronizing unigram data in update().

Packages/vChewing_Megrez/Sources/Megrez/1_Compositor.swift

@@ -172,7 +172,7 @@ extension Megrez {
   }
 }
 
-// MARK: - Internal Methods
+// MARK: - Internal Methods (Maybe Public)
 
 extension Megrez.Compositor {
   // MARK: Internal methods for maintaining the grid.
@@ -242,45 +242,51 @@ extension Megrez.Compositor {
     return true
   }
 
-  func getJointKey(range: Range<Int>) -> String {
-    // 下面這句不能用 contains，不然會要求至少 macOS 13 Ventura。
-    guard range.upperBound <= keys.count, range.lowerBound >= 0 else { return "" }
-    return keys[range].joined(separator: separator)
-  }
-
   func getJointKeyArray(range: Range<Int>) -> [String] {
     // 下面這句不能用 contains，不然會要求至少 macOS 13 Ventura。
     guard range.upperBound <= keys.count, range.lowerBound >= 0 else { return [] }
     return keys[range].map { String($0) }
   }
 
-  func hasNode(at location: Int, length: Int, key: String) -> Bool {
+  func getNode(at location: Int, length: Int, keyArray: [String]) -> Node? {
     let location = max(min(location, spans.count), 0)  // 防呆
-    guard let node = spans[location].nodeOf(length: length) else { return false }
+    guard let node = spans[location].nodeOf(length: length) else { return nil }
-    return key == node.key
+    return keyArray == node.keyArray ? node : nil
   }
 
   /// 根據當前狀況更新整個組字器的節點文脈。
-  /// - Returns: 新增了多少節點。
+  /// - Returns: 新增了多少節點。如果返回「0」則表示可能發生了錯誤。
-  @discardableResult mutating func update() -> Int {
+  @discardableResult public mutating func update(updateExisting: Bool = false) -> Int {
     let maxSpanLength = Megrez.Compositor.maxSpanLength
     let range = max(0, cursor - maxSpanLength)..<min(cursor + maxSpanLength, keys.count)
-    var nodesInserted = 0
+    var nodesChanged = 0
     for position in range {
       for theLength in 1...min(maxSpanLength, range.upperBound - position) {
         let jointKeyArray = getJointKeyArray(range: position..<(position + theLength))
-        let jointKey = getJointKey(range: position..<(position + theLength))
+        let jointKey = jointKeyArray.joined(separator: separator)
-        if hasNode(at: position, length: theLength, key: jointKey) { continue }
+        if let theNode = getNode(at: position, length: theLength, keyArray: jointKeyArray) {
+          if !updateExisting { continue }
+          let unigrams = langModel.unigramsFor(key: jointKey)
+          // 自動銷毀無效的節點。
+          if unigrams.isEmpty {
+            if theNode.keyArray.count == 1 { continue }
+            spans[position].nodes.removeAll { $0 == theNode }
+          } else {
+            theNode.resetUnigrams(using: unigrams)
+          }
+          nodesChanged += 1
+          continue
+        }
         let unigrams = langModel.unigramsFor(key: jointKey)
         guard !unigrams.isEmpty else { continue }
         insertNode(
           .init(keyArray: jointKeyArray, spanLength: theLength, unigrams: unigrams, keySeparator: separator),
           at: position
         )
-        nodesInserted += 1
+        nodesChanged += 1
       }
     }
-    return nodesInserted
+    return nodesChanged
   }
 
   mutating func updateCursorJumpingTables(_ walkedNodes: [Node]) {

Packages/vChewing_Megrez/Sources/Megrez/4_Span.swift

@@ -5,15 +5,15 @@
 
 extension Megrez.Compositor {
   /// 幅位乃指一組共享起點的節點。
-  public struct Span {
+  public class Span {
-    private var nodes: [Node?] = []
+    public var nodes: [Node?] = []
     public private(set) var maxLength = 0
     private var maxSpanLength: Int { Megrez.Compositor.maxSpanLength }
     public init() {
       clear()
     }
 
-    public mutating func clear() {
+    public func clear() {
       nodes.removeAll()
       for _ in 0..<maxSpanLength {
         nodes.append(nil)
@@ -24,7 +24,7 @@ extension Megrez.Compositor {
     /// 往該幅位塞入一個節點。
     /// - Parameter node: 要塞入的節點。
     /// - Returns: 該操作是否成功執行。
-    @discardableResult public mutating func append(node: Node) -> Bool {
+    @discardableResult public func append(node: Node) -> Bool {
       guard (1...maxSpanLength).contains(node.spanLength) else {
         return false
       }
@@ -36,7 +36,7 @@ extension Megrez.Compositor {
     /// 丟掉任何不小於給定幅位長度的節點。
     /// - Parameter length: 給定的幅位長度。
     /// - Returns: 該操作是否成功執行。
-    @discardableResult public mutating func dropNodesOfOrBeyond(length: Int) -> Bool {
+    @discardableResult public func dropNodesOfOrBeyond(length: Int) -> Bool {
       guard (1...maxSpanLength).contains(length) else {
         return false
       }
@@ -66,7 +66,7 @@ extension Megrez.Compositor {
   /// 找出所有與該位置重疊的節點。其返回值為一個節錨陣列（包含節點、以及其起始位置）。
   /// - Parameter location: 游標位置。
   /// - Returns: 一個包含所有與該位置重疊的節點的陣列。
-  func fetchOverlappingNodes(at location: Int) -> [NodeAnchor] {
+  internal func fetchOverlappingNodes(at location: Int) -> [NodeAnchor] {
     var results = [NodeAnchor]()
     guard !spans.isEmpty, location < spans.count else { return results }
 

Packages/vChewing_Megrez/Sources/Megrez/6_Node.swift

@@ -39,7 +39,7 @@ extension Megrez.Compositor {
     public private(set) var spanLength: Int
     public private(set) var unigrams: [Megrez.Unigram]
     public private(set) var currentUnigramIndex: Int = 0 {
-      didSet { currentUnigramIndex = min(max(0, currentUnigramIndex), unigrams.count - 1) }
+      didSet { currentUnigramIndex = max(min(unigrams.count - 1, currentUnigramIndex), 0) }
     }
 
     public var currentPair: Megrez.Compositor.KeyValuePaired { .init(key: key, value: value) }
@@ -53,6 +53,18 @@ extension Megrez.Compositor {
       hasher.combine(overrideType)
     }
 
+    /// 置換掉該節點內的單元圖陣列資料。
+    /// 如果此時影響到了 currentUnigramIndex 所指的內容的話，則將其重設為 0。
+    /// - Parameter source: 新的單元圖陣列資料，必須不能為空（否則必定崩潰）。
+    public func resetUnigrams(using source: [Megrez.Unigram]) {
+      let oldCurrentValue = unigrams[currentUnigramIndex].value
+      unigrams = source
+      // if unigrams.isEmpty { unigrams.append(.init(value: key, score: -114.514)) }  // 保險，請按需啟用。
+      currentUnigramIndex = max(min(unigrams.count - 1, currentUnigramIndex), 0)
+      let newCurrentValue = unigrams[currentUnigramIndex].value
+      if oldCurrentValue != newCurrentValue { currentUnigramIndex = 0 }
+    }
+
     public private(set) var overrideType: Node.OverrideType
 
     public static func == (lhs: Node, rhs: Node) -> Bool {

Packages/vChewing_Megrez/Tests/MegrezTests/LMDataForTests.swift

@@ -36,6 +36,13 @@ class SimpleLM: LangModelProtocol {
   func hasUnigramsFor(key: String) -> Bool {
     mutDatabase.keys.contains(key)
   }
+
+  func trim(key: String, value: String) {
+    guard var arr = mutDatabase[key] else { return }
+    arr = arr.compactMap { $0.value == value ? nil : $0 }
+    guard !arr.isEmpty else { return }
+    mutDatabase[key] = arr
+  }
 }
 
 class MockLM: LangModelProtocol {

Packages/vChewing_Megrez/Tests/MegrezTests/MegrezTests.swift

@@ -11,7 +11,7 @@ import XCTest
 final class MegrezTests: XCTestCase {
   func testSpan() throws {
     let langModel = SimpleLM(input: strSampleData)
-    var span = Megrez.Compositor.Span()
+    let span = Megrez.Compositor.Span()
     let n1 = Megrez.Compositor.Node(keyArray: ["gao1"], spanLength: 1, unigrams: langModel.unigramsFor(key: "gao1"))
     let n3 = Megrez.Compositor.Node(
       keyArray: ["gao1ke1ji4"], spanLength: 3, unigrams: langModel.unigramsFor(key: "gao1ke1ji4")
@@ -518,4 +518,21 @@ final class MegrezTests: XCTestCase {
     result = compositor.walk().0
     XCTAssertEqual(result.values, ["高熱", "🔥", "危險"])
   }
+
+  func testCompositor_updateUnigramData() throws {
+    let theLM = SimpleLM(input: strSampleData)
+    var compositor = Megrez.Compositor(with: theLM)
+    compositor.separator = ""
+    compositor.insertKey("nian2")
+    compositor.insertKey("zhong1")
+    compositor.insertKey("jiang3")
+    compositor.insertKey("jin1")
+    let oldResult = compositor.walk().0.values.joined()
+    print(oldResult)
+    theLM.trim(key: "nian2zhong1", value: "年中")
+    compositor.update(updateExisting: true)
+    let newResult = compositor.walk().0.values.joined()
+    print(newResult)
+    XCTAssertEqual([oldResult, newResult], ["年中獎金", "年終獎金"])
+  }
 }