Add extra security procedures to insertKey(). (#90)

- This also turns Span into struct.
vChewing · Oct 24, 2022 · 8a31f49 · 8a31f49
1 parent 9bc7ef9
commit 8a31f49
Show file tree

Hide file tree

Showing 3 changed files with 20 additions and 9 deletions.
diff --git a/Sources/Megrez/1_Compositor.swift b/Sources/Megrez/1_Compositor.swift
@@ -73,8 +73,14 @@ extension Megrez {
     @discardableResult public mutating func insertKey(_ key: String) -> Bool {
       guard !key.isEmpty, key != separator, langModel.hasUnigramsFor(key: key) else { return false }
       keys.insert(key, at: cursor)
+      let gridBackup = spans
       resizeGrid(at: cursor, do: .expand)
-      update()
+      let nodesInserted = update()
+      // 用來在 langModel.hasUnigramsFor() 結果不準確的時候防呆、恢復被搞壞的 spans。
+      if nodesInserted == 0 {
+        spans = gridBackup
+        return false
+      }
       cursor += 1  // 游標必須得在執行 update() 之後才可以變動。
       return true
     }
@@ -219,7 +225,7 @@ extension Megrez.Compositor {
   ///            (XXXXXXX? <-被砍爛的節點
   /// ```
   /// - Parameter location: 給定的幅位座標。
-  func dropWreckedNodes(at location: Int) {
+  mutating func dropWreckedNodes(at location: Int) {
     let location = max(min(location, spans.count), 0)  // 防呆
     guard !spans.isEmpty else { return }
     let affectedLength = Megrez.Compositor.maxSpanLength - 1
@@ -230,7 +236,7 @@ extension Megrez.Compositor {
     }
   }
 
-  @discardableResult func insertNode(_ node: Node, at location: Int) -> Bool {
+  @discardableResult mutating func insertNode(_ node: Node, at location: Int) -> Bool {
     let location = max(min(location, spans.count - 1), 0)  // 防呆
     spans[location].append(node: node)
     return true
@@ -254,9 +260,12 @@ extension Megrez.Compositor {
     return key == node.key
   }
 
-  func update() {
+  /// 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。
+  @discardableResult mutating func update() -> Int {
     let maxSpanLength = Megrez.Compositor.maxSpanLength
     let range = max(0, cursor - maxSpanLength)..<min(cursor + maxSpanLength, keys.count)
+    var nodesInserted = 0
     for position in range {
       for theLength in 1...min(maxSpanLength, range.upperBound - position) {
         let jointKeyArray = getJointKeyArray(range: position..<(position + theLength))
@@ -268,8 +277,10 @@ extension Megrez.Compositor {
           .init(keyArray: jointKeyArray, spanLength: theLength, unigrams: unigrams, keySeparator: separator),
           at: position
         )
+        nodesInserted += 1
       }
     }
+    return nodesInserted
   }
 
   mutating func updateCursorJumpingTables(_ walkedNodes: [Node]) {

diff --git a/Sources/Megrez/4_Span.swift b/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 class Span {
+  public struct Span {
     private var nodes: [Node?] = []
     public private(set) var maxLength = 0
     private var maxSpanLength: Int { Megrez.Compositor.maxSpanLength }
     public init() {
       clear()
     }
 
-    public func clear() {
+    public mutating 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 func append(node: Node) -> Bool {
+    @discardableResult public mutating 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 func dropNodesOfOrBeyond(length: Int) -> Bool {
+    @discardableResult public mutating func dropNodesOfOrBeyond(length: Int) -> Bool {
       guard (1...maxSpanLength).contains(length) else {
         return false
       }

diff --git a/Tests/MegrezTests/MegrezTests.swift b/Tests/MegrezTests/MegrezTests.swift
@@ -11,7 +11,7 @@ import XCTest
 final class MegrezTests: XCTestCase {
   func testSpan() throws {
     let langModel = SimpleLM(input: strSampleData)
-    let span = Megrez.Compositor.Span()
+    var 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")