From 7db65e6e2c268fa5cfa17ecbf5d2765b116bda54 Mon Sep 17 00:00:00 2001
From: ShikiSuen <shikisuen@outlook.com>
Date: Tue, 10 May 2022 14:23:24 +0800
Subject: [PATCH] Megrez // Weighten nodes with longer span (Megrez v1.0.7).

---
 .../LanguageParsers/Megrez/1_Walker.swift     | 31 ++++++++++++++++---
 .../LanguageParsers/Megrez/2_Grid.swift       | 20 ++++++------
 .../LanguageParsers/Megrez/3_NodeAnchor.swift |  6 ++++
 .../LanguageParsers/Megrez/4_Node.swift       |  4 ++-
 4 files changed, 44 insertions(+), 17 deletions(-)

diff --git a/Source/Modules/LanguageParsers/Megrez/1_Walker.swift b/Source/Modules/LanguageParsers/Megrez/1_Walker.swift
index 7c2a5051..429d78b5 100644
--- a/Source/Modules/LanguageParsers/Megrez/1_Walker.swift
+++ b/Source/Modules/LanguageParsers/Megrez/1_Walker.swift
@@ -31,21 +31,37 @@ extension Megrez {
       mutGrid = grid
     }
 
-    public func reverseWalk(at location: Int, score accumulatedScore: Double = 0.0) -> [NodeAnchor] {
+    public func reverseWalk(at location: Int, score accumulatedScore: Double = 0.0, nodesLimit: Int = 0)
+      -> [NodeAnchor]
+    {
       if location == 0 || location > mutGrid.width() {
         return [] as [NodeAnchor]
       }
 
       var paths: [[NodeAnchor]] = []
-      let nodes: [NodeAnchor] = mutGrid.nodesEndingAt(location: location)
+      var nodes: [NodeAnchor] = mutGrid.nodesEndingAt(location: location)
 
-      for n in nodes {
+      nodes.sort {
+        $0.balancedScore > $1.balancedScore  // 排序規則已經在 NodeAnchor 內定義了。
+      }
+
+      // 只檢查前 X 個 NodeAnchor 是否有 node。
+      // 這裡有 abs 是為了防止有白癡填負數。
+      var border: Int = nodes.count
+      if nodesLimit > 0 {
+        border = min(nodes.count, abs(nodesLimit))
+      }
+
+      for n in nodes[0..<border] {
         var n = n
-        if n.node == nil {
+        guard let nNode = n.node else {
           continue
         }
 
-        n.accumulatedScore = accumulatedScore + n.node!.score()
+        // 利用 Spanning Length 來決定權重。
+        // 這樣一來，例：「再見」比「在」與「見」的權重更高。
+        let weightedScore: Double = (Double(n.spanningLength) - 1) * 2
+        n.accumulatedScore = accumulatedScore + nNode.score() + weightedScore
 
         var path: [NodeAnchor] = reverseWalk(
           at: location - n.spanningLength,
@@ -54,6 +70,11 @@ extension Megrez {
         path.insert(n, at: 0)
 
         paths.append(path)
+
+        // 始終使用固定的候選字
+        if nNode.score() >= 0 {
+          break
+        }
       }
 
       if !paths.isEmpty {
diff --git a/Source/Modules/LanguageParsers/Megrez/2_Grid.swift b/Source/Modules/LanguageParsers/Megrez/2_Grid.swift
index c391b425..355c6bf8 100644
--- a/Source/Modules/LanguageParsers/Megrez/2_Grid.swift
+++ b/Source/Modules/LanguageParsers/Megrez/2_Grid.swift
@@ -140,16 +140,15 @@ extension Megrez {
 
     public func fixNodeSelectedCandidate(location: Int, value: String) -> NodeAnchor {
       var node = NodeAnchor()
-      let nodes = nodesCrossingOrEndingAt(location: location)
-      for nodeAnchor in nodes {
+      for (index, nodeAnchor) in nodesCrossingOrEndingAt(location: location).enumerated() {
         // Reset the candidate-fixed state of every node at the location.
         let candidates = nodeAnchor.node?.candidates() ?? []
-        nodeAnchor.node?.resetCandidate()
+        nodesCrossingOrEndingAt(location: location)[index].node?.resetCandidate()
 
         for (i, candidate) in candidates.enumerated() {
           if candidate.value == value {
-            nodeAnchor.node?.selectCandidateAt(index: i)
-            node = nodeAnchor
+            nodesCrossingOrEndingAt(location: location)[index].node?.selectCandidateAt(index: i)
+            node = nodesCrossingOrEndingAt(location: location)[index]
             break
           }
         }
@@ -158,18 +157,17 @@ extension Megrez {
     }
 
     public func overrideNodeScoreForSelectedCandidate(location: Int, value: String, overridingScore: Double) {
-      for nodeAnchor in nodesCrossingOrEndingAt(location: location) {
-        var nodeAnchor = nodeAnchor
+      for (index, nodeAnchor) in nodesCrossingOrEndingAt(location: location).enumerated() {
         if let theNode = nodeAnchor.node {
           let candidates = theNode.candidates()
           // Reset the candidate-fixed state of every node at the location.
-          theNode.resetCandidate()
-          nodeAnchor.node = theNode
+          nodesCrossingOrEndingAt(location: location)[index].node?.resetCandidate()
 
           for (i, candidate) in candidates.enumerated() {
             if candidate.value == value {
-              theNode.selectFloatingCandidateAt(index: i, score: overridingScore)
-              nodeAnchor.node = theNode
+              nodesCrossingOrEndingAt(location: location)[index].node?.selectFloatingCandidateAt(
+                index: i, score: overridingScore
+              )
               break
             }
           }
diff --git a/Source/Modules/LanguageParsers/Megrez/3_NodeAnchor.swift b/Source/Modules/LanguageParsers/Megrez/3_NodeAnchor.swift
index 0a130a58..938262f4 100644
--- a/Source/Modules/LanguageParsers/Megrez/3_NodeAnchor.swift
+++ b/Source/Modules/LanguageParsers/Megrez/3_NodeAnchor.swift
@@ -32,5 +32,11 @@ extension Megrez {
     public var keyLength: Int {
       node?.key().count ?? 0
     }
+
+    public var balancedScore: Double {
+      let weightedScore: Double = (Double(spanningLength) - 1) * 2
+      let nodeScore: Double = node?.score() ?? 0
+      return weightedScore + nodeScore
+    }
   }
 }
diff --git a/Source/Modules/LanguageParsers/Megrez/4_Node.swift b/Source/Modules/LanguageParsers/Megrez/4_Node.swift
index fdf0838d..be518b3e 100644
--- a/Source/Modules/LanguageParsers/Megrez/4_Node.swift
+++ b/Source/Modules/LanguageParsers/Megrez/4_Node.swift
@@ -36,6 +36,8 @@ extension Megrez {
     var mutCandidateFixed: Bool = false
     var mutSelectedUnigramIndex: Int = 0
 
+    let kSelectedCandidateScore: Double = 99
+
     public init(key: String, unigrams: [Megrez.Unigram], bigrams: [Megrez.Bigram] = []) {
       mutLM = LanguageModel()
 
@@ -112,7 +114,7 @@ extension Megrez {
     public func selectCandidateAt(index: Int = 0, fix: Bool = false) {
       mutSelectedUnigramIndex = index >= mutUnigrams.count ? 0 : index
       mutCandidateFixed = fix
-      mutScore = 99
+      mutScore = kSelectedCandidateScore
     }
 
     public func resetCandidate() {