Add preferences for included character types

kashparty · Jul 26, 2021 · eed8b10 · eed8b10
1 parent db9323e
commit eed8b10
Show file tree

Hide file tree

Showing 3 changed files with 39 additions and 4 deletions.
diff --git a/Preferences.cs b/Preferences.cs
@@ -14,6 +14,10 @@ class Preferences {
         public double learningRate { get; set; }
         public double discount { get; set; }
         public ConsoleColor color { get; set; }
+        public bool includeUppercase { get; set; }
+        public bool includeLowercase { get; set; }
+        public bool includeDigits { get; set; }
+        public bool includePunctuation { get; set; }
 
         public Preferences() {
             // Use the default preferences.
@@ -25,7 +29,12 @@ public Preferences() {
             numRounds = 10;
             learningRate = 0.5;
             discount = 0.9;
+
             color = ConsoleColor.Green;
+            includeUppercase = true;
+            includeLowercase = true;
+            includeDigits = false;
+            includePunctuation = false;
         }
 
         public Preferences(string fileName) : this() {
@@ -77,6 +86,22 @@ public Preferences(string fileName) : this() {
                         Console.ResetColor();
                     }
                 }
+
+                if (data.ContainsKey("includeUppercase")) {
+                    includeUppercase = data["includeUppercase"].GetBoolean();
+                }
+
+                if (data.ContainsKey("includeLowercase")) {
+                    includeLowercase = data["includeLowercase"].GetBoolean();
+                }
+
+                if (data.ContainsKey("includeDigits")) {
+                    includeDigits = data["includeDigits"].GetBoolean();
+                }
+
+                if (data.ContainsKey("includePunctuation")) {
+                    includePunctuation = data["includePunctuation"].GetBoolean();
+                }
             }
         }
     }

diff --git a/Program.cs b/Program.cs
@@ -85,7 +85,13 @@ private static string PreProcess() {
                 trainingData = streamReader.ReadToEnd();
             }
 
-            trainingData = Regex.Replace(trainingData, "[^a-zA-Z ]", "");
+            string regexString = "[^";
+            if (preferences.includeLowercase) regexString += "a-z";
+            if (preferences.includeUppercase) regexString += "A-Z";
+            if (preferences.includeDigits) regexString += "0-9";
+            if (preferences.includePunctuation) regexString += ".,!:;\"\'-()";
+            regexString += " ]";
+            trainingData = Regex.Replace(trainingData, regexString, "");
             return trainingData;
         }
 

diff --git a/QLearn.cs b/QLearn.cs
@@ -20,7 +20,7 @@ class QLearn {
         double discount;
 
         static Random random = new Random();
-        
+
         Dictionary<char, int> charToInt;
         Dictionary<int, char> intToChar;
 
@@ -53,7 +53,9 @@ public QLearn(string trainingData, double explorationLow, double explorationHigh
             for (int i = 0; i < trainingData.Length; i++) {
                 if (i >= 3) current = current.Substring(1);
                 if (current.Length == 2) {
-                    occurrences[charToInt[current[0]], charToInt[current[1]], charToInt[trainingData[i]]]++;
+                    if (charToInt.ContainsKey(current[0]) && charToInt.ContainsKey(current[1]) && charToInt.ContainsKey(trainingData[i])) {
+                        occurrences[charToInt[current[0]], charToInt[current[1]], charToInt[trainingData[i]]]++;
+                    }
                 }
                 current += trainingData[i];
             }
@@ -72,6 +74,8 @@ private void LoadModelIfExists() {
                         string nGram = line.Substring(0, 3);
                         double qValue = double.Parse(line.Substring(4));
 
+                        if (!charToInt.ContainsKey(nGram[0]) || !charToInt.ContainsKey(nGram[1]) || !charToInt.ContainsKey(nGram[2])) continue;
+
                         int charIndex1 = charToInt[nGram[0]];
                         int charIndex2 = charToInt[nGram[1]];
                         int charIndex3 = charToInt[nGram[2]];
@@ -181,7 +185,7 @@ public void FinishRound() {
             currentRound++;
 
             if (currentRound >= numRounds) currentRound = 0;
-            
+
             double roundCompletion = (double)currentRound / (double)numRounds;
             explorationRate = roundCompletion * explorationLow + (1 - roundCompletion) * explorationHigh;