working version

KgK _alr regenerator/Program.cs

@@ -1,17 +1,11 @@
-using System;
-using System.Drawing;
-using System.IO;
-using System.Text;
-using static Program;
-using static System.Net.Mime.MediaTypeNames;
-using static System.Runtime.InteropServices.JavaScript.JSType;
+using System.Text;
 
 class Program
 {
     static void Main()
     {
         //TEST REMOVE LATER
-        string[] args = new string[] { "--all", "C:\\Users\\Lauren\\Desktop\\Ken Ga Kimi\\TEST", "C:\\Users\\Lauren\\Desktop\\Ken Ga Kimi\\TEST\\OUTPUT" };
+        string[] args = new string[] { "--all", "C:\\Users\\Lauren\\Desktop\\Ken Ga Kimi\\PATCH\\EN Files", "C:\\Users\\Lauren\\Desktop\\Ken Ga Kimi\\PATCH\\EN Files\\OUTPUT" };
         
         if (args.Length < 1)
         {
@@ -187,7 +181,7 @@ class Program
         offset = RoundUpToNextMultipleOf4(offset);
         int size = BitConverter.ToInt32(data, offset);
         offset += 4;
-        if (data[offset+1] == 255 && data[offset + 2] == 254) {
+        if (data[offset] == 255 && data[offset + 1] == 254) {
             offset += 2;
         }
         //offset = RoundUpToNextMultipleOf4(offset);
@@ -233,84 +227,86 @@ class Program
             }
         }
         List<string> searchStrings = new List<string>() {
-        "\n＠",
-        "\n<voice name=",
-        "\n<textboxtype type=\"monologue",
-        "\n<textboxtype type=\"dialogue\">",
-        "\n<monologue background=\"on\">",
+        "＠",
+        "<voice name=",
+        "<textboxtype type=\"monologue\">",
+        "<textboxtype type=\"monologue2\">",
+        "<textboxtype type=\"monologue3\">",
+        "<monologue background=\"on\">",
         };
-        List<(string, int)> stringPositions = FindStringPositions(script.ScriptText, searchStrings);
+        List<(string, int)> stringPositions = FindStringPositions(script.ScriptText);
+        List<(string, int)> filteredStringPositions = stringPositions
+            .Where(pos => searchStrings.Any(s => pos.Item1.Contains(s)))
+            .ToList();
         //Add length of data
-        newAlr.AddRange(BitConverter.GetBytes(stringPositions.Count * 8));
+        newAlr.AddRange(BitConverter.GetBytes(filteredStringPositions.Count * 8));
 
+        if (alr.ScriptTitle == "KEN_00_00_00_alr")
+        {
+            Console.WriteLine("Found KEN_00_00_00");
+        }
         //position += 1;
         for (int i = 0; i < stringPositions.Count; i++)
+        {
+            if (searchStrings.Any(s => stringPositions[i].Item1.Contains(s)))
             {
                 newAlr.AddRange(BitConverter.GetBytes(stringPositions[i].Item2));
                 newAlr.AddRange(BitConverter.GetBytes(position));
-            if (stringPositions[i].Item1 != "\n＠")
+                var isNormalTag = searchStrings.Any(t => stringPositions[i].Item1.Contains(t));
+                if (stringPositions[i].Item1.Contains("<voice name="))
+                {
+                    continue;
+                }
+                else if (isNormalTag)
                 {
                     position += 1;
                 }
-            else if (stringPositions[i].Item1 == "\n＠" && i > 0 && stringPositions[i - 1].Item1 != "\n＠")
+                else
                 {
                     //Don't add to position if the previous string was a tag
+                    //Console.WriteLine("OOPS");
                 }
-            else if (stringPositions.Count == 1)
-            {
-                position += 1;
             }
-
         }
         return newAlr.ToArray();
     }
 
-    public static List<(string SearchString, int Position)> FindStringPositions(byte[] scriptBytes, IEnumerable<string> searchStrings)
+    public static List<(string SearchString, int Position)> FindStringPositions(byte[] scriptBytes)
     {
         var results = new List<(string, int)>();
-        // Decode the script bytes to string for line analysis
         string scriptText = Encoding.Unicode.GetString(scriptBytes);
 
-        foreach (var search in searchStrings)
-        {
-            byte[] searchBytes = Encoding.Unicode.GetBytes(search);
-            int index = 0;
-            while (index <= scriptBytes.Length - searchBytes.Length)
+        // 1. Find all <...> tags not commented out
+        var tagRegex = new System.Text.RegularExpressions.Regex(@"<[^>\r\n]+>");
+        foreach (System.Text.RegularExpressions.Match match in tagRegex.Matches(scriptText))
         {
-                bool match = true;
-                for (int j = 0; j < searchBytes.Length; j++)
-                {
-                    if (scriptBytes[index + j] != searchBytes[j])
-                    {
-                        match = false;
-                        break;
-                    }
-                }
-                if (match)
-                {
-                    // Find the character index in the string corresponding to this byte index
-                    int charIndex = Encoding.Unicode.GetString(scriptBytes, 0, index).Length;
-                    // Find the start of the line
-                    int lineStart = scriptText.LastIndexOf('\n', charIndex - 1);
+            if (match.Value == "<pb>") continue; // Skip <pb> tags
+
+            int lineStart = scriptText.LastIndexOf('\n', match.Index);
             if (lineStart == -1) lineStart = 0; else lineStart++;
-                    int lineEnd = scriptText.IndexOf('\n', charIndex);
+            int lineEnd = scriptText.IndexOf('\n', match.Index);
             if (lineEnd == -1) lineEnd = scriptText.Length;
             string line = scriptText.Substring(lineStart, lineEnd - lineStart).TrimStart();
 
-                    // Only add if the line does NOT start with //
             if (!line.StartsWith("//"))
             {
-                        results.Add((search, index));
+                int byteIndex = Encoding.Unicode.GetByteCount(scriptText.Substring(0, match.Index));
+                results.Add((match.Value, byteIndex));
             }
-                    index += searchBytes.Length; // Move past this match
         }
-                else
+
+        // 2. Find all lines that start with ＠ and are not commented out
+        int currentIndex = 0;
+        foreach (var line in scriptText.Split('\n'))
         {
-                    index++;
-                }
+            string trimmedLine = line.TrimStart();
+            if (!trimmedLine.StartsWith("//") && trimmedLine.StartsWith("＠"))
+            {
+                int byteIndex = Encoding.Unicode.GetByteCount(scriptText.Substring(0, currentIndex));
+                results.Add((line, byteIndex));
             }
+            currentIndex += line.Length + 1; // +1 for '\n'
         }
-        // Sort by position in order of appearance
         return results.OrderBy(r => r.Item2).ToList();
     }