add badword filter version 1

2026-03-23 10:46:10 +07:00
parent a91faac68f
commit f87dcbd7f8
10 changed files with 1671 additions and 2 deletions
@@ -0,0 +1,7 @@
+fileFormatVersion: 2
+guid: 6ab087813f7c53349baf2a476e6904f1
+TextScriptImporter:
+  externalObjects: {}
+  userData: 
+  assetBundleName: 
+  assetBundleVariant: 
@@ -1,5 +1,6 @@
 using UnityEngine;
 using System.Collections;
+using BrewMonster.PerfectWorld.Scripts.Utility.ChatFilter;
 using UnityEngine.SceneManagement;

 namespace BrewMonster
@@ -10,6 +11,7 @@ namespace BrewMonster

        IEnumerator Start()
        {
+            ChatFilterService.Init();
            // Load the next scene after 1 second
            yield return new WaitForSeconds(1f);
            LoadNextScene();
@@ -20,8 +22,8 @@ namespace BrewMonster
 #if TESTFAST
            SceneManager.LoadSceneAsync(_nextSceneName,LoadSceneMode.Additive);
 #else
-            SceneManager.LoadScene(_nextSceneName); 
+            SceneManager.LoadScene(_nextSceneName);
 #endif
        }
    }
-}
+}
@@ -0,0 +1,8 @@
+fileFormatVersion: 2
+guid: 89107298c9450ab4fae2d12e2426713e
+folderAsset: yes
+DefaultImporter:
+  externalObjects: {}
+  userData: 
+  assetBundleName: 
+  assetBundleVariant: 
@@ -0,0 +1,183 @@
+using System;
+using System.Collections.Generic;
+using System.Globalization;
+using System.Text;
+
+namespace BrewMonster.PerfectWorld.Scripts.Utility.ChatFilter
+{
+    public class ChatFilter
+    {
+        private HashSet<string> badWordSet = new HashSet<string>();
+
+        // =========================
+        // BUILD
+        // =========================
+        public void Build(string[] words)
+        {
+            badWordSet.Clear();
+
+            foreach (var w in words)
+            {
+                if (!string.IsNullOrWhiteSpace(w))
+                    badWordSet.Add(w.Trim());
+            }
+        }
+
+        // =========================
+        // NORMALIZE
+        // =========================
+        private string NormalizeRuntime(string input, out List<int> map)
+        {
+            map = new List<int>();
+
+            string formD = input.Normalize(NormalizationForm.FormD);
+            StringBuilder sb = new StringBuilder();
+
+            bool lastWasSpace = false;
+
+            for (int i = 0; i < formD.Length; i++)
+            {
+                char c = formD[i];
+
+                var uc = Char.GetUnicodeCategory(c);
+                if (uc == UnicodeCategory.NonSpacingMark)
+                    continue;
+
+                char n = NormalizeChar(c);
+
+                if (n == '\0')
+                {
+                    if (!lastWasSpace)
+                    {
+                        sb.Append(' ');
+                        map.Add(i);
+                        lastWasSpace = true;
+                    }
+                }
+                else
+                {
+                    sb.Append(n);
+                    map.Add(i);
+                    lastWasSpace = false;
+                }
+            }
+
+            return sb.ToString().Trim();
+        }
+
+        private char NormalizeChar(char c)
+        {
+            c = char.ToLower(c);
+
+            if (c == 'đ') return 'd';
+
+            switch (c)
+            {
+                case '@': return 'a';
+                case '4': return 'a';
+                case '0': return 'o';
+                case '1': return 'i';
+                case '!': return 'i';
+                case '$': return 's';
+                case '3': return 'e';
+            }
+
+            if (char.IsLetterOrDigit(c))
+                return c;
+
+            return '\0';
+        }
+
+        // =========================
+        // TOKEN → CHAR INDEX
+        // =========================
+        private int FindCharIndex(string normalized, int tokenIndex, List<int> map)
+        {
+            int tokenCounter = 0;
+
+            for (int i = 0; i < normalized.Length; i++)
+            {
+                if (normalized[i] == ' ')
+                    continue;
+
+                if (tokenCounter == tokenIndex)
+                    return map[i];
+
+                // skip current token
+                while (i < normalized.Length && normalized[i] != ' ')
+                    i++;
+
+                tokenCounter++;
+            }
+
+            return -1;
+        }
+
+        // =========================
+        // MATCH JOINED TOKENS
+        // =========================
+        private bool MatchJoinedTokens(string[] tokens, int index, out int length)
+        {
+            length = 0;
+
+            // thử ghép tối đa 3 token
+            for (int size = 1; size <= 3 && index + size <= tokens.Length; size++)
+            {
+                string merged = "";
+
+                for (int j = 0; j < size; j++)
+                    merged += tokens[index + j];
+
+                if (badWordSet.Contains(merged))
+                {
+                    length = size;
+                    return true;
+                }
+            }
+
+            return false;
+        }
+
+        // =========================
+        // FILTER
+        // =========================
+        public string Filter(string input, out bool isValidWord)
+        {
+            isValidWord = false;
+            if (string.IsNullOrEmpty(input))
+                return input;
+            isValidWord = true;
+            List<int> map;
+            string normalized = NormalizeRuntime(input, out map);
+
+            var tokens = normalized.Split(' ', StringSplitOptions.RemoveEmptyEntries);
+
+            char[] result = input.ToCharArray();
+
+            for (int i = 0; i < tokens.Length; i++)
+            {
+                if (MatchJoinedTokens(tokens, i, out int len))
+                {
+                    int startToken = i;
+                    int endToken = i + len - 1;
+
+                    int startChar = FindCharIndex(normalized, startToken, map);
+                    int endChar = FindCharIndex(normalized, endToken, map);
+
+                    if (startChar >= 0 && endChar >= 0)
+                    {
+                        for (int k = startChar; k <= endChar && k < result.Length; k++)
+                        {
+                            result[k] = '*';
+                            isValidWord = false;
+                        }
+                    }
+
+                    i += len - 1;
+                }
+            }
+
+            return new string(result);
+        }
+    }
+}
@@ -0,0 +1,2 @@
+fileFormatVersion: 2
+guid: 436ad91ec808a7d4d80d00ab164d42b7
@@ -0,0 +1,68 @@
+using System.IO;
+using UnityEngine;
+
+namespace BrewMonster.PerfectWorld.Scripts.Utility.ChatFilter
+{
+    public static class ChatFilterService
+    {
+        private static ChatFilter _filter;
+        private static bool _initialized = false;
+
+        // =========================
+        // INIT
+        // =========================
+        public static void Init()
+        {
+            if (_initialized) return;
+
+            _filter = new ChatFilter();
+
+            string path = Path.Combine(Application.dataPath, "PerfectWorld/Data/clean_words.txt");
+
+            if (!File.Exists(path))
+            {
+                BMLogger.LogError("ChatFilter file not found: " + path);
+                return;
+            }
+
+            var lines = File.ReadAllLines(path);
+
+            _filter.Build(lines);
+
+            _initialized = true;
+
+            Debug.Log("ChatFilterV2 initialized");
+        }
+
+        // =========================
+        // FILTER
+        // =========================
+        public static string Filter(string input, out bool isValidWord)
+        {
+            isValidWord =false;
+            if (!_initialized)
+            {
+                BMLogger.LogError("ChatFilter not initialized!");
+                return input;
+            }
+
+            return _filter.Filter(input, out isValidWord);
+        }
+
+        // =========================
+        // CHECK ONLY (không mask)
+        // =========================
+        public static bool ContainsBadWord(string input)
+        {
+            if (!_initialized)
+            {
+                BMLogger.LogError("ChatFilter not initialized!");
+                return false;
+            }
+
+            _filter.Filter(input, out var isValidWord);
+
+            return !isValidWord;
+        }
+    }
+}
@@ -0,0 +1,2 @@
+fileFormatVersion: 2
+guid: b09bed44157324f46b0f980aa4da0720
@@ -0,0 +1,21 @@
+using BrewMonster.PerfectWorld.Scripts.Utility.ChatFilter;
+using UnityEngine;
+
+namespace BrewMonster
+{
+    public class TestFilter : MonoBehaviour
+    {
+        public string input;
+        [ContextMenu("Filter")]
+        void Test()
+        {
+            ChatFilterService.Init();
+
+            // string msg = "d*m*m mày";
+
+            string filtered = ChatFilterService.Filter(input, out var isValidWord);
+
+            Debug.Log($"Is valid: {isValidWord}: "+filtered);
+        }
+    }
+}
@@ -0,0 +1,2 @@
+fileFormatVersion: 2
+guid: c235df647ffffd54a84fdae3527eedff