Add repeated words filtering to chat filter.

WebContent/js/cah.constants.js

@@ -130,6 +130,7 @@ cah.$.ErrorCode.TOO_MANY_USERS = "tmu";
 cah.$.ErrorCode.NOT_GAME_HOST = "ngh";
 cah.$.ErrorCode.MESSAGE_TOO_LONG = "mtl";
 cah.$.ErrorCode.NO_NICK_SPECIFIED = "nns";
+cah.$.ErrorCode.REPEATED_WORDS = "rW";
 cah.$.ErrorCode.BANNED = "B&";
 cah.$.ErrorCode.WRONG_PASSWORD = "wp";
 cah.$.ErrorCode.RESERVED_NICK = "rn";
@@ -200,6 +201,7 @@ cah.$.ErrorCode_msg['cjag'] = "You cannot join another game.";
 cah.$.ErrorCode_msg['B&'] = "Banned.";
 cah.$.ErrorCode_msg['mtl'] = "Messages cannot be longer than 200 characters.";
 cah.$.ErrorCode_msg['in'] = "Nickname must contain only upper and lower case letters, numbers, or underscores, must be 3 to 30 characters long, and must not start with a number.";
+cah.$.ErrorCode_msg['rW'] = "You must use more unique words in your message.";
 cah.$.ErrorCode_msg['serr'] = "An error occured on the server.";
 cah.$.ErrorCode_msg['CL'] = "Try turning caps lock off.";
 cah.$.ErrorCode_msg['dnhc'] = "You don't have that card.";

build.properties.example

@@ -43,6 +43,10 @@ pyx.global.spaces_min_count=4
 pyx.global.flood_count=3
 # seconds
 pyx.global.flood_time=25
+# Messages with more than repeated_words_min_count words must have at least
+# repeated_words_unique_ratio unique words in them.
+pyx.global.repeated_words_min_count=10
+pyx.global.repeated_words_unique_ratio=.5
 
 # Settings for game chat protection. If it isn't listed here, it isn't supported.
 # same but for game chats

src/main/filtered-resources/WEB-INF/pyx.properties

@@ -16,6 +16,8 @@ pyx.chat.global.spaces_min_len=${pyx.global.spaces_min_len}
 pyx.chat.global.spaces_min_count=${pyx.global.spaces_min_count}
 pyx.chat.global.capslock_min_len=${pyx.global.capslock_min_len}
 pyx.chat.global.capslock_ratio=${pyx.global.capslock_ratio}
+pyx.chat.global.repeated_words_min_count=${pyx.global.repeated_words_min_count}
+pyx.chat.global.repeated_words_unique_ratio=${pyx.global.repeated_words_unique_ratio}
 pyx.chat.game.flood_count=${pyx.game.flood_count}
 pyx.chat.game.flood_time=${pyx.game.flood_time}
 pyx.build=${buildNumber}

src/main/java/net/socialgamer/cah/Constants.java

@@ -399,6 +399,7 @@ public class Constants {
     RESERVED_NICK("rn", "That nick is reserved."),
     REPEAT_MESSAGE("rm",
         "You can't repeat the same message multiple times in a row."),
+    REPEATED_WORDS("rW", "You must use more unique words in your message."),
     SERVER_ERROR("serr", "An error occured on the server."),
     SESSION_EXPIRED("se", "Your session has expired. Refresh the page."),
     TOO_FAST("tf", "You are chatting too fast. Wait a few seconds and try again."),

src/main/java/net/socialgamer/cah/handlers/ChatHandler.java

@@ -110,6 +110,8 @@ public class ChatHandler extends Handler {
           break;
         case REPEAT:
           return error(ErrorCode.REPEAT_MESSAGE);
+        case REPEAT_WORDS:
+          return error(ErrorCode.REPEATED_WORDS);
         case TOO_FAST:
           return error(ErrorCode.TOO_FAST);
         case TOO_LONG:

src/main/java/net/socialgamer/cah/handlers/GameChatHandler.java

@@ -99,6 +99,8 @@ public class GameChatHandler extends GameWithPlayerHandler {
           break;
         case REPEAT:
           return error(ErrorCode.REPEAT_MESSAGE);
+        case REPEAT_WORDS:
+          return error(ErrorCode.REPEATED_WORDS);
         case TOO_FAST:
           return error(ErrorCode.TOO_FAST);
         case TOO_LONG:

src/main/java/net/socialgamer/cah/util/ChatFilter.java

@@ -37,6 +37,7 @@ import java.util.regex.Pattern;
 
 import org.apache.log4j.Logger;
 
+import com.google.common.collect.ImmutableSet;
 import com.google.inject.Inject;
 import com.google.inject.Provider;
 import com.google.inject.Singleton;
@@ -60,6 +61,8 @@ public class ChatFilter {
   private static final int DEFAULT_SPACES_REQUIRED = 3;
   private static final int DEFAULT_CAPSLOCK_MIN_MSG_LENGTH = 50;
   private static final double DEFAULT_CAPSLOCK_RATIO = .5;
+  private static final int DEFAULT_REPEATED_WORDS_MIN_COUNT = 10;
+  private static final double DEFAULT_REPEATED_WORDS_UNIQUE_RATIO = .5;
   public static final String DEFAULT_SHADOWBAN_PROVIDER = DefaultShadowBannedStringsProvider.class
       .getCanonicalName();
 
@@ -70,7 +73,7 @@ public class ChatFilter {
   private final Map<User, FilterData> filterData = Collections.synchronizedMap(new WeakHashMap<>());
 
   public enum Result {
-    CAPSLOCK, DROP_MESSAGE, NO_MESSAGE, NOT_ENOUGH_SPACES, OK, REPEAT, TOO_FAST, TOO_LONG, TOO_MANY_SPECIALS
+    CAPSLOCK, DROP_MESSAGE, NO_MESSAGE, NOT_ENOUGH_SPACES, OK, REPEAT, REPEAT_WORDS, TOO_FAST, TOO_LONG, TOO_MANY_SPECIALS
   }
 
   private enum Scope {
@@ -102,12 +105,21 @@ public class ChatFilter {
       }
     }
 
-    final int spaces = message.split("\\s+").length + 1;
+    final String[] words = message.toLowerCase(Locale.ENGLISH).split("\\s+");
+    final int spaces = words.length + 1;
     if (total >= getIntParameter(Scope.global, "spaces_min_len", DEFAULT_SPACES_MIN_MSG_LENGTH)
         && spaces < getIntParameter(Scope.global, "spaces_min_count", DEFAULT_SPACES_REQUIRED)) {
       return Result.NOT_ENOUGH_SPACES;
     }
 
+    final Set<String> uniqueWords = ImmutableSet.copyOf(words);
+    if (words.length >= getIntParameter(Scope.global, "repeated_words_min_count",
+        DEFAULT_REPEATED_WORDS_MIN_COUNT)
+        && ((double) uniqueWords.size()) / words.length < getDoubleParameter(Scope.global,
+            "repeated_words_unique_ratio", DEFAULT_REPEATED_WORDS_UNIQUE_RATIO)) {
+      return Result.REPEAT_WORDS;
+    }
+
     final long caps = message.codePoints().filter(c -> Character.isUpperCase(c)).count();
     if (total >= getIntParameter(Scope.global, "capslock_min_len", DEFAULT_CAPSLOCK_MIN_MSG_LENGTH)
         && ((double) caps) / total > getDoubleParameter(Scope.global, "capslock_ratio",