Fix the censoring in global

[Brenn]

Support the cause to improve global chat! Join the group: http://forum.supercell.net/group.php?groupid=2602

So... In my time playing this wonderful game, I've noticed that the anti-swear system in global seems to be downright useless. It seems to fairly often censor words that aren't even intentionally put there (like between words...). The other problem is that people can easily get around it by using accented letters (I'm pretty sure everyone who has visited global knows what I'm talking about). Thus, I came up with a way to fix it.

How it works:
The existing list of swear words is separated into 2 lists: One of "light" words, and another with the typical swear words. The "light" list would contain all of those obscure words that end up just getting in the way. The normal one would have all of the "normal" swear words. The two lists would be censored in different ways:

The "light" list only censors words if they are in the same word: spaces or punctuation will prevent censoring. (As I said earlier, these words are the ones that just make it harder to write normal messages.

The other list is like the existing system, but more heavily enforced: It will censor stuff across punctuation marks.
To put it more simply:
Light censor is for obscure/ more tame words that more often than not impede conversation.
Heavy censor is for the truly "bad" words-- f-word, for example.

Both systems ignore accents, which prevent a giant loophole which is currently in the chat system.

For example, if "AAAA" was a light swear word, AAAA or AAÀA would be censored, but AA AA or AA.AA wouldn't.
If it was a (well, I guess "heavy?") swear word, AAAA, AA AA, AAÀA, AA.....AA, and AA$&,]AA would be censored.

Before anyone says how hard it would be to implement, it isn't. I made a (as far as I can tell) working version of it in about 75 lines of Java code. (Just the base light-heavy-isolate functionality) (Java because that's what SC uses for server stuff).

The one issue is that (or at least the version I made) may behave erratically in other languages, as I coded it to recognize only English characters as actual letters. But hey--it's a start.


Edit: According to popular demand, I'm also requesting that many of the obscure words be removed from the list entirely, and most others be put on the "light censor" list

[Brenn]

So... I wrote an actual program to do the kind of censoring I was envisioning, just to prove that it is possible, and feasible, to implement this new kind of censoring for global chat. I wrote it in Java, because I'm pretty sure that the SC servers are written with Java, and I figured chat stuff was a server thing.

The things I tested:
-light censoring as normal word
-LC with space in middle
-LC with punctuation in middle
-HC with accent
-HC as normal
-HC with space
-HC with accent & space
-HC with periods in middle
-HC with random non-letters in middle
-HC spread apart (last thing in string, if you want to see what I mean)
All worked as intended.


So without further ado, the 75 lines of source code. It can probably be improved, but I'll leave that to SC if they decide to use it:


import java.text.Normalizer;
public class Code {

public static String[] lightCensor = {"cde"};
public static String[] heavyCensor = {"auu", "iiv"};

//For variants: First in row is intended character, following ones are the variants
//To add new ones (people start using a new variant): Continue the pattern shown
public static String[][] stringVariants = {
{"a", "/-\\", "/=\\"},
{"u", "|_|", "\\_/", "\\./"}
};
public static char[][] characterVariants = {
{'i', 'l', '1', '!'}
};


public static void main(String[] args) {

String s = "/-\\ |_| |_| l1v";
s = censor(s);
System.out.println(s);
}

public static String replaceStringVariants(String message){
for (int i = 0; i < stringVariants.length; i++){
String replaceWith = stringVariants[i][0];
for(int j = 1; j < stringVariants[i].length; j++){
message = message.replace(stringVariants[i][j], replaceWith);
}
}
return message;
}

public static boolean testIfVariant(char toTest, char testAgainst){
testAgainst = Character.toLowerCase(testAgainst);
if (toTest == testAgainst) return true;

for (int i = 0; i < characterVariants.length; i++){
if (characterVariants[i][0] == toTest) {

for (int j = 1; j < characterVariants[i].length; j++){

if (testAgainst == characterVariants[i][j]){
return true;

}
}
}
}
return false;
}

public static String censor(String message){
message = replaceStringVariants(message);
//Remove all the variant letters that people use to get around the current system
String t = Normalizer.normalize(message.toLowerCase(), Normalizer.Form.NFD);
t = t.replaceAll("\\p{InCombiningDiacriticalMarks}+", "").toLowerCase();

//The light censor list contains the more obscure/less swear-y words: the ones that frequently can mess up normal messages
for (String s : lightCensor){//Words in the light censor list are only censored if it appears in 1 word, not across multiple,
String replaceWith = "";//which lowers the chances of normal messages getting bleeped out.
for (int i = 0; i < s.length(); i++){
replaceWith += "*";
}
while(t.contains(s)){//The while loop replaces the words with the asterisks.
int index = t.indexOf(s);
t = t.substring(0, index) + replaceWith + t.substring(index + replaceWith.length());
message = message.substring(0, index) + replaceWith + message.substring(index + replaceWith.length());
}
}

//The heavy censor list contains the words you don't want showing up in any way (I think you know what I mean...)
//It will bleep these words out across spaces, and even non-letters, making it much harder to swear in global chat
//Note that the use of variant letters (common way to avoid, currently) have already been fixed, so that this will work.
for (String s : heavyCensor){
String censored = isolate(t, s);
while(!censored.equals("")){

String replace = "";
for (int i = 0; i < censored.length(); i++)
replace += "*";

int index = t.indexOf(censored);
t = t.substring(0, index) + replace + t.substring(index + replace.length());
message = message.substring(0, index) + replace + message.substring(index + replace.length());
censored = isolate(t, s);
}
}
return message;
}

public static String isolate(String message, String word){
int wordIndex = 0;
String toReturn = "";
if (message.contains(word))
return word;

for (int i = 0; i < message.length() && wordIndex < word.length(); i++){
char testing = word.charAt(wordIndex);
char current = message.charAt(i);//Get the two letters
if (testIfVariant(testing, current)){
wordIndex++;
toReturn += current;//If the letters match, great! Add them!
}
else if (wordIndex > 0 && testIfVariant(word.charAt(wordIndex - 1), current)){
toReturn += current;
}
else if ((current < 'a' || current > 'z') && (current < 'A' || current > 'Z') && !Character.isDigit(current)){
if (!toReturn.equals(""))//If it is blank (hasn't found START of word yet), don't add anything
toReturn += current;
}
else{
wordIndex = 0;//If it is a letter not in the word, reset
toReturn = "";
}
}
return toReturn;
}
}

[Brenn]

(Reserved)

[AxionXD]

Yea I noticed some weird words being censored, I found the weirdest one yet:

g 600k

I swear, type that into Global, it'll get censored. What is that?!

And the one you see the most common censored....

I nee(D A Go)od clan.

I was never sure what a Dago was, but apparently it's a slang name for an Italian?

[DokuganR]

Awesome idea. Always hate it when it censors casual conversation.

[NuNuHeAd99]

I can't even say my name with an "s" at the end of it... Keanu's Ke-anus Get it???

[Brenn]

I can't even say my name with an "s" at the end of it... Keanu's Ke-anus Get it???
XD

Sorry, but I don't think that there's any easy way to fix that...

Clever name, though :)

By the way, thanks for the support so far!

[NuNuHeAd99]

Sorry, but I don't think that there's any easy way to fix that...

Clever name, though

By the way, thanks for the support so far!

Yea cause it says anus at the end

[Brenn]

Yea cause it says anus at the end

Yup.


Would anyone be interested in actually seeing the code I wrote to do this?

[KingKazuma]

Yup.


Would anyone be interested in actually seeing the code I wrote to do this?

Yesh plz thatll be good