fix: catch punctuation-separated slurs in username censor

Re-add skipNonAlphabeticTransformer to both matcher chains so bypass
attempts like "n.i.g.g.e.r" are detected.

src/server/Privilege.ts

@@ -6,6 +6,7 @@ import {
   pattern,
   resolveConfusablesTransformer,
   resolveLeetSpeakTransformer,
+  skipNonAlphabeticTransformer,
   toAsciiLowerCaseTransformer,
 } from "obscenity";
 import countries from "resources/countries.json";
@@ -71,15 +72,21 @@ export function createMatcher(bannedWords: string[]): RegExpMatcher {
   ];
   // substringMatcher: literal patterns, no collapse — catches "niggertesting" as a substring
   // collapseMatcher: deduped patterns + collapse transformer — catches "niiiigger", "hiiitler"
+  // skipNonAlphabeticTransformer is applied last to catch punctuation-separated bypasses
+  // like "n.i.g.g.e.r".
   const substringMatcher = new RegExpMatcher({
     ...buildDataset(bannedWords, false),
-    blacklistMatcherTransformers: baseTransformers,
+    blacklistMatcherTransformers: [
+      ...baseTransformers,
+      skipNonAlphabeticTransformer(),
+    ],
   });
   const collapseMatcher = new RegExpMatcher({
     ...buildDataset(bannedWords, true),
     blacklistMatcherTransformers: [
       ...baseTransformers,
       collapseDuplicatesTransformer(),
+      skipNonAlphabeticTransformer(),
     ],
   });
   return {

tests/Privilege.test.ts

@@ -114,11 +114,9 @@ describe("UsernameCensor", () => {
       expect(matcher.hasMatch("MyChairName")).toBe(true);
     });
 
-    test("detects banned words with underscores/dots/numbers mixed in", () => {
+    test("detects banned words with non-alphabetic characters mixed in", () => {
-      // These should NOT bypass the filter (skipNonAlphabetic was intentionally removed)
+      expect(matcher.hasMatch("n.i.g.g.e.r")).toBe(true);
-      // Words separated by non-alpha chars are treated as separate tokens
+      expect(matcher.hasMatch("hi_tler")).toBe(true);
-      expect(matcher.hasMatch("n.i.g.g.e.r")).toBe(false); // dots break the word
-      expect(matcher.hasMatch("hi_tler")).toBe(false); // underscore breaks it
     });
 
     test("allows clean usernames", () => {
@@ -141,6 +139,19 @@ describe("UsernameCensor", () => {
       expect(matcher.hasMatch("kkklover")).toBe(true);
       expect(matcher.hasMatch("ilovekkkboys")).toBe(true);
     });
+
+    test("catches slurs separated by periods (bypass attempt)", () => {
+      expect(matcher.hasMatch("n.i.g.g.e.r")).toBe(true);
+      expect(matcher.hasMatch("N.I.G.G.E.R")).toBe(true);
+      expect(matcher.hasMatch("n.i.g.g.a")).toBe(true);
+      expect(matcher.hasMatch("h.i.t.l.e.r")).toBe(true);
+      expect(matcher.hasMatch("hello n.i.g.g.e.r world")).toBe(true);
+    });
+
+    test("censor replaces period-separated slur usernames", () => {
+      const result = checker.censor("n.i.g.g.e.r", null);
+      expect(shadowNames).toContain(result.username);
+    });
   });
 
   describe("censor", () => {