Tune profanity detection (#3214)

* Tune profanity detection

* Remove console log

* Remove profanity

* Add tests for profanity

* Fix test case

backend/__tests__/utils/validation.spec.ts

@@ -0,0 +1,124 @@
+import * as Validation from "../../src/utils/validation";
+
+describe("Validation", () => {
+  it("inRange", () => {
+    const testCases = [
+      {
+        value: 1,
+        min: 1,
+        max: 2,
+        expected: true,
+      },
+      {
+        value: 1,
+        min: 2,
+        max: 2,
+        expected: false,
+      },
+      {
+        value: 1,
+        min: 1,
+        max: 1,
+        expected: true,
+      },
+      {
+        value: 53,
+        min: -100,
+        max: 100,
+        expected: true,
+      },
+      {
+        value: 153,
+        min: -100,
+        max: 100,
+        expected: false,
+      },
+    ];
+
+    testCases.forEach((testCase) => {
+      expect(
+        Validation.inRange(testCase.value, testCase.min, testCase.max)
+      ).toBe(testCase.expected);
+    });
+  });
+
+  it("isUsernameValid", () => {
+    const testCases = [
+      {
+        name: "Miodec",
+        expected: false,
+      },
+      {
+        name: "fucker",
+        expected: false,
+      },
+      {
+        name: "Bruce",
+        expected: true,
+      },
+      {
+        name: "Rizwan_123",
+        expected: true,
+      },
+      {
+        name: "Fe-rotiq._123._",
+        expected: true,
+      },
+      {
+        name: " ",
+        expected: false,
+      },
+      {
+        name: "",
+        expected: false,
+      },
+      {
+        name: "superduperlongnamethatshouldbeinvalid",
+        expected: false,
+      },
+      {
+        name: ".period",
+        expected: false,
+      },
+      {
+        name: "fucking_profane",
+        expected: false,
+      },
+    ];
+
+    testCases.forEach((testCase) => {
+      expect(Validation.isUsernameValid(testCase.name)).toBe(testCase.expected);
+    });
+  });
+
+  it("containsProfanity", () => {
+    const testCases = [
+      {
+        text: "https://www.fuckyou.com",
+        expected: true,
+      },
+      {
+        text: "Hello world!",
+        expected: false,
+      },
+      {
+        text: "I fucking hate you",
+        expected: true,
+      },
+      {
+        text: "I love you",
+        expected: false,
+      },
+      {
+        text: "\n.fuck!",
+        expected: true,
+      },
+    ];
+
+    testCases.forEach((testCase) => {
+      expect(Validation.containsProfanity(testCase.text)).toBe(
+        testCase.expected
+      );
+    });
+  });
+});

backend/src/constants/profanities.ts

@@ -92,16 +92,13 @@ const profanities = [
   "fukker",
   "fukkin",
   "g00k",
-  "gay",
   "gayboy",
   "gaygirl",
-  "gays",
   "gayz",
   "god-damned",
   "h00r",
   "h0ar",
   "h0re",
-  "hells",
   "hoar",
   "hoor",
   "hoore",
@@ -118,7 +115,6 @@ const profanities = [
   "kunt",
   "kunts",
   "kuntz",
-  "lesbian",
   "lezzian",
   "lipshits",
   "lipshitz",
@@ -177,9 +173,6 @@ const profanities = [
   "pussy",
   "puuke",
   "puuker",
-  "queer",
-  "queers",
-  "queerz",
   "qweers",
   "qweerz",
   "qweir",
@@ -269,7 +262,6 @@ const profanities = [
   "kawk",
   "l3itch",
   "l3i+ch",
-  "lesbian",
   "masturbate",
   "masterbat",
   "masterbat3",
@@ -350,7 +342,6 @@ const profanities = [
   "fu(",
   "fuk",
   "futkretzn",
-  "gay",
   "gook",
   "guiena",
   "h0r",

backend/src/utils/validation.ts

@@ -1,6 +1,6 @@
 import _ from "lodash";
 import profanities from "../constants/profanities";
-import { matchesAPattern } from "./misc";
+import { matchesAPattern, sanitizeString } from "./misc";
 
 export function inRange(value: number, min: number, max: number): boolean {
   return value >= min && value <= max;
@@ -31,7 +31,12 @@ export function isUsernameValid(name: string): boolean {
 }
 
 export function containsProfanity(text: string): boolean {
-  const normalizedText = text.toLowerCase().split(" ");
+  const normalizedText = text
+    .toLowerCase()
+    .split(/[.,"/#!?$%^&*;:{}=\-_`~()\s\n]+/g)
+    .map((str) => {
+      return sanitizeString(str) ?? "";
+    });
 
   const hasProfanity = profanities.some((profanity) => {
     const normalizedProfanity = _.escapeRegExp(profanity.toLowerCase());