From b7185c8231a95e66f42186196eb96220045f0d7c Mon Sep 17 00:00:00 2001
From: Eric Eastwood <contact@ericeastwood.com>
Date: Thu, 29 Jun 2023 18:49:25 -0500
Subject: [PATCH] Fix `18+` false positives with NSFW check

Was noticing this with our test room names like: `planet-1688081266353-room-18`
---
 shared/lib/check-text-for-nsfw.js            | 8 +++++++-
 test/shared/lib/check-text-for-nsfw-tests.js | 5 +++++
 2 files changed, 12 insertions(+), 1 deletion(-)

diff --git a/shared/lib/check-text-for-nsfw.js b/shared/lib/check-text-for-nsfw.js
index 445e557..1c20820 100644
--- a/shared/lib/check-text-for-nsfw.js
+++ b/shared/lib/check-text-for-nsfw.js
@@ -1,7 +1,13 @@
 'use strict';
 
+const escapeStringRegexp = require('escape-string-regexp');
+
 const NSFW_WORDS = ['nsfw', 'porn', 'nudes', 'sex', '18+'];
-const NSFW_REGEXES = NSFW_WORDS.map((word) => new RegExp(`(\\b|_)${word}(\\b|_)`, 'i'));
+const NSFW_REGEXES = NSFW_WORDS.map(
+  // We use `(\b|_|-|\s|^)` instead of just `(\b|_)` because the word boundary doesn't
+  // match next to the `+` sign in `18+`
+  (word) => new RegExp(`(\\b|_|-|\\s|^)${escapeStringRegexp(word)}(\\b|_|-|\\s|$)`, 'i')
+);
 
 // A very basic check for NSFW content that just looks for some keywords in the given
 // text
diff --git a/test/shared/lib/check-text-for-nsfw-tests.js b/test/shared/lib/check-text-for-nsfw-tests.js
index 9a19179..3d7ba04 100644
--- a/test/shared/lib/check-text-for-nsfw-tests.js
+++ b/test/shared/lib/check-text-for-nsfw-tests.js
@@ -13,6 +13,11 @@ describe('checkTextForNsfw', () => {
     NSFW_foo: true,
     'NSFW-foo': true,
     'NSFW:foo': true,
+    '18+ only': true,
+    // Previous false positives that we ran into in the wild that should not be flagged
+    // as NSFW
+    '1888-great-blizzard': false,
+    'argon-18-element': false,
   }).forEach(([inputText, expectedNsfw]) => {
     it(`should return ${expectedNsfw} for '${inputText}'`, () => {
       assert.strictEqual(