From e63cd67e7b024cfc74e61bd0c643b785dd0c334c Mon Sep 17 00:00:00 2001
From: Jim Parslow <jimpars@cisco.com>
Date: Thu, 20 Jun 2024 11:53:17 +0100
Subject: [PATCH] feat: support for newline characters

Will now find profanity next to newline characters
---
 ProfanityFilter.Tests.Unit/ProfanityTests.cs  | 46 +++++++++++++++++++
 .../ProfanityFilter/ProfanityFilter.cs        | 10 ++--
 2 files changed, 52 insertions(+), 4 deletions(-)

diff --git a/ProfanityFilter.Tests.Unit/ProfanityTests.cs b/ProfanityFilter.Tests.Unit/ProfanityTests.cs
index 33c48f2..a57b4e9 100644
--- a/ProfanityFilter.Tests.Unit/ProfanityTests.cs
+++ b/ProfanityFilter.Tests.Unit/ProfanityTests.cs
@@ -390,6 +390,52 @@ public void CensoredStringReturnsStringWithProfanitiesBleepedOut4()
             Assert.AreEqual(censored, result);
         }
 
+        [TestMethod]
+        public void CensoredStringReturnsStringWithProfanitiesBleepedOutMultiLine()
+        {
+            var filter = new ProfanityFilter();
+            filter.AllowList.Add("scunthorpe");
+            filter.AllowList.Add("penistone");
+
+            var censored = filter.CensorString("I fucking live in Scunthorpe and it is a shit place to live. I would much rather live in penistone you great big\ncock\r\nfuck.", '*');
+            var result = "I ******* live in Scunthorpe and it is a **** place to live. I would much rather live in penistone you great big\n****\r\n****.";
+
+            Assert.AreEqual(censored, result);
+        }
+
+        [TestMethod]
+        public void CensoredStringReturnsStringWithProfanitiesBleepedOutMultiLine2()
+        {
+            var filter = new ProfanityFilter();
+
+            var censored = filter.CensorString("2 girls 1 cup, is my favourite twatting\ntwat\r\n video.");
+            var result = "* ***** * ***, is my favourite ********\n****\r\n video.";
+
+            Assert.AreEqual(censored, result);
+        }
+
+        [TestMethod]
+        public void CensoredStringReturnsStringWithProfanitiesBleepedOutMultiLine3()
+        {
+            var filter = new ProfanityFilter();
+
+            var censored = filter.CensorString("Mary had a\r\n little shit lamb who was a little \r\nfucker\r\n.");
+            var result = "Mary had a\r\n little **** lamb who was a little \r\n******\r\n.";
+
+            Assert.AreEqual(censored, result);
+        }
+
+        [TestMethod]
+        public void CensoredStringReturnsStringWithProfanitiesBleepedOutMultiLine4()
+        {
+            var filter = new ProfanityFilter();
+
+            var censored = filter.CensorString("You are a \nstupid little twat\n, and you like to blow your load in an alaskan\r\npipeline.");
+            var result = "You are a \n****** little ****\n, and you like to **** **** **** in an alaskan\r\npipeline.";
+
+            Assert.AreEqual(censored, result);
+        }
+
 
         [TestMethod]
         public void CensoredStringReturnsStringWithSingleScunthorpe()
diff --git a/ProfanityFilter/ProfanityFilter/ProfanityFilter.cs b/ProfanityFilter/ProfanityFilter/ProfanityFilter.cs
index 3af0bc6..32913e3 100644
--- a/ProfanityFilter/ProfanityFilter/ProfanityFilter.cs
+++ b/ProfanityFilter/ProfanityFilter/ProfanityFilter.cs
@@ -121,7 +121,8 @@ public ReadOnlyCollection<string> DetectAllProfanities(string sentence, bool rem
             sentence = sentence.Replace(".", "");
             sentence = sentence.Replace(",", "");
 
-            var words = sentence.Split(' ');
+            // Split the string on spaces and newlines
+            var words = sentence.Split(new[] { ' ', '\n', '\r' }, StringSplitOptions.RemoveEmptyEntries);
             var postAllowList = FilterWordListByAllowList(words);
             List<string> swearList = new List<string>();
 
@@ -180,7 +181,8 @@ public string CensorString(string sentence, char censorCharacter, bool ignoreNum
 
             noPunctuation = Regex.Replace(noPunctuation, @"[^\w\s]", "");
 
-            var words = noPunctuation.Split(' ');
+            // Split the string on spaces and newlines
+            var words = noPunctuation.Split(new[] { ' ', '\n', '\r' }, StringSplitOptions.RemoveEmptyEntries);
 
             var postAllowList = FilterWordListByAllowList(words);
             var swearList = new List<string>();
@@ -225,7 +227,7 @@ public string CensorString(string sentence, char censorCharacter, bool ignoreNum
                 // Work backwards in string to get to the start of the word.
                 while (startIndex > 0)
                 {
-                    if (toCheck[startIndex - 1] == ' ' || char.IsPunctuation(toCheck[startIndex - 1]))
+                    if (toCheck[startIndex - 1] == ' ' || toCheck[startIndex - 1] == '\n' || toCheck[startIndex - 1] == '\r' || char.IsPunctuation(toCheck[startIndex - 1]))
                     {
                         break;
                     }
@@ -236,7 +238,7 @@ public string CensorString(string sentence, char censorCharacter, bool ignoreNum
                 // Work forwards to get to the end of the word.
                 while (endIndex < toCheck.Length)
                 {
-                    if (toCheck[endIndex] == ' ' || char.IsPunctuation(toCheck[endIndex]))
+                    if (toCheck[endIndex] == ' ' || toCheck[endIndex] == '\n' || toCheck[endIndex] == '\r' || char.IsPunctuation(toCheck[endIndex]))
                     {
                         break;
                     }