From ed5be399da157abc50b37d33189fff9cf91b79c3 Mon Sep 17 00:00:00 2001
From: jetsparrow <scram.jet.sparrow@gmail.com>
Date: Mon, 19 Aug 2019 21:37:34 +0300
Subject: [PATCH] separate words from punctuation

---
 AntiAntiSwearingBot.Tests/DetectTests.cs      | 40 +++++++++++++++++++
 AntiAntiSwearingBot.Tests/FilterTests.cs      | 34 ++++++++++------
 AntiAntiSwearingBot/Unbleeper.cs              |  4 +-
 .../dict/ObsceneDictionaryRu.txt              |  3 ++
 4 files changed, 67 insertions(+), 14 deletions(-)
 create mode 100644 AntiAntiSwearingBot.Tests/DetectTests.cs

diff --git a/AntiAntiSwearingBot.Tests/DetectTests.cs b/AntiAntiSwearingBot.Tests/DetectTests.cs
new file mode 100644
index 0000000..effa499
--- /dev/null
+++ b/AntiAntiSwearingBot.Tests/DetectTests.cs
@@ -0,0 +1,40 @@
+using System;
+using System.Collections.Generic;
+using System.Text;
+using Xunit;
+
+namespace AntiAntiSwearingBot.Tests
+{
+    public class DetectTests
+    {
+        Unbleeper ubl { get; }
+        Config cfg { get; }
+        SearchDictionary dict { get; }
+
+        public DetectTests()
+        {
+            cfg = Config.Load<Config>("aasb.cfg.json");
+            dict = new SearchDictionary(cfg);
+            ubl = new Unbleeper(dict, cfg.Unbleeper);
+        }
+
+        [Theory]
+        [InlineData("бл**ь", "*блядь")]
+        [InlineData("ж**а", "*жопа")]
+        public void UnbleepSimpleSwears(string word, string expected)
+        {
+            var unbleep = ubl.UnbleepSwears(word).TrimEnd(Environment.NewLine.ToCharArray());
+            Assert.Equal(expected, unbleep);
+        }
+
+        [Theory]
+        [InlineData("Просто пи**ец, как хочется кушать.", "*пиздец")]
+        [InlineData("Ужас на*уй!", "*нахуй")]
+        [InlineData("Сергей опять вы**нулся своим знанием тонкостей русского языка; в окно еб*шил стылый ноябрьский ветер. ", "*выебнулся\n**ебашил")]
+        public void DetectWordsWithPunctuation(string text, string expected)
+        {
+            var unbleep = ubl.UnbleepSwears(text).Replace("\r\n", "\n").Trim();
+            Assert.Equal(expected, unbleep);
+        }
+    }
+}
diff --git a/AntiAntiSwearingBot.Tests/FilterTests.cs b/AntiAntiSwearingBot.Tests/FilterTests.cs
index cb95e09..11ef2a6 100644
--- a/AntiAntiSwearingBot.Tests/FilterTests.cs
+++ b/AntiAntiSwearingBot.Tests/FilterTests.cs
@@ -6,27 +6,35 @@ namespace AntiAntiSwearingBot.Tests
     public class FilterTests
     {
         Unbleeper ubl { get; }
+        Config cfg { get; }
+        SearchDictionary dict { get; }
+
         public FilterTests()
         {
-            var cfg = Config.Load<Config>("aasb.cfg.json", "aasb.cfg.secret.json");
-            var dict = new SearchDictionary(cfg);
+            cfg = Config.Load<Config>("aasb.cfg.json", "aasb.cfg.secret.json");
+            dict = new SearchDictionary(cfg);
             ubl = new Unbleeper(dict, cfg.Unbleeper);
         }
 
-        [Theory]
-        [InlineData("бл**ь", "*блядь")]
-        [InlineData("ж**а", "*жопа")]
-        public void UnbleepSimpleSwears(string word, string expected)
-        {
-            var unbleep = ubl.UnbleepSwears(word).TrimEnd(Environment.NewLine.ToCharArray());
-            Assert.Equal(expected, unbleep);
-        }
-
         [Theory]
         [InlineData("*")]
         [InlineData("**#")]
-        [InlineData("@**#")]
-        public void IgnoreShortGrawlixes(string text) => Assert.Null(ubl.UnbleepSwears(text));
+        [InlineData("@*#")]
+        public void IgnoreShortGrawlixesWithoutLetters(string text)
+        {
+            if (text.Length < cfg.Unbleeper.MinAmbiguousWordLength)
+                Assert.Null(ubl.UnbleepSwears(text));
+        }
+
+        [Theory]
+        [InlineData("*")]
+        [InlineData("*б")]
+        [InlineData("х#")]
+        public void IgnoreShortWords(string text)
+        {
+            if (text.Length < cfg.Unbleeper.MinWordLength)
+                Assert.Null(ubl.UnbleepSwears(text));
+        }
 
         [Theory]
         [InlineData("@pvkuznetsov https://github.com/jacksondunstan/UnityNativeScripting")]
diff --git a/AntiAntiSwearingBot/Unbleeper.cs b/AntiAntiSwearingBot/Unbleeper.cs
index 2ffb392..473ca9e 100644
--- a/AntiAntiSwearingBot/Unbleeper.cs
+++ b/AntiAntiSwearingBot/Unbleeper.cs
@@ -20,6 +20,8 @@ namespace AntiAntiSwearingBot
 
         Regex BleepedSwearsRegex { get; }
 
+        static readonly char[] WORD_SEPARATORS = { ' ', '\t', '\r', '\n', '.', ',', '!', '?', ';' };
+
         public string UnbleepSwears(string text)
         {
             if (string.IsNullOrWhiteSpace(text))
@@ -30,7 +32,7 @@ namespace AntiAntiSwearingBot
             if (text.StartsWith('/')) // is chat command
                 return null;
 
-            var words = text.Split(new char[0], StringSplitOptions.RemoveEmptyEntries);
+            var words = text.Split(WORD_SEPARATORS, StringSplitOptions.RemoveEmptyEntries);
             var candidates = words
                 .Where(w =>
                     !Language.IsTelegramMention(w)
diff --git a/AntiAntiSwearingBot/dict/ObsceneDictionaryRu.txt b/AntiAntiSwearingBot/dict/ObsceneDictionaryRu.txt
index ff62eac..3687723 100644
--- a/AntiAntiSwearingBot/dict/ObsceneDictionaryRu.txt
+++ b/AntiAntiSwearingBot/dict/ObsceneDictionaryRu.txt
@@ -84,6 +84,7 @@
 выебанный
 выебат
 выебаться
+выебнулся
 высрать
 высраться
 выссать
@@ -217,6 +218,7 @@
 ебат
 ебаться
 ебатьс
+ебашил
 ебитесь
 ебло
 еблом
@@ -457,6 +459,7 @@
 натрахаться
 натрахивать
 натрахиваться
+нахуй
 нахуякать
 нахуякаться
 нахуякивать