feat: add sanitizer

Dalamud.Test/Dalamud.Test.csproj

@@ -43,6 +43,7 @@
     </ItemGroup>
     <ItemGroup>
         <Compile Include="LocalizationTests.cs" />
+        <Compile Include="Plugin\Sanitizer\SanitizerTests.cs" />
         <Compile Include="Properties\AssemblyInfo.cs" />
     </ItemGroup>
     <ItemGroup>

Dalamud.Test/Plugin/Sanitizer/SanitizerTests.cs

@@ -0,0 +1,67 @@
+// ReSharper disable StringLiteralTypo
+
+using System.Linq;
+using NUnit.Framework;
+
+namespace Dalamud.Test.Plugin.Sanitizer {
+    [TestFixture]
+    public class SanitizerTests {
+        global::Dalamud.Plugin.Sanitizer.Sanitizer sanitizer;
+
+        [Test]
+        public void Sanitize_NormalString_NoChange() {
+            sanitizer = new global::Dalamud.Plugin.Sanitizer.Sanitizer(ClientLanguage.English);
+            const string str = "Pixie Cotton Hood of Healing";
+            var sanitizedString = sanitizer.Sanitize(str);
+            Assert.AreEqual(str, sanitizedString);
+        }
+
+        [Test]
+        public void Sanitize_DESpecialCharacters_Sanitized() {
+            sanitizer = new global::Dalamud.Plugin.Sanitizer.Sanitizer(ClientLanguage.German);
+            const string str = @"Anemos-Pan
zer
hand
schu
he des Drachenbluts";
+            var sanitizedString = sanitizer.Sanitize(str);
+            Assert.AreEqual(@"Anemos-Panzerhandschuhe des Drachenbluts", sanitizedString);
+        }
+
+        [Test]
+        public void Sanitize_FRSpecialCharacters_Sanitized() {
+            sanitizer = new global::Dalamud.Plugin.Sanitizer.Sanitizer(ClientLanguage.French);
+            const string str = @"Le Diademe
: terrains de chasse|Le Diademe
: terrains de chasse";
+            var sanitizedString = sanitizer.Sanitize(str);
+            Assert.AreEqual(@"Le Diademe: terrains de chasse|Le Diademe: terrains de chasse", sanitizedString);
+        }
+
+        [Test]
+        public void Sanitize_SpecifyLanguage_Sanitized() {
+            sanitizer = new global::Dalamud.Plugin.Sanitizer.Sanitizer(ClientLanguage.French);
+            const string str = @"Le Diademe
: terrains de chasse|Le Diademe
: terrains de chasse";
+            var sanitizedString = sanitizer.Sanitize(str, ClientLanguage.French);
+            Assert.AreEqual(@"Le Diademe: terrains de chasse|Le Diademe: terrains de chasse", sanitizedString);
+        }
+
+        [Test]
+        public void Sanitize_List_Sanitized() {
+            sanitizer = new global::Dalamud.Plugin.Sanitizer.Sanitizer(ClientLanguage.German);
+            const string str = @"Anemos-Pan
zer
hand
schu
he des Drachenbluts";
+            var sanitizedStrings = sanitizer.Sanitize(new[] {str});
+            Assert.AreEqual(@"Anemos-Panzerhandschuhe des Drachenbluts", sanitizedStrings.First());
+        }
+
+        [Test]
+        public void Sanitize_SpecifyLanguageList_Sanitized() {
+            sanitizer = new global::Dalamud.Plugin.Sanitizer.Sanitizer(ClientLanguage.German);
+            const string str = @"Anemos-Pan
zer
hand
schu
he des Drachenbluts";
+            var sanitizedStrings = sanitizer.Sanitize(new[] {str}, ClientLanguage.German);
+            Assert.AreEqual(@"Anemos-Panzerhandschuhe des Drachenbluts", sanitizedStrings.First());
+        }
+        
+        [Test]
+        public void Sanitize_UseAlternateLanguage_Sanitized() {
+            sanitizer = new global::Dalamud.Plugin.Sanitizer.Sanitizer(ClientLanguage.English);
+            const string str = @"Anemos-Pan
zer
hand
schu
he des Drachenbluts";
+            var sanitizedStrings = sanitizer.Sanitize(new[] {str}, ClientLanguage.German);
+            Assert.AreEqual(@"Anemos-Panzerhandschuhe des Drachenbluts", sanitizedStrings.First());
+        }
+    }
+}

Dalamud.sln.DotSettings

@@ -47,4 +47,5 @@
 	<s:Boolean x:Key="/Default/Environment/SettingsMigration/IsMigratorApplied/=JetBrains_002EReSharper_002EPsi_002ECSharp_002ECodeStyle_002ESettingsUpgrade_002EMigrateBlankLinesAroundFieldToBlankLinesAroundProperty/@EntryIndexedValue">True</s:Boolean>
 	<s:Boolean x:Key="/Default/Environment/SettingsMigration/IsMigratorApplied/=JetBrains_002EReSharper_002EPsi_002ECSharp_002ECodeStyle_002ESettingsUpgrade_002EMigrateThisQualifierSettings/@EntryIndexedValue">True</s:Boolean>
 	<s:Boolean x:Key="/Default/Environment/SettingsMigration/IsMigratorApplied/=JetBrains_002EReSharper_002EPsi_002EFormat_002ESettingsUpgrade_002EAlignmentTabFillStyleMigration/@EntryIndexedValue">True</s:Boolean>
-	<s:Boolean x:Key="/Default/UserDictionary/Words/=Dalamud/@EntryIndexedValue">True</s:Boolean></wpf:ResourceDictionary>
+	<s:Boolean x:Key="/Default/UserDictionary/Words/=Dalamud/@EntryIndexedValue">True</s:Boolean>
+	<s:Boolean x:Key="/Default/UserDictionary/Words/=unsanitized/@EntryIndexedValue">True</s:Boolean></wpf:ResourceDictionary>

Dalamud/Plugin/DalamudPluginInterface.cs

@@ -13,6 +13,7 @@ using Dalamud.Game.ClientState;
 using Dalamud.Game.Command;
 using Dalamud.Game.Internal;
 using Dalamud.Interface;
+using Dalamud.Plugin.Sanitizer;
 
 namespace Dalamud.Plugin
 {
@@ -48,6 +49,7 @@ namespace Dalamud.Plugin
             this.pluginName = pluginName;
             this.configs = configs;
 
+            this.Sanitizer = new Sanitizer.Sanitizer(this.Data.Language);
             this.UiLanguage = this.dalamud.Configuration.LanguageOverride;
             dalamud.LocalizationManager.OnLocalizationChanged += this.OnLocalizationChanged;
         }
@@ -132,6 +134,11 @@ namespace Dalamud.Plugin
         /// </summary>
         public string UiLanguage { get; private set; }
 
+        /// <summary>
+        /// Gets serializer class with functions to remove special characters from strings.
+        /// </summary>
+        public ISanitizer Sanitizer { get; }
+
         /// <summary>
         /// Gets the action that should be executed when any plugin sends a message.
         /// </summary>

Dalamud/Plugin/Sanitizer/ISanitizer.cs

@@ -0,0 +1,40 @@
+using System.Collections.Generic;
+
+namespace Dalamud.Plugin.Sanitizer
+{
+    /// <summary>
+    /// Sanitize strings to remove soft hyphens and other special characters.
+    /// </summary>
+    public interface ISanitizer
+    {
+        /// <summary>
+        /// Creates a sanitized string using current clientLanguage.
+        /// </summary>
+        /// <param name="unsanitizedString">An unsanitized string to sanitize.</param>
+        /// <returns>A sanitized string.</returns>
+        string Sanitize(string unsanitizedString);
+
+        /// <summary>
+        /// Creates a sanitized string using request clientLanguage.
+        /// </summary>
+        /// <param name="unsanitizedString">An unsanitized string to sanitize.</param>
+        /// <param name="clientLanguage">Target language for sanitized strings.</param>
+        /// <returns>A sanitized string.</returns>
+        string Sanitize(string unsanitizedString, ClientLanguage clientLanguage);
+
+        /// <summary>
+        /// Creates a list of sanitized strings using current clientLanguage.
+        /// </summary>
+        /// <param name="unsanitizedStrings">List of unsanitized string to sanitize.</param>
+        /// <returns>A list of sanitized strings.</returns>
+        IEnumerable<string> Sanitize(IEnumerable<string> unsanitizedStrings);
+
+        /// <summary>
+        /// Creates a list of sanitized strings using requested clientLanguage.
+        /// </summary>
+        /// <param name="unsanitizedStrings">List of unsanitized string to sanitize.</param>
+        /// <param name="clientLanguage">Target language for sanitized strings.</param>
+        /// <returns>A list of sanitized strings.</returns>
+        IEnumerable<string> Sanitize(IEnumerable<string> unsanitizedStrings, ClientLanguage clientLanguage);
+    }
+}

Dalamud/Plugin/Sanitizer/Sanitizer.cs

@@ -0,0 +1,113 @@
+using System;
+using System.Collections.Generic;
+using System.Linq;
+using System.Text;
+
+namespace Dalamud.Plugin.Sanitizer
+{
+    /// <summary>
+    /// Sanitize strings to remove soft hyphens and other special characters.
+    /// </summary>
+    public class Sanitizer : ISanitizer
+    {
+        private readonly KeyValuePair<string, string> softHyphen = new KeyValuePair<string, string>("\u0002\u0016\u0001\u0003", string.Empty);
+        private readonly KeyValuePair<string, string> emphasisOpen = new KeyValuePair<string, string>("\u0002\u001A\u0003", string.Empty);
+        private readonly KeyValuePair<string, string> emphasisClose = new KeyValuePair<string, string>("\u0002\u001A\u0001\u0003", string.Empty);
+        private readonly KeyValuePair<string, string> indent = new KeyValuePair<string, string>("\u0002\u001D\u0001\u0003", string.Empty);
+        private readonly KeyValuePair<string, string> dagger = new KeyValuePair<string, string>("\u0020\u2020", string.Empty);
+        private readonly KeyValuePair<string, string> ligatureOE = new KeyValuePair<string, string>("\u0153", "\u006F\u0065");
+        private readonly List<KeyValuePair<string, string>> defaultSanitizationList;
+
+        /// <summary>
+        /// Initializes a new instance of the <see cref="Sanitizer"/> class.
+        /// </summary>
+        /// <param name="clientLanguage">Default clientLanguage for sanitizing strings.</param>
+        public Sanitizer(ClientLanguage clientLanguage)
+        {
+            this.defaultSanitizationList = this.BuildSanitizationList(clientLanguage);
+        }
+
+        /// <summary>
+        /// Creates a sanitized string using current clientLanguage.
+        /// </summary>
+        /// <param name="unsanitizedString">An unsanitized string to sanitize.</param>
+        /// <returns>A sanitized string.</returns>
+        public string Sanitize(string unsanitizedString)
+        {
+            return this.defaultSanitizationList == null ? unsanitizedString : ApplySanitizationList(unsanitizedString, this.defaultSanitizationList);
+        }
+
+        /// <summary>
+        /// Creates a sanitized string using request clientLanguage.
+        /// </summary>
+        /// <param name="unsanitizedString">An unsanitized string to sanitize.</param>
+        /// <param name="clientLanguage">Target language for sanitized strings.</param>
+        /// <returns>A sanitized string.</returns>
+        public string Sanitize(string unsanitizedString, ClientLanguage clientLanguage)
+        {
+            var newSanitizationList = this.BuildSanitizationList(clientLanguage);
+            return newSanitizationList == null ? unsanitizedString : ApplySanitizationList(unsanitizedString, newSanitizationList);
+        }
+
+        /// <summary>
+        /// Creates a list of sanitized strings using current clientLanguage.
+        /// </summary>
+        /// <param name="unsanitizedStrings">List of unsanitized string to sanitize.</param>
+        /// <returns>A list of sanitized strings.</returns>
+        public IEnumerable<string> Sanitize(IEnumerable<string> unsanitizedStrings)
+        {
+            return this.defaultSanitizationList == null ? unsanitizedStrings.Select(unsanitizedString => unsanitizedString) :
+                       unsanitizedStrings.Select(unsanitizedString => ApplySanitizationList(unsanitizedString, this.defaultSanitizationList));
+        }
+
+        /// <summary>
+        /// Creates a list of sanitized strings using requested clientLanguage.
+        /// </summary>
+        /// <param name="unsanitizedStrings">List of unsanitized string to sanitize.</param>
+        /// <param name="clientLanguage">Target language for sanitized strings.</param>
+        /// <returns>A list of sanitized strings.</returns>
+        public IEnumerable<string> Sanitize(IEnumerable<string> unsanitizedStrings, ClientLanguage clientLanguage)
+        {
+            var newSanitizationList = this.BuildSanitizationList(clientLanguage);
+            return newSanitizationList == null ? unsanitizedStrings.Select(unsanitizedString => unsanitizedString) :
+                       unsanitizedStrings.Select(unsanitizedString => ApplySanitizationList(unsanitizedString, newSanitizationList));
+        }
+
+        private static string ApplySanitizationList(string unsanitizedString, IEnumerable<KeyValuePair<string, string>> sanitizationList)
+        {
+            var sanitizedValue = new StringBuilder(unsanitizedString);
+            foreach (var item in sanitizationList) sanitizedValue.Replace(item.Key, item.Value);
+            return sanitizedValue.ToString();
+        }
+
+        private List<KeyValuePair<string, string>> BuildSanitizationList(ClientLanguage clientLanguage)
+        {
+            switch (clientLanguage)
+            {
+                case ClientLanguage.Japanese:
+                    break;
+                case ClientLanguage.English:
+                    break;
+                case ClientLanguage.German:
+                    return new List<KeyValuePair<string, string>>
+                    {
+                        this.softHyphen,
+                        this.emphasisOpen,
+                        this.emphasisClose,
+                        this.dagger,
+                    };
+                case ClientLanguage.French:
+                    return new List<KeyValuePair<string, string>>
+                    {
+                        this.softHyphen,
+                        this.indent,
+                        this.ligatureOE,
+                    };
+                default:
+                    throw new ArgumentOutOfRangeException();
+            }
+
+            return null;
+        }
+    }
+}