From f0fc79c142e6d69f90b3b73d6066ecf88d566a90 Mon Sep 17 00:00:00 2001
From: barelyprofessional
 <150058423+barelyprofessional@users.noreply.github.com>
Date: Sun, 9 Feb 2025 15:59:39 +0800
Subject: [PATCH] Attempt to improve the way long strings are split up. There's
 a new extension method called FancySplitMessage to achieve this. Truncation
 options now work on bytes instead of string length too

---
 KfChatDotNetBot/ChatBot.cs                |  9 +--
 KfChatDotNetBot/Commands/AdminCommands.cs |  4 +-
 KfChatDotNetBot/Extensions.cs             | 79 ++++++++++++++++++++++-
 3 files changed, 85 insertions(+), 7 deletions(-)

diff --git a/KfChatDotNetBot/ChatBot.cs b/KfChatDotNetBot/ChatBot.cs
index 0165a75..c1b82a4 100644
--- a/KfChatDotNetBot/ChatBot.cs
+++ b/KfChatDotNetBot/ChatBot.cs
@@ -285,7 +285,7 @@ public class ChatBot
             return messageTracker;
         }
         
-        if (Encoding.UTF8.GetByteCount(messageTracker.Message) > lengthLimit && lengthLimitBehavior != LengthLimitBehavior.DoNothing)
+        if (messageTracker.Message.Utf8LengthBytes() > lengthLimit && lengthLimitBehavior != LengthLimitBehavior.DoNothing)
         {
             if (lengthLimitBehavior == LengthLimitBehavior.RefuseToSend)
             {
@@ -296,7 +296,8 @@ public class ChatBot
             }
             if (lengthLimitBehavior == LengthLimitBehavior.TruncateNicely)
             {
-                messageTracker.Message = messageTracker.Message.Truncate(lengthLimit);
+                // '…' is 3 bytes so we have to make room for it
+                messageTracker.Message = messageTracker.Message.TruncateBytes(lengthLimit - 3).TrimEnd() + "…";
             }
 
             if (lengthLimitBehavior == LengthLimitBehavior.TruncateExactly)
@@ -304,7 +305,7 @@ public class ChatBot
                 // ReSharper disable once ReplaceSubstringWithRangeIndexer
                 // The range indexer is a fucking piece of shit that does not work.
                 // TrimEnd in case you end up truncating on a space (happened during testing) as Sneedchat will trim it
-                messageTracker.Message = messageTracker.Message.Substring(0, lengthLimit).TrimEnd();
+                messageTracker.Message = messageTracker.Message.TruncateBytes(lengthLimit).TrimEnd();
             }
         }
         
@@ -316,7 +317,7 @@ public class ChatBot
     }
 
     public SentMessageTrackerModel SendChatMessage(string message, bool bypassSeshDetect = false,
-        LengthLimitBehavior lengthLimitBehavior = LengthLimitBehavior.TruncateNicely, int lengthLimit = 500)
+        LengthLimitBehavior lengthLimitBehavior = LengthLimitBehavior.TruncateNicely, int lengthLimit = 1023)
     {
         return SendChatMessageAsync(message, bypassSeshDetect, lengthLimitBehavior, lengthLimit).Result;
     }
diff --git a/KfChatDotNetBot/Commands/AdminCommands.cs b/KfChatDotNetBot/Commands/AdminCommands.cs
index 991c450..daae646 100644
--- a/KfChatDotNetBot/Commands/AdminCommands.cs
+++ b/KfChatDotNetBot/Commands/AdminCommands.cs
@@ -123,7 +123,7 @@ public class GmKasinoListCommand : ICommand
             result += $"[br]{i}: {image}";
         }
 
-        await botInstance.SendChatMessagesAsync(result.SplitMessage().ToList(), true);
+        await botInstance.SendChatMessagesAsync(result.FancySplitMessage(), true);
     }
 }
 
@@ -213,7 +213,7 @@ public class GnKasinoListCommand : ICommand
             result += $"[br]{i}: {image}";
         }
 
-        await botInstance.SendChatMessagesAsync(result.SplitMessage().ToList(), true);
+        await botInstance.SendChatMessagesAsync(result.FancySplitMessage(), true);
     }
 }
 
diff --git a/KfChatDotNetBot/Extensions.cs b/KfChatDotNetBot/Extensions.cs
index 27b507b..1b99dab 100644
--- a/KfChatDotNetBot/Extensions.cs
+++ b/KfChatDotNetBot/Extensions.cs
@@ -1,4 +1,5 @@
-using System.Text.RegularExpressions;
+using System.Text;
+using System.Text.RegularExpressions;
 
 namespace KfChatDotNetBot;
 
@@ -44,4 +45,80 @@ public static class Extensions
         }
     }
 
+    /// <summary>
+    /// Split messages to x number of bytes while avoiding splitting mid-word where possible
+    /// </summary>
+    /// <param name="s">String that should get split</param>
+    /// <param name="partLengthBytes">Length limit, no part should be > than the number of bytes specified</param>
+    /// <param name="partLimit">Limit for how many parts to return (returns first n elements). Set to 0 to disable.</param>
+    /// <returns></returns>
+    public static List<string> FancySplitMessage(this string s, int partLengthBytes = 1023, int partLimit = 5)
+    {
+        var output = new List<string>();
+        var part = string.Empty;
+        foreach (var word in s.Split(' '))
+        {
+            if (word.Utf8LengthBytes() > partLengthBytes)
+            {
+                // Add the part already in memory if there is one
+                if (part != string.Empty)
+                {
+                    output.Add(part.TrimEnd());
+                    part = string.Empty;
+                }
+                // Breaks into chunks of x size which will break really long URLs etc. but no other way really
+                output.AddRange(word.ChunkBytes(partLengthBytes));
+                continue;
+            }
+            
+            if (part.Utf8LengthBytes() + word.Utf8LengthBytes() > partLengthBytes)
+            {
+                // TrimEnd() to remove trailing spaces
+                output.Add(part.TrimEnd());
+                part = word + " ";
+                continue;
+            }
+
+            part += word + " ";
+        }
+
+        // Add on whatever remains
+        if (part != string.Empty)
+        {
+            output.Add(part.TrimEnd());
+        }
+
+        if (partLimit != 0 && output.Count > partLimit)
+        {
+            return output.Take(partLimit).ToList();
+        }
+        
+        return output;
+    }
+
+    public static int Utf8LengthBytes(this string s)
+    {
+        return Encoding.UTF8.GetByteCount(s);
+    }
+
+    public static IEnumerable<string> ChunkBytes(this string input, int bytesPerChunk)
+    {
+        var bytes = Encoding.UTF8.GetBytes(input);
+    
+        for (var i = 0; i < bytes.Length; i += bytesPerChunk)
+        {
+            var chunkSize = Math.Min(bytesPerChunk, bytes.Length - i);
+            yield return Encoding.UTF8.GetString(bytes, i, chunkSize);
+        }
+    }
+
+
+    public static string TruncateBytes(this string s, int limitBytes)
+    {
+        return Encoding.UTF8.GetString(
+            Encoding.UTF8.GetBytes(s)
+                .Take(limitBytes)
+                .ToArray()
+        ).TrimEnd();
+    }
 }
\ No newline at end of file