From 48be12a92085b40c4b5253c99ccab05a6db98301 Mon Sep 17 00:00:00 2001
From: Jicheng Lu <franklujc@gmail.com>
Date: Sat, 31 Aug 2024 01:05:53 -0500
Subject: [PATCH] relocate text chopper

---
 .../BotSharp.Abstraction/Knowledges/ITextChopper.cs   | 11 -----------
 .../Knowledges/Helpers}/TextChopperService.cs         | 11 ++++++-----
 .../Knowledges/Services/KnowledgeService.Create.cs    |  5 ++---
 .../KnowledgeBasePlugin.cs                            |  1 -
 4 files changed, 8 insertions(+), 20 deletions(-)
 delete mode 100644 src/Infrastructure/BotSharp.Abstraction/Knowledges/ITextChopper.cs
 rename src/{Plugins/BotSharp.Plugin.KnowledgeBase/Services => Infrastructure/BotSharp.Core/Knowledges/Helpers}/TextChopperService.cs (78%)

diff --git a/src/Infrastructure/BotSharp.Abstraction/Knowledges/ITextChopper.cs b/src/Infrastructure/BotSharp.Abstraction/Knowledges/ITextChopper.cs
deleted file mode 100644
index 7d06a8a4..00000000
--- a/src/Infrastructure/BotSharp.Abstraction/Knowledges/ITextChopper.cs
+++ /dev/null
@@ -1,11 +0,0 @@
-using BotSharp.Abstraction.Knowledges.Models;
-
-namespace BotSharp.Abstraction.Knowledges;
-
-/// <summary>
-/// Chop large content into chunks
-/// </summary>
-public interface ITextChopper
-{
-    List<string> Chop(string content, ChunkOption option);
-}
diff --git a/src/Plugins/BotSharp.Plugin.KnowledgeBase/Services/TextChopperService.cs b/src/Infrastructure/BotSharp.Core/Knowledges/Helpers/TextChopperService.cs
similarity index 78%
rename from src/Plugins/BotSharp.Plugin.KnowledgeBase/Services/TextChopperService.cs
rename to src/Infrastructure/BotSharp.Core/Knowledges/Helpers/TextChopperService.cs
index 88c77641..68b58539 100644
--- a/src/Plugins/BotSharp.Plugin.KnowledgeBase/Services/TextChopperService.cs
+++ b/src/Infrastructure/BotSharp.Core/Knowledges/Helpers/TextChopperService.cs
@@ -1,17 +1,18 @@
+using BotSharp.Abstraction.Knowledges.Models;
 using System.Text.RegularExpressions;
 
-namespace BotSharp.Plugin.KnowledgeBase.Services;
+namespace BotSharp.Core.Knowledges.Helpers;
 
-public class TextChopperService : ITextChopper
+public static class TextChopper
 {
-    public List<string> Chop(string content, ChunkOption option)
+    public static List<string> Chop(string content, ChunkOption option)
     {
         content = Regex.Replace(content, @"\.{2,}", " ");
         content = Regex.Replace(content, @"_{2,}", " ");
         return option.SplitByWord ? ChopByWord(content, option) : ChopByChar(content, option);
     }
 
-    private List<string> ChopByWord(string content, ChunkOption option)
+    private static List<string> ChopByWord(string content, ChunkOption option)
     {
         var chunks = new List<string>();
 
@@ -34,7 +35,7 @@ public class TextChopperService : ITextChopper
         return chunks;
     }
 
-    private List<string> ChopByChar(string content, ChunkOption option)
+    private static List<string> ChopByChar(string content, ChunkOption option)
     {
         var chunks = new List<string>();
         var currentPos = 0;
diff --git a/src/Infrastructure/BotSharp.Core/Knowledges/Services/KnowledgeService.Create.cs b/src/Infrastructure/BotSharp.Core/Knowledges/Services/KnowledgeService.Create.cs
index 2927d952..d8b82fb5 100644
--- a/src/Infrastructure/BotSharp.Core/Knowledges/Services/KnowledgeService.Create.cs
+++ b/src/Infrastructure/BotSharp.Core/Knowledges/Services/KnowledgeService.Create.cs
@@ -1,5 +1,6 @@
 using BotSharp.Abstraction.Knowledges.Models;
 using BotSharp.Abstraction.VectorStorage.Models;
+using BotSharp.Core.Knowledges.Helpers;
 
 namespace BotSharp.Core.Knowledges.Services;
 
@@ -8,9 +9,7 @@ public partial class KnowledgeService
     public async Task FeedVectorKnowledge(string collectionName, KnowledgeCreationModel knowledge)
     {
         var index = 0;
-
-        var textChopper = _services.GetRequiredService<ITextChopper>();
-        var lines = textChopper.Chop(knowledge.Content, new ChunkOption
+        var lines = TextChopper.Chop(knowledge.Content, new ChunkOption
         {
             Size = 1024,
             Conjunction = 32,
diff --git a/src/Plugins/BotSharp.Plugin.KnowledgeBase/KnowledgeBasePlugin.cs b/src/Plugins/BotSharp.Plugin.KnowledgeBase/KnowledgeBasePlugin.cs
index dd864a73..9a955a5b 100644
--- a/src/Plugins/BotSharp.Plugin.KnowledgeBase/KnowledgeBasePlugin.cs
+++ b/src/Plugins/BotSharp.Plugin.KnowledgeBase/KnowledgeBasePlugin.cs
@@ -20,7 +20,6 @@ public class KnowledgeBasePlugin : IBotSharpPlugin
             return settingService.Bind<KnowledgeBaseSettings>("KnowledgeBase");
         });
 
-        services.AddScoped<ITextChopper, TextChopperService>();
         services.AddSingleton<IPdf2TextConverter, PigPdf2TextConverter>();
         services.AddScoped<IAgentUtilityHook, KnowledgeBaseUtilityHook>();
         services.AddScoped<IAgentHook, KnowledgeBaseAgentHook>();