add upload vector knowledge files

src/Infrastructure/BotSharp.Abstraction/Files/IFileStorageService.cs

@@ -61,4 +61,8 @@ public interface IFileStorageService
     bool SaveSpeechFile(string conversationId, string fileName, BinaryData data);
     BinaryData GetSpeechFile(string conversationId, string fileName);
     #endregion
+
+    #region Knowledge
+    bool SaveKnowledgeFiles(string collectionName, string fileId, string fileName, Stream stream);
+    #endregion
 }

src/Infrastructure/BotSharp.Abstraction/Knowledges/IKnowledgeService.cs

@@ -22,6 +22,7 @@ public interface IKnowledgeService
     #endregion
 
     #region Document
+    Task<UploadKnowledgeResponse> UploadVectorKnowledge(string collectionName, IEnumerable<InputFileModel> files);
     #endregion
 
     #region Common

src/Infrastructure/BotSharp.Abstraction/Knowledges/Models/UploadKnowledgeResponse.cs

@@ -0,0 +1,10 @@
+namespace BotSharp.Abstraction.Knowledges.Models;
+
+public class UploadKnowledgeResponse
+{
+    [JsonPropertyName("success")]
+    public IEnumerable<string> Success { get; set; } = new List<string>();
+
+    [JsonPropertyName("failed")]
+    public IEnumerable<string> Failed { get; set; } = new List<string>();
+}

src/Infrastructure/BotSharp.Abstraction/Utilities/StringExtensions.cs

@@ -44,6 +44,13 @@ public static class StringExtensions
         return str1.Equals(str2, option);
     }
 
+    public static string RemoveWhiteSpaces(this string? str)
+    {
+        if (string.IsNullOrWhiteSpace(str)) return string.Empty;
+
+        return str.Replace(" ", "").Replace("\t", "").Replace("\n", "").Replace("\r", "");
+    }
+
     public static string JsonContent(this string text)
     {
         var m = Regex.Match(text, @"\{(?:[^{}]|(?<open>\{)|(?<-open>\}))+(?(open)(?!))\}");

src/Infrastructure/BotSharp.Core/Files/Services/Storage/LocalFileStorageService.Knowledge.cs

@@ -0,0 +1,34 @@
+using System.IO;
+
+namespace BotSharp.Core.Files.Services;
+
+public partial class LocalFileStorageService
+{
+    public bool SaveKnowledgeFiles(string collectionName, string fileId, string fileName, Stream stream)
+    {
+        if (string.IsNullOrWhiteSpace(collectionName) || string.IsNullOrWhiteSpace(fileId))
+        {
+            return false;
+        }
+
+        try
+        {
+            var dir = Path.Combine(_baseDir, KNOWLEDGE_FOLDER, KNOWLEDGE_DOC_FOLDER, collectionName, fileId);
+            if (ExistDirectory(dir))
+            {
+                Directory.Delete(dir);
+            }
+            Directory.CreateDirectory(dir);
+
+            var filePath = Path.Combine(dir, fileName);
+            using var fs = File.Create(filePath);
+            stream.CopyTo(fs);
+            return true;
+        }
+        catch (Exception ex)
+        {
+            _logger.LogWarning($"Error when saving knowledge file (Collection: {collectionName}, File name: {fileName}). {ex.Message}\r\n{ex.InnerException}");
+            return false;
+        }
+    }
+}

src/Infrastructure/BotSharp.Core/Files/Services/Storage/LocalFileStorageService.cs

@@ -19,6 +19,8 @@ public partial class LocalFileStorageService : IFileStorageService
     private const string USER_AVATAR_FOLDER = "avatar";
     private const string SESSION_FOLDER = "sessions";
     private const string TEXT_TO_SPEECH_FOLDER = "speeches";
+    private const string KNOWLEDGE_FOLDER = "knowledgebase";
+    private const string KNOWLEDGE_DOC_FOLDER = "document";
 
     public LocalFileStorageService(
         BotSharpDatabaseSettings dbSettings,

src/Infrastructure/BotSharp.Core/Repository/FileRepository/FileRepository.cs

@@ -40,7 +40,7 @@ public partial class FileRepository : IBotSharpRepository
     private const string AGENT_RESPONSES_FOLDER = "responses";
     private const string AGENT_TASKS_FOLDER = "tasks";
     private const string USERS_FOLDER = "users";
-    private const string KNOWLEDGE_FOLDER = "knowledge";
+    private const string KNOWLEDGE_FOLDER = "knowledgebase";
     private const string VECTOR_FOLDER = "vector";
     private const string COLLECTION_CONFIG_FILE = "collection-config.json";
 

src/Infrastructure/BotSharp.OpenAPI/Controllers/ConversationController.cs

@@ -366,8 +366,7 @@ public class ConversationController : ControllerBase
 
     #region Files and attachments
     [HttpPost("/conversation/{conversationId}/attachments")]
-    public IActionResult UploadAttachments([FromRoute] string conversationId,
-        IFormFile[] files)
+    public IActionResult UploadAttachments([FromRoute] string conversationId, IFormFile[] files)
     {
         if (files != null && files.Length > 0)
         {

src/Infrastructure/BotSharp.OpenAPI/Controllers/KnowledgeBaseController.cs

@@ -99,29 +99,6 @@ public class KnowledgeBaseController : ControllerBase
     {
         return await _knowledgeService.DeleteVectorCollectionData(collection, id);
     }
-
-    [HttpPost("/knowledge/vector/{collection}/upload")]
-    public async Task<IActionResult> UploadVectorKnowledge([FromRoute] string collection, IFormFile file, [FromForm] int? startPageNum, [FromForm] int? endPageNum)
-    {
-        var setttings = _services.GetRequiredService<FileCoreSettings>();
-        var textConverter = _services.GetServices<IPdf2TextConverter>().FirstOrDefault(x => x.Provider == setttings.Pdf2TextConverter.Provider);
-
-        var filePath = Path.GetTempFileName();
-        using (var stream = new FileStream(filePath, FileMode.Create, FileAccess.Write, FileShare.None))
-        {
-            await file.CopyToAsync(stream);
-            await stream.FlushAsync();
-        }
-
-        var content = await textConverter.ConvertPdfToText(filePath, startPageNum, endPageNum);
-        await _knowledgeService.FeedVectorKnowledge(collection, new KnowledgeCreationModel
-        {
-            Content = content
-        });
-
-        System.IO.File.Delete(filePath);
-        return Ok(new { count = 1, file.Length });
-    }
     #endregion
 
 
@@ -144,7 +121,35 @@ public class KnowledgeBaseController : ControllerBase
 
 
     #region Document
+    //[HttpPost("/knowledge/vector/{collection}/upload")]
+    //public async Task<IActionResult> UploadVectorKnowledge([FromRoute] string collection, IFormFile file, [FromForm] int? startPageNum, [FromForm] int? endPageNum)
+    //{
+    //    var setttings = _services.GetRequiredService<FileCoreSettings>();
+    //    var textConverter = _services.GetServices<IPdf2TextConverter>().FirstOrDefault(x => x.Provider == setttings.Pdf2TextConverter.Provider);
 
+    //    var filePath = Path.GetTempFileName();
+    //    using (var stream = new FileStream(filePath, FileMode.Create, FileAccess.Write, FileShare.None))
+    //    {
+    //        await file.CopyToAsync(stream);
+    //        await stream.FlushAsync();
+    //    }
+
+    //    var content = await textConverter.ConvertPdfToText(filePath, startPageNum, endPageNum);
+    //    await _knowledgeService.FeedVectorKnowledge(collection, new KnowledgeCreationModel
+    //    {
+    //        Content = content
+    //    });
+
+    //    System.IO.File.Delete(filePath);
+    //    return Ok(new { count = 1, file.Length });
+    //}
+
+    [HttpPost("/knowledge/vector/{collection}/upload")]
+    public async Task<UploadKnowledgeResponse> UploadVectorKnowledge([FromRoute] string collection, [FromBody] VectorKnowledgeUploadRequest request)
+    {
+        var response = await _knowledgeService.UploadVectorKnowledge(collection, request.Files);
+        return response;
+    }
     #endregion
 
     #region Common

src/Infrastructure/BotSharp.OpenAPI/ViewModels/Knowledges/VectorKnowledgeUploadRequest.cs

@@ -0,0 +1,6 @@
+namespace BotSharp.OpenAPI.ViewModels.Knowledges;
+
+public class VectorKnowledgeUploadRequest
+{
+    public IEnumerable<InputFileModel> Files { get; set; } = new List<InputFileModel>();
+}

src/Plugins/BotSharp.Plugin.AudioHandler/Provider/NativeWhisperProvider.cs

@@ -14,7 +14,7 @@ public class NativeWhisperProvider : IAudioCompletion
     private readonly IFileStorageService _fileStorage;
     private readonly ILogger<NativeWhisperProvider> _logger;
 
-    public string Provider => "native";
+    public string Provider => "native-whisper";
 
     public NativeWhisperProvider(
         BotSharpDatabaseSettings dbSettings,

src/Plugins/BotSharp.Plugin.KnowledgeBase/Services/KnowledgeService.Document.cs

@@ -1,7 +1,89 @@
+using BotSharp.Abstraction.Files;
+using BotSharp.Abstraction.Files.Models;
+using BotSharp.Abstraction.Files.Utilities;
+
 namespace BotSharp.Plugin.KnowledgeBase.Services;
 
 public partial class KnowledgeService
 {
+    public async Task<UploadKnowledgeResponse> UploadVectorKnowledge(string collectionName, IEnumerable<InputFileModel> files)
+    {
+        if (string.IsNullOrWhiteSpace(collectionName))
+        {
+            return new UploadKnowledgeResponse
+            {
+                Success = [],
+                Failed = files.Select(x => x.FileName)
+            };
+        }
+
+        var fileStoreage = _services.GetRequiredService<IFileStorageService>();
+        var cleanCollectionName = collectionName.RemoveWhiteSpaces();
+        var successFiles = new List<string>();
+        var failedFiles = new List<string>();
+
+        foreach (var file in files)
+        {
+            if (string.IsNullOrWhiteSpace(file.FileData) || string.IsNullOrWhiteSpace(file.FileName))
+            {
+                continue;
+            }
+
+            var dataIds = new List<string>();
+
+            try
+            {
+                // Chop text
+                var (contentType, bytes) = FileUtility.GetFileInfoFromData(file.FileData);
+                using var stream = new MemoryStream(bytes);
+                using var reader = new StreamReader(stream);
+                var content = await reader.ReadToEndAsync();
+
+                // Save file
+                var fileId = Guid.NewGuid().ToString();
+                var saved = fileStoreage.SaveKnowledgeFiles(cleanCollectionName, fileId, file.FileName, stream);
+                reader.Close();
+                stream.Close();
+
+                if (!saved)
+                {
+                    failedFiles.Add(file.FileName);
+                    continue;
+                }
+
+                // Text embedding
+                var vectorDb = GetVectorDb();
+                var textEmbedding = GetTextEmbedding(collectionName);
+                var vector = await textEmbedding.GetVectorAsync(content);
+
+                // Save to vector db
+                var dataId = Guid.NewGuid();
+                await vectorDb.Upsert(collectionName, dataId, vector, content, new Dictionary<string, string>
+                {
+                    { "fileName", file.FileName },
+                    { "fileId", fileId },
+                    { "page", "0" }
+                });
+
+                dataIds.Add(dataId.ToString());
+                successFiles.Add(file.FileName);
+            }
+            catch (Exception ex)
+            {
+                _logger.LogError($"Error when processing knowledge file ({file.FileName}). {ex.Message}\r\n{ex.InnerException}");
+                failedFiles.Add(file.FileName);
+                continue;
+            }
+        }
+
+        return new UploadKnowledgeResponse
+        {
+            Success = successFiles,
+            Failed = failedFiles
+        };
+    }
+
+
     public async Task FeedVectorKnowledge(string collectionName, KnowledgeCreationModel knowledge)
     {
         var index = 0;
@@ -24,4 +106,7 @@ public partial class KnowledgeService
             Console.WriteLine($"Saved vector {index}/{lines.Count}: {line}\n");
         }
     }
+
+    #region Private methods
+    #endregion
 }

src/WebStarter/appsettings.json

@@ -141,7 +141,8 @@
       "Enable": true,
       "BatchSize": 50,
       "MessageLimit": 2,
-      "BufferHours": 12
+      "BufferHours": 12,
+      "ExcludeAgentIds": []
     },
     "RateLimit": {
       "MaxConversationPerDay": 100,