Features/add file handler

BotSharp.sln

@@ -97,6 +97,8 @@ Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "BotSharp.Plugin.OpenAI", "s
 EndProject
 Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "BotSharp.Plugin.EmailHandler", "src\Plugins\BotSharp.Plugin.EmailHandler\BotSharp.Plugin.EmailHandler.csproj", "{A72B3BEB-E14B-4917-BE44-97EAE4E122D2}"
 EndProject
+Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "BotSharp.Plugin.FileHandler", "src\Plugins\BotSharp.Plugin.FileHandler\BotSharp.Plugin.FileHandler.csproj", "{D6A99D4F-6248-419E-8A43-B38ADEBABA2C}"
+EndProject
 Global
 	GlobalSection(SolutionConfigurationPlatforms) = preSolution
 		Debug|Any CPU = Debug|Any CPU
@@ -393,6 +395,14 @@ Global
 		{A72B3BEB-E14B-4917-BE44-97EAE4E122D2}.Release|Any CPU.Build.0 = Release|Any CPU
 		{A72B3BEB-E14B-4917-BE44-97EAE4E122D2}.Release|x64.ActiveCfg = Release|Any CPU
 		{A72B3BEB-E14B-4917-BE44-97EAE4E122D2}.Release|x64.Build.0 = Release|Any CPU
+		{D6A99D4F-6248-419E-8A43-B38ADEBABA2C}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
+		{D6A99D4F-6248-419E-8A43-B38ADEBABA2C}.Debug|Any CPU.Build.0 = Debug|Any CPU
+		{D6A99D4F-6248-419E-8A43-B38ADEBABA2C}.Debug|x64.ActiveCfg = Debug|Any CPU
+		{D6A99D4F-6248-419E-8A43-B38ADEBABA2C}.Debug|x64.Build.0 = Debug|Any CPU
+		{D6A99D4F-6248-419E-8A43-B38ADEBABA2C}.Release|Any CPU.ActiveCfg = Release|Any CPU
+		{D6A99D4F-6248-419E-8A43-B38ADEBABA2C}.Release|Any CPU.Build.0 = Release|Any CPU
+		{D6A99D4F-6248-419E-8A43-B38ADEBABA2C}.Release|x64.ActiveCfg = Release|Any CPU
+		{D6A99D4F-6248-419E-8A43-B38ADEBABA2C}.Release|x64.Build.0 = Release|Any CPU
 	EndGlobalSection
 	GlobalSection(SolutionProperties) = preSolution
 		HideSolutionNode = FALSE
@@ -439,6 +449,7 @@ Global
 		{806A0B0E-FEFF-420E-B5B2-C9FCBF890A8C} = {D5293208-2BEF-42FC-A64C-5954F61720BA}
 		{6507D336-3A4D-41D4-81C0-2B900173A5FE} = {D5293208-2BEF-42FC-A64C-5954F61720BA}
 		{A72B3BEB-E14B-4917-BE44-97EAE4E122D2} = {51AFE054-AE99-497D-A593-69BAEFB5106F}
+		{D6A99D4F-6248-419E-8A43-B38ADEBABA2C} = {51AFE054-AE99-497D-A593-69BAEFB5106F}
 	EndGlobalSection
 	GlobalSection(ExtensibilityGlobals) = postSolution
 		SolutionGuid = {A9969D89-C98B-40A5-A12B-FC87E55B3A19}

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

@@ -3,7 +3,9 @@ namespace BotSharp.Abstraction.Files;
 public interface IBotSharpFileService
 {
     string GetDirectory(string conversationId);
-    Task<IEnumerable<MessageFileModel>> GetChatImages(string conversationId, string source, IEnumerable<string> fileTypes, List<RoleDialogModel> conversations, int? offset = null);
+    Task<IEnumerable<MessageFileModel>> GetChatImages(string conversationId, string source,
+        IEnumerable<RoleDialogModel> conversations, IEnumerable<string> contentTypes,
+        bool includeScreenShot = false, int? offset = null);
     IEnumerable<MessageFileModel> GetMessageFiles(string conversationId, IEnumerable<string> messageIds, string source, bool imageOnly = false);
     string GetMessageFile(string conversationId, string messageId, string source, string index, string fileName);
     IEnumerable<MessageFileModel> GetMessagesWithFile(string conversationId, IEnumerable<string> messageIds);

src/Infrastructure/BotSharp.Core/BotSharp.Core.csproj

@@ -48,10 +48,8 @@
   <ItemGroup>
     <None Remove="data\agents\6745151e-6d46-4a02-8de4-1c4f21c7da95\agent.json" />
     <None Remove="data\agents\6745151e-6d46-4a02-8de4-1c4f21c7da95\functions\generate_image.json" />
-    <None Remove="data\agents\6745151e-6d46-4a02-8de4-1c4f21c7da95\functions\read_file.json" />
     <None Remove="data\agents\6745151e-6d46-4a02-8de4-1c4f21c7da95\instruction.liquid" />
     <None Remove="data\agents\6745151e-6d46-4a02-8de4-1c4f21c7da95\templates\generate_image.fn.liquid" />
-    <None Remove="data\agents\6745151e-6d46-4a02-8de4-1c4f21c7da95\templates\read_file.fn.liquid" />
     <None Remove="data\agents\01dcc3e5-0af7-49e6-ad7a-a760bd12dc4b\agent.json" />
     <None Remove="data\agents\01dcc3e5-0af7-49e6-ad7a-a760bd12dc4b\functions.json" />
     <None Remove="data\agents\01dcc3e5-0af7-49e6-ad7a-a760bd12dc4b\functions\human_intervention_needed.json" />
@@ -159,12 +157,6 @@
     <Content Include="data\agents\6745151e-6d46-4a02-8de4-1c4f21c7da95\instruction.liquid">
       <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
     </Content>
-    <Content Include="data\agents\6745151e-6d46-4a02-8de4-1c4f21c7da95\functions\read_file.json">
-      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
-    </Content>
-    <Content Include="data\agents\6745151e-6d46-4a02-8de4-1c4f21c7da95\templates\read_file.fn.liquid">
-      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
-    </Content>
     <Content Include="data\agents\6745151e-6d46-4a02-8de4-1c4f21c7da95\functions\generate_image.json">
       <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
     </Content>

src/Infrastructure/BotSharp.Core/Files/FilePlugin.cs

@@ -17,8 +17,6 @@ public void RegisterDI(IServiceCollection services, IConfiguration config)
     {
         services.AddScoped<IBotSharpFileService, BotSharpFileService>();
 
-        services.AddScoped<IAgentHook, FileReaderHook>();
-        services.AddScoped<IAgentUtilityHook, FileReaderUtilityHook>();
         services.AddScoped<IAgentHook, ImageGeneratorHook>();
         services.AddScoped<IAgentUtilityHook, ImageGeneratorUtilityHook>();
     }

src/Infrastructure/BotSharp.Core/Files/Services/BotSharpFileService.Conversation.cs

@@ -1,124 +1,43 @@
-using AspectInjector.Broker;
 using BotSharp.Abstraction.Files.Converters;
 using Microsoft.EntityFrameworkCore;
 using System.IO;
-using System.Threading;
 
 namespace BotSharp.Core.Files.Services;
 
 public partial class BotSharpFileService
 {
-    public async Task<IEnumerable<MessageFileModel>> GetChatImages(string conversationId, string source, IEnumerable<string> fileTypes,
-        List<RoleDialogModel> conversations, int? offset = null)
+    public async Task<IEnumerable<MessageFileModel>> GetChatImages(string conversationId, string source,
+        IEnumerable<RoleDialogModel> conversations, IEnumerable<string> contentTypes,
+        bool includeScreenShot = false, int? offset = null)
     {
         var files = new List<MessageFileModel>();
         if (string.IsNullOrEmpty(conversationId) || conversations.IsNullOrEmpty())
         {
-            return new List<MessageFileModel>();
+            return files;
         }
 
-        if (offset <= 0)
-        {
-            offset = MIN_OFFSET;
-        }
-        else if (offset > MAX_OFFSET)
-        {
-            offset = MAX_OFFSET;
-        }
+        var messageIds = GetMessageIds(conversations, offset);
+        var pathPrefix = Path.Combine(_baseDir, CONVERSATION_FOLDER, conversationId, FILE_FOLDER);
 
-        var messageIds = new List<string>();
-        if (offset.HasValue)
-        {
-            messageIds = conversations.Select(x => x.MessageId).Distinct().TakeLast(offset.Value).ToList();
-        }
-        else
-        {
-            messageIds = conversations.Select(x => x.MessageId).Distinct().ToList();
-        }
-
-        files = await GetMessageFiles(conversationId, messageIds, source, fileTypes);
-        return files;
-    }
-
-    private async Task<List<MessageFileModel>> GetMessageFiles(string conversationId, IEnumerable<string> messageIds, string source, IEnumerable<string> fileTypes)
-    {
-        var files = new List<MessageFileModel>();
-        if (string.IsNullOrEmpty(conversationId) || messageIds.IsNullOrEmpty() || fileTypes.IsNullOrEmpty()) return files;
-
-        var isNeedScreenShot = fileTypes.Any(x => _allowScreenShotTypes.Contains(x));
-        var onlyScreenShot = fileTypes.All(x => _allowScreenShotTypes.Contains(x));
-
-        try
+        foreach (var messageId in messageIds)
         {
-            var preFixPath = Path.Combine(_baseDir, CONVERSATION_FOLDER, conversationId, FILE_FOLDER);
+            var dir = Path.Combine(pathPrefix, messageId, source);
+            if (!ExistDirectory(dir)) continue;
 
-            foreach (var messageId in messageIds)
+            foreach (var subDir in Directory.GetDirectories(dir))
             {
-                var dir = Path.Combine(preFixPath, messageId, source);
-                if (!ExistDirectory(dir)) continue;
+                var file = Directory.GetFiles(subDir).FirstOrDefault();
+                if (file == null) continue;
 
-                foreach (var subDir in Directory.GetDirectories(dir))
-                {
-                    var file = Directory.GetFiles(subDir).FirstOrDefault();
-                    if (file == null) continue;
+                var contentType = GetFileContentType(file);
+                if (contentTypes?.Contains(contentType) != true) continue;
 
-                    var index = subDir.Split(Path.DirectorySeparatorChar).Last();
-                    var contentType = GetFileContentType(file);
+                var foundFiles = await GetMessageFiles(file, subDir, contentType, messageId, source, includeScreenShot);
+                if (foundFiles.IsNullOrEmpty()) continue;
 
-                    if ((!isNeedScreenShot || (isNeedScreenShot && !onlyScreenShot)) && _allowedImageTypes.Contains(contentType))
-                    {
-                        var model = new MessageFileModel()
-                        {
-                            MessageId = messageId,
-                            FileStorageUrl = file,
-                            ContentType = contentType
-                        };
-                        files.Add(model);
-                    }
-                    else if ((isNeedScreenShot && !onlyScreenShot || onlyScreenShot) && !_allowedImageTypes.Contains(contentType))
-                    {
-                        var screenShotDir = Path.Combine(subDir, SCREENSHOT_FILE_FOLDER);
-                        if (ExistDirectory(screenShotDir) && Directory.GetFiles(screenShotDir).Any())
-                        {
-                            foreach (var screenShot in Directory.GetFiles(screenShotDir))
-                            {
-                                contentType = GetFileContentType(screenShot);
-                                if (!_allowedImageTypes.Contains(contentType)) continue;
-
-                                var model = new MessageFileModel()
-                                {
-                                    MessageId = messageId,
-                                    FileStorageUrl = screenShot,
-                                    ContentType = contentType
-                                };
-                                files.Add(model);
-                            }
-                        }
-                        else
-                        {
-                            var screenShotPath = Path.Combine(subDir, SCREENSHOT_FILE_FOLDER);
-                            var images = await ConvertPdfToImages(file, screenShotPath);
-
-                            foreach (var image in images)
-                            {
-                                contentType = GetFileContentType(image);
-                                var model = new MessageFileModel()
-                                {
-                                    MessageId = messageId,
-                                    FileStorageUrl = image,
-                                    ContentType = contentType
-                                };
-                                files.Add(model);
-                            }
-                        }
-                    }
-                }
+                files.AddRange(foundFiles);
             }
         }
-        catch (Exception ex)
-        {
-            _logger.LogWarning($"Error when reading conversation ({conversationId}) files: {ex.Message}\r\n{ex.InnerException}\r\n{ex.StackTrace}");
-        }
 
         return files;
     }
@@ -144,7 +63,7 @@ public IEnumerable<MessageFileModel> GetMessageFiles(string conversationId, IEnu
                 foreach (var file in Directory.GetFiles(subDir))
                 {
                     var contentType = GetFileContentType(file);
-                    if (imageOnly && !_allowedImageTypes.Contains(contentType))
+                    if (imageOnly && !_imageTypes.Contains(contentType))
                     {
                         continue;
                     }
@@ -327,6 +246,102 @@ private string GetConversationFileDirectory(string? conversationId, string? mess
         return dir;
     }
 
+    private IEnumerable<string> GetMessageIds(IEnumerable<RoleDialogModel> conversations, int? offset = null)
+    {
+        if (conversations.IsNullOrEmpty()) return Enumerable.Empty<string>();
+
+        if (offset <= 0)
+        {
+            offset = MIN_OFFSET;
+        }
+        else if (offset > MAX_OFFSET)
+        {
+            offset = MAX_OFFSET;
+        }
+
+        var messageIds = new List<string>();
+        if (offset.HasValue)
+        {
+            messageIds = conversations.Select(x => x.MessageId).Distinct().TakeLast(offset.Value).ToList();
+        }
+        else
+        {
+            messageIds = conversations.Select(x => x.MessageId).Distinct().ToList();
+        }
+
+        return messageIds;
+    }
+
+
+    private async Task<IEnumerable<MessageFileModel>> GetMessageFiles(string file, string fileDir, string contentType,
+        string messageId, string source, bool includeScreenShot)
+    {
+        var files = new List<MessageFileModel>();
+
+        try
+        {
+            if (!_imageTypes.Contains(contentType) && includeScreenShot)
+            {
+                var screenShotDir = Path.Combine(fileDir, SCREENSHOT_FILE_FOLDER);
+                if (ExistDirectory(screenShotDir) && !Directory.GetFiles(screenShotDir).IsNullOrEmpty())
+                {
+                    foreach (var screenShot in Directory.GetFiles(screenShotDir))
+                    {
+                        contentType = GetFileContentType(screenShot);
+                        if (!_imageTypes.Contains(contentType)) continue;
+
+                        var model = new MessageFileModel()
+                        {
+                            MessageId = messageId,
+                            FileName = Path.GetFileName(screenShot),
+                            FileStorageUrl = screenShot,
+                            ContentType = contentType,
+                            FileSource = source
+                        };
+                        files.Add(model);
+                    }
+                }
+                else if (contentType == MediaTypeNames.Application.Pdf)
+                {
+                    var images = await ConvertPdfToImages(file, screenShotDir);
+                    foreach (var image in images)
+                    {
+                        contentType = GetFileContentType(image);
+                        var model = new MessageFileModel()
+                        {
+                            MessageId = messageId,
+                            FileName = Path.GetFileName(image),
+                            FileStorageUrl = image,
+                            ContentType = contentType,
+                            FileSource = source
+                        };
+                        files.Add(model);
+                    }
+                }
+            }
+            else
+            {
+                var model = new MessageFileModel()
+                {
+                    MessageId = messageId,
+                    FileName = Path.GetFileName(file),
+                    FileStorageUrl = file,
+                    ContentType = contentType,
+                    FileSource = source
+                };
+                files.Add(model);
+            }
+
+            return files;
+        }
+        catch (Exception ex)
+        {
+            _logger.LogWarning($"Error when getting message files {file} (messageId: {messageId}), Error: {ex.Message}\r\n{ex.InnerException}");
+            return files;
+        }
+    }
+
+
     private async Task<IEnumerable<string>> ConvertPdfToImages(string pdfLoc, string imageLoc)
     {
         var converters = _services.GetServices<IPdf2ImageConverter>();