Перепиливание под чистый DDD

This commit is contained in:
Халимов Рустам
2026-03-22 23:59:33 +03:00
parent 5da1a2f45d
commit 6e532b021d
302 changed files with 3595 additions and 3679 deletions
@@ -0,0 +1,97 @@
using Knot.Modules.Conversations.Domain;
using Knot.Modules.Messaging.Domain;
using System;
using System.Collections.Concurrent;
using System.Collections.Generic;
using System.IO;
using System.IO.Compression;
using System.Linq;
using System.Threading;
using System.Threading.Tasks;
using AngleSharp.Html.Parser;
using AngleSharp.Dom;
using Knot.Shared.Kernel;
using MediatR;
namespace Knot.Modules.TelegramImport.Application.TelegramImport;
public static class TelegramImportState
{
public static readonly ConcurrentDictionary<Guid, string> TempZips = new();
}
public record AnalyzeImportResponseDto(Guid Token, List<string> Names);
public record AnalyzeImportCommand(Stream FileStream, string FileName) : ICommand<AnalyzeImportResponseDto>;
internal sealed class AnalyzeImportCommandHandler : ICommandHandler<AnalyzeImportCommand, AnalyzeImportResponseDto>
{
public async Task<Result<AnalyzeImportResponseDto>> Handle(AnalyzeImportCommand request, CancellationToken cancellationToken)
{
if (request.FileStream == null || request.FileStream.Length == 0)
{
return Result.Failure<AnalyzeImportResponseDto>(ChatErrors.FileEmpty);
}
if (!request.FileName.EndsWith(".zip", StringComparison.OrdinalIgnoreCase))
{
return Result.Failure<AnalyzeImportResponseDto>(ChatErrors.FileInvalidExtension);
}
var token = Guid.NewGuid();
var tempPath = Path.Combine(Path.GetTempPath(), $"{token}.zip");
await using (var fs = new FileStream(tempPath, FileMode.Create))
{
await request.FileStream.CopyToAsync(fs, cancellationToken);
}
var names = new HashSet<string>();
using (var archive = ZipFile.OpenRead(tempPath))
{
var htmlEntries = archive.Entries
.Where(e => e.FullName.EndsWith(".html", StringComparison.OrdinalIgnoreCase) && e.Name.StartsWith("messages", StringComparison.OrdinalIgnoreCase));
foreach (var entry in htmlEntries)
{
using var stream = entry.Open();
var parser = new HtmlParser();
var doc = parser.ParseDocument(stream);
var messageNodes = doc.QuerySelectorAll(".message");
if (messageNodes == null)
{
continue;
}
foreach (var node in messageNodes)
{
var fromNameNode = node.QuerySelector(".from_name");
if (fromNameNode != null)
{
var nameNodeText = (IElement)fromNameNode.Clone();
var innerSpans = nameNodeText.QuerySelectorAll("span");
foreach (var span in innerSpans)
{
span.Remove();
}
var name = nameNodeText.TextContent.Trim();
if (!string.IsNullOrWhiteSpace(name))
{
names.Add(name);
}
}
}
}
}
TelegramImportState.TempZips[token] = tempPath;
return Result.Success(new AnalyzeImportResponseDto(token, names.ToList()));
}
}
@@ -0,0 +1,12 @@
using System;
using System.Collections.Generic;
namespace Knot.Modules.TelegramImport.Application.TelegramImport.DTOs;
public record AnalyzeImportResponseDto(
Guid Token,
List<string> Names
);
@@ -0,0 +1,13 @@
using System;
using System.Collections.Generic;
namespace Knot.Modules.TelegramImport.Application.TelegramImport.DTOs;
public record ExecuteImportRequest(
Guid Token,
Dictionary<string, Guid> Mapping,
string? GroupName
);
@@ -0,0 +1,12 @@
using System;
namespace Knot.Modules.TelegramImport.Application.TelegramImport.DTOs;
public record ExecuteImportResponseDto(
bool Success,
int MessagesImported,
Guid ChatId
);
@@ -0,0 +1,500 @@
using System;
using System.Collections.Generic;
using System.IO;
using System.IO.Compression;
using System.Linq;
using System.Threading;
using System.Threading.Tasks;
using AngleSharp.Dom;
using AngleSharp.Html.Parser;
using MediatR;
using Knot.Shared.Kernel;
using Knot.Shared.Kernel.Storage;
using Knot.Modules.Messaging.Domain;
using Knot.Modules.Messaging.Application.Abstractions;
using Knot.Modules.Conversations.Application.Chats.Create;
using Knot.Modules.Conversations.Infrastructure.SignalR;
using Microsoft.AspNetCore.SignalR;
using Knot.Modules.Conversations.Domain;
using Knot.Modules.Conversations.Application.Abstractions;
using Knot.Modules.Messaging.Application.Abstractions;
namespace Knot.Modules.TelegramImport.Application.TelegramImport;
public record ExecuteImportResponseDto(bool Success, int MessagesImported, Guid ChatId);
public record ExecuteImportCommand(
Guid CurrentUserId,
Guid Token,
Dictionary<string, Guid> Mapping,
string? GroupName
) : ICommand<ExecuteImportResponseDto>;
internal sealed class ExecuteImportCommandHandler : ICommandHandler<ExecuteImportCommand, ExecuteImportResponseDto>
{
private readonly ISender _sender;
private readonly IChatsUnitOfWork _uow;
private readonly IChatRepository _chatRepository;
private readonly IMessageRepository _messageRepository;
private readonly IFileStorageService _fileStorage;
private readonly IHubContext<ChatHub> _hubContext;
private readonly IMessageReactionRepository _reactionRepository;
public ExecuteImportCommandHandler(
ISender sender,
IChatsUnitOfWork uow,
IChatRepository chatRepository,
IMessageRepository messageRepository,
IFileStorageService fileStorage,
IHubContext<ChatHub> hubContext,
IMessageReactionRepository reactionRepository)
{
_sender = sender;
_uow = uow;
_chatRepository = chatRepository;
_messageRepository = messageRepository;
_fileStorage = fileStorage;
_hubContext = hubContext;
_reactionRepository = reactionRepository;
}
public async Task<Result<ExecuteImportResponseDto>> Handle(ExecuteImportCommand request, CancellationToken cancellationToken)
{
if (!TelegramImportState.TempZips.TryGetValue(request.Token, out var tempPath))
{
return Result.Failure<ExecuteImportResponseDto>(ChatErrors.ImportExpired);
}
if (!System.IO.File.Exists(tempPath))
{
return Result.Failure<ExecuteImportResponseDto>(ChatErrors.ImportMissing);
}
var myId = request.CurrentUserId;
var targetUserIds = request.Mapping.Values.Distinct().Where(id => id != Guid.Empty).ToList();
if (!targetUserIds.Contains(myId))
{
targetUserIds.Add(myId);
}
Guid chatId = Guid.Empty;
var chatMembers = targetUserIds;
if (chatMembers.Count <= 2)
{
var existingChats = await _chatRepository.GetUserChatsAsync(myId, cancellationToken);
var personalChat = existingChats.FirstOrDefault(c => c.Type == ChatType.Personal && c.Members.All(m => chatMembers.Contains(m.UserId)) && c.Members.Count == chatMembers.Count);
if (personalChat != null)
{
chatId = personalChat.Id;
}
else
{
var friendId = chatMembers.FirstOrDefault(id => id != myId);
if (friendId == Guid.Empty)
{
friendId = myId;
}
var command = new CreateChatCommand(string.Empty, ChatType.Personal, new List<Guid> { myId, friendId });
var res = await _sender.Send(command, cancellationToken);
if (res.IsFailure)
{
return Result.Failure<ExecuteImportResponseDto>(ChatErrors.ImportCreateChatFailed(res.Error.Description ?? res.Error.Code));
}
chatId = res.Value;
}
}
else
{
var command = new CreateChatCommand(request.GroupName ?? "Импортированный чат", ChatType.Group, chatMembers);
var res = await _sender.Send(command, cancellationToken);
if (res.IsFailure)
{
return Result.Failure<ExecuteImportResponseDto>(ChatErrors.ImportCreateChatFailed(res.Error.Description ?? res.Error.Code));
}
chatId = res.Value;
}
int importedCount = 0;
using (var archive = ZipFile.OpenRead(tempPath))
{
var htmlEntries = archive.Entries
.Where(e => e.FullName.EndsWith(".html", StringComparison.OrdinalIgnoreCase) && e.Name.StartsWith("messages", StringComparison.OrdinalIgnoreCase))
.OrderBy(e =>
{
var name = e.Name.ToLower().Replace("messages", "").Replace(".html", "");
return string.IsNullOrEmpty(name) ? 0 : int.TryParse(name, out var num) ? num : 999999;
})
.ToList();
Guid lastSenderGuid = myId;
DateTime lastCreatedAt = DateTime.UtcNow;
Dictionary<string, Guid> messageIdMap = new();
Message? lastSavedMessage = null;
foreach (var entry in htmlEntries)
{
using var stream = entry.Open();
var parser = new HtmlParser();
var doc = parser.ParseDocument(stream);
var messageNodes = doc.QuerySelectorAll(".message");
if (messageNodes == null)
{
continue;
}
var baseDir = Path.GetDirectoryName(entry.FullName)?.Replace("\\", "/") ?? "";
if (!string.IsNullOrEmpty(baseDir) && !baseDir.EndsWith("/"))
{
baseDir += "/";
}
foreach (var node in messageNodes)
{
try
{
var fromNameNode = node.QuerySelector(".from_name");
var textNode = node.QuerySelector(".text");
var dateNode = node.QuerySelector(".date[title]")
?? node.QuerySelector(".pull_right[title]")
?? node.QuerySelector("[title]");
if (fromNameNode != null)
{
var nameNodeText = (AngleSharp.Dom.IElement)fromNameNode.Clone();
var innerSpans = nameNodeText.QuerySelectorAll("span");
foreach (var span in innerSpans)
{
span.Remove();
}
var name = nameNodeText.TextContent.Trim();
if (request.Mapping.TryGetValue(name, out var mappedId) && mappedId != Guid.Empty)
{
lastSenderGuid = mappedId;
}
else
{
lastSenderGuid = myId;
}
}
Guid senderGuid = lastSenderGuid;
string content = "";
var mainBodyNode = node.QuerySelector(".body");
var isForwarded = node.QuerySelector(".forwarded") != null;
var contentTextNode = isForwarded
? (node.QuerySelector(".body > .text") ?? node.QuerySelector(".text:not(.forwarded .text)"))
: node.QuerySelector(".text");
if (contentTextNode != null)
{
var html = contentTextNode.InnerHtml
.Replace("<br>", "\n", StringComparison.OrdinalIgnoreCase)
.Replace("<br/>", "\n", StringComparison.OrdinalIgnoreCase)
.Replace("<br />", "\n", StringComparison.OrdinalIgnoreCase);
var tempParser = new HtmlParser();
var tempDoc = tempParser.ParseDocument("<div>" + html + "</div>");
content = tempDoc.Body?.TextContent.Trim() ?? "";
}
DateTime createdAt = lastCreatedAt;
var titleNodes = node.QuerySelectorAll("[title]");
bool parsed = false;
if (titleNodes != null)
{
foreach (var tnode in titleNodes)
{
var dateStr = tnode.GetAttribute("title")?.Trim() ?? "";
if (dateStr.Length >= 10 && char.IsDigit(dateStr[0]) && char.IsDigit(dateStr[1]))
{
var cleanStr = dateStr.Replace("UTC", "", StringComparison.OrdinalIgnoreCase).Trim();
if (DateTimeOffset.TryParseExact(cleanStr, "dd.MM.yyyy HH:mm:ss zzz", System.Globalization.CultureInfo.InvariantCulture, System.Globalization.DateTimeStyles.None, out var dto))
{
createdAt = dto.UtcDateTime;
parsed = true;
break;
}
else if (DateTime.TryParseExact(cleanStr, "dd.MM.yyyy HH:mm:ss", System.Globalization.CultureInfo.InvariantCulture, System.Globalization.DateTimeStyles.AssumeUniversal | System.Globalization.DateTimeStyles.AdjustToUniversal, out var dt))
{
createdAt = dt;
parsed = true;
break;
}
else if (DateTime.TryParse(cleanStr, out var dFallback))
{
createdAt = dFallback.ToUniversalTime();
parsed = true;
break;
}
}
}
}
if (!parsed)
{
Console.WriteLine("Warning: Could not parse date in imported message! Using lastCreatedAt.");
}
else
{
lastCreatedAt = createdAt;
}
Guid? forwardedFromId = null;
var forwardedNode = node.QuerySelector(".forwarded.body");
if (forwardedNode != null)
{
var fwdNameNode = forwardedNode.QuerySelector(".from_name");
var fwdNameText = fwdNameNode != null ? (AngleSharp.Dom.IElement)fwdNameNode.Clone() : null;
if (fwdNameText != null)
{
var innerSpans = fwdNameText.QuerySelectorAll("span");
foreach (var s in innerSpans)
{
s.Remove();
}
}
var fwdName = fwdNameText != null ? fwdNameText.TextContent.Trim() : "Неизвестного";
if (request.Mapping.TryGetValue(fwdName, out var mappedFwdId) && mappedFwdId != Guid.Empty)
{
forwardedFromId = mappedFwdId;
}
else if (fwdName == "Это я" || fwdName == request.Mapping.FirstOrDefault(x => x.Value == myId).Key)
{
forwardedFromId = myId;
}
var fwdTextNode = forwardedNode.QuerySelector(".text");
string fwdContent = "";
if (fwdTextNode != null)
{
var fHtml = fwdTextNode.InnerHtml
.Replace("<br>", "\n", StringComparison.OrdinalIgnoreCase)
.Replace("<br/>", "\n", StringComparison.OrdinalIgnoreCase)
.Replace("<br />", "\n", StringComparison.OrdinalIgnoreCase);
var tempParser = new HtmlParser();
var tempDoc = tempParser.ParseDocument("<div>" + fHtml + "</div>");
fwdContent = tempDoc.Body?.TextContent.Trim() ?? "";
}
if (forwardedFromId == null)
{
content = string.IsNullOrEmpty(content)
? $"[Переслано от {fwdName}]:\n{fwdContent}"
: $"{content}\n\n[Переслано от {fwdName}]:\n{fwdContent}";
}
else if (string.IsNullOrEmpty(content))
{
content = fwdContent;
}
}
Guid? replyToId = null;
var replyNode = node.QuerySelector(".reply_to a");
if (replyNode != null)
{
var href = replyNode.GetAttribute("href");
if (href != null && href.StartsWith("#go_to_"))
{
var tgId = href.Substring("#go_to_".Length);
if (messageIdMap.TryGetValue(tgId, out var mappedMsgId))
{
replyToId = mappedMsgId;
}
}
}
var mediaNodes = node.QuerySelectorAll("a.photo_wrap, a.animated_wrap, video, audio, a.document, a.media_voice_message, a.media_video, img.sticker").ToList();
if (mediaNodes.Count == 0)
{
var fallback = node.QuerySelectorAll(".media_wrap a[href]");
mediaNodes.AddRange(fallback);
}
var messageType = "text";
(string mType, string cType) GetMediaTypes(string fileUrl)
{
var ext = Path.GetExtension(fileUrl)?.ToLower();
return ext switch
{
".jpg" or ".jpeg" or ".png" or ".webp" => ("image", "image/jpeg"),
".mp4" or ".mov" or ".avi" => ("video", "video/mp4"),
".ogg" or ".mp3" => ("voice", "audio/ogg"),
_ => ("file", "application/octet-stream")
};
}
if (mediaNodes != null && mediaNodes.Count > 0)
{
var firstHref = mediaNodes[0].GetAttribute("href") ?? mediaNodes[0].GetAttribute("src");
if (firstHref != null)
{
messageType = GetMediaTypes(firstHref).mType;
if (mediaNodes[0].ClassName?.Contains("animated") == true || firstHref.EndsWith(".mp4"))
{
if (mediaNodes[0].ClassName?.Contains("animated") == true)
{
messageType = "image";
}
}
}
}
bool isJoined = fromNameNode == null;
bool hasMedia = mediaNodes != null && mediaNodes.Count > 0;
Message? targetMessage = null;
// Check if we should combine this message with the previous one
// We combine if: it's joined AND it's just media/text within 60s AND same sender
// Even if it's forwarded, Telegram exports media groups as joined forwarded messages.
bool shouldCombine = isJoined && lastSavedMessage is MediaMessage
&& Math.Abs((createdAt - lastSavedMessage.CreatedAt).TotalSeconds) <= 60
&& lastSavedMessage.SenderId == senderGuid
&& lastSavedMessage.ForwardedFromId == forwardedFromId;
if (shouldCombine)
{
targetMessage = lastSavedMessage;
if (targetMessage is MediaMessage mm && !string.IsNullOrEmpty(content) && content != mm.Content)
{
// If the joined message has text (caption), append it
mm.AppendImportedCaption(content);
}
}
else
{
string finalContent = content;
if (hasMedia)
{
targetMessage = new MediaMessage(Guid.NewGuid(), chatId, senderGuid, MediaType.File, finalContent, replyToId, forwardedFromId, createdAt, true);
}
else
{
targetMessage = new TextMessage(Guid.NewGuid(), chatId, senderGuid, finalContent, replyToId, null, forwardedFromId, createdAt, true);
}
var idAttr = node.GetAttribute("id");
if (!string.IsNullOrEmpty(idAttr))
{
messageIdMap[idAttr] = targetMessage.Id;
}
}
if (hasMedia)
{
var seenMedia = new HashSet<string>();
var validMediaExtracted = new List<(string href, string finalMType, string cType)>();
foreach (var mediaNode in mediaNodes!)
{
string? href = mediaNode.GetAttribute("href") ?? mediaNode.GetAttribute("src");
if (!string.IsNullOrEmpty(href) && !href.StartsWith("http"))
{
if (!seenMedia.Add(href)) continue;
var types = GetMediaTypes(href);
var finalMType = types.mType;
if (mediaNode.ClassName?.Contains("animated") == true)
{
finalMType = "image";
}
validMediaExtracted.Add((href, finalMType, types.cType));
}
}
// В Telegram экспорте если в одном .message блоке есть и видео, и картинка - картинка это просто миниатюра (thumbnail).
// Реальные альбомы идут отдельными .message div'ами c классом joined.
// Поэтому мы просто удаляем картинку, чтобы она не дублировалась как отдельный файл в галерее!
if (validMediaExtracted.Any(m => m.finalMType == "video") && validMediaExtracted.Any(m => m.finalMType == "image"))
{
validMediaExtracted.RemoveAll(m => m.finalMType == "image");
}
foreach (var mediaTuple in validMediaExtracted)
{
var zipPath = baseDir + mediaTuple.href.Replace("\\", "/");
var zipEntry = archive.GetEntry(zipPath);
if (zipEntry != null)
{
using var ms = new MemoryStream();
using var zipfs = zipEntry.Open();
await zipfs.CopyToAsync(ms, cancellationToken);
ms.Position = 0;
var parsedType = Enum.TryParse<MediaType>(mediaTuple.finalMType, true, out var mTypeEnum) ? mTypeEnum : MediaType.File;
if (targetMessage is MediaMessage mm)
{
var fileId = await _fileStorage.UploadFileAsync(ms, Path.GetFileName(mediaTuple.href), mediaTuple.cType);
mm.AddMedia(parsedType, $"/api/files/{fileId}", Path.GetFileName(mediaTuple.href), zipEntry.Length);
}
}
}
}
var reactionNodes = node.QuerySelectorAll(".reactions .reaction");
foreach (var reactionNode in reactionNodes)
{
var emojiNode = reactionNode.QuerySelector(".emoji");
if (emojiNode == null)
{
continue;
}
var emoji = emojiNode.TextContent.Trim();
var userpicNodes = reactionNode.QuerySelectorAll(".userpics .userpic .initials[title]");
foreach (var userpicNode in userpicNodes)
{
var title = userpicNode.GetAttribute("title")?.Trim();
if (!string.IsNullOrEmpty(title) && request.Mapping.TryGetValue(title, out var rUserId) && rUserId != Guid.Empty && targetMessage != null)
{
var reaction = new MessageReaction(targetMessage.Id, rUserId, emoji);
await _reactionRepository.AddAsync(reaction, cancellationToken);
}
}
}
if (targetMessage != null && targetMessage != lastSavedMessage && (!string.IsNullOrEmpty(content) || targetMessage.Media.Any()))
{
_messageRepository.Add(targetMessage);
lastSavedMessage = targetMessage;
importedCount++;
}
else if (shouldCombine && lastSavedMessage != null)
{
await _messageRepository.UpdateAsync(lastSavedMessage, cancellationToken);
}
}
catch { /* ignore single message parse error */ }
}
}
}
await _uow.SaveChangesAsync(cancellationToken);
try { System.IO.File.Delete(tempPath); TelegramImportState.TempZips.TryRemove(request.Token, out _); } catch { }
await _hubContext.Clients.Users(chatMembers.Select(x => x.ToString())).SendAsync("history_updated", new { chatId });
return Result.Success(new ExecuteImportResponseDto(true, importedCount, chatId));
}
}
@@ -0,0 +1,15 @@
<Project Sdk="Microsoft.NET.Sdk">
<ItemGroup>
<ProjectReference Include="..\..\Shared\Knot.Shared.Kernel\Knot.Shared.Kernel.csproj" />
<ProjectReference Include="..\Conversations\Knot.Modules.Conversations.csproj" />
<ProjectReference Include="..\Messaging\Knot.Modules.Messaging.csproj" />
</ItemGroup>
<PropertyGroup>
<TargetFramework>net10.0</TargetFramework>
<ImplicitUsings>enable</ImplicitUsings>
<Nullable>enable</Nullable>
</PropertyGroup>
</Project>
@@ -0,0 +1,40 @@
using Carter;
using Knot.Shared.Kernel;
using Knot.Modules.TelegramImport.Application.TelegramImport.DTOs;
using Knot.Modules.TelegramImport.Application.TelegramImport;
using MediatR;
using Microsoft.AspNetCore.Builder;
using Microsoft.AspNetCore.Http;
using Microsoft.AspNetCore.Routing;
using Microsoft.AspNetCore.Mvc;
namespace Knot.Modules.TelegramImport.Presentation.Endpoints;
public sealed class TelegramImportEndpoints : ICarterModule
{
public void AddRoutes(IEndpointRouteBuilder app)
{
var group = app.MapGroup("api/import/telegram").RequireAuthorization();
group.MapPost("analyze", async (HttpRequest req, ISender sender, CancellationToken ct) =>
{
if (!req.HasFormContentType) return Results.BadRequest("No file uploaded.");
var form = await req.ReadFormAsync(ct);
var file = form.Files.FirstOrDefault();
if (file == null) return Results.BadRequest("No file uploaded.");
using var stream = file.OpenReadStream();
var result = await sender.Send(new AnalyzeImportCommand(stream, file.FileName), ct);
return result.IsSuccess ? Results.Ok(result.Value) : Results.BadRequest(result.Error.Description ?? result.Error.Code);
}).DisableAntiforgery();
group.MapPost("execute", async ([FromBody] ExecuteImportRequest req, ISender sender, IUserContext userContext, CancellationToken ct) =>
{
var command = new ExecuteImportCommand(userContext.UserId, req.Token, req.Mapping, req.GroupName);
var result = await sender.Send(command, ct);
return result.IsSuccess ? Results.Ok(result.Value) : Results.BadRequest(result.Error.Description ?? result.Error.Code);
});
}
}