Импорт из телеграм
This commit is contained in:
Халимов Рустам
282
apps/server-net/src/Host/Controllers/TelegramImportController.cs
Normal file
282
apps/server-net/src/Host/Controllers/TelegramImportController.cs
Normal file
@@ -0,0 +1,282 @@
|
||||
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 HtmlAgilityPack;
|
||||
using MediatR;
|
||||
using Microsoft.AspNetCore.Authorization;
|
||||
using Microsoft.AspNetCore.Http;
|
||||
using Microsoft.AspNetCore.Mvc;
|
||||
using Microsoft.AspNetCore.SignalR;
|
||||
using Knot.Modules.Chats.Application.Abstractions;
|
||||
using Knot.Modules.Chats.Domain;
|
||||
using Knot.Modules.Identity.Domain;
|
||||
using Knot.Shared.Kernel;
|
||||
using Knot.Shared.Kernel.Storage;
|
||||
using Knot.Modules.Chats.Infrastructure.SignalR;
|
||||
using Knot.Modules.Chats.Application.Chats.Create;
|
||||
|
||||
namespace Host.Controllers;
|
||||
|
||||
[Authorize]
|
||||
[ApiController]
|
||||
[Route("api/import/telegram")]
|
||||
public sealed class TelegramImportController : ControllerBase
|
||||
{
|
||||
private readonly ISender _sender;
|
||||
private readonly IUserContext _userContext;
|
||||
private readonly IChatsUnitOfWork _unitOfWork;
|
||||
private readonly IChatRepository _chatRepository;
|
||||
private readonly IMessageRepository _messageRepository;
|
||||
private readonly IFileStorageService _fileStorage;
|
||||
private readonly IHubContext<ChatHub> _hubContext;
|
||||
|
||||
// In-memory store for uploaded zips (good enough for typical cases, ideally should be removed after use or by timer)
|
||||
private static readonly ConcurrentDictionary<Guid, string> _tempZips = new();
|
||||
|
||||
public TelegramImportController(
|
||||
ISender sender,
|
||||
IUserContext userContext,
|
||||
IChatsUnitOfWork unitOfWork,
|
||||
IChatRepository chatRepository,
|
||||
IMessageRepository messageRepository,
|
||||
IFileStorageService fileStorage,
|
||||
IHubContext<ChatHub> hubContext)
|
||||
{
|
||||
_sender = sender;
|
||||
_userContext = userContext;
|
||||
_unitOfWork = unitOfWork;
|
||||
_chatRepository = chatRepository;
|
||||
_messageRepository = messageRepository;
|
||||
_fileStorage = fileStorage;
|
||||
_hubContext = hubContext;
|
||||
}
|
||||
|
||||
[HttpPost("analyze")]
|
||||
[DisableRequestSizeLimit]
|
||||
[RequestFormLimits(MultipartBodyLengthLimit = 10L * 1024 * 1024 * 1024)] // 10GB for big exports
|
||||
public async Task<IActionResult> Analyze(IFormFile file, CancellationToken ct)
|
||||
{
|
||||
if (file == null || file.Length == 0) return BadRequest("No file uploaded");
|
||||
if (!file.FileName.EndsWith(".zip", StringComparison.OrdinalIgnoreCase)) return BadRequest("Must be a ZIP archive");
|
||||
|
||||
var token = Guid.NewGuid();
|
||||
var tempPath = Path.Combine(Path.GetTempPath(), $"{token}.zip");
|
||||
|
||||
await using (var fs = new FileStream(tempPath, FileMode.Create))
|
||||
{
|
||||
await file.CopyToAsync(fs, ct);
|
||||
}
|
||||
|
||||
var names = new HashSet<string>();
|
||||
|
||||
// Open zip and quickly scan messages.html
|
||||
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 doc = new HtmlDocument();
|
||||
doc.Load(stream);
|
||||
|
||||
var messageNodes = doc.DocumentNode.SelectNodes("//div[contains(@class, 'message ')]");
|
||||
if (messageNodes == null) continue;
|
||||
|
||||
foreach (var node in messageNodes)
|
||||
{
|
||||
var fromNameNode = node.SelectSingleNode(".//div[contains(@class, 'from_name')]");
|
||||
if (fromNameNode != null)
|
||||
{
|
||||
var name = fromNameNode.InnerText.Trim();
|
||||
// Ignore standard system names if obvious (for now everything is recorded)
|
||||
if (!string.IsNullOrWhiteSpace(name))
|
||||
{
|
||||
names.Add(name);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
_tempZips[token] = tempPath;
|
||||
|
||||
return Ok(new
|
||||
{
|
||||
token,
|
||||
names = names.ToList()
|
||||
});
|
||||
}
|
||||
|
||||
public sealed record ExecuteImportRequest(Guid Token, Dictionary<string, Guid> Mapping);
|
||||
|
||||
[HttpPost("execute")]
|
||||
public async Task<IActionResult> Execute([FromBody] ExecuteImportRequest req, CancellationToken ct)
|
||||
{
|
||||
if (!_tempZips.TryGetValue(req.Token, out var tempPath))
|
||||
return BadRequest("Session not found or expired");
|
||||
|
||||
if (!System.IO.File.Exists(tempPath))
|
||||
return BadRequest("ZIP file lost");
|
||||
|
||||
var myId = _userContext.UserId;
|
||||
// Collect targeted users to check whose chat it is. Find the friend.
|
||||
// Usually, the mapping contains MyId and FriendId.
|
||||
var targetUserIds = req.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)
|
||||
{
|
||||
// Find existing personal chat
|
||||
var existingChats = await _chatRepository.GetUserChatsAsync(myId, ct);
|
||||
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
|
||||
{
|
||||
// Create personal chat
|
||||
var friendId = chatMembers.FirstOrDefault(id => id != myId);
|
||||
if (friendId == Guid.Empty) friendId = myId; // Notes to self
|
||||
var command = new CreateChatCommand(string.Empty, ChatType.Personal, new List<Guid> { myId, friendId });
|
||||
var res = await _sender.Send(command, ct);
|
||||
if (res.IsFailure) return BadRequest(res.Error);
|
||||
chatId = res.Value;
|
||||
}
|
||||
}
|
||||
else
|
||||
{
|
||||
// Create a group
|
||||
var command = new CreateChatCommand("Импортированный чат", ChatType.Group, chatMembers);
|
||||
var res = await _sender.Send(command, ct);
|
||||
if (res.IsFailure) return BadRequest(res.Error);
|
||||
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 => e.FullName); // Read chronologically
|
||||
|
||||
foreach (var entry in htmlEntries)
|
||||
{
|
||||
using var stream = entry.Open();
|
||||
var doc = new HtmlDocument();
|
||||
doc.Load(stream);
|
||||
|
||||
var messageNodes = doc.DocumentNode.SelectNodes("//div[contains(@class, '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.SelectSingleNode(".//div[contains(@class, 'from_name')]");
|
||||
var textNode = node.SelectSingleNode(".//div[contains(@class, 'text')]");
|
||||
var dateNode = node.SelectSingleNode(".//div[contains(@class, 'pull_right date')]");
|
||||
|
||||
// Default values
|
||||
var senderGuid = myId; // Fallback
|
||||
if (fromNameNode != null)
|
||||
{
|
||||
var name = fromNameNode.InnerText.Trim();
|
||||
if (req.Mapping.TryGetValue(name, out var mappedId) && mappedId != Guid.Empty)
|
||||
senderGuid = mappedId;
|
||||
}
|
||||
|
||||
var content = textNode?.InnerText?.Trim() ?? "";
|
||||
|
||||
DateTime createdAt = DateTime.UtcNow;
|
||||
if (dateNode != null && dateNode.Attributes["title"] != null)
|
||||
{
|
||||
var dateStr = dateNode.Attributes["title"].Value;
|
||||
if (DateTime.TryParse(dateStr, out var d))
|
||||
createdAt = d.ToUniversalTime();
|
||||
}
|
||||
|
||||
var mediaNodes = node.SelectNodes(".//a[contains(@class, 'photo_wrap')] | .//video | .//a[contains(@class, 'document')] | .//a[contains(@class, 'media_voice_message')]");
|
||||
|
||||
var messageType = "text";
|
||||
if (mediaNodes != null && mediaNodes.Count > 0)
|
||||
{
|
||||
var firstHref = mediaNodes[0].Attributes["href"]?.Value ?? mediaNodes[0].Attributes["src"]?.Value;
|
||||
if (firstHref != null)
|
||||
{
|
||||
if (firstHref.EndsWith(".jpg") || firstHref.EndsWith(".png")) messageType = "image";
|
||||
else if (firstHref.EndsWith(".mp4")) messageType = "video";
|
||||
else if (firstHref.EndsWith(".ogg")) messageType = "voice";
|
||||
else messageType = "file";
|
||||
}
|
||||
}
|
||||
|
||||
var newMessage = Message.Import(chatId, senderGuid, content, messageType, createdAt);
|
||||
|
||||
if (mediaNodes != null)
|
||||
{
|
||||
foreach (var mediaNode in mediaNodes)
|
||||
{
|
||||
string? href = mediaNode.Attributes["href"]?.Value ?? mediaNode.Attributes["src"]?.Value;
|
||||
if (!string.IsNullOrEmpty(href) && !href.StartsWith("http"))
|
||||
{
|
||||
// Local file in zip
|
||||
var zipPath = baseDir + href.Replace("\\", "/");
|
||||
var zipEntry = archive.GetEntry(zipPath);
|
||||
if (zipEntry != null)
|
||||
{
|
||||
using var ms = new MemoryStream();
|
||||
using var zipfs = zipEntry.Open();
|
||||
await zipfs.CopyToAsync(ms, ct);
|
||||
ms.Position = 0;
|
||||
|
||||
string cType = "application/octet-stream";
|
||||
var mType = "file";
|
||||
if (href.EndsWith(".jpg") || href.EndsWith(".png")) { cType = "image/jpeg"; mType = "image"; messageType = "image"; }
|
||||
else if (href.EndsWith(".mp4")) { cType = "video/mp4"; mType = "video"; messageType = "video"; }
|
||||
else if (href.EndsWith(".ogg")) { cType = "audio/ogg"; mType = "voice"; messageType = "voice"; }
|
||||
|
||||
var fileId = await _fileStorage.UploadFileAsync(ms, Path.GetFileName(href), cType);
|
||||
newMessage.AddMedia(mType, $"/api/files/{fileId}", Path.GetFileName(href), zipEntry.Length);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// Save text message without media if content is just empty, but wait, if it's empty and has no media, it was probably a system message (like pinned, joined).
|
||||
if (!string.IsNullOrEmpty(content) || newMessage.Media.Any())
|
||||
{
|
||||
_messageRepository.Add(newMessage);
|
||||
importedCount++;
|
||||
}
|
||||
}
|
||||
catch { /* ignore single message parse error */ }
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
await _unitOfWork.SaveChangesAsync(ct);
|
||||
|
||||
// Delete temp zip
|
||||
try { System.IO.File.Delete(tempPath); _tempZips.TryRemove(req.Token, out _); } catch { }
|
||||
|
||||
// Notify UI for all members of the chat
|
||||
await _hubContext.Clients.Users(chatMembers.Select(x => x.ToString())).SendAsync("history_updated", new { chatId });
|
||||
|
||||
return Ok(new { success = true, messagesImported = importedCount, chatId });
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user