From af1c76b3155aa25d9f474b3253e73fc41ad047b2 Mon Sep 17 00:00:00 2001
From: KotaHv <92137267+KotaHv@users.noreply.github.com>
Date: Tue, 28 May 2024 11:51:25 +0800
Subject: [PATCH] fix: Mikan some torrents are not a valid torrent file

---
 backend/requirements.txt                      |  2 ++
 .../src/module/downloader/download_client.py  | 34 ++++++++++++++++---
 .../src/module/network/request_contents.py    |  8 +++++
 backend/src/module/utils/__init__.py          |  3 +-
 backend/src/module/utils/torrent.py           | 11 ++++++
 5 files changed, 52 insertions(+), 6 deletions(-)
 create mode 100644 backend/src/module/utils/torrent.py

diff --git a/backend/requirements.txt b/backend/requirements.txt
index 02ed21578..ec4be9699 100644
--- a/backend/requirements.txt
+++ b/backend/requirements.txt
@@ -27,3 +27,5 @@ sqlmodel==0.0.8
 sse-starlette==1.6.5
 semver==3.0.1
 openai==0.28.1
+torrentool==1.2.0
+lxml==5.2.2
diff --git a/backend/src/module/downloader/download_client.py b/backend/src/module/downloader/download_client.py
index d01d4fa3c..609e8d96e 100644
--- a/backend/src/module/downloader/download_client.py
+++ b/backend/src/module/downloader/download_client.py
@@ -1,8 +1,11 @@
 import logging
 
+from lxml import etree
+
 from module.conf import settings
 from module.models import Bangumi, Torrent
 from module.network import RequestContent
+from module.utils import check_torrent
 
 from .path import TorrentPath
 
@@ -118,23 +121,44 @@ def add_torrent(self, torrent: Torrent | list, bangumi: Bangumi) -> bool:
         if not bangumi.save_path:
             bangumi.save_path = self._gen_save_path(bangumi)
         with RequestContent() as req:
+
+            def get_torrent_or_magnet(_torrent: Torrent):
+                content = req.get_content(_torrent.url)
+                if check_torrent(content):
+                    return content, None
+                if _torrent.homepage:
+                    magnet = req.get_magnet(_torrent.homepage)
+                    if magnet:
+                        return None, magnet
+                logger.error(
+                    f'[Downloader] {_torrent.name} torrent is corrupted; it is recommended to manually add the magnet link to qBittorrent, with the save path: "{bangumi.save_path}".'
+                )
+                return None, None
+
             if isinstance(torrent, list):
                 if len(torrent) == 0:
-                    logger.debug(f"[Downloader] No torrent found: {bangumi.official_title}")
+                    logger.debug(
+                        f"[Downloader] No torrent found: {bangumi.official_title}"
+                    )
                     return False
                 if "magnet" in torrent[0].url:
                     torrent_url = [t.url for t in torrent]
                     torrent_file = None
                 else:
-                    torrent_file = [req.get_content(t.url) for t in torrent]
-                    torrent_url = None
+                    torrent_file = []
+                    torrent_url = []
+                    for t in torrent:
+                        file, magnet = get_torrent_or_magnet(t)
+                        if file:
+                            torrent_file.append(file)
+                        if magnet:
+                            torrent_url.append(magnet)
             else:
                 if "magnet" in torrent.url:
                     torrent_url = torrent.url
                     torrent_file = None
                 else:
-                    torrent_file = req.get_content(torrent.url)
-                    torrent_url = None
+                    torrent_file, torrent_url = get_torrent_or_magnet(torrent)
         if self.client.add_torrents(
             torrent_urls=torrent_url,
             torrent_files=torrent_file,
diff --git a/backend/src/module/network/request_contents.py b/backend/src/module/network/request_contents.py
index 05abca023..4d3a74416 100644
--- a/backend/src/module/network/request_contents.py
+++ b/backend/src/module/network/request_contents.py
@@ -2,6 +2,8 @@
 import re
 import xml.etree.ElementTree
 
+from lxml import etree
+
 from module.conf import settings
 from module.models import Torrent
 
@@ -73,3 +75,9 @@ def get_rss_title(self, _url):
         soup = self.get_xml(_url)
         if soup:
             return soup.find("./channel/title").text
+
+    def get_magnet(self, _url) -> str | None:
+        html = etree.HTML(self.get_html(_url))
+        magnet = html.xpath('//a[starts-with(@href, "magnet")]/@href')
+        if magnet:
+            return magnet[0]
diff --git a/backend/src/module/utils/__init__.py b/backend/src/module/utils/__init__.py
index a95499d60..75a082393 100644
--- a/backend/src/module/utils/__init__.py
+++ b/backend/src/module/utils/__init__.py
@@ -1 +1,2 @@
-from .cache_image import save_image, load_image
\ No newline at end of file
+from .cache_image import save_image, load_image
+from .torrent import check_torrent
diff --git a/backend/src/module/utils/torrent.py b/backend/src/module/utils/torrent.py
new file mode 100644
index 000000000..10cee08a7
--- /dev/null
+++ b/backend/src/module/utils/torrent.py
@@ -0,0 +1,11 @@
+from torrentool.bencode import Bencode
+from torrentool.exceptions import BencodeDecodingError
+
+
+def check_torrent(content: bytes) -> bool:
+    try:
+        if Bencode.decode(content):
+            return True
+        return False
+    except BencodeDecodingError:
+        return False