修复 RAR 字幕包下载识别

app/chain/download.py

@@ -36,6 +36,11 @@ class DownloadChain(ChainBase):
     下载处理链
     """
 
+    _SUBTITLE_ARCHIVE_FORMATS = {
+        ".zip": "zip",
+        ".rar": "rar",
+    }
+
     @staticmethod
     def _safe_subtitle_file_name(file_name: str, fallback_name: str) -> str:
         """
@@ -51,7 +56,14 @@ class DownloadChain(ChainBase):
         """
         判断是否为字幕压缩包。
         """
-        return Path(file_name).suffix.lower() == ".zip"
+        return Path(file_name).suffix.lower() in DownloadChain._SUBTITLE_ARCHIVE_FORMATS
+
+    @classmethod
+    def _subtitle_archive_format(cls, file_name: str) -> Optional[str]:
+        """
+        获取字幕压缩包格式。
+        """
+        return cls._SUBTITLE_ARCHIVE_FORMATS.get(Path(file_name).suffix.lower())
 
     @staticmethod
     def _is_subtitle_file(file_name: str) -> bool:
@@ -154,7 +166,15 @@ class DownloadChain(ChainBase):
         try:
             temp_file.write_bytes(response.content)
             if self._is_subtitle_archive(file_name):
-                shutil.unpack_archive(temp_file, temp_extract_dir, format='zip')
+                try:
+                    SystemUtils.unpack_archive(
+                        temp_file,
+                        temp_extract_dir,
+                        archive_format=self._subtitle_archive_format(file_name),
+                    )
+                except Exception as err:
+                    logger.error(f"字幕压缩包解压失败：{temp_file} - {str(err)}")
+                    return []
                 for sub_file in SystemUtils.list_files(temp_extract_dir, settings.RMT_SUBEXT):
                     uploaded_path = self._upload_subtitle_file(
                         storage_chain=storage_chain,

app/modules/subtitle/__init__.py

@@ -28,6 +28,11 @@ class SubtitleModule(_ModuleBase):
     字幕下载模块
     """
 
+    _SUBTITLE_ARCHIVE_FORMATS = {
+        ".zip": "zip",
+        ".rar": "rar",
+    }
+
     # 站点详情页字幕下载元素识别XPATH
     _SITE_SUBTITLE_XPATH = [
         '//td[@class="rowhead"][text()="字幕"]/following-sibling::td//a[not(@class)]',
@@ -233,40 +238,52 @@ class SubtitleModule(_ModuleBase):
             ua=torrent.site_ua,
             proxies=settings.PROXY if torrent.site_proxy else None,
         )
+        settings.TEMP_PATH.mkdir(parents=True, exist_ok=True)
         for sublink in sublink_list:
             logger.info(f"找到字幕下载链接：{sublink}，开始下载...")
             # 下载
             ret = request.get_res(sublink)
             if ret and ret.status_code == 200:
-                # 保存ZIP
                 file_name = TorrentHelper.get_url_filename(ret, sublink)
                 if not file_name:
                     logger.warn(f"链接不是字幕文件：{sublink}")
                     continue
-                if file_name.lower().endswith(".zip"):
-                    # ZIP包
-                    zip_file = settings.TEMP_PATH / file_name
+                archive_format = self._SUBTITLE_ARCHIVE_FORMATS.get(Path(file_name).suffix.lower())
+                if archive_format:
+                    archive_file = settings.TEMP_PATH / file_name
                     # 保存
-                    zip_file.write_bytes(ret.content)
+                    archive_file.write_bytes(ret.content)
                     # 解压路径
-                    zip_path = zip_file.with_name(zip_file.stem)
-                    # 解压文件
-                    shutil.unpack_archive(zip_file, zip_path, format='zip')
-                    # 遍历转移文件
-                    for sub_file in SystemUtils.list_files(zip_path, settings.RMT_SUBEXT):
-                        target_sub_file = Path(working_dir_item.path) / Path(sub_file.name)
-                        if storageChain.get_file_item(storage, target_sub_file):
-                            logger.info(f"字幕文件已存在：{target_sub_file}")
-                            continue
-                        logger.info(f"转移字幕 {sub_file} 到 {target_sub_file} ...")
-                        storageChain.upload_file(working_dir_item, sub_file)
+                    archive_path = archive_file.with_name(archive_file.stem)
+                    try:
+                        # 解压文件
+                        SystemUtils.unpack_archive(
+                            archive_file,
+                            archive_path,
+                            archive_format=archive_format,
+                        )
+                        # 遍历转移文件
+                        for sub_file in SystemUtils.list_files(archive_path, settings.RMT_SUBEXT):
+                            target_sub_file = Path(working_dir_item.path) / Path(sub_file.name)
+                            if storageChain.get_file_item(storage, target_sub_file):
+                                logger.info(f"字幕文件已存在：{target_sub_file}")
+                                continue
+                            logger.info(f"转移字幕 {sub_file} 到 {target_sub_file} ...")
+                            storageChain.upload_file(working_dir_item, sub_file)
+                    except Exception as err:
+                        logger.error(f"字幕压缩包解压失败：{archive_file} - {str(err)}")
                     # 删除临时文件
                     try:
-                        shutil.rmtree(zip_path)
-                        zip_file.unlink()
+                        if archive_path.exists():
+                            shutil.rmtree(archive_path)
+                        if archive_file.exists():
+                            archive_file.unlink()
                     except Exception as err:
                         logger.error(f"删除临时文件失败：{str(err)}")
                 else:
+                    if Path(file_name).suffix.lower() not in settings.RMT_SUBEXT:
+                        logger.warn(f"链接不是支持的字幕文件：{sublink} - {file_name}")
+                        continue
                     sub_file = settings.TEMP_PATH / file_name
                     # 保存
                     sub_file.write_bytes(ret.content)

app/utils/system.py

@@ -262,6 +262,84 @@ class SystemUtils:
         _scan_directory(directory, recursive)
         return files
 
+    @staticmethod
+    def unpack_archive(archive_file: Path, extract_dir: Path, archive_format: Optional[str] = None) -> None:
+        """
+        解压压缩包，并补充标准库未覆盖的 RAR 格式支持。
+
+        :param archive_file: 待解压的压缩包文件
+        :param extract_dir: 解压目标目录
+        :param archive_format: 压缩包格式，未指定时按文件后缀推断
+        """
+        if archive_format == "rar" or (not archive_format and archive_file.suffix.lower() == ".rar"):
+            SystemUtils.__unpack_rar_archive(archive_file, extract_dir)
+            return
+        shutil.unpack_archive(archive_file, extract_dir, format=archive_format)
+
+    @staticmethod
+    def __unpack_rar_archive(archive_file: Path, extract_dir: Path) -> None:
+        """
+        调用系统解压工具处理 RAR 压缩包。
+        """
+        extract_dir.mkdir(parents=True, exist_ok=True)
+        commands = []
+        if shutil.which("unar"):
+            commands.append([
+                "unar",
+                "-quiet",
+                "-force-overwrite",
+                "-output-directory",
+                extract_dir.as_posix(),
+                archive_file.as_posix(),
+            ])
+        if shutil.which("unrar"):
+            commands.append([
+                "unrar",
+                "x",
+                "-o+",
+                "-idq",
+                archive_file.as_posix(),
+                f"{extract_dir.as_posix()}/",
+            ])
+        if shutil.which("7z"):
+            commands.append([
+                "7z",
+                "x",
+                "-y",
+                f"-o{extract_dir.as_posix()}",
+                archive_file.as_posix(),
+            ])
+        if shutil.which("bsdtar"):
+            commands.append([
+                "bsdtar",
+                "-xf",
+                archive_file.as_posix(),
+                "-C",
+                extract_dir.as_posix(),
+            ])
+        if not commands:
+            raise RuntimeError("未找到可用的 RAR 解压工具，请安装 unar、unrar、7z 或 bsdtar")
+
+        errors = []
+        for command in commands:
+            try:
+                result = subprocess.run(
+                    command,
+                    check=False,
+                    text=True,
+                    stdout=subprocess.PIPE,
+                    stderr=subprocess.PIPE,
+                    timeout=120,
+                )
+            except Exception as err:
+                errors.append(f"{command[0]}：{str(err)}")
+                continue
+            if result.returncode == 0:
+                return
+            output = (result.stderr or result.stdout or "").strip()
+            errors.append(f"{command[0]}：{output or f'返回码 {result.returncode}'}")
+        raise RuntimeError(f"RAR 压缩包解压失败：{'；'.join(errors)}")
+
     @staticmethod
     def exits_files(directory: Path, extensions: list, min_filesize: int = 0, recursive: bool = True) -> bool:
         """