feat: 新增小黑盒(XiaoHeiHe)平台爬虫支持

cmd_arg/arg.py

@@ -47,6 +47,7 @@ class PlatformEnum(str, Enum):
     WEIBO = "wb"
     TIEBA = "tieba"
     ZHIHU = "zhihu"
+    XIAOHEIHE = "xhh"
 
 
 class LoginTypeEnum(str, Enum):
@@ -106,7 +107,7 @@ def _coerce_enum(
         return enum_cls(value)
     except ValueError:
         typer.secho(
-            f"⚠️ Config value '{value}' is not within the supported range of {enum_cls.__name__}, falling back to default value '{default.value}'.",
+            f"\u26a0\ufe0f Config value '{value}' is not within the supported range of {enum_cls.__name__}, falling back to default value '{default.value}'.",
             fg=typer.colors.YELLOW,
         )
         return default
@@ -162,7 +163,7 @@ def main(
             PlatformEnum,
             typer.Option(
                 "--platform",
-                help="Media platform selection (xhs=XiaoHongShu | dy=Douyin | ks=Kuaishou | bili=Bilibili | wb=Weibo | tieba=Baidu Tieba | zhihu=Zhihu)",
+                help="Media platform selection (xhs=XiaoHongShu | dy=Douyin | ks=Kuaishou | bili=Bilibili | wb=Weibo | tieba=Baidu Tieba | zhihu=Zhihu | xhh=XiaoHeiHe)",
                 rich_help_panel="Basic Configuration",
             ),
         ] = _coerce_enum(PlatformEnum, config.PLATFORM, PlatformEnum.XHS),

config/base_config.py

@@ -18,7 +18,7 @@
 # 使用本代码即表示您同意遵守上述原则和LICENSE中的所有条款。
 
 # Basic configuration
-PLATFORM = "xhs"  # Platform, xhs | dy | ks | bili | wb | tieba | zhihu
+PLATFORM = "xhs"  # Platform, xhs | dy | ks | bili | wb | tieba | zhihu | xhh
 
 # 是否使用海外版小红书 (rednote.com)
 # 开启后 API 走 webapi.rednote.com，cookie 域使用 .rednote.com
@@ -143,3 +143,4 @@
 from .weibo_config import *
 from .tieba_config import *
 from .zhihu_config import *
+from .xhh_config import *

config/xhh_config.py

@@ -0,0 +1,16 @@
+# -*- coding: utf-8 -*-
+# Copyright (c) 2025 relakkes@gmail.com
+#
+# This file is part of MediaCrawler project.
+# Licensed under NON-COMMERCIAL LEARNING LICENSE 1.1
+#
+
+# Xiaoheihe (小黑盒) platform configuration
+
+# 指定帖子 link_id 列表（detail 模式）
+XHH_SPECIFIED_ID_LIST = [
+    "179403000",
+]
+
+# 创作者 user_id 列表（creator 模式）
+XHH_CREATOR_ID_LIST: list = ["93126766"]

main.py

@@ -41,6 +41,7 @@
 from media_platform.kuaishou import KuaishouCrawler
 from media_platform.tieba import TieBaCrawler
 from media_platform.weibo import WeiboCrawler
+from media_platform.xhh import XiaoHeiHeCrawler
 from media_platform.xhs import XiaoHongShuCrawler
 from media_platform.zhihu import ZhihuCrawler
 from tools.async_file_writer import AsyncFileWriter
@@ -56,6 +57,7 @@ class CrawlerFactory:
         "wb": WeiboCrawler,
         "tieba": TieBaCrawler,
         "zhihu": ZhihuCrawler,
+        "xhh": XiaoHeiHeCrawler,
     }
 
     @staticmethod

media_platform/xhh/__init__.py

@@ -0,0 +1,10 @@
+# -*- coding: utf-8 -*-
+# Copyright (c) 2025 relakkes@gmail.com
+#
+# This file is part of MediaCrawler project.
+# Licensed under NON-COMMERCIAL LEARNING LICENSE 1.1
+#
+
+from .core import XiaoHeiHeCrawler
+
+__all__ = ["XiaoHeiHeCrawler"]