yt-dlp/yt_dlp/extractor/xiaohongshu.py
sepro add96eb9f8
[cleanup] Add more ruff rules (#10149)
Authored by: seproDev

Reviewed-by: bashonly <88596187+bashonly@users.noreply.github.com>
Reviewed-by: Simon Sawicki <contact@grub4k.xyz>
2024-06-12 01:09:58 +02:00

84 lines
3.5 KiB
Python

import functools
from .common import InfoExtractor
from ..utils import (
float_or_none,
int_or_none,
js_to_json,
url_or_none,
)
from ..utils.traversal import traverse_obj
class XiaoHongShuIE(InfoExtractor):
_VALID_URL = r'https?://www\.xiaohongshu\.com/explore/(?P<id>[\da-f]+)'
IE_DESC = '小红书'
_TESTS = [{
'url': 'https://www.xiaohongshu.com/explore/6411cf99000000001300b6d9',
'md5': '2a87a77ddbedcaeeda8d7eae61b61228',
'info_dict': {
'id': '6411cf99000000001300b6d9',
'ext': 'mp4',
'uploader_id': '5c31698d0000000007018a31',
'description': '#今日快乐今日发[话题]# #吃货薯看这里[话题]# #香妃蛋糕[话题]# #小五卷蛋糕[话题]# #新手蛋糕卷[话题]#',
'title': '香妃蛋糕也太香了吧🔥不需要卷❗️绝对的友好',
'tags': ['今日快乐今日发', '吃货薯看这里', '香妃蛋糕', '小五卷蛋糕', '新手蛋糕卷'],
'duration': 101.726,
'thumbnail': r're:https?://sns-webpic-qc\.xhscdn\.com/\d+/[a-z0-9]+/[\w]+',
},
}]
def _real_extract(self, url):
display_id = self._match_id(url)
webpage = self._download_webpage(url, display_id)
initial_state = self._search_json(
r'window\.__INITIAL_STATE__\s*=', webpage, 'initial state', display_id, transform_source=js_to_json)
note_info = traverse_obj(initial_state, ('note', 'noteDetailMap', display_id, 'note'))
video_info = traverse_obj(note_info, ('video', 'media', 'stream', ('h264', 'av1', 'h265'), ...))
formats = []
for info in video_info:
format_info = traverse_obj(info, {
'fps': ('fps', {int_or_none}),
'width': ('width', {int_or_none}),
'height': ('height', {int_or_none}),
'vcodec': ('videoCodec', {str}),
'acodec': ('audioCodec', {str}),
'abr': ('audioBitrate', {int_or_none}),
'vbr': ('videoBitrate', {int_or_none}),
'audio_channels': ('audioChannels', {int_or_none}),
'tbr': ('avgBitrate', {int_or_none}),
'format': ('qualityType', {str}),
'filesize': ('size', {int_or_none}),
'duration': ('duration', {functools.partial(float_or_none, scale=1000)}),
})
formats.extend(traverse_obj(info, (('mediaUrl', ('backupUrls', ...)), {
lambda u: url_or_none(u) and {'url': u, **format_info}})))
thumbnails = []
for image_info in traverse_obj(note_info, ('imageList', ...)):
thumbnail_info = traverse_obj(image_info, {
'height': ('height', {int_or_none}),
'width': ('width', {int_or_none}),
})
for thumb_url in traverse_obj(image_info, (('urlDefault', 'urlPre'), {url_or_none})):
thumbnails.append({
'url': thumb_url,
**thumbnail_info,
})
return {
'id': display_id,
'formats': formats,
'thumbnails': thumbnails,
'title': self._html_search_meta(['og:title'], webpage, default=None),
**traverse_obj(note_info, {
'title': ('title', {str}),
'description': ('desc', {str}),
'tags': ('tagList', ..., 'name', {str}),
'uploader_id': ('user', 'userId', {str}),
}),
}