2 from __future__ import unicode_literals
4 from .common import InfoExtractor
10 from ..compat import (
19 class WeiboIE(InfoExtractor):
20 _VALID_URL = r'https?://weibo\.com/[0-9]+/(?P<id>[a-zA-Z0-9]+)'
22 'url': 'https://weibo.com/6275294458/Fp6RGfbff?type=comment',
26 'title': 'You should have servants to massage you,... 来自Hosico_猫 - 微博',
30 def _real_extract(self, url):
31 video_id = self._match_id(url)
32 # to get Referer url for genvisitor
33 webpage, urlh = self._download_webpage_handle(url, video_id, note="first visit the page")
35 visitor_url = urlh.geturl()
37 'Referer': visitor_url
42 "browser": "Gecko57,0,0,0",
44 "screenInfo": "1440*900*24",
47 data = urlencode_postdata({
52 genvisitor_url = 'https://passport.weibo.com/visitor/genvisitor'
53 webpage, _ = self._download_webpage_handle(genvisitor_url, video_id, data=data, headers=headers, note="gen visitor")
55 p = webpage.split("&&")[1] # split "gen_callback && gen_callback(...)"
58 j = p[i1:i2 + 1] # get JSON object
60 tid = d["data"]["tid"]
61 cnfd = "%03d" % d["data"]["confidence"]
70 '_rand': random.random()
72 gencallback_url = "https://passport.weibo.com/visitor/visitor"
73 self._download_webpage_handle(gencallback_url, video_id, note="gen callback", query=query)
75 webpage, _ = self._download_webpage_handle(url, video_id, note="retry to visit the page")
77 title = self._html_search_regex(r'<title>(.+?)</title>', webpage, 'title')
79 video_sources_text = self._search_regex(r'video-sources=\\\"(.+?)\"', webpage, 'video_sources')
81 video_formats = compat_urlparse.parse_qs(video_sources_text)
84 supported_resolutions = ('720', '480')
85 for res in supported_resolutions:
86 f = video_formats.get(res)
87 if isinstance(f, list):
95 self._sort_formats(formats)
96 uploader = self._og_search_property('nick-name', webpage, 'uploader', default=None)
100 'uploader': uploader,
105 class WeiboMobileIE(InfoExtractor):
106 _VALID_URL = r'https?://m\.weibo\.cn/status/(?P<id>[0-9]+)(\?.+)?'
108 'url': 'https://m.weibo.cn/status/4189191225395228?wm=3333_2001&sourcetype=weixin&featurecode=newtitle&from=singlemessage&isappinstalled=0',
110 'id': '4189191225395228',
112 'title': '午睡当然是要甜甜蜜蜜的啦',
117 def _real_extract(self, url):
118 video_id = self._match_id(url)
119 # to get Referer url for genvisitor
120 webpage, _ = self._download_webpage_handle(url, video_id, note="visit the page")
121 js_code = self._search_regex(r'var\s+\$render_data\s*=\s*\[({.*})\]\[0\] \|\| {};', webpage, 'js_code', flags=re.DOTALL)
122 weibo_info = self._parse_json(js_code, video_id, transform_source=js_to_json)
123 page_info = weibo_info['status']['page_info']
124 title = weibo_info.get('status').get('status_title')
125 uploader = weibo_info.get('status').get('user').get('screen_name')
130 'uploader': uploader,
131 'url': page_info['media_info']['stream_url']