[ie/weibo] Fix extractor and support user extraction (#7657)

Closes #3964, Closes #4673, Closes #6979
Authored by: c-basalt

yt_dlp/extractor/_extractors.py

@@ -2371,7 +2371,8 @@
 )
 from .weibo import (
     WeiboIE,
-    WeiboMobileIE
+    WeiboVideoIE,
+    WeiboUserIE,
 )
 from .weiqitv import WeiqiTVIE
 from .weverse import (

yt_dlp/extractor/weibo.py

@@ -1,134 +1,241 @@
-from .common import InfoExtractor
-
-import json
 import random
-import re
+import itertools
+import urllib.parse
 
-from ..compat import (
+from .common import InfoExtractor
-    compat_parse_qs,
-    compat_str,
-)
 from ..utils import (
-    js_to_json,
+    int_or_none,
+    make_archive_id,
+    mimetype2ext,
+    parse_resolution,
+    str_or_none,
     strip_jsonp,
+    traverse_obj,
+    url_or_none,
     urlencode_postdata,
+    urljoin,
 )
 
 
-class WeiboIE(InfoExtractor):
+class WeiboBaseIE(InfoExtractor):
-    _VALID_URL = r'https?://(?:www\.)?weibo\.com/[0-9]+/(?P<id>[a-zA-Z0-9]+)'
+    def _update_visitor_cookies(self, video_id):
-    _TEST = {
+        visitor_data = self._download_json(
-        'url': 'https://weibo.com/6275294458/Fp6RGfbff?type=comment',
+            'https://passport.weibo.com/visitor/genvisitor', video_id,
-        'info_dict': {
+            note='Generating first-visit guest request',
-            'id': 'Fp6RGfbff',
+            transform_source=strip_jsonp,
-            'ext': 'mp4',
+            data=urlencode_postdata({
-            'title': 'You should have servants to massage you,... 来自Hosico_猫 - 微博',
+                'cb': 'gen_callback',
-        }
+                'fp': '{"os":"2","browser":"Gecko57,0,0,0","fonts":"undefined","screenInfo":"1440*900*24","plugins":""}',
-    }
+            }))
 
-    def _real_extract(self, url):
+        self._download_webpage(
-        video_id = self._match_id(url)
+            'https://passport.weibo.com/visitor/visitor', video_id,
-        # to get Referer url for genvisitor
+            note='Running first-visit callback to get guest cookies',
-        webpage, urlh = self._download_webpage_handle(url, video_id)
+            query={
-
+                'a': 'incarnate',
-        visitor_url = urlh.url
+                't': visitor_data['data']['tid'],
-
+                'w': 2,
-        if 'passport.weibo.com' in visitor_url:
+                'c': '%03d' % visitor_data['data']['confidence'],
-            # first visit
+                'cb': 'cross_domain',
-            visitor_data = self._download_json(
+                'from': 'weibo',
-                'https://passport.weibo.com/visitor/genvisitor', video_id,
+                '_rand': random.random(),
-                note='Generating first-visit data',
-                transform_source=strip_jsonp,
-                headers={'Referer': visitor_url},
-                data=urlencode_postdata({
-                    'cb': 'gen_callback',
-                    'fp': json.dumps({
-                        'os': '2',
-                        'browser': 'Gecko57,0,0,0',
-                        'fonts': 'undefined',
-                        'screenInfo': '1440*900*24',
-                        'plugins': '',
-                    }),
-                }))
-
-            tid = visitor_data['data']['tid']
-            cnfd = '%03d' % visitor_data['data']['confidence']
-
-            self._download_webpage(
-                'https://passport.weibo.com/visitor/visitor', video_id,
-                note='Running first-visit callback',
-                query={
-                    'a': 'incarnate',
-                    't': tid,
-                    'w': 2,
-                    'c': cnfd,
-                    'cb': 'cross_domain',
-                    'from': 'weibo',
-                    '_rand': random.random(),
-                })
-
-            webpage = self._download_webpage(
-                url, video_id, note='Revisiting webpage')
-
-        title = self._html_extract_title(webpage)
-
-        video_formats = compat_parse_qs(self._search_regex(
-            r'video-sources=\\\"(.+?)\"', webpage, 'video_sources'))
-
-        formats = []
-        supported_resolutions = (480, 720)
-        for res in supported_resolutions:
-            vid_urls = video_formats.get(compat_str(res))
-            if not vid_urls or not isinstance(vid_urls, list):
-                continue
-
-            vid_url = vid_urls[0]
-            formats.append({
-                'url': vid_url,
-                'height': res,
             })
 
-        uploader = self._og_search_property(
+    def _weibo_download_json(self, url, video_id, *args, fatal=True, note='Downloading JSON metadata', **kwargs):
-            'nick-name', webpage, 'uploader', default=None)
+        webpage, urlh = self._download_webpage_handle(url, video_id, *args, fatal=fatal, note=note, **kwargs)
+        if urllib.parse.urlparse(urlh.url).netloc == 'passport.weibo.com':
+            self._update_visitor_cookies(video_id)
+            webpage = self._download_webpage(url, video_id, *args, fatal=fatal, note=note, **kwargs)
+        return self._parse_json(webpage, video_id, fatal=fatal)
 
+    def _extract_formats(self, video_info):
+        media_info = traverse_obj(video_info, ('page_info', 'media_info'))
+        formats = traverse_obj(media_info, (
+            'playback_list', lambda _, v: url_or_none(v['play_info']['url']), 'play_info', {
+                'url': 'url',
+                'format': ('quality_desc', {str}),
+                'format_id': ('label', {str}),
+                'ext': ('mime', {mimetype2ext}),
+                'tbr': ('bitrate', {int_or_none}, {lambda x: x or None}),
+                'vcodec': ('video_codecs', {str}),
+                'fps': ('fps', {int_or_none}),
+                'width': ('width', {int_or_none}),
+                'height': ('height', {int_or_none}),
+                'filesize': ('size', {int_or_none}),
+                'acodec': ('audio_codecs', {str}),
+                'asr': ('audio_sample_rate', {int_or_none}),
+                'audio_channels': ('audio_channels', {int_or_none}),
+            }))
+        if not formats:  # fallback, should be barely used
+            for url in set(traverse_obj(media_info, (..., {url_or_none}))):
+                if 'label=' in url:  # filter out non-video urls
+                    format_id, resolution = self._search_regex(
+                        r'label=(\w+)&template=(\d+x\d+)', url, 'format info',
+                        group=(1, 2), default=(None, None))
+                    formats.append({
+                        'url': url,
+                        'format_id': format_id,
+                        **parse_resolution(resolution),
+                        **traverse_obj(media_info, (
+                            'video_details', lambda _, v: v['label'].startswith(format_id), {
+                                'size': ('size', {int_or_none}),
+                                'tbr': ('bitrate', {int_or_none}),
+                            }
+                        ), get_all=False),
+                    })
+        return formats
+
+    def _parse_video_info(self, video_info, video_id=None):
         return {
             'id': video_id,
-            'title': title,
+            'extractor_key': WeiboIE.ie_key(),
-            'uploader': uploader,
+            'extractor': WeiboIE.IE_NAME,
-            'formats': formats
+            'formats': self._extract_formats(video_info),
+            'http_headers': {'Referer': 'https://weibo.com/'},
+            '_old_archive_ids': [make_archive_id('WeiboMobile', video_id)],
+            **traverse_obj(video_info, {
+                'id': (('id', 'id_str', 'mid'), {str_or_none}),
+                'display_id': ('mblogid', {str_or_none}),
+                'title': ('page_info', 'media_info', ('video_title', 'kol_title', 'name'), {str}, {lambda x: x or None}),
+                'description': ('text_raw', {str}),
+                'duration': ('page_info', 'media_info', 'duration', {int_or_none}),
+                'timestamp': ('page_info', 'media_info', 'video_publish_time', {int_or_none}),
+                'thumbnail': ('page_info', 'page_pic', {url_or_none}),
+                'uploader': ('user', 'screen_name', {str}),
+                'uploader_id': ('user', ('id', 'id_str'), {str_or_none}),
+                'uploader_url': ('user', 'profile_url', {lambda x: urljoin('https://weibo.com/', x)}),
+                'view_count': ('page_info', 'media_info', 'online_users_number', {int_or_none}),
+                'like_count': ('attitudes_count', {int_or_none}),
+                'repost_count': ('reposts_count', {int_or_none}),
+            }, get_all=False),
+            'tags': traverse_obj(video_info, ('topic_struct', ..., 'topic_title', {str})) or None,
         }
 
 
-class WeiboMobileIE(InfoExtractor):
+class WeiboIE(WeiboBaseIE):
-    _VALID_URL = r'https?://m\.weibo\.cn/status/(?P<id>[0-9]+)(\?.+)?'
+    _VALID_URL = r'https?://(?:m\.weibo\.cn/status|(?:www\.)?weibo\.com/\d+)/(?P<id>[a-zA-Z0-9]+)'
-    _TEST = {
+    _TESTS = [{
-        'url': 'https://m.weibo.cn/status/4189191225395228?wm=3333_2001&sourcetype=weixin&featurecode=newtitle&from=singlemessage&isappinstalled=0',
+        'url': 'https://weibo.com/7827771738/N4xlMvjhI',
+        'info_dict': {
+            'id': '4910815147462302',
+            'ext': 'mp4',
+            'display_id': 'N4xlMvjhI',
+            'title': '【睡前消息暑假版第一期：拉泰国一把  对中国有好处】',
+            'description': 'md5:e2637a7673980d68694ea7c43cf12a5f',
+            'duration': 918,
+            'timestamp': 1686312819,
+            'upload_date': '20230609',
+            'thumbnail': r're:https://.*\.jpg',
+            'uploader': '睡前视频基地',
+            'uploader_id': '7827771738',
+            'uploader_url': 'https://weibo.com/u/7827771738',
+            'view_count': int,
+            'like_count': int,
+            'repost_count': int,
+            'tags': ['泰国大选远进党获胜', '睡前消息', '暑期版'],
+        },
+    }, {
+        'url': 'https://m.weibo.cn/status/4189191225395228',
         'info_dict': {
             'id': '4189191225395228',
             'ext': 'mp4',
-            'title': '午睡当然是要甜甜蜜蜜的啦',
+            'display_id': 'FBqgOmDxO',
-            'uploader': '柴犬柴犬'
+            'title': '柴犬柴犬的秒拍视频',
+            'description': 'md5:80f461ab5cdae6bbdb70efbf5a1db24f',
+            'duration': 53,
+            'timestamp': 1514264429,
+            'upload_date': '20171226',
+            'thumbnail': r're:https://.*\.jpg',
+            'uploader': '柴犬柴犬',
+            'uploader_id': '5926682210',
+            'uploader_url': 'https://weibo.com/u/5926682210',
+            'view_count': int,
+            'like_count': int,
+            'repost_count': int,
         }
-    }
+    }, {
+        'url': 'https://weibo.com/0/4224132150961381',
+        'note': 'no playback_list example',
+        'only_matching': True,
+    }]
 
     def _real_extract(self, url):
         video_id = self._match_id(url)
-        # to get Referer url for genvisitor
-        webpage = self._download_webpage(url, video_id, note='visit the page')
 
-        weibo_info = self._parse_json(self._search_regex(
+        return self._parse_video_info(self._weibo_download_json(
-            r'var\s+\$render_data\s*=\s*\[({.*})\]\[0\]\s*\|\|\s*{};',
+            f'https://weibo.com/ajax/statuses/show?id={video_id}', video_id))
-            webpage, 'js_code', flags=re.DOTALL),
-            video_id, transform_source=js_to_json)
 
-        status_data = weibo_info.get('status', {})
-        page_info = status_data.get('page_info')
-        title = status_data['status_title']
-        uploader = status_data.get('user', {}).get('screen_name')
 
-        return {
+class WeiboVideoIE(WeiboBaseIE):
-            'id': video_id,
+    _VALID_URL = r'https?://(?:www\.)?weibo\.com/tv/show/(?P<id>\d+:\d+)'
-            'title': title,
+    _TESTS = [{
-            'uploader': uploader,
+        'url': 'https://weibo.com/tv/show/1034:4797699866951785?from=old_pc_videoshow',
-            'url': page_info['media_info']['stream_url']
+        'info_dict': {
+            'id': '4797700463137878',
+            'ext': 'mp4',
+            'display_id': 'LEZDodaiW',
+            'title': '呃，稍微了解了一下靡烟miya，感觉这东西也太二了',
+            'description': '呃，稍微了解了一下靡烟miya，感觉这东西也太二了 http://t.cn/A6aerGsM ​​​',
+            'duration': 76,
+            'timestamp': 1659344278,
+            'upload_date': '20220801',
+            'thumbnail': r're:https://.*\.jpg',
+            'uploader': '君子爱财陈平安',
+            'uploader_id': '3905382233',
+            'uploader_url': 'https://weibo.com/u/3905382233',
+            'view_count': int,
+            'like_count': int,
+            'repost_count': int,
         }
+    }]
+
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
+
+        post_data = f'data={{"Component_Play_Playinfo":{{"oid":"{video_id}"}}}}'.encode()
+        video_info = self._weibo_download_json(
+            f'https://weibo.com/tv/api/component?page=%2Ftv%2Fshow%2F{video_id.replace(":", "%3A")}',
+            video_id, headers={'Referer': url}, data=post_data)['data']['Component_Play_Playinfo']
+        return self.url_result(f'https://weibo.com/0/{video_info["mid"]}', WeiboIE)
+
+
+class WeiboUserIE(WeiboBaseIE):
+    _VALID_URL = r'https?://(?:www\.)?weibo\.com/u/(?P<id>\d+)'
+    _TESTS = [{
+        'url': 'https://weibo.com/u/2066652961?tabtype=video',
+        'info_dict': {
+            'id': '2066652961',
+            'title': '萧影殿下的视频',
+            'description': '萧影殿下的全部视频',
+            'uploader': '萧影殿下',
+        },
+        'playlist_mincount': 195,
+    }]
+
+    def _fetch_page(self, uid, cursor=0, page=1):
+        return self._weibo_download_json(
+            'https://weibo.com/ajax/profile/getWaterFallContent',
+            uid, note=f'Downloading videos page {page}',
+            query={'uid': uid, 'cursor': cursor})['data']
+
+    def _entries(self, uid, first_page):
+        cursor = 0
+        for page in itertools.count(1):
+            response = first_page if page == 1 else self._fetch_page(uid, cursor, page)
+            for video_info in traverse_obj(response, ('list', ..., {dict})):
+                yield self._parse_video_info(video_info)
+            cursor = response.get('next_cursor')
+            if (int_or_none(cursor) or -1) < 0:
+                break
+
+    def _real_extract(self, url):
+        uid = self._match_id(url)
+        first_page = self._fetch_page(uid)
+        uploader = traverse_obj(first_page, ('list', ..., 'user', 'screen_name', {str}), get_all=False)
+        metainfo = {
+            'title': f'{uploader}的视频',
+            'description': f'{uploader}的全部视频',
+            'uploader': uploader,
+        } if uploader else {}
+
+        return self.playlist_result(self._entries(uid, first_page), uid, **metainfo)