bellingcat/snscrape
1
0
Fork 0
mirror of https://github.com/bellingcat/snscrape.git synced 2026-06-13 04:48:28 +03:00
Code Issues Packages Projects Releases Wiki Activity

Extract video view count

Browse Source 
Also fix the broken ext values sent to Twitter

Closes #246
This commit is contained in:
JustAnotherArchivist
2021-07-01 17:58:45 +00:00
parent 2bb706feda
commit 0c22608dc7
1 changed files with 6 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
snscrape/modules/twitter.py
View File

@@ -80,6 +80,7 @@ class Video(Medium):
thumbnailUrl: str thumbnailUrl: str
variants: typing.List[VideoVariant] variants: typing.List[VideoVariant]
duration: float duration: float
views: typing.Optional[int] = None
@dataclasses.dataclass @dataclasses.dataclass
@@ -364,6 +365,8 @@ class TwitterAPIScraper(snscrape.base.Scraper):
} }
if medium['type'] == 'video': if medium['type'] == 'video':
mKwargs['duration'] = medium['video_info']['duration_millis'] / 1000 mKwargs['duration'] = medium['video_info']['duration_millis'] / 1000
if (ext := medium['ext']) and (mediaStats := ext['mediaStats']) and isinstance(r := mediaStats['r'], dict) and 'ok' in r and isinstance(r['ok'], dict):
mKwargs['views'] = int(r['ok']['viewCount'])
cls = Video cls = Video
elif medium['type'] == 'animated_gif': elif medium['type'] == 'animated_gif':
cls = Gif cls = Gif
@@ -497,7 +500,7 @@ class TwitterSearchScraper(TwitterAPIScraper):
'cursor': None, 'cursor': None,
'pc': '1', 'pc': '1',
'spelling_corrections': '1', 'spelling_corrections': '1',
'ext': 'ext=mediaStats%2ChighlightedLabel', 'ext': 'mediaStats,highlightedLabel',
} }
params = paginationParams.copy() params = paginationParams.copy()
del params['cursor'] del params['cursor']
@@ -630,7 +633,7 @@ class TwitterProfileScraper(TwitterUserScraper):
'userId': userId, 'userId': userId,
'count': '100', 'count': '100',
'cursor': None, 'cursor': None,
'ext': 'ext=mediaStats%2ChighlightedLabel', 'ext': 'mediaStats,highlightedLabel',
} }
params = paginationParams.copy() params = paginationParams.copy()
del params['cursor'] del params['cursor']
@@ -703,7 +706,7 @@ class TwitterTweetScraper(TwitterAPIScraper):
'count': '20', 'count': '20',
'cursor': None, 'cursor': None,
'include_ext_has_birdwatch_notes': 'false', 'include_ext_has_birdwatch_notes': 'false',
'ext': 'mediaStats%2ChighlightedLabel', 'ext': 'mediaStats,highlightedLabel',
} }
params = paginationParams.copy() params = paginationParams.copy()
del params['cursor'] del params['cursor']