bellingcat/snscrape
1
0
Fork 0
mirror of https://github.com/bellingcat/snscrape.git synced 2026-06-12 20:38:29 +03:00
Code Issues Packages Projects Releases Wiki Activity

Extract alt text for media on Twitter

Browse Source 
Closes #588
This commit is contained in:
JustAnotherArchivist
2023-01-13 03:13:10 +00:00
parent ff5e2d61ee
commit a0414d92cf
1 changed files with 12 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
snscrape/modules/twitter.py
View File

@@ -111,6 +111,7 @@ class Medium:
class Photo(Medium): class Photo(Medium):
previewUrl: str previewUrl: str
fullUrl: str fullUrl: str
altText: typing.Optional[str] = None
@dataclasses.dataclass @dataclasses.dataclass
@@ -126,12 +127,14 @@ class Video(Medium):
variants: typing.List[VideoVariant] variants: typing.List[VideoVariant]
duration: typing.Optional[float] = None duration: typing.Optional[float] = None
views: typing.Optional[int] = None views: typing.Optional[int] = None
altText: typing.Optional[str] = None
@dataclasses.dataclass @dataclasses.dataclass
class Gif(Medium): class Gif(Medium):
thumbnailUrl: str thumbnailUrl: str
variants: typing.List[VideoVariant] variants: typing.List[VideoVariant]
altText: typing.Optional[str] = None
@dataclasses.dataclass @dataclasses.dataclass
@@ -949,10 +952,13 @@ class _TwitterAPIScraper(snscrape.base.Scraper):
if format not in ('jpg', 'png'): if format not in ('jpg', 'png'):
_logger.warning(f'Skipping photo with unknown format on tweet {tweetId}: {format!r}') _logger.warning(f'Skipping photo with unknown format on tweet {tweetId}: {format!r}')
return return
return Photo( mKwargs = {
previewUrl = f'{baseUrl}?format={format}&name=small', 'previewUrl': f'{baseUrl}?format={format}&name=small',
fullUrl = f'{baseUrl}?format={format}&name=large', 'fullUrl': f'{baseUrl}?format={format}&name=large',
) }
if medium.get('ext_alt_text'):
mKwargs['altText'] = medium['ext_alt_text']
return Photo(**mKwargs)
elif medium['type'] == 'video' or medium['type'] == 'animated_gif': elif medium['type'] == 'video' or medium['type'] == 'animated_gif':
variants = [] variants = []
for variant in medium['video_info']['variants']: for variant in medium['video_info']['variants']:
@@ -970,6 +976,8 @@ class _TwitterAPIScraper(snscrape.base.Scraper):
cls = Video cls = Video
elif medium['type'] == 'animated_gif': elif medium['type'] == 'animated_gif':
cls = Gif cls = Gif
if medium.get('ext_alt_text'):
mKwargs['altText'] = medium['ext_alt_text']
return cls(**mKwargs) return cls(**mKwargs)
else: else:
_logger.warning(f'Unsupported medium type on tweet {tweetId}: {medium["type"]!r}') _logger.warning(f'Unsupported medium type on tweet {tweetId}: {medium["type"]!r}')