1
0
mirror of https://github.com/yt-dlp/yt-dlp.git synced 2024-11-18 00:45:03 +01:00
yt-dlp/yt_dlp/extractor/imgur.py
pukkandan 86e5f3ed2e
[cleanup] Upgrade syntax
Using https://github.com/asottile/pyupgrade

1. `__future__` imports and `coding: utf-8` were removed
2. Files were rewritten with `pyupgrade --py36-plus --keep-percent-format`
3. f-strings were cherry-picked from `pyupgrade --py36-plus`

Extractors are left untouched (except removing header) to avoid unnecessary merge conflicts
2022-04-12 05:32:51 +05:30

153 lines
4.9 KiB
Python

import re
from .common import InfoExtractor
from ..utils import (
int_or_none,
js_to_json,
mimetype2ext,
ExtractorError,
)
class ImgurIE(InfoExtractor):
_VALID_URL = r'https?://(?:i\.)?imgur\.com/(?!(?:a|gallery|(?:t(?:opic)?|r)/[^/]+)/)(?P<id>[a-zA-Z0-9]+)'
_TESTS = [{
'url': 'https://i.imgur.com/A61SaA1.gifv',
'info_dict': {
'id': 'A61SaA1',
'ext': 'mp4',
'title': 're:Imgur GIF$|MRW gifv is up and running without any bugs$',
},
}, {
'url': 'https://imgur.com/A61SaA1',
'only_matching': True,
}, {
'url': 'https://i.imgur.com/crGpqCV.mp4',
'only_matching': True,
}, {
# no title
'url': 'https://i.imgur.com/jxBXAMC.gifv',
'only_matching': True,
}]
def _real_extract(self, url):
video_id = self._match_id(url)
webpage = self._download_webpage(
'https://i.imgur.com/{id}.gifv'.format(id=video_id), video_id)
width = int_or_none(self._og_search_property(
'video:width', webpage, default=None))
height = int_or_none(self._og_search_property(
'video:height', webpage, default=None))
video_elements = self._search_regex(
r'(?s)<div class="video-elements">(.*?)</div>',
webpage, 'video elements', default=None)
if not video_elements:
raise ExtractorError(
'No sources found for video %s. Maybe an image?' % video_id,
expected=True)
formats = []
for m in re.finditer(r'<source\s+src="(?P<src>[^"]+)"\s+type="(?P<type>[^"]+)"', video_elements):
formats.append({
'format_id': m.group('type').partition('/')[2],
'url': self._proto_relative_url(m.group('src')),
'ext': mimetype2ext(m.group('type')),
'width': width,
'height': height,
'http_headers': {
'User-Agent': 'yt-dlp (like wget)',
},
})
gif_json = self._search_regex(
r'(?s)var\s+videoItem\s*=\s*(\{.*?\})',
webpage, 'GIF code', fatal=False)
if gif_json:
gifd = self._parse_json(
gif_json, video_id, transform_source=js_to_json)
formats.append({
'format_id': 'gif',
'preference': -10, # gifs are worse than videos
'width': width,
'height': height,
'ext': 'gif',
'acodec': 'none',
'vcodec': 'gif',
'container': 'gif',
'url': self._proto_relative_url(gifd['gifUrl']),
'filesize': gifd.get('size'),
'http_headers': {
'User-Agent': 'yt-dlp (like wget)',
},
})
self._sort_formats(formats)
return {
'id': video_id,
'formats': formats,
'title': self._og_search_title(webpage, default=video_id),
}
class ImgurGalleryIE(InfoExtractor):
IE_NAME = 'imgur:gallery'
_VALID_URL = r'https?://(?:i\.)?imgur\.com/(?:gallery|(?:t(?:opic)?|r)/[^/]+)/(?P<id>[a-zA-Z0-9]+)'
_TESTS = [{
'url': 'http://imgur.com/gallery/Q95ko',
'info_dict': {
'id': 'Q95ko',
'title': 'Adding faces make every GIF better',
},
'playlist_count': 25,
}, {
'url': 'http://imgur.com/topic/Aww/ll5Vk',
'only_matching': True,
}, {
'url': 'https://imgur.com/gallery/YcAQlkx',
'info_dict': {
'id': 'YcAQlkx',
'ext': 'mp4',
'title': 'Classic Steve Carell gif...cracks me up everytime....damn the repost downvotes....',
}
}, {
'url': 'http://imgur.com/topic/Funny/N8rOudd',
'only_matching': True,
}, {
'url': 'http://imgur.com/r/aww/VQcQPhM',
'only_matching': True,
}]
def _real_extract(self, url):
gallery_id = self._match_id(url)
data = self._download_json(
'https://imgur.com/gallery/%s.json' % gallery_id,
gallery_id)['data']['image']
if data.get('is_album'):
entries = [
self.url_result('http://imgur.com/%s' % image['hash'], ImgurIE.ie_key(), image['hash'])
for image in data['album_images']['images'] if image.get('hash')]
return self.playlist_result(entries, gallery_id, data.get('title'), data.get('description'))
return self.url_result('http://imgur.com/%s' % gallery_id, ImgurIE.ie_key(), gallery_id)
class ImgurAlbumIE(ImgurGalleryIE):
IE_NAME = 'imgur:album'
_VALID_URL = r'https?://(?:i\.)?imgur\.com/a/(?P<id>[a-zA-Z0-9]+)'
_TESTS = [{
'url': 'http://imgur.com/a/j6Orj',
'info_dict': {
'id': 'j6Orj',
'title': 'A Literary Analysis of "Star Wars: The Force Awakens"',
},
'playlist_count': 12,
}]