mirror of https://github.com/yt-dlp/yt-dlp.git
Compare commits
5 Commits
22650f4ff3
...
771d7e74f6
Author | SHA1 | Date |
---|---|---|
Renan D | 771d7e74f6 | |
sepro | 96da952504 | |
bashonly | bec9a59e8e | |
bashonly | 036e0d92c6 | |
Renan D. | f80ba18ee9 |
|
@ -1449,6 +1449,8 @@
|
||||||
- **ThisVid**
|
- **ThisVid**
|
||||||
- **ThisVidMember**
|
- **ThisVidMember**
|
||||||
- **ThisVidPlaylist**
|
- **ThisVidPlaylist**
|
||||||
|
- **Threads**
|
||||||
|
- **ThreadsIOS**: Threads' iOS `barcelona://` URL
|
||||||
- **ThreeSpeak**
|
- **ThreeSpeak**
|
||||||
- **ThreeSpeakUser**
|
- **ThreeSpeakUser**
|
||||||
- **TikTok**
|
- **TikTok**
|
||||||
|
|
|
@ -785,6 +785,25 @@ class TestHTTPImpersonateRequestHandler(TestRequestHandlerBase):
|
||||||
assert res.status == 200
|
assert res.status == 200
|
||||||
assert std_headers['user-agent'].lower() not in res.read().decode().lower()
|
assert std_headers['user-agent'].lower() not in res.read().decode().lower()
|
||||||
|
|
||||||
|
def test_response_extensions(self, handler):
|
||||||
|
with handler() as rh:
|
||||||
|
for target in rh.supported_targets:
|
||||||
|
request = Request(
|
||||||
|
f'http://127.0.0.1:{self.http_port}/gen_200', extensions={'impersonate': target})
|
||||||
|
res = validate_and_send(rh, request)
|
||||||
|
assert res.extensions['impersonate'] == rh._get_request_target(request)
|
||||||
|
|
||||||
|
def test_http_error_response_extensions(self, handler):
|
||||||
|
with handler() as rh:
|
||||||
|
for target in rh.supported_targets:
|
||||||
|
request = Request(
|
||||||
|
f'http://127.0.0.1:{self.http_port}/gen_404', extensions={'impersonate': target})
|
||||||
|
try:
|
||||||
|
validate_and_send(rh, request)
|
||||||
|
except HTTPError as e:
|
||||||
|
res = e.response
|
||||||
|
assert res.extensions['impersonate'] == rh._get_request_target(request)
|
||||||
|
|
||||||
|
|
||||||
class TestRequestHandlerMisc:
|
class TestRequestHandlerMisc:
|
||||||
"""Misc generic tests for request handlers, not related to request or validation testing"""
|
"""Misc generic tests for request handlers, not related to request or validation testing"""
|
||||||
|
|
|
@ -2136,6 +2136,11 @@ class YoutubeDL:
|
||||||
|
|
||||||
def _check_formats(self, formats):
|
def _check_formats(self, formats):
|
||||||
for f in formats:
|
for f in formats:
|
||||||
|
working = f.get('__working')
|
||||||
|
if working is not None:
|
||||||
|
if working:
|
||||||
|
yield f
|
||||||
|
continue
|
||||||
self.to_screen('[info] Testing format %s' % f['format_id'])
|
self.to_screen('[info] Testing format %s' % f['format_id'])
|
||||||
path = self.get_output_path('temp')
|
path = self.get_output_path('temp')
|
||||||
if not self._ensure_dir_exists(f'{path}/'):
|
if not self._ensure_dir_exists(f'{path}/'):
|
||||||
|
@ -2152,33 +2157,44 @@ class YoutubeDL:
|
||||||
os.remove(temp_file.name)
|
os.remove(temp_file.name)
|
||||||
except OSError:
|
except OSError:
|
||||||
self.report_warning('Unable to delete temporary file "%s"' % temp_file.name)
|
self.report_warning('Unable to delete temporary file "%s"' % temp_file.name)
|
||||||
|
f['__working'] = success
|
||||||
if success:
|
if success:
|
||||||
yield f
|
yield f
|
||||||
else:
|
else:
|
||||||
self.to_screen('[info] Unable to download format %s. Skipping...' % f['format_id'])
|
self.to_screen('[info] Unable to download format %s. Skipping...' % f['format_id'])
|
||||||
|
|
||||||
|
def _select_formats(self, formats, selector):
|
||||||
|
return list(selector({
|
||||||
|
'formats': formats,
|
||||||
|
'has_merged_format': any('none' not in (f.get('acodec'), f.get('vcodec')) for f in formats),
|
||||||
|
'incomplete_formats': (all(f.get('vcodec') == 'none' for f in formats) # No formats with video
|
||||||
|
or all(f.get('acodec') == 'none' for f in formats)), # OR, No formats with audio
|
||||||
|
}))
|
||||||
|
|
||||||
def _default_format_spec(self, info_dict, download=True):
|
def _default_format_spec(self, info_dict, download=True):
|
||||||
|
download = download and not self.params.get('simulate')
|
||||||
|
prefer_best = download and (
|
||||||
|
self.params['outtmpl']['default'] == '-'
|
||||||
|
or info_dict.get('is_live') and not self.params.get('live_from_start'))
|
||||||
|
|
||||||
def can_merge():
|
def can_merge():
|
||||||
merger = FFmpegMergerPP(self)
|
merger = FFmpegMergerPP(self)
|
||||||
return merger.available and merger.can_merge()
|
return merger.available and merger.can_merge()
|
||||||
|
|
||||||
prefer_best = (
|
if not prefer_best and download and not can_merge():
|
||||||
not self.params.get('simulate')
|
prefer_best = True
|
||||||
and download
|
formats = self._get_formats(info_dict)
|
||||||
and (
|
evaluate_formats = lambda spec: self._select_formats(formats, self.build_format_selector(spec))
|
||||||
not can_merge()
|
if evaluate_formats('b/bv+ba') != evaluate_formats('bv*+ba/b'):
|
||||||
or info_dict.get('is_live') and not self.params.get('live_from_start')
|
self.report_warning('ffmpeg not found. The downloaded format may not be the best available. '
|
||||||
or self.params['outtmpl']['default'] == '-'))
|
'Installing ffmpeg is strongly recommended: https://github.com/yt-dlp/yt-dlp#dependencies')
|
||||||
compat = (
|
|
||||||
prefer_best
|
|
||||||
or self.params.get('allow_multiple_audio_streams', False)
|
|
||||||
or 'format-spec' in self.params['compat_opts'])
|
|
||||||
|
|
||||||
return (
|
compat = (self.params.get('allow_multiple_audio_streams')
|
||||||
'best/bestvideo+bestaudio' if prefer_best
|
or 'format-spec' in self.params['compat_opts'])
|
||||||
else 'bestvideo*+bestaudio/best' if not compat
|
|
||||||
else 'bestvideo+bestaudio/best')
|
return ('best/bestvideo+bestaudio' if prefer_best
|
||||||
|
else 'bestvideo+bestaudio/best' if compat
|
||||||
|
else 'bestvideo*+bestaudio/best')
|
||||||
|
|
||||||
def build_format_selector(self, format_spec):
|
def build_format_selector(self, format_spec):
|
||||||
def syntax_error(note, start):
|
def syntax_error(note, start):
|
||||||
|
@ -2928,12 +2944,7 @@ class YoutubeDL:
|
||||||
self.write_debug(f'Default format spec: {req_format}')
|
self.write_debug(f'Default format spec: {req_format}')
|
||||||
format_selector = self.build_format_selector(req_format)
|
format_selector = self.build_format_selector(req_format)
|
||||||
|
|
||||||
formats_to_download = list(format_selector({
|
formats_to_download = self._select_formats(formats, format_selector)
|
||||||
'formats': formats,
|
|
||||||
'has_merged_format': any('none' not in (f.get('acodec'), f.get('vcodec')) for f in formats),
|
|
||||||
'incomplete_formats': (all(f.get('vcodec') == 'none' for f in formats) # No formats with video
|
|
||||||
or all(f.get('acodec') == 'none' for f in formats)), # OR, No formats with audio
|
|
||||||
}))
|
|
||||||
if interactive_format_selection and not formats_to_download:
|
if interactive_format_selection and not formats_to_download:
|
||||||
self.report_error('Requested format is not available', tb=False, is_error=False)
|
self.report_error('Requested format is not available', tb=False, is_error=False)
|
||||||
continue
|
continue
|
||||||
|
|
|
@ -1988,6 +1988,10 @@ from .thisvid import (
|
||||||
ThisVidMemberIE,
|
ThisVidMemberIE,
|
||||||
ThisVidPlaylistIE,
|
ThisVidPlaylistIE,
|
||||||
)
|
)
|
||||||
|
from .threads import (
|
||||||
|
ThreadsIE,
|
||||||
|
ThreadsIOSIE
|
||||||
|
)
|
||||||
from .threespeak import (
|
from .threespeak import (
|
||||||
ThreeSpeakIE,
|
ThreeSpeakIE,
|
||||||
ThreeSpeakUserIE,
|
ThreeSpeakUserIE,
|
||||||
|
|
|
@ -219,7 +219,29 @@ class PatreonIE(PatreonBaseIE):
|
||||||
'thumbnail': r're:^https?://.+',
|
'thumbnail': r're:^https?://.+',
|
||||||
},
|
},
|
||||||
'params': {'skip_download': 'm3u8'},
|
'params': {'skip_download': 'm3u8'},
|
||||||
|
}, {
|
||||||
|
# multiple attachments/embeds
|
||||||
|
'url': 'https://www.patreon.com/posts/holy-wars-solos-100601977',
|
||||||
|
'playlist_count': 3,
|
||||||
|
'info_dict': {
|
||||||
|
'id': '100601977',
|
||||||
|
'title': '"Holy Wars" (Megadeth) Solos Transcription & Lesson/Analysis',
|
||||||
|
'description': 'md5:d099ab976edfce6de2a65c2b169a88d3',
|
||||||
|
'uploader': 'Bradley Hall',
|
||||||
|
'uploader_id': '24401883',
|
||||||
|
'uploader_url': 'https://www.patreon.com/bradleyhallguitar',
|
||||||
|
'channel_id': '3193932',
|
||||||
|
'channel_url': 'https://www.patreon.com/bradleyhallguitar',
|
||||||
|
'channel_follower_count': int,
|
||||||
|
'timestamp': 1710777855,
|
||||||
|
'upload_date': '20240318',
|
||||||
|
'like_count': int,
|
||||||
|
'comment_count': int,
|
||||||
|
'thumbnail': r're:^https?://.+',
|
||||||
|
},
|
||||||
|
'skip': 'Patron-only content',
|
||||||
}]
|
}]
|
||||||
|
_RETURN_TYPE = 'video'
|
||||||
|
|
||||||
def _real_extract(self, url):
|
def _real_extract(self, url):
|
||||||
video_id = self._match_id(url)
|
video_id = self._match_id(url)
|
||||||
|
@ -234,58 +256,54 @@ class PatreonIE(PatreonBaseIE):
|
||||||
'include': 'audio,user,user_defined_tags,campaign,attachments_media',
|
'include': 'audio,user,user_defined_tags,campaign,attachments_media',
|
||||||
})
|
})
|
||||||
attributes = post['data']['attributes']
|
attributes = post['data']['attributes']
|
||||||
title = attributes['title'].strip()
|
info = traverse_obj(attributes, {
|
||||||
image = attributes.get('image') or {}
|
'title': ('title', {str.strip}),
|
||||||
info = {
|
'description': ('content', {clean_html}),
|
||||||
'id': video_id,
|
'thumbnail': ('image', ('large_url', 'url'), {url_or_none}, any),
|
||||||
'title': title,
|
'timestamp': ('published_at', {parse_iso8601}),
|
||||||
'description': clean_html(attributes.get('content')),
|
'like_count': ('like_count', {int_or_none}),
|
||||||
'thumbnail': image.get('large_url') or image.get('url'),
|
'comment_count': ('comment_count', {int_or_none}),
|
||||||
'timestamp': parse_iso8601(attributes.get('published_at')),
|
})
|
||||||
'like_count': int_or_none(attributes.get('like_count')),
|
|
||||||
'comment_count': int_or_none(attributes.get('comment_count')),
|
|
||||||
}
|
|
||||||
can_view_post = traverse_obj(attributes, 'current_user_can_view')
|
|
||||||
if can_view_post and info['comment_count']:
|
|
||||||
info['__post_extractor'] = self.extract_comments(video_id)
|
|
||||||
|
|
||||||
for i in post.get('included', []):
|
entries = []
|
||||||
i_type = i.get('type')
|
idx = 0
|
||||||
if i_type == 'media':
|
for include in traverse_obj(post, ('included', lambda _, v: v['type'])):
|
||||||
media_attributes = i.get('attributes') or {}
|
include_type = include['type']
|
||||||
download_url = media_attributes.get('download_url')
|
if include_type == 'media':
|
||||||
|
media_attributes = traverse_obj(include, ('attributes', {dict})) or {}
|
||||||
|
download_url = url_or_none(media_attributes.get('download_url'))
|
||||||
ext = mimetype2ext(media_attributes.get('mimetype'))
|
ext = mimetype2ext(media_attributes.get('mimetype'))
|
||||||
|
|
||||||
# if size_bytes is None, this media file is likely unavailable
|
# if size_bytes is None, this media file is likely unavailable
|
||||||
# See: https://github.com/yt-dlp/yt-dlp/issues/4608
|
# See: https://github.com/yt-dlp/yt-dlp/issues/4608
|
||||||
size_bytes = int_or_none(media_attributes.get('size_bytes'))
|
size_bytes = int_or_none(media_attributes.get('size_bytes'))
|
||||||
if download_url and ext in KNOWN_EXTENSIONS and size_bytes is not None:
|
if download_url and ext in KNOWN_EXTENSIONS and size_bytes is not None:
|
||||||
# XXX: what happens if there are multiple attachments?
|
idx += 1
|
||||||
return {
|
entries.append({
|
||||||
**info,
|
'id': f'{video_id}-{idx}',
|
||||||
'ext': ext,
|
'ext': ext,
|
||||||
'filesize': size_bytes,
|
'filesize': size_bytes,
|
||||||
'url': download_url,
|
'url': download_url,
|
||||||
}
|
|
||||||
elif i_type == 'user':
|
|
||||||
user_attributes = i.get('attributes')
|
|
||||||
if user_attributes:
|
|
||||||
info.update({
|
|
||||||
'uploader': user_attributes.get('full_name'),
|
|
||||||
'uploader_id': str_or_none(i.get('id')),
|
|
||||||
'uploader_url': user_attributes.get('url'),
|
|
||||||
})
|
})
|
||||||
|
|
||||||
elif i_type == 'post_tag':
|
elif include_type == 'user':
|
||||||
info.setdefault('tags', []).append(traverse_obj(i, ('attributes', 'value')))
|
info.update(traverse_obj(include, {
|
||||||
|
'uploader': ('attributes', 'full_name', {str}),
|
||||||
|
'uploader_id': ('id', {str_or_none}),
|
||||||
|
'uploader_url': ('attributes', 'url', {url_or_none}),
|
||||||
|
}))
|
||||||
|
|
||||||
elif i_type == 'campaign':
|
elif include_type == 'post_tag':
|
||||||
info.update({
|
if post_tag := traverse_obj(include, ('attributes', 'value', {str})):
|
||||||
'channel': traverse_obj(i, ('attributes', 'title')),
|
info.setdefault('tags', []).append(post_tag)
|
||||||
'channel_id': str_or_none(i.get('id')),
|
|
||||||
'channel_url': traverse_obj(i, ('attributes', 'url')),
|
elif include_type == 'campaign':
|
||||||
'channel_follower_count': int_or_none(traverse_obj(i, ('attributes', 'patron_count'))),
|
info.update(traverse_obj(include, {
|
||||||
})
|
'channel': ('attributes', 'title', {str}),
|
||||||
|
'channel_id': ('id', {str_or_none}),
|
||||||
|
'channel_url': ('attributes', 'url', {url_or_none}),
|
||||||
|
'channel_follower_count': ('attributes', 'patron_count', {int_or_none}),
|
||||||
|
}))
|
||||||
|
|
||||||
# handle Vimeo embeds
|
# handle Vimeo embeds
|
||||||
if traverse_obj(attributes, ('embed', 'provider')) == 'Vimeo':
|
if traverse_obj(attributes, ('embed', 'provider')) == 'Vimeo':
|
||||||
|
@ -296,36 +314,50 @@ class PatreonIE(PatreonBaseIE):
|
||||||
v_url, video_id, 'Checking Vimeo embed URL',
|
v_url, video_id, 'Checking Vimeo embed URL',
|
||||||
headers={'Referer': 'https://patreon.com/'},
|
headers={'Referer': 'https://patreon.com/'},
|
||||||
fatal=False, errnote=False):
|
fatal=False, errnote=False):
|
||||||
return self.url_result(
|
entries.append(self.url_result(
|
||||||
VimeoIE._smuggle_referrer(v_url, 'https://patreon.com/'),
|
VimeoIE._smuggle_referrer(v_url, 'https://patreon.com/'),
|
||||||
VimeoIE, url_transparent=True, **info)
|
VimeoIE, url_transparent=True))
|
||||||
|
|
||||||
embed_url = traverse_obj(attributes, ('embed', 'url', {url_or_none}))
|
embed_url = traverse_obj(attributes, ('embed', 'url', {url_or_none}))
|
||||||
if embed_url and self._request_webpage(embed_url, video_id, 'Checking embed URL', fatal=False, errnote=False):
|
if embed_url and self._request_webpage(embed_url, video_id, 'Checking embed URL', fatal=False, errnote=False):
|
||||||
return self.url_result(embed_url, **info)
|
entries.append(self.url_result(embed_url))
|
||||||
|
|
||||||
post_file = traverse_obj(attributes, 'post_file')
|
post_file = traverse_obj(attributes, ('post_file', {dict}))
|
||||||
if post_file:
|
if post_file:
|
||||||
name = post_file.get('name')
|
name = post_file.get('name')
|
||||||
ext = determine_ext(name)
|
ext = determine_ext(name)
|
||||||
if ext in KNOWN_EXTENSIONS:
|
if ext in KNOWN_EXTENSIONS:
|
||||||
return {
|
entries.append({
|
||||||
**info,
|
'id': video_id,
|
||||||
'ext': ext,
|
'ext': ext,
|
||||||
'url': post_file['url'],
|
'url': post_file['url'],
|
||||||
}
|
})
|
||||||
elif name == 'video' or determine_ext(post_file.get('url')) == 'm3u8':
|
elif name == 'video' or determine_ext(post_file.get('url')) == 'm3u8':
|
||||||
formats, subtitles = self._extract_m3u8_formats_and_subtitles(post_file['url'], video_id)
|
formats, subtitles = self._extract_m3u8_formats_and_subtitles(post_file['url'], video_id)
|
||||||
return {
|
entries.append({
|
||||||
**info,
|
'id': video_id,
|
||||||
'formats': formats,
|
'formats': formats,
|
||||||
'subtitles': subtitles,
|
'subtitles': subtitles,
|
||||||
}
|
})
|
||||||
|
|
||||||
if can_view_post is False:
|
can_view_post = traverse_obj(attributes, 'current_user_can_view')
|
||||||
|
comments = None
|
||||||
|
if can_view_post and info.get('comment_count'):
|
||||||
|
comments = self.extract_comments(video_id)
|
||||||
|
|
||||||
|
if not entries and can_view_post is False:
|
||||||
self.raise_no_formats('You do not have access to this post', video_id=video_id, expected=True)
|
self.raise_no_formats('You do not have access to this post', video_id=video_id, expected=True)
|
||||||
else:
|
elif not entries:
|
||||||
self.raise_no_formats('No supported media found in this post', video_id=video_id, expected=True)
|
self.raise_no_formats('No supported media found in this post', video_id=video_id, expected=True)
|
||||||
|
elif len(entries) == 1:
|
||||||
|
info.update(entries[0])
|
||||||
|
else:
|
||||||
|
for entry in entries:
|
||||||
|
entry.update(info)
|
||||||
|
return self.playlist_result(entries, video_id, **info, __post_extractor=comments)
|
||||||
|
|
||||||
|
info['id'] = video_id
|
||||||
|
info['__post_extractor'] = comments
|
||||||
return info
|
return info
|
||||||
|
|
||||||
def _get_comments(self, post_id):
|
def _get_comments(self, post_id):
|
||||||
|
|
|
@ -0,0 +1,157 @@
|
||||||
|
from .common import InfoExtractor
|
||||||
|
from ..utils import (
|
||||||
|
strftime_or_none,
|
||||||
|
traverse_obj,
|
||||||
|
remove_end,
|
||||||
|
strip_or_none
|
||||||
|
)
|
||||||
|
|
||||||
|
|
||||||
|
class ThreadsIE(InfoExtractor):
|
||||||
|
_VALID_URL = r'https?://(?:www\.)?threads\.net/(?P<uploader>[^/]+)/post/(?P<id>[^/?#&]+)/?(?P<embed>embed.*?)?'
|
||||||
|
|
||||||
|
_TESTS = [{
|
||||||
|
'url': 'https://www.threads.net/@tntsportsbr/post/C6cqebdCfBi',
|
||||||
|
'info_dict': {
|
||||||
|
'id': 'C6cqebdCfBi',
|
||||||
|
'ext': 'mp4',
|
||||||
|
'title': 'md5:062673d04195aa2d99b8d7a11798cb9d',
|
||||||
|
'description': 'md5:fe0c73f9a892fb92efcc67cc075561b0',
|
||||||
|
'uploader': 'TNT Sports Brasil',
|
||||||
|
'uploader_id': 'tntsportsbr',
|
||||||
|
'uploader_url': 'https://www.threads.net/@tntsportsbr',
|
||||||
|
'channel': 'tntsportsbr',
|
||||||
|
'channel_url': 'https://www.threads.net/@tntsportsbr',
|
||||||
|
'timestamp': 1714613811,
|
||||||
|
'upload_date': '20240502',
|
||||||
|
'like_count': int,
|
||||||
|
'channel_is_verified': bool,
|
||||||
|
'thumbnail': r're:^https?://.*\.jpg'
|
||||||
|
}
|
||||||
|
}, {
|
||||||
|
'url': 'https://www.threads.net/@felipebecari/post/C6cM_yNPHCF',
|
||||||
|
'info_dict': {
|
||||||
|
'id': 'C6cM_yNPHCF',
|
||||||
|
'ext': 'mp4',
|
||||||
|
'title': '@felipebecari • Sobre o futuro dos dois últimos resgatados: tem muita notícia boa! 🐶❤️',
|
||||||
|
'description': 'Sobre o futuro dos dois últimos resgatados: tem muita notícia boa! 🐶❤️',
|
||||||
|
'uploader': 'Felipe Becari',
|
||||||
|
'uploader_id': 'felipebecari',
|
||||||
|
'uploader_url': 'https://www.threads.net/@felipebecari',
|
||||||
|
'channel': 'felipebecari',
|
||||||
|
'channel_url': 'https://www.threads.net/@felipebecari',
|
||||||
|
'timestamp': 1714598318,
|
||||||
|
'upload_date': '20240501',
|
||||||
|
'like_count': int,
|
||||||
|
'channel_is_verified': bool,
|
||||||
|
'thumbnail': r're:^https?://.*\.jpg'
|
||||||
|
}
|
||||||
|
}]
|
||||||
|
|
||||||
|
def _real_extract(self, url):
|
||||||
|
video_id = self._match_id(url)
|
||||||
|
webpage = self._download_webpage(url, video_id)
|
||||||
|
metadata = {}
|
||||||
|
|
||||||
|
# Try getting videos from json
|
||||||
|
json_data = self._search_regex(
|
||||||
|
r'<script[^>]+>(.*"code":"%s".*)</script>' % video_id,
|
||||||
|
webpage, 'main json', fatal=True)
|
||||||
|
|
||||||
|
result = self._search_json(
|
||||||
|
r'"result":', json_data,
|
||||||
|
'result data', video_id, fatal=True)
|
||||||
|
|
||||||
|
edges = traverse_obj(result, ('data', 'data', 'edges'))
|
||||||
|
|
||||||
|
for node in edges:
|
||||||
|
items = traverse_obj(node, ('node', 'thread_items'))
|
||||||
|
|
||||||
|
for item in items:
|
||||||
|
post = item.get('post')
|
||||||
|
|
||||||
|
if post and post.get('code') == video_id:
|
||||||
|
formats = []
|
||||||
|
thumbnails = []
|
||||||
|
|
||||||
|
# Videos
|
||||||
|
if (post.get('carousel_media') is not None): # Handle multiple videos posts
|
||||||
|
media_list = post.get('carousel_media')
|
||||||
|
else:
|
||||||
|
media_list = [post]
|
||||||
|
|
||||||
|
for media in media_list:
|
||||||
|
videos = media.get('video_versions')
|
||||||
|
|
||||||
|
for video in videos:
|
||||||
|
formats.append({
|
||||||
|
'format_id': '%s-%s' % (media.get('pk'), video['type']), # id-type
|
||||||
|
'url': video['url'],
|
||||||
|
'width': media.get('original_width'),
|
||||||
|
'height': media.get('original_height'),
|
||||||
|
})
|
||||||
|
|
||||||
|
# Thumbnails
|
||||||
|
thumbs = traverse_obj(post, ('image_versions2', 'candidates'))
|
||||||
|
|
||||||
|
for thumb in thumbs:
|
||||||
|
thumbnails.append({
|
||||||
|
'url': thumb['url'],
|
||||||
|
'width': thumb['width'],
|
||||||
|
'height': thumb['height'],
|
||||||
|
})
|
||||||
|
|
||||||
|
# Metadata
|
||||||
|
metadata.setdefault('uploader_id', traverse_obj(post, ('user', 'username')))
|
||||||
|
metadata.setdefault('channel_is_verified', traverse_obj(post, ('user', 'is_verified')))
|
||||||
|
metadata.setdefault('uploader_url', 'https://www.threads.net/@%s' % traverse_obj(post, ('user', 'username')))
|
||||||
|
metadata.setdefault('timestamp', post.get('taken_at'))
|
||||||
|
metadata.setdefault('like_count', post.get('like_count'))
|
||||||
|
|
||||||
|
# Try getting metadata
|
||||||
|
metadata['id'] = video_id
|
||||||
|
metadata['title'] = strip_or_none(remove_end(self._html_extract_title(webpage), '• Threads'))
|
||||||
|
metadata['description'] = self._og_search_description(webpage)
|
||||||
|
|
||||||
|
metadata['channel'] = metadata.get('uploader_id')
|
||||||
|
metadata['channel_url'] = metadata.get('uploader_url')
|
||||||
|
metadata['uploader'] = self._search_regex(r'(.*?) \(', self._og_search_title(webpage), 'uploader')
|
||||||
|
metadata['upload_date'] = strftime_or_none(metadata.get('timestamp'))
|
||||||
|
|
||||||
|
return {
|
||||||
|
**metadata,
|
||||||
|
'formats': formats,
|
||||||
|
'thumbnails': thumbnails
|
||||||
|
}
|
||||||
|
|
||||||
|
|
||||||
|
class ThreadsIOSIE(InfoExtractor):
|
||||||
|
IE_DESC = 'IOS barcelona:// URL'
|
||||||
|
_VALID_URL = r'barcelona://media\?shortcode=(?P<id>[^/?#&]+)'
|
||||||
|
_TESTS = [{
|
||||||
|
'url': 'barcelona://media?shortcode=C6fDehepo5D',
|
||||||
|
'info_dict': {
|
||||||
|
'id': 'C6fDehepo5D',
|
||||||
|
'ext': 'mp4',
|
||||||
|
'title': 'md5:dc92f960981b8b3a33eba9681e9fdfc6',
|
||||||
|
'description': 'md5:0c36a7e67e1517459bc0334dba932164',
|
||||||
|
'uploader': 'Sa\u0303o Paulo Futebol Clube',
|
||||||
|
'uploader_id': 'saopaulofc',
|
||||||
|
'uploader_url': 'https://www.threads.net/@saopaulofc',
|
||||||
|
'channel': 'saopaulofc',
|
||||||
|
'channel_url': 'https://www.threads.net/@saopaulofc',
|
||||||
|
'timestamp': 1714694014,
|
||||||
|
'upload_date': '20240502',
|
||||||
|
'like_count': int,
|
||||||
|
'channel_is_verified': bool,
|
||||||
|
'thumbnail': r're:^https?://.*\.jpg'
|
||||||
|
},
|
||||||
|
'add_ie': ['Threads']
|
||||||
|
}]
|
||||||
|
|
||||||
|
def _real_extract(self, url):
|
||||||
|
video_id = self._match_id(url)
|
||||||
|
|
||||||
|
# Threads doesn't care about the user url, it redirects to the right one
|
||||||
|
# So we use ** instead so that we don't need to find it
|
||||||
|
return self.url_result(f'http://www.threads.net/**/post/{video_id}', ThreadsIE, video_id)
|
|
@ -132,6 +132,16 @@ class CurlCFFIRH(ImpersonateRequestHandler, InstanceStoreMixin):
|
||||||
extensions.pop('cookiejar', None)
|
extensions.pop('cookiejar', None)
|
||||||
extensions.pop('timeout', None)
|
extensions.pop('timeout', None)
|
||||||
|
|
||||||
|
def send(self, request: Request) -> Response:
|
||||||
|
target = self._get_request_target(request)
|
||||||
|
try:
|
||||||
|
response = super().send(request)
|
||||||
|
except HTTPError as e:
|
||||||
|
e.response.extensions['impersonate'] = target
|
||||||
|
raise
|
||||||
|
response.extensions['impersonate'] = target
|
||||||
|
return response
|
||||||
|
|
||||||
def _send(self, request: Request):
|
def _send(self, request: Request):
|
||||||
max_redirects_exceeded = False
|
max_redirects_exceeded = False
|
||||||
session: curl_cffi.requests.Session = self._get_instance(
|
session: curl_cffi.requests.Session = self._get_instance(
|
||||||
|
|
|
@ -497,6 +497,7 @@ class Response(io.IOBase):
|
||||||
@param headers: response headers.
|
@param headers: response headers.
|
||||||
@param status: Response HTTP status code. Default is 200 OK.
|
@param status: Response HTTP status code. Default is 200 OK.
|
||||||
@param reason: HTTP status reason. Will use built-in reasons based on status code if not provided.
|
@param reason: HTTP status reason. Will use built-in reasons based on status code if not provided.
|
||||||
|
@param extensions: Dictionary of handler-specific response extensions.
|
||||||
"""
|
"""
|
||||||
|
|
||||||
def __init__(
|
def __init__(
|
||||||
|
@ -505,7 +506,9 @@ class Response(io.IOBase):
|
||||||
url: str,
|
url: str,
|
||||||
headers: Mapping[str, str],
|
headers: Mapping[str, str],
|
||||||
status: int = 200,
|
status: int = 200,
|
||||||
reason: str = None):
|
reason: str = None,
|
||||||
|
extensions: dict = None
|
||||||
|
):
|
||||||
|
|
||||||
self.fp = fp
|
self.fp = fp
|
||||||
self.headers = Message()
|
self.headers = Message()
|
||||||
|
@ -517,6 +520,7 @@ class Response(io.IOBase):
|
||||||
self.reason = reason or HTTPStatus(status).phrase
|
self.reason = reason or HTTPStatus(status).phrase
|
||||||
except ValueError:
|
except ValueError:
|
||||||
self.reason = None
|
self.reason = None
|
||||||
|
self.extensions = extensions or {}
|
||||||
|
|
||||||
def readable(self):
|
def readable(self):
|
||||||
return self.fp.readable()
|
return self.fp.readable()
|
||||||
|
|
Loading…
Reference in New Issue