restructured manifest and plugins loading; updated plugins

This commit is contained in:
2025-12-29 22:50:05 -06:00
parent c74f97aca7
commit 21120cd61e
324 changed files with 18088 additions and 15974 deletions

View File

@@ -99,7 +99,7 @@ def _get_suitable_downloader(info_dict, protocol, params, default):
if external_downloader is None:
if info_dict['to_stdout'] and FFmpegFD.can_merge_formats(info_dict, params):
return FFmpegFD
elif external_downloader.lower() != 'native':
elif external_downloader.lower() != 'native' and info_dict.get('impersonate') is None:
ed = get_external_downloader(external_downloader)
if ed.can_download(info_dict, external_downloader):
return ed

View File

@@ -62,7 +62,6 @@ class FileDownloader:
test: Download only first bytes to test the downloader.
min_filesize: Skip files smaller than this size
max_filesize: Skip files larger than this size
xattr_set_filesize: Set ytdl.filesize user xattribute with expected size.
progress_delta: The minimum time between progress output, in seconds
external_downloader_args: A dictionary of downloader keys (in lower case)
and a list of additional command-line arguments for the
@@ -455,14 +454,26 @@ class FileDownloader:
self._finish_multiline_status()
return True, False
sleep_note = ''
if subtitle:
sleep_interval = self.params.get('sleep_interval_subtitles') or 0
else:
min_sleep_interval = self.params.get('sleep_interval') or 0
max_sleep_interval = self.params.get('max_sleep_interval') or 0
if available_at := info_dict.get('available_at'):
forced_sleep_interval = available_at - int(time.time())
if forced_sleep_interval > min_sleep_interval:
sleep_note = 'as required by the site'
min_sleep_interval = forced_sleep_interval
if forced_sleep_interval > max_sleep_interval:
max_sleep_interval = forced_sleep_interval
sleep_interval = random.uniform(
min_sleep_interval, self.params.get('max_sleep_interval') or min_sleep_interval)
min_sleep_interval, max_sleep_interval or min_sleep_interval)
if sleep_interval > 0:
self.to_screen(f'[download] Sleeping {sleep_interval:.2f} seconds ...')
self.to_screen(f'[download] Sleeping {sleep_interval:.2f} seconds {sleep_note}...')
time.sleep(sleep_interval)
ret = self.real_download(filename, info_dict)
@@ -495,3 +506,14 @@ class FileDownloader:
exe = os.path.basename(args[0])
self.write_debug(f'{exe} command line: {shell_quote(args)}')
def _get_impersonate_target(self, info_dict):
impersonate = info_dict.get('impersonate')
if impersonate is None:
return None
available_target, requested_targets = self.ydl._parse_impersonate_targets(impersonate)
if available_target:
return available_target
elif requested_targets:
self.report_warning(self.ydl._unavailable_targets_message(requested_targets))
return None

View File

@@ -3,7 +3,7 @@ import urllib.parse
from . import get_suitable_downloader
from .fragment import FragmentFD
from ..utils import update_url_query, urljoin
from ..utils import ReExtractInfo, update_url_query, urljoin
class DashSegmentsFD(FragmentFD):
@@ -28,6 +28,11 @@ class DashSegmentsFD(FragmentFD):
requested_formats = [{**info_dict, **fmt} for fmt in info_dict.get('requested_formats', [])]
args = []
for fmt in requested_formats or [info_dict]:
# Re-extract if --load-info-json is used and 'fragments' was originally a generator
# See https://github.com/yt-dlp/yt-dlp/issues/13906
if isinstance(fmt['fragments'], str):
raise ReExtractInfo('the stream needs to be re-extracted', expected=True)
try:
fragment_count = 1 if self.params.get('test') else len(fmt['fragments'])
except TypeError:

View File

@@ -563,7 +563,7 @@ class FFmpegFD(ExternalFD):
f'{cookie.name}={cookie.value}; path={cookie.path}; domain={cookie.domain};\r\n'
for cookie in cookies)])
if fmt.get('http_headers') and is_http:
# Trailing \r\n after each HTTP header is important to prevent warning from ffmpeg/avconv:
# Trailing \r\n after each HTTP header is important to prevent warning from ffmpeg:
# [http @ 00000000003d2fa0] No trailing CRLF found in HTTP header.
args.extend(['-headers', ''.join(f'{key}: {val}\r\n' for key, val in fmt['http_headers'].items())])
@@ -572,7 +572,21 @@ class FFmpegFD(ExternalFD):
if end_time:
args += ['-t', str(end_time - start_time)]
args += [*self._configuration_args((f'_i{i + 1}', '_i')), '-i', fmt['url']]
url = fmt['url']
if self.params.get('enable_file_urls') and url.startswith('file:'):
# The default protocol_whitelist is 'file,crypto,data' when reading local m3u8 URLs,
# so only local segments can be read unless we also include 'http,https,tcp,tls'
args += ['-protocol_whitelist', 'file,crypto,data,http,https,tcp,tls']
# ffmpeg incorrectly handles 'file:' URLs by only removing the
# 'file:' prefix and treating the rest as if it's a normal filepath.
# FFmpegPostProcessor also depends on this behavior, so we need to fixup the URLs:
# - On Windows/Cygwin, replace 'file:///' and 'file://localhost/' with 'file:'
# - On *nix, replace 'file://localhost/' with 'file:/'
# Ref: https://github.com/yt-dlp/yt-dlp/issues/13781
# https://trac.ffmpeg.org/ticket/2702
url = re.sub(r'^file://(?:localhost)?/', 'file:' if os.name == 'nt' else 'file:/', url)
args += [*self._configuration_args((f'_i{i + 1}', '_i')), '-i', url]
if not (start_time or end_time) or not self.params.get('force_keyframes_at_cuts'):
args += ['-c', 'copy']
@@ -640,10 +654,6 @@ class FFmpegFD(ExternalFD):
return retval
class AVconvFD(FFmpegFD):
pass
_BY_NAME = {
klass.get_basename(): klass
for name, klass in globals().items()

View File

@@ -149,14 +149,14 @@ class FlvReader(io.BytesIO):
segments_count = self.read_unsigned_char()
segments = []
for _ in range(segments_count):
box_size, box_type, box_data = self.read_box_info()
_box_size, box_type, box_data = self.read_box_info()
assert box_type == b'asrt'
segment = FlvReader(box_data).read_asrt()
segments.append(segment)
fragments_run_count = self.read_unsigned_char()
fragments = []
for _ in range(fragments_run_count):
box_size, box_type, box_data = self.read_box_info()
_box_size, box_type, box_data = self.read_box_info()
assert box_type == b'afrt'
fragments.append(FlvReader(box_data).read_afrt())
@@ -167,7 +167,7 @@ class FlvReader(io.BytesIO):
}
def read_bootstrap_info(self):
total_size, box_type, box_data = self.read_box_info()
_, box_type, box_data = self.read_box_info()
assert box_type == b'abst'
return FlvReader(box_data).read_abst()
@@ -324,9 +324,9 @@ class F4mFD(FragmentFD):
if requested_bitrate is None or len(formats) == 1:
# get the best format
formats = sorted(formats, key=lambda f: f[0])
rate, media = formats[-1]
_, media = formats[-1]
else:
rate, media = next(filter(
_, media = next(filter(
lambda f: int(f[0]) == requested_bitrate, formats))
# Prefer baseURL for relative URLs as per 11.2 of F4M 3.0 spec.

View File

@@ -302,7 +302,7 @@ class FragmentFD(FileDownloader):
elif to_file:
self.try_rename(ctx['tmpfilename'], ctx['filename'])
filetime = ctx.get('fragment_filetime')
if self.params.get('updatetime', True) and filetime:
if self.params.get('updatetime') and filetime:
with contextlib.suppress(Exception):
os.utime(ctx['filename'], (time.time(), filetime))

View File

@@ -94,12 +94,19 @@ class HlsFD(FragmentFD):
can_download, message = self.can_download(s, info_dict, self.params.get('allow_unplayable_formats')), None
if can_download:
has_ffmpeg = FFmpegFD.available()
no_crypto = not Cryptodome.AES and '#EXT-X-KEY:METHOD=AES-128' in s
if no_crypto and has_ffmpeg:
can_download, message = False, 'The stream has AES-128 encryption and pycryptodomex is not available'
elif no_crypto:
message = ('The stream has AES-128 encryption and neither ffmpeg nor pycryptodomex are available; '
'Decryption will be performed natively, but will be extremely slow')
if not Cryptodome.AES and '#EXT-X-KEY:METHOD=AES-128' in s:
# Even if pycryptodomex isn't available, force HlsFD for m3u8s that won't work with ffmpeg
ffmpeg_can_dl = not traverse_obj(info_dict, ((
'extra_param_to_segment_url', 'extra_param_to_key_url',
'hls_media_playlist_data', ('hls_aes', ('uri', 'key', 'iv')),
), any))
message = 'The stream has AES-128 encryption and {} available'.format(
'neither ffmpeg nor pycryptodomex are' if ffmpeg_can_dl and not has_ffmpeg else
'pycryptodomex is not')
if has_ffmpeg and ffmpeg_can_dl:
can_download = False
else:
message += '; decryption will be performed natively, but will be extremely slow'
elif info_dict.get('extractor_key') == 'Generic' and re.search(r'(?m)#EXT-X-MEDIA-SEQUENCE:(?!0$)', s):
install_ffmpeg = '' if has_ffmpeg else 'install ffmpeg and '
message = ('Live HLS streams are not supported by the native downloader. If this is a livestream, '
@@ -198,7 +205,7 @@ class HlsFD(FragmentFD):
line = line.strip()
if line:
if not line.startswith('#'):
if format_index and discontinuity_count != format_index:
if format_index is not None and discontinuity_count != format_index:
continue
if ad_frag_next:
continue
@@ -224,7 +231,7 @@ class HlsFD(FragmentFD):
byte_range = {}
elif line.startswith('#EXT-X-MAP'):
if format_index and discontinuity_count != format_index:
if format_index is not None and discontinuity_count != format_index:
continue
if frag_index > 0:
self.report_error(

View File

@@ -13,12 +13,9 @@ from ..utils import (
ContentTooShortError,
RetryManager,
ThrottledDownload,
XAttrMetadataError,
XAttrUnavailableError,
int_or_none,
parse_http_range,
try_call,
write_xattr,
)
from ..utils.networking import HTTPHeaderDict
@@ -27,6 +24,10 @@ class HttpFD(FileDownloader):
def real_download(self, filename, info_dict):
url = info_dict['url']
request_data = info_dict.get('request_data', None)
request_extensions = {}
impersonate_target = self._get_impersonate_target(info_dict)
if impersonate_target is not None:
request_extensions['impersonate'] = impersonate_target
class DownloadContext(dict):
__getattr__ = dict.get
@@ -109,7 +110,7 @@ class HttpFD(FileDownloader):
if try_call(lambda: range_end >= ctx.content_len):
range_end = ctx.content_len - 1
request = Request(url, request_data, headers)
request = Request(url, request_data, headers, extensions=request_extensions)
has_range = range_start is not None
if has_range:
request.headers['Range'] = f'bytes={int(range_start)}-{int_or_none(range_end) or ""}'
@@ -269,12 +270,6 @@ class HttpFD(FileDownloader):
self.report_error(f'unable to open for writing: {err}')
return False
if self.params.get('xattr_set_filesize', False) and data_len is not None:
try:
write_xattr(ctx.tmpfilename, 'user.ytdl.filesize', str(data_len).encode())
except (XAttrUnavailableError, XAttrMetadataError) as err:
self.report_error(f'unable to set filesize xattr: {err}')
try:
ctx.stream.write(data_block)
except OSError as err:
@@ -348,7 +343,7 @@ class HttpFD(FileDownloader):
self.try_rename(ctx.tmpfilename, ctx.filename)
# Update file modification time
if self.params.get('updatetime', True):
if self.params.get('updatetime'):
info_dict['filetime'] = self.try_utime(ctx.filename, ctx.data.headers.get('last-modified', None))
self._hook_progress({

View File

@@ -5,47 +5,46 @@ import time
from .common import FileDownloader
from .external import FFmpegFD
from ..networking import Request
from ..utils import DownloadError, str_or_none, try_get
from ..networking.websocket import WebSocketResponse
from ..utils import DownloadError, str_or_none, truncate_string
from ..utils.traversal import traverse_obj
class NiconicoLiveFD(FileDownloader):
""" Downloads niconico live without being stopped """
def real_download(self, filename, info_dict):
video_id = info_dict['video_id']
ws_url = info_dict['url']
ws_extractor = info_dict['ws']
ws_origin_host = info_dict['origin']
live_quality = info_dict.get('live_quality', 'high')
live_latency = info_dict.get('live_latency', 'high')
video_id = info_dict['id']
opts = info_dict['downloader_options']
quality, ws_extractor, ws_url = opts['max_quality'], opts['ws'], opts['ws_url']
dl = FFmpegFD(self.ydl, self.params or {})
new_info_dict = info_dict.copy()
new_info_dict.update({
'protocol': 'm3u8',
})
new_info_dict['protocol'] = 'm3u8'
def communicate_ws(reconnect):
if reconnect:
ws = self.ydl.urlopen(Request(ws_url, headers={'Origin': f'https://{ws_origin_host}'}))
# Support --load-info-json as if it is a reconnect attempt
if reconnect or not isinstance(ws_extractor, WebSocketResponse):
ws = self.ydl.urlopen(Request(
ws_url, headers={'Origin': 'https://live.nicovideo.jp'}))
if self.ydl.params.get('verbose', False):
self.to_screen('[debug] Sending startWatching request')
self.write_debug('Sending startWatching request')
ws.send(json.dumps({
'type': 'startWatching',
'data': {
'reconnect': True,
'room': {
'commentable': True,
'protocol': 'webSocket',
},
'stream': {
'quality': live_quality,
'protocol': 'hls+fmp4',
'latency': live_latency,
'accessRightMethod': 'single_cookie',
'chasePlay': False,
'latency': 'high',
'protocol': 'hls',
'quality': quality,
},
'room': {
'protocol': 'webSocket',
'commentable': True,
},
'reconnect': True,
},
'type': 'startWatching',
}))
else:
ws = ws_extractor
@@ -58,7 +57,6 @@ class NiconicoLiveFD(FileDownloader):
if not data or not isinstance(data, dict):
continue
if data.get('type') == 'ping':
# pong back
ws.send(r'{"type":"pong"}')
ws.send(r'{"type":"keepSeat"}')
elif data.get('type') == 'disconnect':
@@ -66,12 +64,10 @@ class NiconicoLiveFD(FileDownloader):
return True
elif data.get('type') == 'error':
self.write_debug(data)
message = try_get(data, lambda x: x['body']['code'], str) or recv
message = traverse_obj(data, ('body', 'code', {str_or_none}), default=recv)
return DownloadError(message)
elif self.ydl.params.get('verbose', False):
if len(recv) > 100:
recv = recv[:100] + '...'
self.to_screen(f'[debug] Server said: {recv}')
self.write_debug(f'Server response: {truncate_string(recv, 100)}')
def ws_main():
reconnect = False
@@ -81,7 +77,8 @@ class NiconicoLiveFD(FileDownloader):
if ret is True:
return
except BaseException as e:
self.to_screen('[{}] {}: Connection error occured, reconnecting after 10 seconds: {}'.format('niconico:live', video_id, str_or_none(e)))
self.to_screen(
f'[niconico:live] {video_id}: Connection error occured, reconnecting after 10 seconds: {e}')
time.sleep(10)
continue
finally: