-from __future__ import division, unicode_literals
-
+import concurrent.futures
+import contextlib
+import json
+import math
import os
+import struct
import time
-import json
-from math import ceil
-
-try:
- import concurrent.futures
- can_threaded_download = True
-except ImportError:
- can_threaded_download = False
from .common import FileDownloader
from .http import HttpFD
-from ..aes import aes_cbc_decrypt_bytes
-from ..compat import (
- compat_urllib_error,
- compat_struct_pack,
-)
-from ..utils import (
- DownloadError,
- error_to_compat_str,
- encodeFilename,
- sanitize_open,
- sanitized_Request,
-)
+from ..aes import aes_cbc_decrypt_bytes, unpad_pkcs7
+from ..compat import compat_os_name
+from ..networking import Request
+from ..networking.exceptions import HTTPError, IncompleteRead
+from ..utils import DownloadError, RetryManager, encodeFilename, traverse_obj
+from ..utils.networking import HTTPHeaderDict
+from ..utils.progress import ProgressCalculator
class HttpQuietDownloader(HttpFD):
def to_screen(self, *args, **kargs):
pass
+ to_console_title = to_screen
+
class FragmentFD(FileDownloader):
"""
Available options:
- fragment_retries: Number of times to retry a fragment for HTTP error (DASH
- and hlsnative only)
+ fragment_retries: Number of times to retry a fragment for HTTP error
+ (DASH and hlsnative only). Default is 0 for API, but 10 for CLI
skip_unavailable_fragments:
Skip unavailable fragments (DASH and hlsnative only)
keep_fragments: Keep downloaded fragments on disk after downloading is
finished
+ concurrent_fragment_downloads: The number of threads to use for native hls and dash downloads
_no_ytdl_file: Don't use .ytdl file
For each incomplete fragment download yt-dlp keeps on disk a special
"""
def report_retry_fragment(self, err, frag_index, count, retries):
- self.to_screen(
- '\r[download] Got server HTTP error: %s. Retrying fragment %d (attempt %d of %s) ...'
- % (error_to_compat_str(err), frag_index, count, self.format_retries(retries)))
+ self.deprecation_warning('yt_dlp.downloader.FragmentFD.report_retry_fragment is deprecated. '
+ 'Use yt_dlp.downloader.FileDownloader.report_retry instead')
+ return self.report_retry(err, count, retries, frag_index)
- def report_skip_fragment(self, frag_index):
- self.to_screen('[download] Skipping fragment %d ...' % frag_index)
+ def report_skip_fragment(self, frag_index, err=None):
+ err = f' {err};' if err else ''
+ self.to_screen(f'[download]{err} Skipping fragment {frag_index:d} ...')
def _prepare_url(self, info_dict, url):
headers = info_dict.get('http_headers')
- return sanitized_Request(url, None, headers) if headers else url
+ return Request(url, None, headers) if headers else url
def _prepare_and_start_frag_download(self, ctx, info_dict):
self._prepare_frag_download(ctx)
self._start_frag_download(ctx, info_dict)
def __do_ytdl_file(self, ctx):
- return not ctx['live'] and not ctx['tmpfilename'] == '-' and not self.params.get('_no_ytdl_file')
+ return ctx['live'] is not True and ctx['tmpfilename'] != '-' and not self.params.get('_no_ytdl_file')
def _read_ytdl_file(self, ctx):
assert 'ytdl_corrupt' not in ctx
- stream, _ = sanitize_open(self.ytdl_filename(ctx['filename']), 'r')
+ stream, _ = self.sanitize_open(self.ytdl_filename(ctx['filename']), 'r')
try:
ytdl_data = json.loads(stream.read())
ctx['fragment_index'] = ytdl_data['downloader']['current_fragment']['index']
stream.close()
def _write_ytdl_file(self, ctx):
- frag_index_stream, _ = sanitize_open(self.ytdl_filename(ctx['filename']), 'w')
+ frag_index_stream, _ = self.sanitize_open(self.ytdl_filename(ctx['filename']), 'w')
try:
downloader = {
'current_fragment': {
'request_data': request_data,
'ctx_id': ctx.get('ctx_id'),
}
- success = ctx['dl'].download(fragment_filename, fragment_info_dict)
+ frag_resume_len = 0
+ if ctx['dl'].params.get('continuedl', True):
+ frag_resume_len = self.filesize_or_none(self.temp_name(fragment_filename))
+ fragment_info_dict['frag_resume_len'] = ctx['frag_resume_len'] = frag_resume_len
+
+ success, _ = ctx['dl'].download(fragment_filename, fragment_info_dict)
if not success:
- return False, None
+ return False
if fragment_info_dict.get('filetime'):
ctx['fragment_filetime'] = fragment_info_dict.get('filetime')
ctx['fragment_filename_sanitized'] = fragment_filename
- return True, self._read_fragment(ctx)
+ return True
def _read_fragment(self, ctx):
- down, frag_sanitized = sanitize_open(ctx['fragment_filename_sanitized'], 'rb')
+ if not ctx.get('fragment_filename_sanitized'):
+ return None
+ try:
+ down, frag_sanitized = self.sanitize_open(ctx['fragment_filename_sanitized'], 'rb')
+ except FileNotFoundError:
+ if ctx.get('live'):
+ return None
+ raise
ctx['fragment_filename_sanitized'] = frag_sanitized
frag_content = down.read()
down.close()
if self.__do_ytdl_file(ctx):
self._write_ytdl_file(ctx)
if not self.params.get('keep_fragments', False):
- os.remove(encodeFilename(ctx['fragment_filename_sanitized']))
+ self.try_remove(encodeFilename(ctx['fragment_filename_sanitized']))
del ctx['fragment_filename_sanitized']
def _prepare_frag_download(self, ctx):
- if 'live' not in ctx:
- ctx['live'] = False
- if not ctx['live']:
+ if not ctx.setdefault('live', False):
total_frags_str = '%d' % ctx['total_frags']
ad_frags = ctx.get('ad_frags', 0)
if ad_frags:
total_frags_str += ' (not including %d ad)' % ad_frags
else:
total_frags_str = 'unknown (live)'
- self.to_screen(
- '[%s] Total fragments: %s' % (self.FD_NAME, total_frags_str))
+ self.to_screen(f'[{self.FD_NAME}] Total fragments: {total_frags_str}')
self.report_destination(ctx['filename'])
- dl = HttpQuietDownloader(
- self.ydl,
- {
- 'continuedl': True,
- 'quiet': True,
- 'noprogress': True,
- 'ratelimit': self.params.get('ratelimit'),
- 'retries': self.params.get('retries', 0),
- 'nopart': self.params.get('nopart', False),
- 'test': self.params.get('test', False),
- }
- )
+ dl = HttpQuietDownloader(self.ydl, {
+ **self.params,
+ 'noprogress': True,
+ 'test': False,
+ 'sleep_interval': 0,
+ 'max_sleep_interval': 0,
+ 'sleep_interval_subtitles': 0,
+ })
tmpfilename = self.temp_name(ctx['filename'])
open_mode = 'wb'
- resume_len = 0
# Establish possible resume length
- if os.path.isfile(encodeFilename(tmpfilename)):
+ resume_len = self.filesize_or_none(tmpfilename)
+ if resume_len > 0:
open_mode = 'ab'
- resume_len = os.path.getsize(encodeFilename(tmpfilename))
# Should be initialized before ytdl file check
ctx.update({
})
if self.__do_ytdl_file(ctx):
- if os.path.isfile(encodeFilename(self.ytdl_filename(ctx['filename']))):
+ ytdl_file_exists = os.path.isfile(encodeFilename(self.ytdl_filename(ctx['filename'])))
+ continuedl = self.params.get('continuedl', True)
+ if continuedl and ytdl_file_exists:
self._read_ytdl_file(ctx)
is_corrupt = ctx.get('ytdl_corrupt') is True
is_inconsistent = ctx['fragment_index'] > 0 and resume_len == 0
if 'ytdl_corrupt' in ctx:
del ctx['ytdl_corrupt']
self._write_ytdl_file(ctx)
+
else:
+ if not continuedl:
+ if ytdl_file_exists:
+ self._read_ytdl_file(ctx)
+ ctx['fragment_index'] = resume_len = 0
self._write_ytdl_file(ctx)
assert ctx['fragment_index'] == 0
- dest_stream, tmpfilename = sanitize_open(tmpfilename, open_mode)
+ dest_stream, tmpfilename = self.sanitize_open(tmpfilename, open_mode)
ctx.update({
'dl': dl,
resume_len = ctx['complete_frags_downloaded_bytes']
total_frags = ctx['total_frags']
ctx_id = ctx.get('ctx_id')
- # This dict stores the download progress, it's updated by the progress
- # hook
+ # Stores the download progress, updated by the progress hook
state = {
'status': 'downloading',
'downloaded_bytes': resume_len,
'tmpfilename': ctx['tmpfilename'],
}
- start = time.time()
- ctx.update({
- 'started': start,
- # Amount of fragment's bytes downloaded by the time of the previous
- # frag progress hook invocation
- 'prev_frag_downloaded_bytes': 0,
- })
+ ctx['started'] = time.time()
+ progress = ProgressCalculator(resume_len)
def frag_progress_hook(s):
if s['status'] not in ('downloading', 'finished'):
return
+ if not total_frags and ctx.get('fragment_count'):
+ state['fragment_count'] = ctx['fragment_count']
+
if ctx_id is not None and s.get('ctx_id') != ctx_id:
return
state['max_progress'] = ctx.get('max_progress')
state['progress_idx'] = ctx.get('progress_idx')
- time_now = time.time()
- state['elapsed'] = time_now - start
+ state['elapsed'] = progress.elapsed
frag_total_bytes = s.get('total_bytes') or 0
s['fragment_info_dict'] = s.pop('info_dict', {})
+
+ # XXX: Fragment resume is not accounted for here
if not ctx['live']:
estimated_size = (
(ctx['complete_frags_downloaded_bytes'] + frag_total_bytes)
/ (state['fragment_index'] + 1) * total_frags)
- state['total_bytes_estimate'] = estimated_size
+ progress.total = estimated_size
+ progress.update(s.get('downloaded_bytes'))
+ state['total_bytes_estimate'] = progress.total
+ else:
+ progress.update(s.get('downloaded_bytes'))
if s['status'] == 'finished':
state['fragment_index'] += 1
ctx['fragment_index'] = state['fragment_index']
- state['downloaded_bytes'] += frag_total_bytes - ctx['prev_frag_downloaded_bytes']
- ctx['complete_frags_downloaded_bytes'] = state['downloaded_bytes']
- ctx['prev_frag_downloaded_bytes'] = 0
- else:
- frag_downloaded_bytes = s['downloaded_bytes']
- state['downloaded_bytes'] += frag_downloaded_bytes - ctx['prev_frag_downloaded_bytes']
- if not ctx['live']:
- state['eta'] = self.calc_eta(
- start, time_now, estimated_size - resume_len,
- state['downloaded_bytes'] - resume_len)
- state['speed'] = s.get('speed') or ctx.get('speed')
- ctx['speed'] = state['speed']
- ctx['prev_frag_downloaded_bytes'] = frag_downloaded_bytes
+ progress.thread_reset()
+
+ state['downloaded_bytes'] = ctx['complete_frags_downloaded_bytes'] = progress.downloaded
+ state['speed'] = ctx['speed'] = progress.speed.smooth
+ state['eta'] = progress.eta.smooth
+
self._hook_progress(state, info_dict)
ctx['dl'].add_progress_hook(frag_progress_hook)
- return start
+ return ctx['started']
def _finish_frag_download(self, ctx, info_dict):
ctx['dest_stream'].close()
if self.__do_ytdl_file(ctx):
- ytdl_filename = encodeFilename(self.ytdl_filename(ctx['filename']))
- if os.path.isfile(ytdl_filename):
- os.remove(ytdl_filename)
+ self.try_remove(self.ytdl_filename(ctx['filename']))
elapsed = time.time() - ctx['started']
- if ctx['tmpfilename'] == '-':
- downloaded_bytes = ctx['complete_frags_downloaded_bytes']
+ to_file = ctx['tmpfilename'] != '-'
+ if to_file:
+ downloaded_bytes = self.filesize_or_none(ctx['tmpfilename'])
else:
+ downloaded_bytes = ctx['complete_frags_downloaded_bytes']
+
+ if not downloaded_bytes:
+ if to_file:
+ self.try_remove(ctx['tmpfilename'])
+ self.report_error('The downloaded file is empty')
+ return False
+ elif to_file:
self.try_rename(ctx['tmpfilename'], ctx['filename'])
- if self.params.get('updatetime', True):
- filetime = ctx.get('fragment_filetime')
- if filetime:
- try:
- os.utime(ctx['filename'], (time.time(), filetime))
- except Exception:
- pass
- downloaded_bytes = os.path.getsize(encodeFilename(ctx['filename']))
+ filetime = ctx.get('fragment_filetime')
+ if self.params.get('updatetime', True) and filetime:
+ with contextlib.suppress(Exception):
+ os.utime(ctx['filename'], (time.time(), filetime))
self._hook_progress({
'downloaded_bytes': downloaded_bytes,
'max_progress': ctx.get('max_progress'),
'progress_idx': ctx.get('progress_idx'),
}, info_dict)
+ return True
def _prepare_external_frag_download(self, ctx):
if 'live' not in ctx:
total_frags_str += ' (not including %d ad)' % ad_frags
else:
total_frags_str = 'unknown (live)'
- self.to_screen(
- '[%s] Total fragments: %s' % (self.FD_NAME, total_frags_str))
+ self.to_screen(f'[{self.FD_NAME}] Total fragments: {total_frags_str}')
tmpfilename = self.temp_name(ctx['filename'])
return _key_cache[url]
def decrypt_fragment(fragment, frag_content):
+ if frag_content is None:
+ return
decrypt_info = fragment.get('decrypt_info')
if not decrypt_info or decrypt_info['METHOD'] != 'AES-128':
return frag_content
- iv = decrypt_info.get('IV') or compat_struct_pack('>8xq', fragment['media_sequence'])
- decrypt_info['KEY'] = decrypt_info.get('KEY') or _get_key(info_dict.get('_decryption_key_url') or decrypt_info['URI'])
+ iv = decrypt_info.get('IV') or struct.pack('>8xq', fragment['media_sequence'])
+ decrypt_info['KEY'] = (decrypt_info.get('KEY')
+ or _get_key(traverse_obj(info_dict, ('hls_aes', 'uri')) or decrypt_info['URI']))
# Don't decrypt the content in tests since the data is explicitly truncated and it's not to a valid block
# size (see https://github.com/ytdl-org/youtube-dl/pull/27660). Tests only care that the correct data downloaded,
# not what it decrypts to.
if self.params.get('test', False):
return frag_content
- return aes_cbc_decrypt_bytes(frag_content, decrypt_info['KEY'], iv)
+ return unpad_pkcs7(aes_cbc_decrypt_bytes(frag_content, decrypt_info['KEY'], iv))
return decrypt_fragment
- def download_and_append_fragments_multiple(self, *args, pack_func=None, finish_func=None):
+ def download_and_append_fragments_multiple(self, *args, **kwargs):
'''
@params (ctx1, fragments1, info_dict1), (ctx2, fragments2, info_dict2), ...
all args must be either tuple or list
'''
+ interrupt_trigger = [True]
max_progress = len(args)
if max_progress == 1:
- return self.download_and_append_fragments(*args[0], pack_func=pack_func, finish_func=finish_func)
- max_workers = self.params.get('concurrent_fragment_downloads', max_progress)
- self._prepare_multiline_status(max_progress)
+ return self.download_and_append_fragments(*args[0], **kwargs)
+ max_workers = self.params.get('concurrent_fragment_downloads', 1)
+ if max_progress > 1:
+ self._prepare_multiline_status(max_progress)
+ is_live = any(traverse_obj(args, (..., 2, 'is_live')))
def thread_func(idx, ctx, fragments, info_dict, tpe):
ctx['max_progress'] = max_progress
ctx['progress_idx'] = idx
- return self.download_and_append_fragments(ctx, fragments, info_dict, pack_func=pack_func, finish_func=finish_func, tpe=tpe)
+ return self.download_and_append_fragments(
+ ctx, fragments, info_dict, **kwargs, tpe=tpe, interrupt_trigger=interrupt_trigger)
class FTPE(concurrent.futures.ThreadPoolExecutor):
# has to stop this or it's going to wait on the worker thread itself
def __exit__(self, exc_type, exc_val, exc_tb):
pass
+ if compat_os_name == 'nt':
+ def future_result(future):
+ while True:
+ try:
+ return future.result(0.1)
+ except KeyboardInterrupt:
+ raise
+ except concurrent.futures.TimeoutError:
+ continue
+ else:
+ def future_result(future):
+ return future.result()
+
+ def interrupt_trigger_iter(fg):
+ for f in fg:
+ if not interrupt_trigger[0]:
+ break
+ yield f
+
spins = []
for idx, (ctx, fragments, info_dict) in enumerate(args):
- tpe = FTPE(ceil(max_workers / max_progress))
- job = tpe.submit(thread_func, idx, ctx, fragments, info_dict, tpe)
+ tpe = FTPE(math.ceil(max_workers / max_progress))
+ job = tpe.submit(thread_func, idx, ctx, interrupt_trigger_iter(fragments), info_dict, tpe)
spins.append((tpe, job))
result = True
for tpe, job in spins:
try:
- result = result and job.result()
+ result = result and future_result(job)
+ except KeyboardInterrupt:
+ interrupt_trigger[0] = False
finally:
tpe.shutdown(wait=True)
+ if not interrupt_trigger[0] and not is_live:
+ raise KeyboardInterrupt()
+ # we expect the user wants to stop and DO WANT the preceding postprocessors to run;
+ # so returning a intermediate result here instead of KeyboardInterrupt on live
+ return result
- self._finish_multiline_status()
- return True
+ def download_and_append_fragments(
+ self, ctx, fragments, info_dict, *, is_fatal=(lambda idx: False),
+ pack_func=(lambda content, idx: content), finish_func=None,
+ tpe=None, interrupt_trigger=(True, )):
- def download_and_append_fragments(self, ctx, fragments, info_dict, *, pack_func=None, finish_func=None, tpe=None):
- fragment_retries = self.params.get('fragment_retries', 0)
- is_fatal = (lambda idx: idx == 0) if self.params.get('skip_unavailable_fragments', True) else (lambda _: True)
- if not pack_func:
- pack_func = lambda frag_content, _: frag_content
+ if not self.params.get('skip_unavailable_fragments', True):
+ is_fatal = lambda _: True
def download_fragment(fragment, ctx):
+ if not interrupt_trigger[0]:
+ return
+
frag_index = ctx['fragment_index'] = fragment['frag_index']
- headers = info_dict.get('http_headers', {}).copy()
+ ctx['last_error'] = None
+ headers = HTTPHeaderDict(info_dict.get('http_headers'))
byte_range = fragment.get('byte_range')
if byte_range:
headers['Range'] = 'bytes=%d-%d' % (byte_range['start'], byte_range['end'] - 1)
# Never skip the first fragment
fatal = is_fatal(fragment.get('index') or (frag_index - 1))
- count, frag_content = 0, None
- while count <= fragment_retries:
- try:
- success, frag_content = self._download_fragment(ctx, fragment['url'], info_dict, headers)
- if not success:
- return False, frag_index
- break
- except compat_urllib_error.HTTPError as err:
- # Unavailable (possibly temporary) fragments may be served.
- # First we try to retry then either skip or abort.
- # See https://github.com/ytdl-org/youtube-dl/issues/10165,
- # https://github.com/ytdl-org/youtube-dl/issues/10448).
- count += 1
- if count <= fragment_retries:
- self.report_retry_fragment(err, frag_index, count, fragment_retries)
- except DownloadError:
- # Don't retry fragment if error occurred during HTTP downloading
- # itself since it has own retry settings
- if not fatal:
- break
- raise
- if count > fragment_retries:
- if not fatal:
- return False, frag_index
- ctx['dest_stream'].close()
- self.report_error('Giving up after %s fragment retries' % fragment_retries)
- return False, frag_index
- return frag_content, frag_index
+ def error_callback(err, count, retries):
+ if fatal and count > retries:
+ ctx['dest_stream'].close()
+ self.report_retry(err, count, retries, frag_index, fatal)
+ ctx['last_error'] = err
+
+ for retry in RetryManager(self.params.get('fragment_retries'), error_callback):
+ try:
+ ctx['fragment_count'] = fragment.get('fragment_count')
+ if not self._download_fragment(
+ ctx, fragment['url'], info_dict, headers, info_dict.get('request_data')):
+ return
+ except (HTTPError, IncompleteRead) as err:
+ retry.error = err
+ continue
+ except DownloadError: # has own retry settings
+ if fatal:
+ raise
def append_fragment(frag_content, frag_index, ctx):
- if not frag_content:
- if not is_fatal(frag_index - 1):
- self.report_skip_fragment(frag_index)
- return True
- else:
- ctx['dest_stream'].close()
- self.report_error(
- 'fragment %s not found, unable to continue' % frag_index)
- return False
- self._append_fragment(ctx, pack_func(frag_content, frag_index))
+ if frag_content:
+ self._append_fragment(ctx, pack_func(frag_content, frag_index))
+ elif not is_fatal(frag_index - 1):
+ self.report_skip_fragment(frag_index, 'fragment not found')
+ else:
+ ctx['dest_stream'].close()
+ self.report_error(f'fragment {frag_index} not found, unable to continue')
+ return False
return True
decrypt_fragment = self.decrypter(info_dict)
- max_workers = self.params.get('concurrent_fragment_downloads', 1)
- if can_threaded_download and max_workers > 1:
-
+ max_workers = math.ceil(
+ self.params.get('concurrent_fragment_downloads', 1) / ctx.get('max_progress', 1))
+ if max_workers > 1:
def _download_fragment(fragment):
ctx_copy = ctx.copy()
- frag_content, frag_index = download_fragment(fragment, ctx_copy)
- return fragment, frag_content, frag_index, ctx_copy.get('fragment_filename_sanitized')
+ download_fragment(fragment, ctx_copy)
+ return fragment, fragment['frag_index'], ctx_copy.get('fragment_filename_sanitized')
- self.report_warning('The download speed shown is only of one thread. This is a known issue and patches are welcome')
with tpe or concurrent.futures.ThreadPoolExecutor(max_workers) as pool:
- for fragment, frag_content, frag_index, frag_filename in pool.map(_download_fragment, fragments):
- ctx['fragment_filename_sanitized'] = frag_filename
- ctx['fragment_index'] = frag_index
- result = append_fragment(decrypt_fragment(fragment, frag_content), frag_index, ctx)
- if not result:
- return False
+ try:
+ for fragment, frag_index, frag_filename in pool.map(_download_fragment, fragments):
+ ctx.update({
+ 'fragment_filename_sanitized': frag_filename,
+ 'fragment_index': frag_index,
+ })
+ if not append_fragment(decrypt_fragment(fragment, self._read_fragment(ctx)), frag_index, ctx):
+ return False
+ except KeyboardInterrupt:
+ self._finish_multiline_status()
+ self.report_error(
+ 'Interrupted by user. Waiting for all threads to shutdown...', is_error=False, tb=False)
+ pool.shutdown(wait=False)
+ raise
else:
for fragment in fragments:
- frag_content, frag_index = download_fragment(fragment, ctx)
- result = append_fragment(decrypt_fragment(fragment, frag_content), frag_index, ctx)
+ if not interrupt_trigger[0]:
+ break
+ try:
+ download_fragment(fragment, ctx)
+ result = append_fragment(
+ decrypt_fragment(fragment, self._read_fragment(ctx)), fragment['frag_index'], ctx)
+ except KeyboardInterrupt:
+ if info_dict.get('is_live'):
+ break
+ raise
if not result:
return False
if finish_func is not None:
ctx['dest_stream'].write(finish_func())
ctx['dest_stream'].flush()
- self._finish_frag_download(ctx, info_dict)
- return True
+ return self._finish_frag_download(ctx, info_dict)