import contextlib
import errno
+import functools
import os
import random
import re
QuietMultilinePrinter,
)
from ..utils import (
- NUMBER_RE,
+ IDENTITY,
+ NO_DEFAULT,
LockingUnsupportedError,
Namespace,
+ RetryManager,
classproperty,
decodeArgument,
+ deprecation_warning,
encodeFilename,
- error_to_compat_str,
- float_or_none,
format_bytes,
+ join_nonempty,
+ parse_bytes,
+ remove_start,
sanitize_open,
shell_quote,
timeconvert,
timetuple_from_msec,
+ try_call,
)
ratelimit: Download speed limit, in bytes/sec.
continuedl: Attempt to continue downloads if possible
throttledratelimit: Assume the download is being throttled below this speed (bytes/sec)
- retries: Number of times to retry for HTTP error 5xx
- file_access_retries: Number of times to retry on file access error
+ retries: Number of times to retry for expected network errors.
+ Default is 0 for API, but 10 for CLI
+ file_access_retries: Number of times to retry on file access error (default: 3)
buffersize: Size of download buffer in bytes.
noresizebuffer: Do not automatically resize the download buffer.
continuedl: Try to continue downloads if possible.
for func in (
'deprecation_warning',
+ 'deprecated_feature',
'report_error',
'report_file_already_downloaded',
'report_warning',
@classproperty
def FD_NAME(cls):
- return re.sub(r'(?<!^)(?=[A-Z])', '_', cls.__name__[:-2]).lower()
+ return re.sub(r'(?<=[a-z])(?=[A-Z])', '_', cls.__name__[:-2]).lower()
@staticmethod
def format_seconds(seconds):
+ if seconds is None:
+ return ' Unknown'
time = timetuple_from_msec(seconds * 1000)
if time.hours > 99:
return '--:--:--'
- if not time.hours:
- return '%02d:%02d' % time[1:-1]
return '%02d:%02d:%02d' % time[:-1]
+ @classmethod
+ def format_eta(cls, seconds):
+ return f'{remove_start(cls.format_seconds(seconds), "00:"):>8s}'
+
@staticmethod
def calc_percent(byte_counter, data_len):
if data_len is None:
@staticmethod
def format_percent(percent):
- if percent is None:
- return '---.-%'
- elif percent == 100:
- return '100%'
- return '%6s' % ('%3.1f%%' % percent)
+ return ' N/A%' if percent is None else f'{percent:>5.1f}%'
@staticmethod
def calc_eta(start, now, total, current):
rate = float(current) / dif
return int((float(total) - float(current)) / rate)
- @staticmethod
- def format_eta(eta):
- if eta is None:
- return '--:--'
- return FileDownloader.format_seconds(eta)
-
@staticmethod
def calc_speed(start, now, bytes):
dif = now - start
@staticmethod
def format_speed(speed):
- if speed is None:
- return '%10s' % '---b/s'
- return '%10s' % ('%s/s' % format_bytes(speed))
+ return ' Unknown B/s' if speed is None else f'{format_bytes(speed):>10s}/s'
@staticmethod
def format_retries(retries):
- return 'inf' if retries == float('inf') else '%.0f' % retries
+ return 'inf' if retries == float('inf') else int(retries)
@staticmethod
def best_block_size(elapsed_time, bytes):
@staticmethod
def parse_bytes(bytestr):
"""Parse a string indicating a byte quantity into an integer."""
- matchobj = re.match(rf'(?i)^({NUMBER_RE})([kMGTPEZY]?)$', bytestr)
- if matchobj is None:
- return None
- number = float(matchobj.group(1))
- multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower())
- return int(round(number * multiplier))
+ deprecation_warning('yt_dlp.FileDownloader.parse_bytes is deprecated and '
+ 'may be removed in the future. Use yt_dlp.utils.parse_bytes instead')
+ return parse_bytes(bytestr)
def slow_down(self, start_time, now, byte_counter):
"""Sleep if the download speed is over the rate limit."""
return filename + '.ytdl'
def wrap_file_access(action, *, fatal=False):
- def outer(func):
- def inner(self, *args, **kwargs):
- file_access_retries = self.params.get('file_access_retries', 0)
- retry = 0
- while True:
- try:
- return func(self, *args, **kwargs)
- except OSError as err:
- retry = retry + 1
- if retry > file_access_retries or err.errno not in (errno.EACCES, errno.EINVAL):
- if not fatal:
- self.report_error(f'unable to {action} file: {err}')
- return
- raise
- self.to_screen(
- f'[download] Unable to {action} file due to file access error. '
- f'Retrying (attempt {retry} of {self.format_retries(file_access_retries)}) ...')
- if not self.sleep_retry('file_access', retry):
- time.sleep(0.01)
- return inner
- return outer
+ def error_callback(err, count, retries, *, fd):
+ return RetryManager.report_retry(
+ err, count, retries, info=fd.__to_screen,
+ warn=lambda e: (time.sleep(0.01), fd.to_screen(f'[download] Unable to {action} file: {e}')),
+ error=None if fatal else lambda e: fd.report_error(f'Unable to {action} file: {e}'),
+ sleep_func=fd.params.get('retry_sleep_functions', {}).get('file_access'))
+
+ def wrapper(self, func, *args, **kwargs):
+ for retry in RetryManager(self.params.get('file_access_retries', 3), error_callback, fd=self):
+ try:
+ return func(self, *args, **kwargs)
+ except OSError as err:
+ if err.errno in (errno.EACCES, errno.EINVAL):
+ retry.error = err
+ continue
+ retry.error_callback(err, 1, 0)
+
+ return functools.partial(functools.partialmethod, wrapper)
@wrap_file_access('open', fatal=True)
def sanitize_open(self, filename, open_mode):
elif self.ydl.params.get('logger'):
self._multiline = MultilineLogger(self.ydl.params['logger'], lines)
elif self.params.get('progress_with_newline'):
- self._multiline = BreaklineStatusPrinter(self.ydl._out_files.screen, lines)
+ self._multiline = BreaklineStatusPrinter(self.ydl._out_files.out, lines)
else:
- self._multiline = MultilinePrinter(self.ydl._out_files.screen, lines, not self.params.get('quiet'))
+ self._multiline = MultilinePrinter(self.ydl._out_files.out, lines, not self.params.get('quiet'))
self._multiline.allow_colors = self._multiline._HAVE_FULLCAP and not self.params.get('no_color')
def _finish_multiline_status(self):
)
def _report_progress_status(self, s, default_template):
- for name, style in self.ProgressStyles:
+ for name, style in self.ProgressStyles.items_:
name = f'_{name}_str'
if name not in s:
continue
self._multiline.stream, self._multiline.allow_colors, *args, **kwargs)
def report_progress(self, s):
+ def with_fields(*tups, default=''):
+ for *fields, tmpl in tups:
+ if all(s.get(f) is not None for f in fields):
+ return tmpl
+ return default
+
+ _format_bytes = lambda k: f'{format_bytes(s.get(k)):>10s}'
+
if s['status'] == 'finished':
if self.params.get('noprogress'):
self.to_screen('[download] Download completed')
- msg_template = '100%%'
- if s.get('total_bytes') is not None:
- s['_total_bytes_str'] = format_bytes(s['total_bytes'])
- msg_template += ' of %(_total_bytes_str)s'
- if s.get('elapsed') is not None:
- s['_elapsed_str'] = self.format_seconds(s['elapsed'])
- msg_template += ' in %(_elapsed_str)s'
- s['_percent_str'] = self.format_percent(100)
- self._report_progress_status(s, msg_template)
- return
+ speed = try_call(lambda: s['total_bytes'] / s['elapsed'])
+ s.update({
+ 'speed': speed,
+ '_speed_str': self.format_speed(speed).strip(),
+ '_total_bytes_str': _format_bytes('total_bytes'),
+ '_elapsed_str': self.format_seconds(s.get('elapsed')),
+ '_percent_str': self.format_percent(100),
+ })
+ self._report_progress_status(s, join_nonempty(
+ '100%%',
+ with_fields(('total_bytes', 'of %(_total_bytes_str)s')),
+ with_fields(('elapsed', 'in %(_elapsed_str)s')),
+ with_fields(('speed', 'at %(_speed_str)s')),
+ delim=' '))
if s['status'] != 'downloading':
return
- if s.get('eta') is not None:
- s['_eta_str'] = self.format_eta(s['eta'])
- else:
- s['_eta_str'] = 'Unknown'
-
- if s.get('total_bytes') and s.get('downloaded_bytes') is not None:
- s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes'])
- elif s.get('total_bytes_estimate') and s.get('downloaded_bytes') is not None:
- s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes_estimate'])
- else:
- if s.get('downloaded_bytes') == 0:
- s['_percent_str'] = self.format_percent(0)
- else:
- s['_percent_str'] = 'Unknown %'
-
- if s.get('speed') is not None:
- s['_speed_str'] = self.format_speed(s['speed'])
- else:
- s['_speed_str'] = 'Unknown speed'
-
- if s.get('total_bytes') is not None:
- s['_total_bytes_str'] = format_bytes(s['total_bytes'])
- msg_template = '%(_percent_str)s of %(_total_bytes_str)s at %(_speed_str)s ETA %(_eta_str)s'
- elif s.get('total_bytes_estimate') is not None:
- s['_total_bytes_estimate_str'] = format_bytes(s['total_bytes_estimate'])
- msg_template = '%(_percent_str)s of ~%(_total_bytes_estimate_str)s at %(_speed_str)s ETA %(_eta_str)s'
- else:
- if s.get('downloaded_bytes') is not None:
- s['_downloaded_bytes_str'] = format_bytes(s['downloaded_bytes'])
- if s.get('elapsed'):
- s['_elapsed_str'] = self.format_seconds(s['elapsed'])
- msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s (%(_elapsed_str)s)'
- else:
- msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s'
- else:
- msg_template = '%(_percent_str)s at %(_speed_str)s ETA %(_eta_str)s'
- if s.get('fragment_index') and s.get('fragment_count'):
- msg_template += ' (frag %(fragment_index)s/%(fragment_count)s)'
- elif s.get('fragment_index'):
- msg_template += ' (frag %(fragment_index)s)'
+ s.update({
+ '_eta_str': self.format_eta(s.get('eta')).strip(),
+ '_speed_str': self.format_speed(s.get('speed')),
+ '_percent_str': self.format_percent(try_call(
+ lambda: 100 * s['downloaded_bytes'] / s['total_bytes'],
+ lambda: 100 * s['downloaded_bytes'] / s['total_bytes_estimate'],
+ lambda: s['downloaded_bytes'] == 0 and 0)),
+ '_total_bytes_str': _format_bytes('total_bytes'),
+ '_total_bytes_estimate_str': _format_bytes('total_bytes_estimate'),
+ '_downloaded_bytes_str': _format_bytes('downloaded_bytes'),
+ '_elapsed_str': self.format_seconds(s.get('elapsed')),
+ })
+
+ msg_template = with_fields(
+ ('total_bytes', '%(_percent_str)s of %(_total_bytes_str)s at %(_speed_str)s ETA %(_eta_str)s'),
+ ('total_bytes_estimate', '%(_percent_str)s of ~%(_total_bytes_estimate_str)s at %(_speed_str)s ETA %(_eta_str)s'),
+ ('downloaded_bytes', 'elapsed', '%(_downloaded_bytes_str)s at %(_speed_str)s (%(_elapsed_str)s)'),
+ ('downloaded_bytes', '%(_downloaded_bytes_str)s at %(_speed_str)s'),
+ default='%(_percent_str)s at %(_speed_str)s ETA %(_eta_str)s')
+
+ msg_template += with_fields(
+ ('fragment_index', 'fragment_count', ' (frag %(fragment_index)s/%(fragment_count)s)'),
+ ('fragment_index', ' (frag %(fragment_index)s)'))
self._report_progress_status(s, msg_template)
def report_resuming_byte(self, resume_len):
"""Report attempt to resume at given byte."""
self.to_screen('[download] Resuming download at byte %s' % resume_len)
- def report_retry(self, err, count, retries):
- """Report retry in case of HTTP error 5xx"""
- self.__to_screen(
- '[download] Got server HTTP error: %s. Retrying (attempt %d of %s) ...'
- % (error_to_compat_str(err), count, self.format_retries(retries)))
- self.sleep_retry('http', count)
+ def report_retry(self, err, count, retries, frag_index=NO_DEFAULT, fatal=True):
+ """Report retry"""
+ is_frag = False if frag_index is NO_DEFAULT else 'fragment'
+ RetryManager.report_retry(
+ err, count, retries, info=self.__to_screen,
+ warn=lambda msg: self.__to_screen(f'[download] Got error: {msg}'),
+ error=IDENTITY if not fatal else lambda e: self.report_error(f'\r[download] Got error: {e}'),
+ sleep_func=self.params.get('retry_sleep_functions', {}).get(is_frag or 'http'),
+ suffix=f'fragment{"s" if frag_index is None else f" {frag_index}"}' if is_frag else None)
def report_unable_to_resume(self):
"""Report it was impossible to resume download."""
self.to_screen('[download] Unable to resume')
- def sleep_retry(self, retry_type, count):
- sleep_func = self.params.get('retry_sleep_functions', {}).get(retry_type)
- delay = float_or_none(sleep_func(n=count - 1)) if sleep_func else None
- if delay:
- self.__to_screen(f'Sleeping {delay:.2f} seconds ...')
- time.sleep(delay)
- return sleep_func is not None
-
@staticmethod
def supports_manifest(manifest):
""" Whether the downloader can download the fragments from the manifest.
raise NotImplementedError('This method must be implemented by subclasses')
def _hook_progress(self, status, info_dict):
- if not self._progress_hooks:
- return
+ # Ideally we want to make a copy of the dict, but that is too slow
status['info_dict'] = info_dict
# youtube-dl passes the same status object to all the hooks.
# Some third party scripts seems to be relying on this.