yt_dlp/downloader/common.py

   1 from __future__ import division, unicode_literals
   2
   3 import os
   4 import re
   5 import time
   6 import random
   7
   8 from ..utils import (
   9     decodeArgument,
  10     encodeFilename,
  11     error_to_compat_str,
  12     format_bytes,
  13     shell_quote,
  14     timeconvert,
  15     timetuple_from_msec,
  16 )
  17 from ..minicurses import (
  18     MultilineLogger,
  19     MultilinePrinter,
  20     QuietMultilinePrinter,
  21     BreaklineStatusPrinter
  22 )
  23
  24
  25 class FileDownloader(object):
  26     """File Downloader class.
  27
  28     File downloader objects are the ones responsible of downloading the
  29     actual video file and writing it to disk.
  30
  31     File downloaders accept a lot of parameters. In order not to saturate
  32     the object constructor with arguments, it receives a dictionary of
  33     options instead.
  34
  35     Available options:
  36
  37     verbose:            Print additional info to stdout.
  38     quiet:              Do not print messages to stdout.
  39     ratelimit:          Download speed limit, in bytes/sec.
  40     throttledratelimit: Assume the download is being throttled below this speed (bytes/sec)
  41     retries:            Number of times to retry for HTTP error 5xx
  42     buffersize:         Size of download buffer in bytes.
  43     noresizebuffer:     Do not automatically resize the download buffer.
  44     continuedl:         Try to continue downloads if possible.
  45     noprogress:         Do not print the progress bar.
  46     nopart:             Do not use temporary .part files.
  47     updatetime:         Use the Last-modified header to set output file timestamps.
  48     test:               Download only first bytes to test the downloader.
  49     min_filesize:       Skip files smaller than this size
  50     max_filesize:       Skip files larger than this size
  51     xattr_set_filesize: Set ytdl.filesize user xattribute with expected size.
  52     external_downloader_args:  A dictionary of downloader keys (in lower case)
  53                         and a list of additional command-line arguments for the
  54                         executable. Use 'default' as the name for arguments to be
  55                         passed to all downloaders. For compatibility with youtube-dl,
  56                         a single list of args can also be used
  57     hls_use_mpegts:     Use the mpegts container for HLS videos.
  58     http_chunk_size:    Size of a chunk for chunk-based HTTP downloading. May be
  59                         useful for bypassing bandwidth throttling imposed by
  60                         a webserver (experimental)
  61     progress_template:  See YoutubeDL.py
  62
  63     Subclasses of this one must re-define the real_download method.
  64     """
  65
  66     _TEST_FILE_SIZE = 10241
  67     params = None
  68
  69     def __init__(self, ydl, params):
  70         """Create a FileDownloader object with the given options."""
  71         self.ydl = ydl
  72         self._progress_hooks = []
  73         self.params = params
  74         self._prepare_multiline_status()
  75         self.add_progress_hook(self.report_progress)
  76
  77     @staticmethod
  78     def format_seconds(seconds):
  79         time = timetuple_from_msec(seconds * 1000)
  80         if time.hours > 99:
  81             return '--:--:--'
  82         if not time.hours:
  83             return '%02d:%02d' % time[1:-1]
  84         return '%02d:%02d:%02d' % time[:-1]
  85
  86     @staticmethod
  87     def calc_percent(byte_counter, data_len):
  88         if data_len is None:
  89             return None
  90         return float(byte_counter) / float(data_len) * 100.0
  91
  92     @staticmethod
  93     def format_percent(percent):
  94         if percent is None:
  95             return '---.-%'
  96         elif percent == 100:
  97             return '100%'
  98         return '%6s' % ('%3.1f%%' % percent)
  99
 100     @staticmethod
 101     def calc_eta(start, now, total, current):
 102         if total is None:
 103             return None
 104         if now is None:
 105             now = time.time()
 106         dif = now - start
 107         if current == 0 or dif < 0.001:  # One millisecond
 108             return None
 109         rate = float(current) / dif
 110         return int((float(total) - float(current)) / rate)
 111
 112     @staticmethod
 113     def format_eta(eta):
 114         if eta is None:
 115             return '--:--'
 116         return FileDownloader.format_seconds(eta)
 117
 118     @staticmethod
 119     def calc_speed(start, now, bytes):
 120         dif = now - start
 121         if bytes == 0 or dif < 0.001:  # One millisecond
 122             return None
 123         return float(bytes) / dif
 124
 125     @staticmethod
 126     def format_speed(speed):
 127         if speed is None:
 128             return '%10s' % '---b/s'
 129         return '%10s' % ('%s/s' % format_bytes(speed))
 130
 131     @staticmethod
 132     def format_retries(retries):
 133         return 'inf' if retries == float('inf') else '%.0f' % retries
 134
 135     @staticmethod
 136     def best_block_size(elapsed_time, bytes):
 137         new_min = max(bytes / 2.0, 1.0)
 138         new_max = min(max(bytes * 2.0, 1.0), 4194304)  # Do not surpass 4 MB
 139         if elapsed_time < 0.001:
 140             return int(new_max)
 141         rate = bytes / elapsed_time
 142         if rate > new_max:
 143             return int(new_max)
 144         if rate < new_min:
 145             return int(new_min)
 146         return int(rate)
 147
 148     @staticmethod
 149     def parse_bytes(bytestr):
 150         """Parse a string indicating a byte quantity into an integer."""
 151         matchobj = re.match(r'(?i)^(\d+(?:\.\d+)?)([kMGTPEZY]?)$', bytestr)
 152         if matchobj is None:
 153             return None
 154         number = float(matchobj.group(1))
 155         multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower())
 156         return int(round(number * multiplier))
 157
 158     def to_screen(self, *args, **kargs):
 159         self.ydl.to_stdout(*args, quiet=self.params.get('quiet'), **kargs)
 160
 161     def to_stderr(self, message):
 162         self.ydl.to_stderr(message)
 163
 164     def to_console_title(self, message):
 165         self.ydl.to_console_title(message)
 166
 167     def trouble(self, *args, **kargs):
 168         self.ydl.trouble(*args, **kargs)
 169
 170     def report_warning(self, *args, **kargs):
 171         self.ydl.report_warning(*args, **kargs)
 172
 173     def report_error(self, *args, **kargs):
 174         self.ydl.report_error(*args, **kargs)
 175
 176     def write_debug(self, *args, **kargs):
 177         self.ydl.write_debug(*args, **kargs)
 178
 179     def slow_down(self, start_time, now, byte_counter):
 180         """Sleep if the download speed is over the rate limit."""
 181         rate_limit = self.params.get('ratelimit')
 182         if rate_limit is None or byte_counter == 0:
 183             return
 184         if now is None:
 185             now = time.time()
 186         elapsed = now - start_time
 187         if elapsed <= 0.0:
 188             return
 189         speed = float(byte_counter) / elapsed
 190         if speed > rate_limit:
 191             sleep_time = float(byte_counter) / rate_limit - elapsed
 192             if sleep_time > 0:
 193                 time.sleep(sleep_time)
 194
 195     def temp_name(self, filename):
 196         """Returns a temporary filename for the given filename."""
 197         if self.params.get('nopart', False) or filename == '-' or \
 198                 (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))):
 199             return filename
 200         return filename + '.part'
 201
 202     def undo_temp_name(self, filename):
 203         if filename.endswith('.part'):
 204             return filename[:-len('.part')]
 205         return filename
 206
 207     def ytdl_filename(self, filename):
 208         return filename + '.ytdl'
 209
 210     def try_rename(self, old_filename, new_filename):
 211         if old_filename == new_filename:
 212             return
 213         try:
 214             os.replace(old_filename, new_filename)
 215         except (IOError, OSError) as err:
 216             self.report_error(f'unable to rename file: {err}')
 217
 218     def try_utime(self, filename, last_modified_hdr):
 219         """Try to set the last-modified time of the given file."""
 220         if last_modified_hdr is None:
 221             return
 222         if not os.path.isfile(encodeFilename(filename)):
 223             return
 224         timestr = last_modified_hdr
 225         if timestr is None:
 226             return
 227         filetime = timeconvert(timestr)
 228         if filetime is None:
 229             return filetime
 230         # Ignore obviously invalid dates
 231         if filetime == 0:
 232             return
 233         try:
 234             os.utime(filename, (time.time(), filetime))
 235         except Exception:
 236             pass
 237         return filetime
 238
 239     def report_destination(self, filename):
 240         """Report destination filename."""
 241         self.to_screen('[download] Destination: ' + filename)
 242
 243     def _prepare_multiline_status(self, lines=1):
 244         if self.params.get('noprogress'):
 245             self._multiline = QuietMultilinePrinter()
 246         elif self.ydl.params.get('logger'):
 247             self._multiline = MultilineLogger(self.ydl.params['logger'], lines)
 248         elif self.params.get('progress_with_newline'):
 249             self._multiline = BreaklineStatusPrinter(self.ydl._screen_file, lines)
 250         else:
 251             self._multiline = MultilinePrinter(self.ydl._screen_file, lines, not self.params.get('quiet'))
 252         self._multiline.allow_colors = self._multiline._HAVE_FULLCAP and not self.params.get('no_color')
 253
 254     def _finish_multiline_status(self):
 255         self._multiline.end()
 256
 257     _progress_styles = {
 258         'downloaded_bytes': 'light blue',
 259         'percent': 'light blue',
 260         'eta': 'yellow',
 261         'speed': 'green',
 262         'elapsed': 'bold white',
 263         'total_bytes': '',
 264         'total_bytes_estimate': '',
 265     }
 266
 267     def _report_progress_status(self, s, default_template):
 268         for name, style in self._progress_styles.items():
 269             name = f'_{name}_str'
 270             if name not in s:
 271                 continue
 272             s[name] = self._format_progress(s[name], style)
 273         s['_default_template'] = default_template % s
 274
 275         progress_dict = s.copy()
 276         progress_dict.pop('info_dict')
 277         progress_dict = {'info': s['info_dict'], 'progress': progress_dict}
 278
 279         progress_template = self.params.get('progress_template', {})
 280         self._multiline.print_at_line(self.ydl.evaluate_outtmpl(
 281             progress_template.get('download') or '[download] %(progress._default_template)s',
 282             progress_dict), s.get('progress_idx') or 0)
 283         self.to_console_title(self.ydl.evaluate_outtmpl(
 284             progress_template.get('download-title') or 'yt-dlp %(progress._default_template)s',
 285             progress_dict))
 286
 287     def _format_progress(self, *args, **kwargs):
 288         return self.ydl._format_text(
 289             self._multiline.stream, self._multiline.allow_colors, *args, **kwargs)
 290
 291     def report_progress(self, s):
 292         if s['status'] == 'finished':
 293             if self.params.get('noprogress'):
 294                 self.to_screen('[download] Download completed')
 295             msg_template = '100%%'
 296             if s.get('total_bytes') is not None:
 297                 s['_total_bytes_str'] = format_bytes(s['total_bytes'])
 298                 msg_template += ' of %(_total_bytes_str)s'
 299             if s.get('elapsed') is not None:
 300                 s['_elapsed_str'] = self.format_seconds(s['elapsed'])
 301                 msg_template += ' in %(_elapsed_str)s'
 302             s['_percent_str'] = self.format_percent(100)
 303             self._report_progress_status(s, msg_template)
 304             return
 305
 306         if s['status'] != 'downloading':
 307             return
 308
 309         if s.get('eta') is not None:
 310             s['_eta_str'] = self.format_eta(s['eta'])
 311         else:
 312             s['_eta_str'] = 'Unknown'
 313
 314         if s.get('total_bytes') and s.get('downloaded_bytes') is not None:
 315             s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes'])
 316         elif s.get('total_bytes_estimate') and s.get('downloaded_bytes') is not None:
 317             s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes_estimate'])
 318         else:
 319             if s.get('downloaded_bytes') == 0:
 320                 s['_percent_str'] = self.format_percent(0)
 321             else:
 322                 s['_percent_str'] = 'Unknown %'
 323
 324         if s.get('speed') is not None:
 325             s['_speed_str'] = self.format_speed(s['speed'])
 326         else:
 327             s['_speed_str'] = 'Unknown speed'
 328
 329         if s.get('total_bytes') is not None:
 330             s['_total_bytes_str'] = format_bytes(s['total_bytes'])
 331             msg_template = '%(_percent_str)s of %(_total_bytes_str)s at %(_speed_str)s ETA %(_eta_str)s'
 332         elif s.get('total_bytes_estimate') is not None:
 333             s['_total_bytes_estimate_str'] = format_bytes(s['total_bytes_estimate'])
 334             msg_template = '%(_percent_str)s of ~%(_total_bytes_estimate_str)s at %(_speed_str)s ETA %(_eta_str)s'
 335         else:
 336             if s.get('downloaded_bytes') is not None:
 337                 s['_downloaded_bytes_str'] = format_bytes(s['downloaded_bytes'])
 338                 if s.get('elapsed'):
 339                     s['_elapsed_str'] = self.format_seconds(s['elapsed'])
 340                     msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s (%(_elapsed_str)s)'
 341                 else:
 342                     msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s'
 343             else:
 344                 msg_template = '%(_percent_str)s at %(_speed_str)s ETA %(_eta_str)s'
 345         if s.get('fragment_index') and s.get('fragment_count'):
 346             msg_template += ' (frag %(fragment_index)s/%(fragment_count)s)'
 347         elif s.get('fragment_index'):
 348             msg_template += ' (frag %(fragment_index)s)'
 349         self._report_progress_status(s, msg_template)
 350
 351     def report_resuming_byte(self, resume_len):
 352         """Report attempt to resume at given byte."""
 353         self.to_screen('[download] Resuming download at byte %s' % resume_len)
 354
 355     def report_retry(self, err, count, retries):
 356         """Report retry in case of HTTP error 5xx"""
 357         self.to_screen(
 358             '[download] Got server HTTP error: %s. Retrying (attempt %d of %s) ...'
 359             % (error_to_compat_str(err), count, self.format_retries(retries)))
 360
 361     def report_file_already_downloaded(self, *args, **kwargs):
 362         """Report file has already been fully downloaded."""
 363         return self.ydl.report_file_already_downloaded(*args, **kwargs)
 364
 365     def report_unable_to_resume(self):
 366         """Report it was impossible to resume download."""
 367         self.to_screen('[download] Unable to resume')
 368
 369     @staticmethod
 370     def supports_manifest(manifest):
 371         """ Whether the downloader can download the fragments from the manifest.
 372         Redefine in subclasses if needed. """
 373         pass
 374
 375     def download(self, filename, info_dict, subtitle=False):
 376         """Download to a filename using the info from info_dict
 377         Return True on success and False otherwise
 378         """
 379
 380         nooverwrites_and_exists = (
 381             not self.params.get('overwrites', True)
 382             and os.path.exists(encodeFilename(filename))
 383         )
 384
 385         if not hasattr(filename, 'write'):
 386             continuedl_and_exists = (
 387                 self.params.get('continuedl', True)
 388                 and os.path.isfile(encodeFilename(filename))
 389                 and not self.params.get('nopart', False)
 390             )
 391
 392             # Check file already present
 393             if filename != '-' and (nooverwrites_and_exists or continuedl_and_exists):
 394                 self.report_file_already_downloaded(filename)
 395                 self._hook_progress({
 396                     'filename': filename,
 397                     'status': 'finished',
 398                     'total_bytes': os.path.getsize(encodeFilename(filename)),
 399                 }, info_dict)
 400                 return True, False
 401
 402         if subtitle is False:
 403             min_sleep_interval = self.params.get('sleep_interval')
 404             if min_sleep_interval:
 405                 max_sleep_interval = self.params.get('max_sleep_interval', min_sleep_interval)
 406                 sleep_interval = random.uniform(min_sleep_interval, max_sleep_interval)
 407                 self.to_screen(
 408                     '[download] Sleeping %s seconds ...' % (
 409                         int(sleep_interval) if sleep_interval.is_integer()
 410                         else '%.2f' % sleep_interval))
 411                 time.sleep(sleep_interval)
 412         else:
 413             sleep_interval_sub = 0
 414             if type(self.params.get('sleep_interval_subtitles')) is int:
 415                 sleep_interval_sub = self.params.get('sleep_interval_subtitles')
 416             if sleep_interval_sub > 0:
 417                 self.to_screen(
 418                     '[download] Sleeping %s seconds ...' % (
 419                         sleep_interval_sub))
 420                 time.sleep(sleep_interval_sub)
 421         ret = self.real_download(filename, info_dict)
 422         self._finish_multiline_status()
 423         return ret, True
 424
 425     def real_download(self, filename, info_dict):
 426         """Real download process. Redefine in subclasses."""
 427         raise NotImplementedError('This method must be implemented by subclasses')
 428
 429     def _hook_progress(self, status, info_dict):
 430         if not self._progress_hooks:
 431             return
 432         status['info_dict'] = info_dict
 433         # youtube-dl passes the same status object to all the hooks.
 434         # Some third party scripts seems to be relying on this.
 435         # So keep this behavior if possible
 436         for ph in self._progress_hooks:
 437             ph(status)
 438
 439     def add_progress_hook(self, ph):
 440         # See YoutubeDl.py (search for progress_hooks) for a description of
 441         # this interface
 442         self._progress_hooks.append(ph)
 443
 444     def _debug_cmd(self, args, exe=None):
 445         if not self.params.get('verbose', False):
 446             return
 447
 448         str_args = [decodeArgument(a) for a in args]
 449
 450         if exe is None:
 451             exe = os.path.basename(str_args[0])
 452
 453         self.write_debug('%s command line: %s' % (exe, shell_quote(str_args)))