Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/downloader/common.py

   1 from __future__ import division, unicode_literals
   2
   3 import os
   4 import re
   5 import sys
   6 import time
   7 import random
   8
   9 from ..compat import compat_os_name
  10 from ..utils import (
  11     encodeFilename,
  12     error_to_compat_str,
  13     decodeArgument,
  14     format_bytes,
  15     timeconvert,
  16 )
  17
  18
  19 class FileDownloader(object):
  20     """File Downloader class.
  21
  22     File downloader objects are the ones responsible of downloading the
  23     actual video file and writing it to disk.
  24
  25     File downloaders accept a lot of parameters. In order not to saturate
  26     the object constructor with arguments, it receives a dictionary of
  27     options instead.
  28
  29     Available options:
  30
  31     verbose:            Print additional info to stdout.
  32     quiet:              Do not print messages to stdout.
  33     ratelimit:          Download speed limit, in bytes/sec.
  34     retries:            Number of times to retry for HTTP error 5xx
  35     buffersize:         Size of download buffer in bytes.
  36     noresizebuffer:     Do not automatically resize the download buffer.
  37     continuedl:         Try to continue downloads if possible.
  38     noprogress:         Do not print the progress bar.
  39     logtostderr:        Log messages to stderr instead of stdout.
  40     consoletitle:       Display progress in console window's titlebar.
  41     nopart:             Do not use temporary .part files.
  42     updatetime:         Use the Last-modified header to set output file timestamps.
  43     test:               Download only first bytes to test the downloader.
  44     min_filesize:       Skip files smaller than this size
  45     max_filesize:       Skip files larger than this size
  46     xattr_set_filesize: Set ytdl.filesize user xattribute with expected size.
  47                         (experimental)
  48     external_downloader_args:  A list of additional command-line arguments for the
  49                         external downloader.
  50     hls_use_mpegts:     Use the mpegts container for HLS videos.
  51
  52     Subclasses of this one must re-define the real_download method.
  53     """
  54
  55     _TEST_FILE_SIZE = 10241
  56     params = None
  57
  58     def __init__(self, ydl, params):
  59         """Create a FileDownloader object with the given options."""
  60         self.ydl = ydl
  61         self._progress_hooks = []
  62         self.params = params
  63         self.add_progress_hook(self.report_progress)
  64
  65     @staticmethod
  66     def format_seconds(seconds):
  67         (mins, secs) = divmod(seconds, 60)
  68         (hours, mins) = divmod(mins, 60)
  69         if hours > 99:
  70             return '--:--:--'
  71         if hours == 0:
  72             return '%02d:%02d' % (mins, secs)
  73         else:
  74             return '%02d:%02d:%02d' % (hours, mins, secs)
  75
  76     @staticmethod
  77     def calc_percent(byte_counter, data_len):
  78         if data_len is None:
  79             return None
  80         return float(byte_counter) / float(data_len) * 100.0
  81
  82     @staticmethod
  83     def format_percent(percent):
  84         if percent is None:
  85             return '---.-%'
  86         return '%6s' % ('%3.1f%%' % percent)
  87
  88     @staticmethod
  89     def calc_eta(start, now, total, current):
  90         if total is None:
  91             return None
  92         if now is None:
  93             now = time.time()
  94         dif = now - start
  95         if current == 0 or dif < 0.001:  # One millisecond
  96             return None
  97         rate = float(current) / dif
  98         return int((float(total) - float(current)) / rate)
  99
 100     @staticmethod
 101     def format_eta(eta):
 102         if eta is None:
 103             return '--:--'
 104         return FileDownloader.format_seconds(eta)
 105
 106     @staticmethod
 107     def calc_speed(start, now, bytes):
 108         dif = now - start
 109         if bytes == 0 or dif < 0.001:  # One millisecond
 110             return None
 111         return float(bytes) / dif
 112
 113     @staticmethod
 114     def format_speed(speed):
 115         if speed is None:
 116             return '%10s' % '---b/s'
 117         return '%10s' % ('%s/s' % format_bytes(speed))
 118
 119     @staticmethod
 120     def format_retries(retries):
 121         return 'inf' if retries == float('inf') else '%.0f' % retries
 122
 123     @staticmethod
 124     def best_block_size(elapsed_time, bytes):
 125         new_min = max(bytes / 2.0, 1.0)
 126         new_max = min(max(bytes * 2.0, 1.0), 4194304)  # Do not surpass 4 MB
 127         if elapsed_time < 0.001:
 128             return int(new_max)
 129         rate = bytes / elapsed_time
 130         if rate > new_max:
 131             return int(new_max)
 132         if rate < new_min:
 133             return int(new_min)
 134         return int(rate)
 135
 136     @staticmethod
 137     def parse_bytes(bytestr):
 138         """Parse a string indicating a byte quantity into an integer."""
 139         matchobj = re.match(r'(?i)^(\d+(?:\.\d+)?)([kMGTPEZY]?)$', bytestr)
 140         if matchobj is None:
 141             return None
 142         number = float(matchobj.group(1))
 143         multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower())
 144         return int(round(number * multiplier))
 145
 146     def to_screen(self, *args, **kargs):
 147         self.ydl.to_screen(*args, **kargs)
 148
 149     def to_stderr(self, message):
 150         self.ydl.to_screen(message)
 151
 152     def to_console_title(self, message):
 153         self.ydl.to_console_title(message)
 154
 155     def trouble(self, *args, **kargs):
 156         self.ydl.trouble(*args, **kargs)
 157
 158     def report_warning(self, *args, **kargs):
 159         self.ydl.report_warning(*args, **kargs)
 160
 161     def report_error(self, *args, **kargs):
 162         self.ydl.report_error(*args, **kargs)
 163
 164     def slow_down(self, start_time, now, byte_counter):
 165         """Sleep if the download speed is over the rate limit."""
 166         rate_limit = self.params.get('ratelimit')
 167         if rate_limit is None or byte_counter == 0:
 168             return
 169         if now is None:
 170             now = time.time()
 171         elapsed = now - start_time
 172         if elapsed <= 0.0:
 173             return
 174         speed = float(byte_counter) / elapsed
 175         if speed > rate_limit:
 176             time.sleep(max((byte_counter // rate_limit) - elapsed, 0))
 177
 178     def temp_name(self, filename):
 179         """Returns a temporary filename for the given filename."""
 180         if self.params.get('nopart', False) or filename == '-' or \
 181                 (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))):
 182             return filename
 183         return filename + '.part'
 184
 185     def undo_temp_name(self, filename):
 186         if filename.endswith('.part'):
 187             return filename[:-len('.part')]
 188         return filename
 189
 190     def try_rename(self, old_filename, new_filename):
 191         try:
 192             if old_filename == new_filename:
 193                 return
 194             os.rename(encodeFilename(old_filename), encodeFilename(new_filename))
 195         except (IOError, OSError) as err:
 196             self.report_error('unable to rename file: %s' % error_to_compat_str(err))
 197
 198     def try_utime(self, filename, last_modified_hdr):
 199         """Try to set the last-modified time of the given file."""
 200         if last_modified_hdr is None:
 201             return
 202         if not os.path.isfile(encodeFilename(filename)):
 203             return
 204         timestr = last_modified_hdr
 205         if timestr is None:
 206             return
 207         filetime = timeconvert(timestr)
 208         if filetime is None:
 209             return filetime
 210         # Ignore obviously invalid dates
 211         if filetime == 0:
 212             return
 213         try:
 214             os.utime(filename, (time.time(), filetime))
 215         except Exception:
 216             pass
 217         return filetime
 218
 219     def report_destination(self, filename):
 220         """Report destination filename."""
 221         self.to_screen('[download] Destination: ' + filename)
 222
 223     def _report_progress_status(self, msg, is_last_line=False):
 224         fullmsg = '[download] ' + msg
 225         if self.params.get('progress_with_newline', False):
 226             self.to_screen(fullmsg)
 227         else:
 228             if compat_os_name == 'nt':
 229                 prev_len = getattr(self, '_report_progress_prev_line_length',
 230                                    0)
 231                 if prev_len > len(fullmsg):
 232                     fullmsg += ' ' * (prev_len - len(fullmsg))
 233                 self._report_progress_prev_line_length = len(fullmsg)
 234                 clear_line = '\r'
 235             else:
 236                 clear_line = ('\r\x1b[K' if sys.stderr.isatty() else '\r')
 237             self.to_screen(clear_line + fullmsg, skip_eol=not is_last_line)
 238         self.to_console_title('youtube-dl ' + msg)
 239
 240     def report_progress(self, s):
 241         if s['status'] == 'finished':
 242             if self.params.get('noprogress', False):
 243                 self.to_screen('[download] Download completed')
 244             else:
 245                 s['_total_bytes_str'] = format_bytes(s['total_bytes'])
 246                 if s.get('elapsed') is not None:
 247                     s['_elapsed_str'] = self.format_seconds(s['elapsed'])
 248                     msg_template = '100%% of %(_total_bytes_str)s in %(_elapsed_str)s'
 249                 else:
 250                     msg_template = '100%% of %(_total_bytes_str)s'
 251                 self._report_progress_status(
 252                     msg_template % s, is_last_line=True)
 253
 254         if self.params.get('noprogress'):
 255             return
 256
 257         if s['status'] != 'downloading':
 258             return
 259
 260         if s.get('eta') is not None:
 261             s['_eta_str'] = self.format_eta(s['eta'])
 262         else:
 263             s['_eta_str'] = 'Unknown ETA'
 264
 265         if s.get('total_bytes') and s.get('downloaded_bytes') is not None:
 266             s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes'])
 267         elif s.get('total_bytes_estimate') and s.get('downloaded_bytes') is not None:
 268             s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes_estimate'])
 269         else:
 270             if s.get('downloaded_bytes') == 0:
 271                 s['_percent_str'] = self.format_percent(0)
 272             else:
 273                 s['_percent_str'] = 'Unknown %'
 274
 275         if s.get('speed') is not None:
 276             s['_speed_str'] = self.format_speed(s['speed'])
 277         else:
 278             s['_speed_str'] = 'Unknown speed'
 279
 280         if s.get('total_bytes') is not None:
 281             s['_total_bytes_str'] = format_bytes(s['total_bytes'])
 282             msg_template = '%(_percent_str)s of %(_total_bytes_str)s at %(_speed_str)s ETA %(_eta_str)s'
 283         elif s.get('total_bytes_estimate') is not None:
 284             s['_total_bytes_estimate_str'] = format_bytes(s['total_bytes_estimate'])
 285             msg_template = '%(_percent_str)s of ~%(_total_bytes_estimate_str)s at %(_speed_str)s ETA %(_eta_str)s'
 286         else:
 287             if s.get('downloaded_bytes') is not None:
 288                 s['_downloaded_bytes_str'] = format_bytes(s['downloaded_bytes'])
 289                 if s.get('elapsed'):
 290                     s['_elapsed_str'] = self.format_seconds(s['elapsed'])
 291                     msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s (%(_elapsed_str)s)'
 292                 else:
 293                     msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s'
 294             else:
 295                 msg_template = '%(_percent_str)s % at %(_speed_str)s ETA %(_eta_str)s'
 296
 297         self._report_progress_status(msg_template % s)
 298
 299     def report_resuming_byte(self, resume_len):
 300         """Report attempt to resume at given byte."""
 301         self.to_screen('[download] Resuming download at byte %s' % resume_len)
 302
 303     def report_retry(self, count, retries):
 304         """Report retry in case of HTTP error 5xx"""
 305         self.to_screen(
 306             '[download] Got server HTTP error. Retrying (attempt %d of %s)...'
 307             % (count, self.format_retries(retries)))
 308
 309     def report_file_already_downloaded(self, file_name):
 310         """Report file has already been fully downloaded."""
 311         try:
 312             self.to_screen('[download] %s has already been downloaded' % file_name)
 313         except UnicodeEncodeError:
 314             self.to_screen('[download] The file has already been downloaded')
 315
 316     def report_unable_to_resume(self):
 317         """Report it was impossible to resume download."""
 318         self.to_screen('[download] Unable to resume')
 319
 320     def download(self, filename, info_dict):
 321         """Download to a filename using the info from info_dict
 322         Return True on success and False otherwise
 323         """
 324
 325         nooverwrites_and_exists = (
 326             self.params.get('nooverwrites', False) and
 327             os.path.exists(encodeFilename(filename))
 328         )
 329
 330         continuedl_and_exists = (
 331             self.params.get('continuedl', True) and
 332             os.path.isfile(encodeFilename(filename)) and
 333             not self.params.get('nopart', False)
 334         )
 335
 336         # Check file already present
 337         if filename != '-' and (nooverwrites_and_exists or continuedl_and_exists):
 338             self.report_file_already_downloaded(filename)
 339             self._hook_progress({
 340                 'filename': filename,
 341                 'status': 'finished',
 342                 'total_bytes': os.path.getsize(encodeFilename(filename)),
 343             })
 344             return True
 345
 346         min_sleep_interval = self.params.get('sleep_interval')
 347         if min_sleep_interval:
 348             max_sleep_interval = self.params.get('max_sleep_interval', min_sleep_interval)
 349             sleep_interval = random.uniform(min_sleep_interval, max_sleep_interval)
 350             self.to_screen(
 351                 '[download] Sleeping %s seconds...' % (
 352                     int(sleep_interval) if sleep_interval.is_integer()
 353                     else '%.2f' % sleep_interval))
 354             time.sleep(sleep_interval)
 355
 356         return self.real_download(filename, info_dict)
 357
 358     def real_download(self, filename, info_dict):
 359         """Real download process. Redefine in subclasses."""
 360         raise NotImplementedError('This method must be implemented by subclasses')
 361
 362     def _hook_progress(self, status):
 363         for ph in self._progress_hooks:
 364             ph(status)
 365
 366     def add_progress_hook(self, ph):
 367         # See YoutubeDl.py (search for progress_hooks) for a description of
 368         # this interface
 369         self._progress_hooks.append(ph)
 370
 371     def _debug_cmd(self, args, exe=None):
 372         if not self.params.get('verbose', False):
 373             return
 374
 375         str_args = [decodeArgument(a) for a in args]
 376
 377         if exe is None:
 378             exe = os.path.basename(str_args[0])
 379
 380         try:
 381             import pipes
 382             shell_quote = lambda args: ' '.join(map(pipes.quote, str_args))
 383         except ImportError:
 384             shell_quote = repr
 385         self.to_screen('[debug] %s command line: %s' % (
 386             exe, shell_quote(str_args)))