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