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