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