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