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