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