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