Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/FileDownloader.py

   1 import math
   2 import os
   3 import re
   4 import subprocess
   5 import sys
   6 import time
   7 import traceback
   8
   9 if os.name == 'nt':
  10     import ctypes
  11
  12 from .utils import *
  13
  14
  15 class FileDownloader(object):
  16     """File Downloader class.
  17
  18     File downloader objects are the ones responsible of downloading the
  19     actual video file and writing it to disk.
  20
  21     File downloaders accept a lot of parameters. In order not to saturate
  22     the object constructor with arguments, it receives a dictionary of
  23     options instead.
  24
  25     Available options:
  26
  27     verbose:           Print additional info to stdout.
  28     quiet:             Do not print messages to stdout.
  29     ratelimit:         Download speed limit, in bytes/sec.
  30     retries:           Number of times to retry for HTTP error 5xx
  31     buffersize:        Size of download buffer in bytes.
  32     noresizebuffer:    Do not automatically resize the download buffer.
  33     continuedl:        Try to continue downloads if possible.
  34     noprogress:        Do not print the progress bar.
  35     logtostderr:       Log messages to stderr instead of stdout.
  36     consoletitle:      Display progress in console window's titlebar.
  37     nopart:            Do not use temporary .part files.
  38     updatetime:        Use the Last-modified header to set output file timestamps.
  39     test:              Download only first bytes to test the downloader.
  40     min_filesize:      Skip files smaller than this size
  41     max_filesize:      Skip files larger than this size
  42     """
  43
  44     params = None
  45
  46     def __init__(self, ydl, params):
  47         """Create a FileDownloader object with the given options."""
  48         self.ydl = ydl
  49         self._progress_hooks = []
  50         self.params = params
  51
  52     @staticmethod
  53     def format_bytes(bytes):
  54         if bytes is None:
  55             return 'N/A'
  56         if type(bytes) is str:
  57             bytes = float(bytes)
  58         if bytes == 0.0:
  59             exponent = 0
  60         else:
  61             exponent = int(math.log(bytes, 1024.0))
  62         suffix = ['B','KiB','MiB','GiB','TiB','PiB','EiB','ZiB','YiB'][exponent]
  63         converted = float(bytes) / float(1024 ** exponent)
  64         return '%.2f%s' % (converted, suffix)
  65
  66     @staticmethod
  67     def format_seconds(seconds):
  68         (mins, secs) = divmod(seconds, 60)
  69         (hours, eta_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 '---.-%'
  81         return '%6s' % ('%3.1f%%' % (float(byte_counter) / float(data_len) * 100.0))
  82
  83     @staticmethod
  84     def calc_eta(start, now, total, current):
  85         if total is None:
  86             return '--:--'
  87         dif = now - start
  88         if current == 0 or dif < 0.001: # One millisecond
  89             return '--:--'
  90         rate = float(current) / dif
  91         eta = int((float(total) - float(current)) / rate)
  92         return FileDownloader.format_seconds(eta)
  93
  94     @staticmethod
  95     def calc_speed(start, now, bytes):
  96         dif = now - start
  97         if bytes == 0 or dif < 0.001: # One millisecond
  98             return '%10s' % '---b/s'
  99         return '%10s' % ('%s/s' % FileDownloader.format_bytes(float(bytes) / dif))
 100
 101     @staticmethod
 102     def best_block_size(elapsed_time, bytes):
 103         new_min = max(bytes / 2.0, 1.0)
 104         new_max = min(max(bytes * 2.0, 1.0), 4194304) # Do not surpass 4 MB
 105         if elapsed_time < 0.001:
 106             return int(new_max)
 107         rate = bytes / elapsed_time
 108         if rate > new_max:
 109             return int(new_max)
 110         if rate < new_min:
 111             return int(new_min)
 112         return int(rate)
 113
 114     @staticmethod
 115     def parse_bytes(bytestr):
 116         """Parse a string indicating a byte quantity into an integer."""
 117         matchobj = re.match(r'(?i)^(\d+(?:\.\d+)?)([kMGTPEZY]?)$', bytestr)
 118         if matchobj is None:
 119             return None
 120         number = float(matchobj.group(1))
 121         multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower())
 122         return int(round(number * multiplier))
 123
 124     def to_screen(self, *args, **kargs):
 125         self.ydl.to_screen(*args, **kargs)
 126
 127     def to_stderr(self, message):
 128         self.ydl.to_screen(message)
 129
 130     def to_cons_title(self, message):
 131         """Set console/terminal window title to message."""
 132         if not self.params.get('consoletitle', False):
 133             return
 134         if os.name == 'nt' and ctypes.windll.kernel32.GetConsoleWindow():
 135             # c_wchar_p() might not be necessary if `message` is
 136             # already of type unicode()
 137             ctypes.windll.kernel32.SetConsoleTitleW(ctypes.c_wchar_p(message))
 138         elif 'TERM' in os.environ:
 139             self.to_screen('\033]0;%s\007' % message, skip_eol=True)
 140
 141     def trouble(self, *args, **kargs):
 142         self.ydl.trouble(*args, **kargs)
 143
 144     def report_warning(self, *args, **kargs):
 145         self.ydl.report_warning(*args, **kargs)
 146
 147     def report_error(self, *args, **kargs):
 148         self.ydl.report_error(*args, **kargs)
 149
 150     def slow_down(self, start_time, byte_counter):
 151         """Sleep if the download speed is over the rate limit."""
 152         rate_limit = self.params.get('ratelimit', None)
 153         if rate_limit is None or byte_counter == 0:
 154             return
 155         now = time.time()
 156         elapsed = now - start_time
 157         if elapsed <= 0.0:
 158             return
 159         speed = float(byte_counter) / elapsed
 160         if speed > rate_limit:
 161             time.sleep((byte_counter - rate_limit * (now - start_time)) / rate_limit)
 162
 163     def temp_name(self, filename):
 164         """Returns a temporary filename for the given filename."""
 165         if self.params.get('nopart', False) or filename == u'-' or \
 166                 (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))):
 167             return filename
 168         return filename + u'.part'
 169
 170     def undo_temp_name(self, filename):
 171         if filename.endswith(u'.part'):
 172             return filename[:-len(u'.part')]
 173         return filename
 174
 175     def try_rename(self, old_filename, new_filename):
 176         try:
 177             if old_filename == new_filename:
 178                 return
 179             os.rename(encodeFilename(old_filename), encodeFilename(new_filename))
 180         except (IOError, OSError) as err:
 181             self.report_error(u'unable to rename file')
 182
 183     def try_utime(self, filename, last_modified_hdr):
 184         """Try to set the last-modified time of the given file."""
 185         if last_modified_hdr is None:
 186             return
 187         if not os.path.isfile(encodeFilename(filename)):
 188             return
 189         timestr = last_modified_hdr
 190         if timestr is None:
 191             return
 192         filetime = timeconvert(timestr)
 193         if filetime is None:
 194             return filetime
 195         # Ignore obviously invalid dates
 196         if filetime == 0:
 197             return
 198         try:
 199             os.utime(filename, (time.time(), filetime))
 200         except:
 201             pass
 202         return filetime
 203
 204     def report_destination(self, filename):
 205         """Report destination filename."""
 206         self.to_screen(u'[download] Destination: ' + filename)
 207
 208     def report_progress(self, percent_str, data_len_str, speed_str, eta_str):
 209         """Report download progress."""
 210         if self.params.get('noprogress', False):
 211             return
 212         clear_line = (u'\x1b[K' if sys.stderr.isatty() and os.name != 'nt' else u'')
 213         if self.params.get('progress_with_newline', False):
 214             self.to_screen(u'[download] %s of %s at %s ETA %s' %
 215                 (percent_str, data_len_str, speed_str, eta_str))
 216         else:
 217             self.to_screen(u'\r%s[download] %s of %s at %s ETA %s' %
 218                 (clear_line, percent_str, data_len_str, speed_str, eta_str), skip_eol=True)
 219         self.to_cons_title(u'youtube-dl - %s of %s at %s ETA %s' %
 220                 (percent_str.strip(), data_len_str.strip(), speed_str.strip(), eta_str.strip()))
 221
 222     def report_resuming_byte(self, resume_len):
 223         """Report attempt to resume at given byte."""
 224         self.to_screen(u'[download] Resuming download at byte %s' % resume_len)
 225
 226     def report_retry(self, count, retries):
 227         """Report retry in case of HTTP error 5xx"""
 228         self.to_screen(u'[download] Got server HTTP error. Retrying (attempt %d of %d)...' % (count, retries))
 229
 230     def report_file_already_downloaded(self, file_name):
 231         """Report file has already been fully downloaded."""
 232         try:
 233             self.to_screen(u'[download] %s has already been downloaded' % file_name)
 234         except (UnicodeEncodeError) as err:
 235             self.to_screen(u'[download] The file has already been downloaded')
 236
 237     def report_unable_to_resume(self):
 238         """Report it was impossible to resume download."""
 239         self.to_screen(u'[download] Unable to resume')
 240
 241     def report_finish(self, data_len_str, tot_time):
 242         """Report download finished."""
 243         if self.params.get('noprogress', False):
 244             self.to_screen(u'[download] Download completed')
 245         else:
 246             clear_line = (u'\x1b[K' if sys.stderr.isatty() and os.name != 'nt' else u'')
 247             self.to_screen(u'\r%s[download] 100%% of %s in %s' %
 248                 (clear_line, data_len_str, self.format_seconds(tot_time)))
 249
 250     def _download_with_rtmpdump(self, filename, url, player_url, page_url, play_path, tc_url):
 251         self.report_destination(filename)
 252         tmpfilename = self.temp_name(filename)
 253
 254         # Check for rtmpdump first
 255         try:
 256             subprocess.call(['rtmpdump', '-h'], stdout=(open(os.path.devnull, 'w')), stderr=subprocess.STDOUT)
 257         except (OSError, IOError):
 258             self.report_error(u'RTMP download detected but "rtmpdump" could not be run')
 259             return False
 260         verbosity_option = '--verbose' if self.params.get('verbose', False) else '--quiet'
 261
 262         # Download using rtmpdump. rtmpdump returns exit code 2 when
 263         # the connection was interrumpted and resuming appears to be
 264         # possible. This is part of rtmpdump's normal usage, AFAIK.
 265         basic_args = ['rtmpdump', verbosity_option, '-r', url, '-o', tmpfilename]
 266         if player_url is not None:
 267             basic_args += ['--swfVfy', player_url]
 268         if page_url is not None:
 269             basic_args += ['--pageUrl', page_url]
 270         if play_path is not None:
 271             basic_args += ['--playpath', play_path]
 272         if tc_url is not None:
 273             basic_args += ['--tcUrl', url]
 274         args = basic_args + [[], ['--resume', '--skip', '1']][self.params.get('continuedl', False)]
 275         if self.params.get('verbose', False):
 276             try:
 277                 import pipes
 278                 shell_quote = lambda args: ' '.join(map(pipes.quote, args))
 279             except ImportError:
 280                 shell_quote = repr
 281             self.to_screen(u'[debug] rtmpdump command line: ' + shell_quote(args))
 282         retval = subprocess.call(args)
 283         while retval == 2 or retval == 1:
 284             prevsize = os.path.getsize(encodeFilename(tmpfilename))
 285             self.to_screen(u'\r[rtmpdump] %s bytes' % prevsize, skip_eol=True)
 286             time.sleep(5.0) # This seems to be needed
 287             retval = subprocess.call(basic_args + ['-e'] + [[], ['-k', '1']][retval == 1])
 288             cursize = os.path.getsize(encodeFilename(tmpfilename))
 289             if prevsize == cursize and retval == 1:
 290                 break
 291              # Some rtmp streams seem abort after ~ 99.8%. Don't complain for those
 292             if prevsize == cursize and retval == 2 and cursize > 1024:
 293                 self.to_screen(u'\r[rtmpdump] Could not download the whole video. This can happen for some advertisements.')
 294                 retval = 0
 295                 break
 296         if retval == 0:
 297             fsize = os.path.getsize(encodeFilename(tmpfilename))
 298             self.to_screen(u'\r[rtmpdump] %s bytes' % fsize)
 299             self.try_rename(tmpfilename, filename)
 300             self._hook_progress({
 301                 'downloaded_bytes': fsize,
 302                 'total_bytes': fsize,
 303                 'filename': filename,
 304                 'status': 'finished',
 305             })
 306             return True
 307         else:
 308             self.to_stderr(u"\n")
 309             self.report_error(u'rtmpdump exited with code %d' % retval)
 310             return False
 311
 312     def _download_with_mplayer(self, filename, url):
 313         self.report_destination(filename)
 314         tmpfilename = self.temp_name(filename)
 315
 316         args = ['mplayer', '-really-quiet', '-vo', 'null', '-vc', 'dummy', '-dumpstream', '-dumpfile', tmpfilename, url]
 317         # Check for mplayer first
 318         try:
 319             subprocess.call(['mplayer', '-h'], stdout=(open(os.path.devnull, 'w')), stderr=subprocess.STDOUT)
 320         except (OSError, IOError):
 321             self.report_error(u'MMS or RTSP download detected but "%s" could not be run' % args[0] )
 322             return False
 323
 324         # Download using mplayer.
 325         retval = subprocess.call(args)
 326         if retval == 0:
 327             fsize = os.path.getsize(encodeFilename(tmpfilename))
 328             self.to_screen(u'\r[%s] %s bytes' % (args[0], fsize))
 329             self.try_rename(tmpfilename, filename)
 330             self._hook_progress({
 331                 'downloaded_bytes': fsize,
 332                 'total_bytes': fsize,
 333                 'filename': filename,
 334                 'status': 'finished',
 335             })
 336             return True
 337         else:
 338             self.to_stderr(u"\n")
 339             self.report_error(u'mplayer exited with code %d' % retval)
 340             return False
 341
 342     def _download_m3u8_with_ffmpeg(self, filename, url):
 343         self.report_destination(filename)
 344         tmpfilename = self.temp_name(filename)
 345
 346         args = ['ffmpeg', '-y', '-i', url, '-f', 'mp4', tmpfilename]
 347         # Check for ffmpeg first
 348         try:
 349             subprocess.call(['ffmpeg', '-h'], stdout=(open(os.path.devnull, 'w')), stderr=subprocess.STDOUT)
 350         except (OSError, IOError):
 351             self.report_error(u'm3u8 download detected but "%s" could not be run' % args[0] )
 352             return False
 353
 354         retval = subprocess.call(args)
 355         if retval == 0:
 356             fsize = os.path.getsize(encodeFilename(tmpfilename))
 357             self.to_screen(u'\r[%s] %s bytes' % (args[0], fsize))
 358             self.try_rename(tmpfilename, filename)
 359             self._hook_progress({
 360                 'downloaded_bytes': fsize,
 361                 'total_bytes': fsize,
 362                 'filename': filename,
 363                 'status': 'finished',
 364             })
 365             return True
 366         else:
 367             self.to_stderr(u"\n")
 368             self.report_error(u'ffmpeg exited with code %d' % retval)
 369             return False
 370
 371
 372     def _do_download(self, filename, info_dict):
 373         url = info_dict['url']
 374
 375         # Check file already present
 376         if self.params.get('continuedl', False) and os.path.isfile(encodeFilename(filename)) and not self.params.get('nopart', False):
 377             self.report_file_already_downloaded(filename)
 378             self._hook_progress({
 379                 'filename': filename,
 380                 'status': 'finished',
 381             })
 382             return True
 383
 384         # Attempt to download using rtmpdump
 385         if url.startswith('rtmp'):
 386             return self._download_with_rtmpdump(filename, url,
 387                                                 info_dict.get('player_url', None),
 388                                                 info_dict.get('page_url', None),
 389                                                 info_dict.get('play_path', None),
 390                                                 info_dict.get('tc_url', None))
 391
 392         # Attempt to download using mplayer
 393         if url.startswith('mms') or url.startswith('rtsp'):
 394             return self._download_with_mplayer(filename, url)
 395
 396         # m3u8 manifest are downloaded with ffmpeg
 397         if determine_ext(url) == u'm3u8':
 398             return self._download_m3u8_with_ffmpeg(filename, url)
 399
 400         tmpfilename = self.temp_name(filename)
 401         stream = None
 402
 403         # Do not include the Accept-Encoding header
 404         headers = {'Youtubedl-no-compression': 'True'}
 405         if 'user_agent' in info_dict:
 406             headers['Youtubedl-user-agent'] = info_dict['user_agent']
 407         basic_request = compat_urllib_request.Request(url, None, headers)
 408         request = compat_urllib_request.Request(url, None, headers)
 409
 410         if self.params.get('test', False):
 411             request.add_header('Range','bytes=0-10240')
 412
 413         # Establish possible resume length
 414         if os.path.isfile(encodeFilename(tmpfilename)):
 415             resume_len = os.path.getsize(encodeFilename(tmpfilename))
 416         else:
 417             resume_len = 0
 418
 419         open_mode = 'wb'
 420         if resume_len != 0:
 421             if self.params.get('continuedl', False):
 422                 self.report_resuming_byte(resume_len)
 423                 request.add_header('Range','bytes=%d-' % resume_len)
 424                 open_mode = 'ab'
 425             else:
 426                 resume_len = 0
 427
 428         count = 0
 429         retries = self.params.get('retries', 0)
 430         while count <= retries:
 431             # Establish connection
 432             try:
 433                 if count == 0 and 'urlhandle' in info_dict:
 434                     data = info_dict['urlhandle']
 435                 data = compat_urllib_request.urlopen(request)
 436                 break
 437             except (compat_urllib_error.HTTPError, ) as err:
 438                 if (err.code < 500 or err.code >= 600) and err.code != 416:
 439                     # Unexpected HTTP error
 440                     raise
 441                 elif err.code == 416:
 442                     # Unable to resume (requested range not satisfiable)
 443                     try:
 444                         # Open the connection again without the range header
 445                         data = compat_urllib_request.urlopen(basic_request)
 446                         content_length = data.info()['Content-Length']
 447                     except (compat_urllib_error.HTTPError, ) as err:
 448                         if err.code < 500 or err.code >= 600:
 449                             raise
 450                     else:
 451                         # Examine the reported length
 452                         if (content_length is not None and
 453                                 (resume_len - 100 < int(content_length) < resume_len + 100)):
 454                             # The file had already been fully downloaded.
 455                             # Explanation to the above condition: in issue #175 it was revealed that
 456                             # YouTube sometimes adds or removes a few bytes from the end of the file,
 457                             # changing the file size slightly and causing problems for some users. So
 458                             # I decided to implement a suggested change and consider the file
 459                             # completely downloaded if the file size differs less than 100 bytes from
 460                             # the one in the hard drive.
 461                             self.report_file_already_downloaded(filename)
 462                             self.try_rename(tmpfilename, filename)
 463                             self._hook_progress({
 464                                 'filename': filename,
 465                                 'status': 'finished',
 466                             })
 467                             return True
 468                         else:
 469                             # The length does not match, we start the download over
 470                             self.report_unable_to_resume()
 471                             open_mode = 'wb'
 472                             break
 473             # Retry
 474             count += 1
 475             if count <= retries:
 476                 self.report_retry(count, retries)
 477
 478         if count > retries:
 479             self.report_error(u'giving up after %s retries' % retries)
 480             return False
 481
 482         data_len = data.info().get('Content-length', None)
 483         if data_len is not None:
 484             data_len = int(data_len) + resume_len
 485             min_data_len = self.params.get("min_filesize", None)
 486             max_data_len =  self.params.get("max_filesize", None)
 487             if min_data_len is not None and data_len < min_data_len:
 488                 self.to_screen(u'\r[download] File is smaller than min-filesize (%s bytes < %s bytes). Aborting.' % (data_len, min_data_len))
 489                 return False
 490             if max_data_len is not None and data_len > max_data_len:
 491                 self.to_screen(u'\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len, max_data_len))
 492                 return False
 493
 494         data_len_str = self.format_bytes(data_len)
 495         byte_counter = 0 + resume_len
 496         block_size = self.params.get('buffersize', 1024)
 497         start = time.time()
 498         while True:
 499             # Download and write
 500             before = time.time()
 501             data_block = data.read(block_size)
 502             after = time.time()
 503             if len(data_block) == 0:
 504                 break
 505             byte_counter += len(data_block)
 506
 507             # Open file just in time
 508             if stream is None:
 509                 try:
 510                     (stream, tmpfilename) = sanitize_open(tmpfilename, open_mode)
 511                     assert stream is not None
 512                     filename = self.undo_temp_name(tmpfilename)
 513                     self.report_destination(filename)
 514                 except (OSError, IOError) as err:
 515                     self.report_error(u'unable to open for writing: %s' % str(err))
 516                     return False
 517             try:
 518                 stream.write(data_block)
 519             except (IOError, OSError) as err:
 520                 self.to_stderr(u"\n")
 521                 self.report_error(u'unable to write data: %s' % str(err))
 522                 return False
 523             if not self.params.get('noresizebuffer', False):
 524                 block_size = self.best_block_size(after - before, len(data_block))
 525
 526             # Progress message
 527             speed_str = self.calc_speed(start, time.time(), byte_counter - resume_len)
 528             if data_len is None:
 529                 self.report_progress('Unknown %', data_len_str, speed_str, 'Unknown ETA')
 530             else:
 531                 percent_str = self.calc_percent(byte_counter, data_len)
 532                 eta_str = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len)
 533                 self.report_progress(percent_str, data_len_str, speed_str, eta_str)
 534
 535             self._hook_progress({
 536                 'downloaded_bytes': byte_counter,
 537                 'total_bytes': data_len,
 538                 'tmpfilename': tmpfilename,
 539                 'filename': filename,
 540                 'status': 'downloading',
 541             })
 542
 543             # Apply rate limit
 544             self.slow_down(start, byte_counter - resume_len)
 545
 546         if stream is None:
 547             self.to_stderr(u"\n")
 548             self.report_error(u'Did not get any data blocks')
 549             return False
 550         stream.close()
 551         self.report_finish(data_len_str, (time.time() - start))
 552         if data_len is not None and byte_counter != data_len:
 553             raise ContentTooShortError(byte_counter, int(data_len))
 554         self.try_rename(tmpfilename, filename)
 555
 556         # Update file modification time
 557         if self.params.get('updatetime', True):
 558             info_dict['filetime'] = self.try_utime(filename, data.info().get('last-modified', None))
 559
 560         self._hook_progress({
 561             'downloaded_bytes': byte_counter,
 562             'total_bytes': byte_counter,
 563             'filename': filename,
 564             'status': 'finished',
 565         })
 566
 567         return True
 568
 569     def _hook_progress(self, status):
 570         for ph in self._progress_hooks:
 571             ph(status)
 572
 573     def add_progress_hook(self, ph):
 574         """ ph gets called on download progress, with a dictionary with the entries
 575         * filename: The final filename
 576         * status: One of "downloading" and "finished"
 577
 578         It can also have some of the following entries:
 579
 580         * downloaded_bytes: Bytes on disks
 581         * total_bytes: Total bytes, None if unknown
 582         * tmpfilename: The filename we're currently writing to
 583
 584         Hooks are guaranteed to be called at least once (with status "finished")
 585         if the download is successful.
 586         """
 587         self._progress_hooks.append(ph)