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, 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 None
  81         return float(byte_counter) / float(data_len) * 100.0
  82
  83     @staticmethod
  84     def format_percent(percent):
  85         if percent is None:
  86             return '---.-%'
  87         return '%6s' % ('%3.1f%%' % percent)
  88
  89     @staticmethod
  90     def calc_eta(start, now, total, current):
  91         if total is None:
  92             return None
  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' % FileDownloader.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_cons_title(self, message):
 148         """Set console/terminal window title to message."""
 149         if not self.params.get('consoletitle', False):
 150             return
 151         if os.name == 'nt' and ctypes.windll.kernel32.GetConsoleWindow():
 152             # c_wchar_p() might not be necessary if `message` is
 153             # already of type unicode()
 154             ctypes.windll.kernel32.SetConsoleTitleW(ctypes.c_wchar_p(message))
 155         elif 'TERM' in os.environ:
 156             self.to_screen('\033]0;%s\007' % message, skip_eol=True)
 157
 158     def trouble(self, *args, **kargs):
 159         self.ydl.trouble(*args, **kargs)
 160
 161     def report_warning(self, *args, **kargs):
 162         self.ydl.report_warning(*args, **kargs)
 163
 164     def report_error(self, *args, **kargs):
 165         self.ydl.report_error(*args, **kargs)
 166
 167     def slow_down(self, start_time, byte_counter):
 168         """Sleep if the download speed is over the rate limit."""
 169         rate_limit = self.params.get('ratelimit', None)
 170         if rate_limit is None or byte_counter == 0:
 171             return
 172         now = time.time()
 173         elapsed = now - start_time
 174         if elapsed <= 0.0:
 175             return
 176         speed = float(byte_counter) / elapsed
 177         if speed > rate_limit:
 178             time.sleep((byte_counter - rate_limit * (now - start_time)) / rate_limit)
 179
 180     def temp_name(self, filename):
 181         """Returns a temporary filename for the given filename."""
 182         if self.params.get('nopart', False) or filename == u'-' or \
 183                 (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))):
 184             return filename
 185         return filename + u'.part'
 186
 187     def undo_temp_name(self, filename):
 188         if filename.endswith(u'.part'):
 189             return filename[:-len(u'.part')]
 190         return filename
 191
 192     def try_rename(self, old_filename, new_filename):
 193         try:
 194             if old_filename == new_filename:
 195                 return
 196             os.rename(encodeFilename(old_filename), encodeFilename(new_filename))
 197         except (IOError, OSError) as err:
 198             self.report_error(u'unable to rename file')
 199
 200     def try_utime(self, filename, last_modified_hdr):
 201         """Try to set the last-modified time of the given file."""
 202         if last_modified_hdr is None:
 203             return
 204         if not os.path.isfile(encodeFilename(filename)):
 205             return
 206         timestr = last_modified_hdr
 207         if timestr is None:
 208             return
 209         filetime = timeconvert(timestr)
 210         if filetime is None:
 211             return filetime
 212         # Ignore obviously invalid dates
 213         if filetime == 0:
 214             return
 215         try:
 216             os.utime(filename, (time.time(), filetime))
 217         except:
 218             pass
 219         return filetime
 220
 221     def report_destination(self, filename):
 222         """Report destination filename."""
 223         self.to_screen(u'[download] Destination: ' + filename)
 224
 225     def report_progress(self, percent, data_len_str, speed, eta):
 226         """Report download progress."""
 227         if self.params.get('noprogress', False):
 228             return
 229         clear_line = (u'\x1b[K' if sys.stderr.isatty() and os.name != 'nt' else u'')
 230         eta_str = self.format_eta(eta)
 231         percent_str = self.format_percent(percent)
 232         speed_str = self.format_speed(speed)
 233         if self.params.get('progress_with_newline', False):
 234             self.to_screen(u'[download] %s of %s at %s ETA %s' %
 235                 (percent_str, data_len_str, speed_str, eta_str))
 236         else:
 237             self.to_screen(u'\r%s[download] %s of %s at %s ETA %s' %
 238                 (clear_line, percent_str, data_len_str, speed_str, eta_str), skip_eol=True)
 239         self.to_cons_title(u'youtube-dl - %s of %s at %s ETA %s' %
 240                 (percent_str.strip(), data_len_str.strip(), speed_str.strip(), eta_str.strip()))
 241
 242     def report_resuming_byte(self, resume_len):
 243         """Report attempt to resume at given byte."""
 244         self.to_screen(u'[download] Resuming download at byte %s' % resume_len)
 245
 246     def report_retry(self, count, retries):
 247         """Report retry in case of HTTP error 5xx"""
 248         self.to_screen(u'[download] Got server HTTP error. Retrying (attempt %d of %d)...' % (count, retries))
 249
 250     def report_file_already_downloaded(self, file_name):
 251         """Report file has already been fully downloaded."""
 252         try:
 253             self.to_screen(u'[download] %s has already been downloaded' % file_name)
 254         except (UnicodeEncodeError) as err:
 255             self.to_screen(u'[download] The file has already been downloaded')
 256
 257     def report_unable_to_resume(self):
 258         """Report it was impossible to resume download."""
 259         self.to_screen(u'[download] Unable to resume')
 260
 261     def report_finish(self, data_len_str, tot_time):
 262         """Report download finished."""
 263         if self.params.get('noprogress', False):
 264             self.to_screen(u'[download] Download completed')
 265         else:
 266             clear_line = (u'\x1b[K' if sys.stderr.isatty() and os.name != 'nt' else u'')
 267             self.to_screen(u'\r%s[download] 100%% of %s in %s' %
 268                 (clear_line, data_len_str, self.format_seconds(tot_time)))
 269
 270     def _download_with_rtmpdump(self, filename, url, player_url, page_url, play_path, tc_url):
 271         self.report_destination(filename)
 272         tmpfilename = self.temp_name(filename)
 273         test = self.params.get('test', False)
 274
 275         # Check for rtmpdump first
 276         try:
 277             subprocess.call(['rtmpdump', '-h'], stdout=(open(os.path.devnull, 'w')), stderr=subprocess.STDOUT)
 278         except (OSError, IOError):
 279             self.report_error(u'RTMP download detected but "rtmpdump" could not be run')
 280             return False
 281         verbosity_option = '--verbose' if self.params.get('verbose', False) else '--quiet'
 282
 283         # Download using rtmpdump. rtmpdump returns exit code 2 when
 284         # the connection was interrumpted and resuming appears to be
 285         # possible. This is part of rtmpdump's normal usage, AFAIK.
 286         basic_args = ['rtmpdump', verbosity_option, '-r', url, '-o', tmpfilename]
 287         if player_url is not None:
 288             basic_args += ['--swfVfy', player_url]
 289         if page_url is not None:
 290             basic_args += ['--pageUrl', page_url]
 291         if play_path is not None:
 292             basic_args += ['--playpath', play_path]
 293         if tc_url is not None:
 294             basic_args += ['--tcUrl', url]
 295         if test:
 296             basic_args += ['--stop', '1']
 297         args = basic_args + [[], ['--resume', '--skip', '1']][self.params.get('continuedl', False)]
 298         if self.params.get('verbose', False):
 299             try:
 300                 import pipes
 301                 shell_quote = lambda args: ' '.join(map(pipes.quote, args))
 302             except ImportError:
 303                 shell_quote = repr
 304             self.to_screen(u'[debug] rtmpdump command line: ' + shell_quote(args))
 305         retval = subprocess.call(args)
 306         while (retval == 2 or retval == 1) and not test:
 307             prevsize = os.path.getsize(encodeFilename(tmpfilename))
 308             self.to_screen(u'\r[rtmpdump] %s bytes' % prevsize, skip_eol=True)
 309             time.sleep(5.0) # This seems to be needed
 310             retval = subprocess.call(basic_args + ['-e'] + [[], ['-k', '1']][retval == 1])
 311             cursize = os.path.getsize(encodeFilename(tmpfilename))
 312             if prevsize == cursize and retval == 1:
 313                 break
 314              # Some rtmp streams seem abort after ~ 99.8%. Don't complain for those
 315             if prevsize == cursize and retval == 2 and cursize > 1024:
 316                 self.to_screen(u'\r[rtmpdump] Could not download the whole video. This can happen for some advertisements.')
 317                 retval = 0
 318                 break
 319         if retval == 0 or (test and retval == 2):
 320             fsize = os.path.getsize(encodeFilename(tmpfilename))
 321             self.to_screen(u'\r[rtmpdump] %s bytes' % fsize)
 322             self.try_rename(tmpfilename, filename)
 323             self._hook_progress({
 324                 'downloaded_bytes': fsize,
 325                 'total_bytes': fsize,
 326                 'filename': filename,
 327                 'status': 'finished',
 328             })
 329             return True
 330         else:
 331             self.to_stderr(u"\n")
 332             self.report_error(u'rtmpdump exited with code %d' % retval)
 333             return False
 334
 335     def _download_with_mplayer(self, filename, url):
 336         self.report_destination(filename)
 337         tmpfilename = self.temp_name(filename)
 338
 339         args = ['mplayer', '-really-quiet', '-vo', 'null', '-vc', 'dummy', '-dumpstream', '-dumpfile', tmpfilename, url]
 340         # Check for mplayer first
 341         try:
 342             subprocess.call(['mplayer', '-h'], stdout=(open(os.path.devnull, 'w')), stderr=subprocess.STDOUT)
 343         except (OSError, IOError):
 344             self.report_error(u'MMS or RTSP download detected but "%s" could not be run' % args[0] )
 345             return False
 346
 347         # Download using mplayer.
 348         retval = subprocess.call(args)
 349         if retval == 0:
 350             fsize = os.path.getsize(encodeFilename(tmpfilename))
 351             self.to_screen(u'\r[%s] %s bytes' % (args[0], fsize))
 352             self.try_rename(tmpfilename, filename)
 353             self._hook_progress({
 354                 'downloaded_bytes': fsize,
 355                 'total_bytes': fsize,
 356                 'filename': filename,
 357                 'status': 'finished',
 358             })
 359             return True
 360         else:
 361             self.to_stderr(u"\n")
 362             self.report_error(u'mplayer exited with code %d' % retval)
 363             return False
 364
 365     def _download_m3u8_with_ffmpeg(self, filename, url):
 366         self.report_destination(filename)
 367         tmpfilename = self.temp_name(filename)
 368
 369         args = ['ffmpeg', '-y', '-i', url, '-f', 'mp4', tmpfilename]
 370         # Check for ffmpeg first
 371         try:
 372             subprocess.call(['ffmpeg', '-h'], stdout=(open(os.path.devnull, 'w')), stderr=subprocess.STDOUT)
 373         except (OSError, IOError):
 374             self.report_error(u'm3u8 download detected but "%s" could not be run' % args[0] )
 375             return False
 376
 377         retval = subprocess.call(args)
 378         if retval == 0:
 379             fsize = os.path.getsize(encodeFilename(tmpfilename))
 380             self.to_screen(u'\r[%s] %s bytes' % (args[0], fsize))
 381             self.try_rename(tmpfilename, filename)
 382             self._hook_progress({
 383                 'downloaded_bytes': fsize,
 384                 'total_bytes': fsize,
 385                 'filename': filename,
 386                 'status': 'finished',
 387             })
 388             return True
 389         else:
 390             self.to_stderr(u"\n")
 391             self.report_error(u'ffmpeg exited with code %d' % retval)
 392             return False
 393
 394
 395     def _do_download(self, filename, info_dict):
 396         url = info_dict['url']
 397
 398         # Check file already present
 399         if self.params.get('continuedl', False) and os.path.isfile(encodeFilename(filename)) and not self.params.get('nopart', False):
 400             self.report_file_already_downloaded(filename)
 401             self._hook_progress({
 402                 'filename': filename,
 403                 'status': 'finished',
 404                 'total_bytes': os.path.getsize(encodeFilename(filename)),
 405             })
 406             return True
 407
 408         # Attempt to download using rtmpdump
 409         if url.startswith('rtmp'):
 410             return self._download_with_rtmpdump(filename, url,
 411                                                 info_dict.get('player_url', None),
 412                                                 info_dict.get('page_url', None),
 413                                                 info_dict.get('play_path', None),
 414                                                 info_dict.get('tc_url', None))
 415
 416         # Attempt to download using mplayer
 417         if url.startswith('mms') or url.startswith('rtsp'):
 418             return self._download_with_mplayer(filename, url)
 419
 420         # m3u8 manifest are downloaded with ffmpeg
 421         if determine_ext(url) == u'm3u8':
 422             return self._download_m3u8_with_ffmpeg(filename, url)
 423
 424         tmpfilename = self.temp_name(filename)
 425         stream = None
 426
 427         # Do not include the Accept-Encoding header
 428         headers = {'Youtubedl-no-compression': 'True'}
 429         if 'user_agent' in info_dict:
 430             headers['Youtubedl-user-agent'] = info_dict['user_agent']
 431         basic_request = compat_urllib_request.Request(url, None, headers)
 432         request = compat_urllib_request.Request(url, None, headers)
 433
 434         if self.params.get('test', False):
 435             request.add_header('Range','bytes=0-10240')
 436
 437         # Establish possible resume length
 438         if os.path.isfile(encodeFilename(tmpfilename)):
 439             resume_len = os.path.getsize(encodeFilename(tmpfilename))
 440         else:
 441             resume_len = 0
 442
 443         open_mode = 'wb'
 444         if resume_len != 0:
 445             if self.params.get('continuedl', False):
 446                 self.report_resuming_byte(resume_len)
 447                 request.add_header('Range','bytes=%d-' % resume_len)
 448                 open_mode = 'ab'
 449             else:
 450                 resume_len = 0
 451
 452         count = 0
 453         retries = self.params.get('retries', 0)
 454         while count <= retries:
 455             # Establish connection
 456             try:
 457                 if count == 0 and 'urlhandle' in info_dict:
 458                     data = info_dict['urlhandle']
 459                 data = compat_urllib_request.urlopen(request)
 460                 break
 461             except (compat_urllib_error.HTTPError, ) as err:
 462                 if (err.code < 500 or err.code >= 600) and err.code != 416:
 463                     # Unexpected HTTP error
 464                     raise
 465                 elif err.code == 416:
 466                     # Unable to resume (requested range not satisfiable)
 467                     try:
 468                         # Open the connection again without the range header
 469                         data = compat_urllib_request.urlopen(basic_request)
 470                         content_length = data.info()['Content-Length']
 471                     except (compat_urllib_error.HTTPError, ) as err:
 472                         if err.code < 500 or err.code >= 600:
 473                             raise
 474                     else:
 475                         # Examine the reported length
 476                         if (content_length is not None and
 477                                 (resume_len - 100 < int(content_length) < resume_len + 100)):
 478                             # The file had already been fully downloaded.
 479                             # Explanation to the above condition: in issue #175 it was revealed that
 480                             # YouTube sometimes adds or removes a few bytes from the end of the file,
 481                             # changing the file size slightly and causing problems for some users. So
 482                             # I decided to implement a suggested change and consider the file
 483                             # completely downloaded if the file size differs less than 100 bytes from
 484                             # the one in the hard drive.
 485                             self.report_file_already_downloaded(filename)
 486                             self.try_rename(tmpfilename, filename)
 487                             self._hook_progress({
 488                                 'filename': filename,
 489                                 'status': 'finished',
 490                             })
 491                             return True
 492                         else:
 493                             # The length does not match, we start the download over
 494                             self.report_unable_to_resume()
 495                             open_mode = 'wb'
 496                             break
 497             # Retry
 498             count += 1
 499             if count <= retries:
 500                 self.report_retry(count, retries)
 501
 502         if count > retries:
 503             self.report_error(u'giving up after %s retries' % retries)
 504             return False
 505
 506         data_len = data.info().get('Content-length', None)
 507         if data_len is not None:
 508             data_len = int(data_len) + resume_len
 509             min_data_len = self.params.get("min_filesize", None)
 510             max_data_len =  self.params.get("max_filesize", None)
 511             if min_data_len is not None and data_len < min_data_len:
 512                 self.to_screen(u'\r[download] File is smaller than min-filesize (%s bytes < %s bytes). Aborting.' % (data_len, min_data_len))
 513                 return False
 514             if max_data_len is not None and data_len > max_data_len:
 515                 self.to_screen(u'\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len, max_data_len))
 516                 return False
 517
 518         data_len_str = self.format_bytes(data_len)
 519         byte_counter = 0 + resume_len
 520         block_size = self.params.get('buffersize', 1024)
 521         start = time.time()
 522         while True:
 523             # Download and write
 524             before = time.time()
 525             data_block = data.read(block_size)
 526             after = time.time()
 527             if len(data_block) == 0:
 528                 break
 529             byte_counter += len(data_block)
 530
 531             # Open file just in time
 532             if stream is None:
 533                 try:
 534                     (stream, tmpfilename) = sanitize_open(tmpfilename, open_mode)
 535                     assert stream is not None
 536                     filename = self.undo_temp_name(tmpfilename)
 537                     self.report_destination(filename)
 538                 except (OSError, IOError) as err:
 539                     self.report_error(u'unable to open for writing: %s' % str(err))
 540                     return False
 541             try:
 542                 stream.write(data_block)
 543             except (IOError, OSError) as err:
 544                 self.to_stderr(u"\n")
 545                 self.report_error(u'unable to write data: %s' % str(err))
 546                 return False
 547             if not self.params.get('noresizebuffer', False):
 548                 block_size = self.best_block_size(after - before, len(data_block))
 549
 550             # Progress message
 551             speed = self.calc_speed(start, time.time(), byte_counter - resume_len)
 552             if data_len is None:
 553                 self.report_progress('Unknown %', data_len_str, speed_str, 'Unknown ETA')
 554                 eta = None
 555             else:
 556                 percent = self.calc_percent(byte_counter, data_len)
 557                 eta = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len)
 558                 self.report_progress(percent, data_len_str, speed, eta)
 559
 560             self._hook_progress({
 561                 'downloaded_bytes': byte_counter,
 562                 'total_bytes': data_len,
 563                 'tmpfilename': tmpfilename,
 564                 'filename': filename,
 565                 'status': 'downloading',
 566                 'eta': eta,
 567                 'speed': speed,
 568             })
 569
 570             # Apply rate limit
 571             self.slow_down(start, byte_counter - resume_len)
 572
 573         if stream is None:
 574             self.to_stderr(u"\n")
 575             self.report_error(u'Did not get any data blocks')
 576             return False
 577         stream.close()
 578         self.report_finish(data_len_str, (time.time() - start))
 579         if data_len is not None and byte_counter != data_len:
 580             raise ContentTooShortError(byte_counter, int(data_len))
 581         self.try_rename(tmpfilename, filename)
 582
 583         # Update file modification time
 584         if self.params.get('updatetime', True):
 585             info_dict['filetime'] = self.try_utime(filename, data.info().get('last-modified', None))
 586
 587         self._hook_progress({
 588             'downloaded_bytes': byte_counter,
 589             'total_bytes': byte_counter,
 590             'filename': filename,
 591             'status': 'finished',
 592         })
 593
 594         return True
 595
 596     def _hook_progress(self, status):
 597         for ph in self._progress_hooks:
 598             ph(status)
 599
 600     def add_progress_hook(self, ph):
 601         """ ph gets called on download progress, with a dictionary with the entries
 602         * filename: The final filename
 603         * status: One of "downloading" and "finished"
 604
 605         It can also have some of the following entries:
 606
 607         * downloaded_bytes: Bytes on disks
 608         * total_bytes: Total bytes, None if unknown
 609         * tmpfilename: The filename we're currently writing to
 610         * eta: The estimated time in seconds, None if unknown
 611         * speed: The download speed in bytes/second, None if unknown
 612
 613         Hooks are guaranteed to be called at least once (with status "finished")
 614         if the download is successful.
 615         """
 616         self._progress_hooks.append(ph)