youtube_dl/downloader/http.py

   1 from __future__ import unicode_literals
   2
   3 import errno
   4 import os
   5 import socket
   6 import time
   7
   8 from .common import FileDownloader
   9 from ..compat import (
  10     compat_urllib_request,
  11     compat_urllib_error,
  12 )
  13 from ..utils import (
  14     ContentTooShortError,
  15     encodeFilename,
  16     sanitize_open,
  17 )
  18
  19
  20 class HttpFD(FileDownloader):
  21     def real_download(self, filename, info_dict):
  22         url = info_dict['url']
  23         tmpfilename = self.temp_name(filename)
  24         stream = None
  25
  26         # Do not include the Accept-Encoding header
  27         headers = {'Youtubedl-no-compression': 'True'}
  28         add_headers = info_dict.get('http_headers')
  29         if add_headers:
  30             headers.update(add_headers)
  31         data = info_dict.get('http_post_data')
  32         http_method = info_dict.get('http_method')
  33         basic_request = compat_urllib_request.Request(url, data, headers)
  34         request = compat_urllib_request.Request(url, data, headers)
  35         if http_method is not None:
  36             basic_request.get_method = lambda: http_method
  37             request.get_method = lambda: http_method
  38
  39         is_test = self.params.get('test', False)
  40
  41         if is_test:
  42             request.add_header('Range', 'bytes=0-%s' % str(self._TEST_FILE_SIZE - 1))
  43
  44         # Establish possible resume length
  45         if os.path.isfile(encodeFilename(tmpfilename)):
  46             resume_len = os.path.getsize(encodeFilename(tmpfilename))
  47         else:
  48             resume_len = 0
  49
  50         open_mode = 'wb'
  51         if resume_len != 0:
  52             if self.params.get('continuedl', False):
  53                 self.report_resuming_byte(resume_len)
  54                 request.add_header('Range', 'bytes=%d-' % resume_len)
  55                 open_mode = 'ab'
  56             else:
  57                 resume_len = 0
  58
  59         count = 0
  60         retries = self.params.get('retries', 0)
  61         while count <= retries:
  62             # Establish connection
  63             try:
  64                 data = self.ydl.urlopen(request)
  65                 break
  66             except (compat_urllib_error.HTTPError, ) as err:
  67                 if (err.code < 500 or err.code >= 600) and err.code != 416:
  68                     # Unexpected HTTP error
  69                     raise
  70                 elif err.code == 416:
  71                     # Unable to resume (requested range not satisfiable)
  72                     try:
  73                         # Open the connection again without the range header
  74                         data = self.ydl.urlopen(basic_request)
  75                         content_length = data.info()['Content-Length']
  76                     except (compat_urllib_error.HTTPError, ) as err:
  77                         if err.code < 500 or err.code >= 600:
  78                             raise
  79                     else:
  80                         # Examine the reported length
  81                         if (content_length is not None and
  82                                 (resume_len - 100 < int(content_length) < resume_len + 100)):
  83                             # The file had already been fully downloaded.
  84                             # Explanation to the above condition: in issue #175 it was revealed that
  85                             # YouTube sometimes adds or removes a few bytes from the end of the file,
  86                             # changing the file size slightly and causing problems for some users. So
  87                             # I decided to implement a suggested change and consider the file
  88                             # completely downloaded if the file size differs less than 100 bytes from
  89                             # the one in the hard drive.
  90                             self.report_file_already_downloaded(filename)
  91                             self.try_rename(tmpfilename, filename)
  92                             self._hook_progress({
  93                                 'filename': filename,
  94                                 'status': 'finished',
  95                                 'downloaded_bytes': resume_len,
  96                                 'total_bytes': resume_len,
  97                             })
  98                             return True
  99                         else:
 100                             # The length does not match, we start the download over
 101                             self.report_unable_to_resume()
 102                             resume_len = 0
 103                             open_mode = 'wb'
 104                             break
 105             except socket.error as e:
 106                 if e.errno != errno.ECONNRESET:
 107                     # Connection reset is no problem, just retry
 108                     raise
 109
 110             # Retry
 111             count += 1
 112             if count <= retries:
 113                 self.report_retry(count, retries)
 114
 115         if count > retries:
 116             self.report_error('giving up after %s retries' % retries)
 117             return False
 118
 119         data_len = data.info().get('Content-length', None)
 120
 121         # Range HTTP header may be ignored/unsupported by a webserver
 122         # (e.g. extractor/scivee.py, extractor/bambuser.py).
 123         # However, for a test we still would like to download just a piece of a file.
 124         # To achieve this we limit data_len to _TEST_FILE_SIZE and manually control
 125         # block size when downloading a file.
 126         if is_test and (data_len is None or int(data_len) > self._TEST_FILE_SIZE):
 127             data_len = self._TEST_FILE_SIZE
 128
 129         if data_len is not None:
 130             data_len = int(data_len) + resume_len
 131             min_data_len = self.params.get("min_filesize", None)
 132             max_data_len = self.params.get("max_filesize", None)
 133             if min_data_len is not None and data_len < min_data_len:
 134                 self.to_screen('\r[download] File is smaller than min-filesize (%s bytes < %s bytes). Aborting.' % (data_len, min_data_len))
 135                 return False
 136             if max_data_len is not None and data_len > max_data_len:
 137                 self.to_screen('\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len, max_data_len))
 138                 return False
 139
 140         byte_counter = 0 + resume_len
 141         block_size = self.params.get('buffersize', 1024)
 142         start = time.time()
 143
 144         # measure time over whole while-loop, so slow_down() and best_block_size() work together properly
 145         now = None  # needed for slow_down() in the first loop run
 146         before = start  # start measuring
 147         while True:
 148
 149             # Download and write
 150             data_block = data.read(block_size if not is_test else min(block_size, data_len - byte_counter))
 151             byte_counter += len(data_block)
 152
 153             # exit loop when download is finished
 154             if len(data_block) == 0:
 155                 break
 156
 157             # Open destination file just in time
 158             if stream is None:
 159                 try:
 160                     (stream, tmpfilename) = sanitize_open(tmpfilename, open_mode)
 161                     assert stream is not None
 162                     filename = self.undo_temp_name(tmpfilename)
 163                     self.report_destination(filename)
 164                 except (OSError, IOError) as err:
 165                     self.report_error('unable to open for writing: %s' % str(err))
 166                     return False
 167
 168                 if self.params.get('xattr_set_filesize', False) and data_len is not None:
 169                     try:
 170                         import xattr
 171                         xattr.setxattr(tmpfilename, 'user.ytdl.filesize', str(data_len))
 172                     except(OSError, IOError, ImportError) as err:
 173                         self.report_error('unable to set filesize xattr: %s' % str(err))
 174
 175             try:
 176                 stream.write(data_block)
 177             except (IOError, OSError) as err:
 178                 self.to_stderr('\n')
 179                 self.report_error('unable to write data: %s' % str(err))
 180                 return False
 181
 182             # Apply rate limit
 183             self.slow_down(start, now, byte_counter - resume_len)
 184
 185             # end measuring of one loop run
 186             now = time.time()
 187             after = now
 188
 189             # Adjust block size
 190             if not self.params.get('noresizebuffer', False):
 191                 block_size = self.best_block_size(after - before, len(data_block))
 192
 193             before = after
 194
 195             # Progress message
 196             speed = self.calc_speed(start, now, byte_counter - resume_len)
 197             if data_len is None:
 198                 eta = None
 199             else:
 200                 eta = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len)
 201
 202             self._hook_progress({
 203                 'status': 'downloading',
 204                 'downloaded_bytes': byte_counter,
 205                 'total_bytes': data_len,
 206                 'tmpfilename': tmpfilename,
 207                 'filename': filename,
 208                 'eta': eta,
 209                 'speed': speed,
 210                 'elapsed': now - start,
 211             })
 212
 213             if is_test and byte_counter == data_len:
 214                 break
 215
 216         if stream is None:
 217             self.to_stderr('\n')
 218             self.report_error('Did not get any data blocks')
 219             return False
 220         if tmpfilename != '-':
 221             stream.close()
 222
 223         if data_len is not None and byte_counter != data_len:
 224             raise ContentTooShortError(byte_counter, int(data_len))
 225         self.try_rename(tmpfilename, filename)
 226
 227         # Update file modification time
 228         if self.params.get('updatetime', True):
 229             info_dict['filetime'] = self.try_utime(filename, data.info().get('last-modified', None))
 230
 231         self._hook_progress({
 232             'downloaded_bytes': byte_counter,
 233             'total_bytes': byte_counter,
 234             'filename': filename,
 235             'status': 'finished',
 236             'elapsed': time.time() - start,
 237         })
 238
 239         return True