Raphaël G. Git Repositories - youtubedl/blob - 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 import re
   8
   9 from .common import FileDownloader
  10 from ..compat import compat_urllib_error
  11 from ..utils import (
  12     ContentTooShortError,
  13     encodeFilename,
  14     sanitize_open,
  15     sanitized_Request,
  16     write_xattr,
  17     XAttrMetadataError,
  18     XAttrUnavailableError,
  19 )
  20
  21
  22 class HttpFD(FileDownloader):
  23     def real_download(self, filename, info_dict):
  24         url = info_dict['url']
  25         tmpfilename = self.temp_name(filename)
  26         stream = None
  27
  28         # Do not include the Accept-Encoding header
  29         headers = {'Youtubedl-no-compression': 'True'}
  30         add_headers = info_dict.get('http_headers')
  31         if add_headers:
  32             headers.update(add_headers)
  33         basic_request = sanitized_Request(url, None, headers)
  34         request = sanitized_Request(url, None, headers)
  35
  36         is_test = self.params.get('test', False)
  37
  38         if is_test:
  39             request.add_header('Range', 'bytes=0-%s' % str(self._TEST_FILE_SIZE - 1))
  40
  41         # Establish possible resume length
  42         if os.path.isfile(encodeFilename(tmpfilename)):
  43             resume_len = os.path.getsize(encodeFilename(tmpfilename))
  44         else:
  45             resume_len = 0
  46
  47         open_mode = 'wb'
  48         if resume_len != 0:
  49             if self.params.get('continuedl', True):
  50                 self.report_resuming_byte(resume_len)
  51                 request.add_header('Range', 'bytes=%d-' % resume_len)
  52                 open_mode = 'ab'
  53             else:
  54                 resume_len = 0
  55
  56         count = 0
  57         retries = self.params.get('retries', 0)
  58         while count <= retries:
  59             # Establish connection
  60             try:
  61                 data = self.ydl.urlopen(request)
  62                 # When trying to resume, Content-Range HTTP header of response has to be checked
  63                 # to match the value of requested Range HTTP header. This is due to a webservers
  64                 # that don't support resuming and serve a whole file with no Content-Range
  65                 # set in response despite of requested Range (see
  66                 # https://github.com/rg3/youtube-dl/issues/6057#issuecomment-126129799)
  67                 if resume_len > 0:
  68                     content_range = data.headers.get('Content-Range')
  69                     if content_range:
  70                         content_range_m = re.search(r'bytes (\d+)-', content_range)
  71                         # Content-Range is present and matches requested Range, resume is possible
  72                         if content_range_m and resume_len == int(content_range_m.group(1)):
  73                             break
  74                     # Content-Range is either not present or invalid. Assuming remote webserver is
  75                     # trying to send the whole file, resume is not possible, so wiping the local file
  76                     # and performing entire redownload
  77                     self.report_unable_to_resume()
  78                     resume_len = 0
  79                     open_mode = 'wb'
  80                 break
  81             except (compat_urllib_error.HTTPError, ) as err:
  82                 if (err.code < 500 or err.code >= 600) and err.code != 416:
  83                     # Unexpected HTTP error
  84                     raise
  85                 elif err.code == 416:
  86                     # Unable to resume (requested range not satisfiable)
  87                     try:
  88                         # Open the connection again without the range header
  89                         data = self.ydl.urlopen(basic_request)
  90                         content_length = data.info()['Content-Length']
  91                     except (compat_urllib_error.HTTPError, ) as err:
  92                         if err.code < 500 or err.code >= 600:
  93                             raise
  94                     else:
  95                         # Examine the reported length
  96                         if (content_length is not None and
  97                                 (resume_len - 100 < int(content_length) < resume_len + 100)):
  98                             # The file had already been fully downloaded.
  99                             # Explanation to the above condition: in issue #175 it was revealed that
 100                             # YouTube sometimes adds or removes a few bytes from the end of the file,
 101                             # changing the file size slightly and causing problems for some users. So
 102                             # I decided to implement a suggested change and consider the file
 103                             # completely downloaded if the file size differs less than 100 bytes from
 104                             # the one in the hard drive.
 105                             self.report_file_already_downloaded(filename)
 106                             self.try_rename(tmpfilename, filename)
 107                             self._hook_progress({
 108                                 'filename': filename,
 109                                 'status': 'finished',
 110                                 'downloaded_bytes': resume_len,
 111                                 'total_bytes': resume_len,
 112                             })
 113                             return True
 114                         else:
 115                             # The length does not match, we start the download over
 116                             self.report_unable_to_resume()
 117                             resume_len = 0
 118                             open_mode = 'wb'
 119                             break
 120             except socket.error as e:
 121                 if e.errno != errno.ECONNRESET:
 122                     # Connection reset is no problem, just retry
 123                     raise
 124
 125             # Retry
 126             count += 1
 127             if count <= retries:
 128                 self.report_retry(count, retries)
 129
 130         if count > retries:
 131             self.report_error('giving up after %s retries' % retries)
 132             return False
 133
 134         data_len = data.info().get('Content-length', None)
 135
 136         # Range HTTP header may be ignored/unsupported by a webserver
 137         # (e.g. extractor/scivee.py, extractor/bambuser.py).
 138         # However, for a test we still would like to download just a piece of a file.
 139         # To achieve this we limit data_len to _TEST_FILE_SIZE and manually control
 140         # block size when downloading a file.
 141         if is_test and (data_len is None or int(data_len) > self._TEST_FILE_SIZE):
 142             data_len = self._TEST_FILE_SIZE
 143
 144         if data_len is not None:
 145             data_len = int(data_len) + resume_len
 146             min_data_len = self.params.get('min_filesize')
 147             max_data_len = self.params.get('max_filesize')
 148             if min_data_len is not None and data_len < min_data_len:
 149                 self.to_screen('\r[download] File is smaller than min-filesize (%s bytes < %s bytes). Aborting.' % (data_len, min_data_len))
 150                 return False
 151             if max_data_len is not None and data_len > max_data_len:
 152                 self.to_screen('\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len, max_data_len))
 153                 return False
 154
 155         byte_counter = 0 + resume_len
 156         block_size = self.params.get('buffersize', 1024)
 157         start = time.time()
 158
 159         # measure time over whole while-loop, so slow_down() and best_block_size() work together properly
 160         now = None  # needed for slow_down() in the first loop run
 161         before = start  # start measuring
 162         while True:
 163
 164             # Download and write
 165             data_block = data.read(block_size if not is_test else min(block_size, data_len - byte_counter))
 166             byte_counter += len(data_block)
 167
 168             # exit loop when download is finished
 169             if len(data_block) == 0:
 170                 break
 171
 172             # Open destination file just in time
 173             if stream is None:
 174                 try:
 175                     (stream, tmpfilename) = sanitize_open(tmpfilename, open_mode)
 176                     assert stream is not None
 177                     filename = self.undo_temp_name(tmpfilename)
 178                     self.report_destination(filename)
 179                 except (OSError, IOError) as err:
 180                     self.report_error('unable to open for writing: %s' % str(err))
 181                     return False
 182
 183                 if self.params.get('xattr_set_filesize', False) and data_len is not None:
 184                     try:
 185                         write_xattr(tmpfilename, 'user.ytdl.filesize', str(data_len).encode('utf-8'))
 186                     except (XAttrUnavailableError, XAttrMetadataError) as err:
 187                         self.report_error('unable to set filesize xattr: %s' % str(err))
 188
 189             try:
 190                 stream.write(data_block)
 191             except (IOError, OSError) as err:
 192                 self.to_stderr('\n')
 193                 self.report_error('unable to write data: %s' % str(err))
 194                 return False
 195
 196             # Apply rate limit
 197             self.slow_down(start, now, byte_counter - resume_len)
 198
 199             # end measuring of one loop run
 200             now = time.time()
 201             after = now
 202
 203             # Adjust block size
 204             if not self.params.get('noresizebuffer', False):
 205                 block_size = self.best_block_size(after - before, len(data_block))
 206
 207             before = after
 208
 209             # Progress message
 210             speed = self.calc_speed(start, now, byte_counter - resume_len)
 211             if data_len is None:
 212                 eta = None
 213             else:
 214                 eta = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len)
 215
 216             self._hook_progress({
 217                 'status': 'downloading',
 218                 'downloaded_bytes': byte_counter,
 219                 'total_bytes': data_len,
 220                 'tmpfilename': tmpfilename,
 221                 'filename': filename,
 222                 'eta': eta,
 223                 'speed': speed,
 224                 'elapsed': now - start,
 225             })
 226
 227             if is_test and byte_counter == data_len:
 228                 break
 229
 230         if stream is None:
 231             self.to_stderr('\n')
 232             self.report_error('Did not get any data blocks')
 233             return False
 234         if tmpfilename != '-':
 235             stream.close()
 236
 237         if data_len is not None and byte_counter != data_len:
 238             raise ContentTooShortError(byte_counter, int(data_len))
 239         self.try_rename(tmpfilename, filename)
 240
 241         # Update file modification time
 242         if self.params.get('updatetime', True):
 243             info_dict['filetime'] = self.try_utime(filename, data.info().get('last-modified', None))
 244
 245         self._hook_progress({
 246             'downloaded_bytes': byte_counter,
 247             'total_bytes': byte_counter,
 248             'filename': filename,
 249             'status': 'finished',
 250             'elapsed': time.time() - start,
 251         })
 252
 253         return True