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