Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/downloader/http.py

   1 from __future__ import unicode_literals
   2
   3 import os
   4 import time
   5
   6 from socket import error as SocketError
   7 import errno
   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     format_bytes,
  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         data = info_dict.get('http_post_data')
  34         http_method = info_dict.get('http_method')
  35         basic_request = compat_urllib_request.Request(url, data, headers)
  36         request = compat_urllib_request.Request(url, data, headers)
  37         if http_method is not None:
  38             basic_request.get_method = lambda: http_method
  39             request.get_method = lambda: http_method
  40
  41         is_test = self.params.get('test', False)
  42
  43         if is_test:
  44             request.add_header('Range', 'bytes=0-%s' % str(self._TEST_FILE_SIZE - 1))
  45
  46         # Establish possible resume length
  47         if os.path.isfile(encodeFilename(tmpfilename)):
  48             resume_len = os.path.getsize(encodeFilename(tmpfilename))
  49         else:
  50             resume_len = 0
  51
  52         open_mode = 'wb'
  53         if resume_len != 0:
  54             if self.params.get('continuedl', False):
  55                 self.report_resuming_byte(resume_len)
  56                 request.add_header('Range', 'bytes=%d-' % resume_len)
  57                 open_mode = 'ab'
  58             else:
  59                 resume_len = 0
  60
  61         count = 0
  62         retries = self.params.get('retries', 0)
  63         while count <= retries:
  64             # Establish connection
  65             try:
  66                 data = self.ydl.urlopen(request)
  67                 break
  68             except (compat_urllib_error.HTTPError, ) as err:
  69                 if (err.code < 500 or err.code >= 600) and err.code != 416:
  70                     # Unexpected HTTP error
  71                     raise
  72                 elif err.code == 416:
  73                     # Unable to resume (requested range not satisfiable)
  74                     try:
  75                         # Open the connection again without the range header
  76                         data = self.ydl.urlopen(basic_request)
  77                         content_length = data.info()['Content-Length']
  78                     except (compat_urllib_error.HTTPError, ) as err:
  79                         if err.code < 500 or err.code >= 600:
  80                             raise
  81                     else:
  82                         # Examine the reported length
  83                         if (content_length is not None and
  84                                 (resume_len - 100 < int(content_length) < resume_len + 100)):
  85                             # The file had already been fully downloaded.
  86                             # Explanation to the above condition: in issue #175 it was revealed that
  87                             # YouTube sometimes adds or removes a few bytes from the end of the file,
  88                             # changing the file size slightly and causing problems for some users. So
  89                             # I decided to implement a suggested change and consider the file
  90                             # completely downloaded if the file size differs less than 100 bytes from
  91                             # the one in the hard drive.
  92                             self.report_file_already_downloaded(filename)
  93                             self.try_rename(tmpfilename, filename)
  94                             self._hook_progress({
  95                                 'filename': filename,
  96                                 'status': 'finished',
  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 SocketError 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         data_len_str = format_bytes(data_len)
 141         byte_counter = 0 + resume_len
 142         block_size = self.params.get('buffersize', 1024)
 143         start = time.time()
 144
 145         # measure time over whole while-loop, so slow_down() and best_block_size() work together properly
 146         now = None  # needed for slow_down() in the first loop run
 147         before = start  # start measuring
 148         while True:
 149
 150             # Download and write
 151             data_block = data.read(block_size if not is_test else min(block_size, data_len - byte_counter))
 152             byte_counter += len(data_block)
 153
 154             # exit loop when download is finished
 155             if len(data_block) == 0:
 156                 break
 157
 158             # Open destination file just in time
 159             if stream is None:
 160                 try:
 161                     (stream, tmpfilename) = sanitize_open(tmpfilename, open_mode)
 162                     assert stream is not None
 163                     filename = self.undo_temp_name(tmpfilename)
 164                     self.report_destination(filename)
 165                 except (OSError, IOError) as err:
 166                     self.report_error('unable to open for writing: %s' % str(err))
 167                     return False
 168
 169                 if self.params.get('xattr_set_filesize', False) and data_len is not None:
 170                     try:
 171                         import xattr
 172                         xattr.setxattr(tmpfilename, 'user.ytdl.filesize', str(data_len))
 173                     except(OSError, IOError, ImportError) as err:
 174                         self.report_error('unable to set filesize xattr: %s' % str(err))
 175
 176             try:
 177                 stream.write(data_block)
 178             except (IOError, OSError) as err:
 179                 self.to_stderr('\n')
 180                 self.report_error('unable to write data: %s' % str(err))
 181                 return False
 182
 183             # Apply rate limit
 184             self.slow_down(start, now, byte_counter - resume_len)
 185
 186             # end measuring of one loop run
 187             now = time.time()
 188             after = now
 189
 190             # Adjust block size
 191             if not self.params.get('noresizebuffer', False):
 192                 block_size = self.best_block_size(after - before, len(data_block))
 193
 194             before = after
 195
 196             # Progress message
 197             speed = self.calc_speed(start, now, byte_counter - resume_len)
 198             if data_len is None:
 199                 eta = percent = None
 200             else:
 201                 percent = self.calc_percent(byte_counter, data_len)
 202                 eta = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len)
 203             self.report_progress(percent, data_len_str, speed, eta)
 204
 205             self._hook_progress({
 206                 'downloaded_bytes': byte_counter,
 207                 'total_bytes': data_len,
 208                 'tmpfilename': tmpfilename,
 209                 'filename': filename,
 210                 'status': 'downloading',
 211                 'eta': eta,
 212                 'speed': speed,
 213             })
 214
 215             if is_test and byte_counter == data_len:
 216                 break
 217
 218         if stream is None:
 219             self.to_stderr('\n')
 220             self.report_error('Did not get any data blocks')
 221             return False
 222         if tmpfilename != '-':
 223             stream.close()
 224         self.report_finish(data_len_str, (time.time() - start))
 225         if data_len is not None and byte_counter != data_len:
 226             raise ContentTooShortError(byte_counter, int(data_len))
 227         self.try_rename(tmpfilename, filename)
 228
 229         # Update file modification time
 230         if self.params.get('updatetime', True):
 231             info_dict['filetime'] = self.try_utime(filename, data.info().get('last-modified', None))
 232
 233         self._hook_progress({
 234             'downloaded_bytes': byte_counter,
 235             'total_bytes': byte_counter,
 236             'filename': filename,
 237             'status': 'finished',
 238         })
 239
 240         return True