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