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
  62                 if resume_len > 0:
  63                     content_range = data.headers.get('Content-Range')
  64                     if content_range:
  65                         content_range_m = re.search(r'bytes (\d+)-', content_range)
  66                         if content_range_m:
  67                             # Content-Range is correct - go on
  68                             if resume_len == int(content_range_m.group(1)):
  69                                 break
  70
  71                     # Content-Range is invalid - wipe the file and do entire redownload
  72                     resume_len = 0
  73                     open_mode = 'wb'
  74
  75                 break
  76             except (compat_urllib_error.HTTPError, ) as err:
  77                 if (err.code < 500 or err.code >= 600) and err.code != 416:
  78                     # Unexpected HTTP error
  79                     raise
  80                 elif err.code == 416:
  81                     # Unable to resume (requested range not satisfiable)
  82                     try:
  83                         # Open the connection again without the range header
  84                         data = self.ydl.urlopen(basic_request)
  85                         content_length = data.info()['Content-Length']
  86                     except (compat_urllib_error.HTTPError, ) as err:
  87                         if err.code < 500 or err.code >= 600:
  88                             raise
  89                     else:
  90                         # Examine the reported length
  91                         if (content_length is not None and
  92                                 (resume_len - 100 < int(content_length) < resume_len + 100)):
  93                             # The file had already been fully downloaded.
  94                             # Explanation to the above condition: in issue #175 it was revealed that
  95                             # YouTube sometimes adds or removes a few bytes from the end of the file,
  96                             # changing the file size slightly and causing problems for some users. So
  97                             # I decided to implement a suggested change and consider the file
  98                             # completely downloaded if the file size differs less than 100 bytes from
  99                             # the one in the hard drive.
 100                             self.report_file_already_downloaded(filename)
 101                             self.try_rename(tmpfilename, filename)
 102                             self._hook_progress({
 103                                 'filename': filename,
 104                                 'status': 'finished',
 105                                 'downloaded_bytes': resume_len,
 106                                 'total_bytes': resume_len,
 107                             })
 108                             return True
 109                         else:
 110                             # The length does not match, we start the download over
 111                             self.report_unable_to_resume()
 112                             resume_len = 0
 113                             open_mode = 'wb'
 114                             break
 115             except socket.error as e:
 116                 if e.errno != errno.ECONNRESET:
 117                     # Connection reset is no problem, just retry
 118                     raise
 119
 120             # Retry
 121             count += 1
 122             if count <= retries:
 123                 self.report_retry(count, retries)
 124
 125         if count > retries:
 126             self.report_error('giving up after %s retries' % retries)
 127             return False
 128
 129         data_len = data.info().get('Content-length', None)
 130
 131         # Range HTTP header may be ignored/unsupported by a webserver
 132         # (e.g. extractor/scivee.py, extractor/bambuser.py).
 133         # However, for a test we still would like to download just a piece of a file.
 134         # To achieve this we limit data_len to _TEST_FILE_SIZE and manually control
 135         # block size when downloading a file.
 136         if is_test and (data_len is None or int(data_len) > self._TEST_FILE_SIZE):
 137             data_len = self._TEST_FILE_SIZE
 138
 139         if data_len is not None:
 140             data_len = int(data_len) + resume_len
 141             min_data_len = self.params.get("min_filesize", None)
 142             max_data_len = self.params.get("max_filesize", None)
 143             if min_data_len is not None and data_len < min_data_len:
 144                 self.to_screen('\r[download] File is smaller than min-filesize (%s bytes < %s bytes). Aborting.' % (data_len, min_data_len))
 145                 return False
 146             if max_data_len is not None and data_len > max_data_len:
 147                 self.to_screen('\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len, max_data_len))
 148                 return False
 149
 150         byte_counter = 0 + resume_len
 151         block_size = self.params.get('buffersize', 1024)
 152         start = time.time()
 153
 154         # measure time over whole while-loop, so slow_down() and best_block_size() work together properly
 155         now = None  # needed for slow_down() in the first loop run
 156         before = start  # start measuring
 157         while True:
 158
 159             # Download and write
 160             data_block = data.read(block_size if not is_test else min(block_size, data_len - byte_counter))
 161             byte_counter += len(data_block)
 162
 163             # exit loop when download is finished
 164             if len(data_block) == 0:
 165                 break
 166
 167             # Open destination file just in time
 168             if stream is None:
 169                 try:
 170                     (stream, tmpfilename) = sanitize_open(tmpfilename, open_mode)
 171                     assert stream is not None
 172                     filename = self.undo_temp_name(tmpfilename)
 173                     self.report_destination(filename)
 174                 except (OSError, IOError) as err:
 175                     self.report_error('unable to open for writing: %s' % str(err))
 176                     return False
 177
 178                 if self.params.get('xattr_set_filesize', False) and data_len is not None:
 179                     try:
 180                         import xattr
 181                         xattr.setxattr(tmpfilename, 'user.ytdl.filesize', str(data_len))
 182                     except(OSError, IOError, ImportError) as err:
 183                         self.report_error('unable to set filesize xattr: %s' % str(err))
 184
 185             try:
 186                 stream.write(data_block)
 187             except (IOError, OSError) as err:
 188                 self.to_stderr('\n')
 189                 self.report_error('unable to write data: %s' % str(err))
 190                 return False
 191
 192             # Apply rate limit
 193             self.slow_down(start, now, byte_counter - resume_len)
 194
 195             # end measuring of one loop run
 196             now = time.time()
 197             after = now
 198
 199             # Adjust block size
 200             if not self.params.get('noresizebuffer', False):
 201                 block_size = self.best_block_size(after - before, len(data_block))
 202
 203             before = after
 204
 205             # Progress message
 206             speed = self.calc_speed(start, now, byte_counter - resume_len)
 207             if data_len is None:
 208                 eta = None
 209             else:
 210                 eta = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len)
 211
 212             self._hook_progress({
 213                 'status': 'downloading',
 214                 'downloaded_bytes': byte_counter,
 215                 'total_bytes': data_len,
 216                 'tmpfilename': tmpfilename,
 217                 'filename': filename,
 218                 'eta': eta,
 219                 'speed': speed,
 220                 'elapsed': now - start,
 221             })
 222
 223             if is_test and byte_counter == data_len:
 224                 break
 225
 226         if stream is None:
 227             self.to_stderr('\n')
 228             self.report_error('Did not get any data blocks')
 229             return False
 230         if tmpfilename != '-':
 231             stream.close()
 232
 233         if data_len is not None and byte_counter != data_len:
 234             raise ContentTooShortError(byte_counter, int(data_len))
 235         self.try_rename(tmpfilename, filename)
 236
 237         # Update file modification time
 238         if self.params.get('updatetime', True):
 239             info_dict['filetime'] = self.try_utime(filename, data.info().get('last-modified', None))
 240
 241         self._hook_progress({
 242             'downloaded_bytes': byte_counter,
 243             'total_bytes': byte_counter,
 244             'filename': filename,
 245             'status': 'finished',
 246             'elapsed': time.time() - start,
 247         })
 248
 249         return True