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