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
   8 from .common import FileDownloader
   9 from ..compat import (
  10     compat_urllib_request,
  11     compat_urllib_error,
  12 )
  13 from ..utils import (
  14     ContentTooShortError,
  15     encodeFilename,
  16     sanitize_open,
  17 )
  18
  19
  20 class HttpFD(FileDownloader):
  21     def real_download(self, filename, info_dict):
  22         url = info_dict['url']
  23         tmpfilename = self.temp_name(filename)
  24         stream = None
  25
  26         # Do not include the Accept-Encoding header
  27         headers = {'Youtubedl-no-compression': 'True'}
  28         add_headers = info_dict.get('http_headers')
  29         if add_headers:
  30             headers.update(add_headers)
  31         basic_request = compat_urllib_request.Request(url, None, headers)
  32         request = compat_urllib_request.Request(url, None, headers)
  33
  34         is_test = self.params.get('test', False)
  35
  36         if is_test:
  37             request.add_header('Range', 'bytes=0-%s' % str(self._TEST_FILE_SIZE - 1))
  38
  39         # Establish possible resume length
  40         if os.path.isfile(encodeFilename(tmpfilename)):
  41             resume_len = os.path.getsize(encodeFilename(tmpfilename))
  42         else:
  43             resume_len = 0
  44
  45         open_mode = 'wb'
  46         if resume_len != 0:
  47             if self.params.get('continuedl', True):
  48                 self.report_resuming_byte(resume_len)
  49                 request.add_header('Range', 'bytes=%d-' % resume_len)
  50                 open_mode = 'ab'
  51             else:
  52                 resume_len = 0
  53
  54         count = 0
  55         retries = self.params.get('retries', 0)
  56         while count <= retries:
  57             # Establish connection
  58             try:
  59                 data = self.ydl.urlopen(request)
  60                 break
  61             except (compat_urllib_error.HTTPError, ) as err:
  62                 if (err.code < 500 or err.code >= 600) and err.code != 416:
  63                     # Unexpected HTTP error
  64                     raise
  65                 elif err.code == 416:
  66                     # Unable to resume (requested range not satisfiable)
  67                     try:
  68                         # Open the connection again without the range header
  69                         data = self.ydl.urlopen(basic_request)
  70                         content_length = data.info()['Content-Length']
  71                     except (compat_urllib_error.HTTPError, ) as err:
  72                         if err.code < 500 or err.code >= 600:
  73                             raise
  74                     else:
  75                         # Examine the reported length
  76                         if (content_length is not None and
  77                                 (resume_len - 100 < int(content_length) < resume_len + 100)):
  78                             # The file had already been fully downloaded.
  79                             # Explanation to the above condition: in issue #175 it was revealed that
  80                             # YouTube sometimes adds or removes a few bytes from the end of the file,
  81                             # changing the file size slightly and causing problems for some users. So
  82                             # I decided to implement a suggested change and consider the file
  83                             # completely downloaded if the file size differs less than 100 bytes from
  84                             # the one in the hard drive.
  85                             self.report_file_already_downloaded(filename)
  86                             self.try_rename(tmpfilename, filename)
  87                             self._hook_progress({
  88                                 'filename': filename,
  89                                 'status': 'finished',
  90                                 'downloaded_bytes': resume_len,
  91                                 'total_bytes': resume_len,
  92                             })
  93                             return True
  94                         else:
  95                             # The length does not match, we start the download over
  96                             self.report_unable_to_resume()
  97                             resume_len = 0
  98                             open_mode = 'wb'
  99                             break
 100             except socket.error as e:
 101                 if e.errno != errno.ECONNRESET:
 102                     # Connection reset is no problem, just retry
 103                     raise
 104
 105             # Retry
 106             count += 1
 107             if count <= retries:
 108                 self.report_retry(count, retries)
 109
 110         if count > retries:
 111             self.report_error('giving up after %s retries' % retries)
 112             return False
 113
 114         data_len = data.info().get('Content-length', None)
 115
 116         # Range HTTP header may be ignored/unsupported by a webserver
 117         # (e.g. extractor/scivee.py, extractor/bambuser.py).
 118         # However, for a test we still would like to download just a piece of a file.
 119         # To achieve this we limit data_len to _TEST_FILE_SIZE and manually control
 120         # block size when downloading a file.
 121         if is_test and (data_len is None or int(data_len) > self._TEST_FILE_SIZE):
 122             data_len = self._TEST_FILE_SIZE
 123
 124         if data_len is not None:
 125             data_len = int(data_len) + resume_len
 126             min_data_len = self.params.get("min_filesize", None)
 127             max_data_len = self.params.get("max_filesize", None)
 128             if min_data_len is not None and data_len < min_data_len:
 129                 self.to_screen('\r[download] File is smaller than min-filesize (%s bytes < %s bytes). Aborting.' % (data_len, min_data_len))
 130                 return False
 131             if max_data_len is not None and data_len > max_data_len:
 132                 self.to_screen('\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len, max_data_len))
 133                 return False
 134
 135         byte_counter = 0 + resume_len
 136         block_size = self.params.get('buffersize', 1024)
 137         start = time.time()
 138
 139         # measure time over whole while-loop, so slow_down() and best_block_size() work together properly
 140         now = None  # needed for slow_down() in the first loop run
 141         before = start  # start measuring
 142         while True:
 143
 144             # Download and write
 145             data_block = data.read(block_size if not is_test else min(block_size, data_len - byte_counter))
 146             byte_counter += len(data_block)
 147
 148             # exit loop when download is finished
 149             if len(data_block) == 0:
 150                 break
 151
 152             # Open destination file just in time
 153             if stream is None:
 154                 try:
 155                     (stream, tmpfilename) = sanitize_open(tmpfilename, open_mode)
 156                     assert stream is not None
 157                     filename = self.undo_temp_name(tmpfilename)
 158                     self.report_destination(filename)
 159                 except (OSError, IOError) as err:
 160                     self.report_error('unable to open for writing: %s' % str(err))
 161                     return False
 162
 163                 if self.params.get('xattr_set_filesize', False) and data_len is not None:
 164                     try:
 165                         import xattr
 166                         xattr.setxattr(tmpfilename, 'user.ytdl.filesize', str(data_len))
 167                     except(OSError, IOError, ImportError) as err:
 168                         self.report_error('unable to set filesize xattr: %s' % str(err))
 169
 170             try:
 171                 stream.write(data_block)
 172             except (IOError, OSError) as err:
 173                 self.to_stderr('\n')
 174                 self.report_error('unable to write data: %s' % str(err))
 175                 return False
 176
 177             # Apply rate limit
 178             self.slow_down(start, now, byte_counter - resume_len)
 179
 180             # end measuring of one loop run
 181             now = time.time()
 182             after = now
 183
 184             # Adjust block size
 185             if not self.params.get('noresizebuffer', False):
 186                 block_size = self.best_block_size(after - before, len(data_block))
 187
 188             before = after
 189
 190             # Progress message
 191             speed = self.calc_speed(start, now, byte_counter - resume_len)
 192             if data_len is None:
 193                 eta = None
 194             else:
 195                 eta = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len)
 196
 197             self._hook_progress({
 198                 'status': 'downloading',
 199                 'downloaded_bytes': byte_counter,
 200                 'total_bytes': data_len,
 201                 'tmpfilename': tmpfilename,
 202                 'filename': filename,
 203                 'eta': eta,
 204                 'speed': speed,
 205                 'elapsed': now - start,
 206             })
 207
 208             if is_test and byte_counter == data_len:
 209                 break
 210
 211         if stream is None:
 212             self.to_stderr('\n')
 213             self.report_error('Did not get any data blocks')
 214             return False
 215         if tmpfilename != '-':
 216             stream.close()
 217
 218         if data_len is not None and byte_counter != data_len:
 219             raise ContentTooShortError(byte_counter, int(data_len))
 220         self.try_rename(tmpfilename, filename)
 221
 222         # Update file modification time
 223         if self.params.get('updatetime', True):
 224             info_dict['filetime'] = self.try_utime(filename, data.info().get('last-modified', None))
 225
 226         self._hook_progress({
 227             'downloaded_bytes': byte_counter,
 228             'total_bytes': byte_counter,
 229             'filename': filename,
 230             'status': 'finished',
 231             'elapsed': time.time() - start,
 232         })
 233
 234         return True