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

   1 import os
   2 import time
   3
   4 from .common import FileDownloader
   5 from ..utils import (
   6     compat_urllib_request,
   7     compat_urllib_error,
   8     ContentTooShortError,
   9
  10     encodeFilename,
  11     sanitize_open,
  12     format_bytes,
  13 )
  14
  15
  16 class HttpFD(FileDownloader):
  17     _TEST_FILE_SIZE = 10241
  18
  19     def real_download(self, filename, info_dict):
  20         url = info_dict['url']
  21         tmpfilename = self.temp_name(filename)
  22         stream = None
  23
  24         # Do not include the Accept-Encoding header
  25         headers = {'Youtubedl-no-compression': 'True'}
  26         if 'user_agent' in info_dict:
  27             headers['Youtubedl-user-agent'] = info_dict['user_agent']
  28         if 'http_referer' in info_dict:
  29             headers['Referer'] = info_dict['http_referer']
  30         basic_request = compat_urllib_request.Request(url, None, headers)
  31         request = compat_urllib_request.Request(url, None, headers)
  32
  33         is_test = self.params.get('test', False)
  34
  35         if is_test:
  36             request.add_header('Range', 'bytes=0-%s' % str(self._TEST_FILE_SIZE - 1))
  37
  38         # Establish possible resume length
  39         if os.path.isfile(encodeFilename(tmpfilename)):
  40             resume_len = os.path.getsize(encodeFilename(tmpfilename))
  41         else:
  42             resume_len = 0
  43
  44         open_mode = 'wb'
  45         if resume_len != 0:
  46             if self.params.get('continuedl', False):
  47                 self.report_resuming_byte(resume_len)
  48                 request.add_header('Range', 'bytes=%d-' % resume_len)
  49                 open_mode = 'ab'
  50             else:
  51                 resume_len = 0
  52
  53         count = 0
  54         retries = self.params.get('retries', 0)
  55         while count <= retries:
  56             # Establish connection
  57             try:
  58                 data = self.ydl.urlopen(request)
  59                 break
  60             except (compat_urllib_error.HTTPError, ) as err:
  61                 if (err.code < 500 or err.code >= 600) and err.code != 416:
  62                     # Unexpected HTTP error
  63                     raise
  64                 elif err.code == 416:
  65                     # Unable to resume (requested range not satisfiable)
  66                     try:
  67                         # Open the connection again without the range header
  68                         data = self.ydl.urlopen(basic_request)
  69                         content_length = data.info()['Content-Length']
  70                     except (compat_urllib_error.HTTPError, ) as err:
  71                         if err.code < 500 or err.code >= 600:
  72                             raise
  73                     else:
  74                         # Examine the reported length
  75                         if (content_length is not None and
  76                                 (resume_len - 100 < int(content_length) < resume_len + 100)):
  77                             # The file had already been fully downloaded.
  78                             # Explanation to the above condition: in issue #175 it was revealed that
  79                             # YouTube sometimes adds or removes a few bytes from the end of the file,
  80                             # changing the file size slightly and causing problems for some users. So
  81                             # I decided to implement a suggested change and consider the file
  82                             # completely downloaded if the file size differs less than 100 bytes from
  83                             # the one in the hard drive.
  84                             self.report_file_already_downloaded(filename)
  85                             self.try_rename(tmpfilename, filename)
  86                             self._hook_progress({
  87                                 'filename': filename,
  88                                 'status': 'finished',
  89                             })
  90                             return True
  91                         else:
  92                             # The length does not match, we start the download over
  93                             self.report_unable_to_resume()
  94                             resume_len = 0
  95                             open_mode = 'wb'
  96                             break
  97             # Retry
  98             count += 1
  99             if count <= retries:
 100                 self.report_retry(count, retries)
 101
 102         if count > retries:
 103             self.report_error(u'giving up after %s retries' % retries)
 104             return False
 105
 106         data_len = data.info().get('Content-length', None)
 107
 108         # Range HTTP header may be ignored/unsupported by a webserver
 109         # (e.g. extractor/scivee.py, extractor/bambuser.py).
 110         # However, for a test we still would like to download just a piece of a file.
 111         # To achieve this we limit data_len to _TEST_FILE_SIZE and manually control
 112         # block size when downloading a file.
 113         if is_test and (data_len is None or int(data_len) > self._TEST_FILE_SIZE):
 114             data_len = self._TEST_FILE_SIZE
 115
 116         if data_len is not None:
 117             data_len = int(data_len) + resume_len
 118             min_data_len = self.params.get("min_filesize", None)
 119             max_data_len = self.params.get("max_filesize", None)
 120             if min_data_len is not None and data_len < min_data_len:
 121                 self.to_screen(u'\r[download] File is smaller than min-filesize (%s bytes < %s bytes). Aborting.' % (data_len, min_data_len))
 122                 return False
 123             if max_data_len is not None and data_len > max_data_len:
 124                 self.to_screen(u'\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len, max_data_len))
 125                 return False
 126
 127         data_len_str = format_bytes(data_len)
 128         byte_counter = 0 + resume_len
 129         block_size = self.params.get('buffersize', 1024)
 130         start = time.time()
 131         while True:
 132             # Download and write
 133             before = time.time()
 134             data_block = data.read(block_size if not is_test else min(block_size, data_len - byte_counter))
 135             after = time.time()
 136             if len(data_block) == 0:
 137                 break
 138             byte_counter += len(data_block)
 139
 140             # Open file just in time
 141             if stream is None:
 142                 try:
 143                     (stream, tmpfilename) = sanitize_open(tmpfilename, open_mode)
 144                     assert stream is not None
 145                     filename = self.undo_temp_name(tmpfilename)
 146                     self.report_destination(filename)
 147                 except (OSError, IOError) as err:
 148                     self.report_error(u'unable to open for writing: %s' % str(err))
 149                     return False
 150             try:
 151                 stream.write(data_block)
 152             except (IOError, OSError) as err:
 153                 self.to_stderr(u"\n")
 154                 self.report_error(u'unable to write data: %s' % str(err))
 155                 return False
 156             if not self.params.get('noresizebuffer', False):
 157                 block_size = self.best_block_size(after - before, len(data_block))
 158
 159             # Progress message
 160             speed = self.calc_speed(start, time.time(), byte_counter - resume_len)
 161             if data_len is None:
 162                 eta = percent = None
 163             else:
 164                 percent = self.calc_percent(byte_counter, data_len)
 165                 eta = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len)
 166             self.report_progress(percent, data_len_str, speed, eta)
 167
 168             self._hook_progress({
 169                 'downloaded_bytes': byte_counter,
 170                 'total_bytes': data_len,
 171                 'tmpfilename': tmpfilename,
 172                 'filename': filename,
 173                 'status': 'downloading',
 174                 'eta': eta,
 175                 'speed': speed,
 176             })
 177
 178             if is_test and byte_counter == data_len:
 179                 break
 180
 181             # Apply rate limit
 182             self.slow_down(start, byte_counter - resume_len)
 183
 184         if stream is None:
 185             self.to_stderr(u"\n")
 186             self.report_error(u'Did not get any data blocks')
 187             return False
 188         stream.close()
 189         self.report_finish(data_len_str, (time.time() - start))
 190         if data_len is not None and byte_counter != data_len:
 191             raise ContentTooShortError(byte_counter, int(data_len))
 192         self.try_rename(tmpfilename, filename)
 193
 194         # Update file modification time
 195         if self.params.get('updatetime', True):
 196             info_dict['filetime'] = self.try_utime(filename, data.info().get('last-modified', None))
 197
 198         self._hook_progress({
 199             'downloaded_bytes': byte_counter,
 200             'total_bytes': byte_counter,
 201             'filename': filename,
 202             'status': 'finished',
 203         })
 204
 205         return True