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

   1 from __future__ import unicode_literals
   2
   3 import os
   4 import time
   5
   6 from .common import FileDownloader
   7 from ..utils import (
   8     compat_urllib_request,
   9     compat_urllib_error,
  10     ContentTooShortError,
  11
  12     encodeFilename,
  13     sanitize_open,
  14     format_bytes,
  15 )
  16
  17
  18 class HttpFD(FileDownloader):
  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         add_headers = info_dict.get('http_headers')
  31         if add_headers:
  32             headers.update(add_headers)
  33         data = info_dict.get('http_post_data')
  34         http_method = info_dict.get('http_method')
  35         basic_request = compat_urllib_request.Request(url, data, headers)
  36         request = compat_urllib_request.Request(url, data, headers)
  37         if http_method is not None:
  38             basic_request.get_method = lambda: http_method
  39             request.get_method = lambda: http_method
  40
  41         is_test = self.params.get('test', False)
  42
  43         if is_test:
  44             request.add_header('Range', 'bytes=0-%s' % str(self._TEST_FILE_SIZE - 1))
  45
  46         # Establish possible resume length
  47         if os.path.isfile(encodeFilename(tmpfilename)):
  48             resume_len = os.path.getsize(encodeFilename(tmpfilename))
  49         else:
  50             resume_len = 0
  51
  52         open_mode = 'wb'
  53         if resume_len != 0:
  54             if self.params.get('continuedl', False):
  55                 self.report_resuming_byte(resume_len)
  56                 request.add_header('Range', 'bytes=%d-' % resume_len)
  57                 open_mode = 'ab'
  58             else:
  59                 resume_len = 0
  60
  61         count = 0
  62         retries = self.params.get('retries', 0)
  63         while count <= retries:
  64             # Establish connection
  65             try:
  66                 data = self.ydl.urlopen(request)
  67                 break
  68             except (compat_urllib_error.HTTPError, ) as err:
  69                 if (err.code < 500 or err.code >= 600) and err.code != 416:
  70                     # Unexpected HTTP error
  71                     raise
  72                 elif err.code == 416:
  73                     # Unable to resume (requested range not satisfiable)
  74                     try:
  75                         # Open the connection again without the range header
  76                         data = self.ydl.urlopen(basic_request)
  77                         content_length = data.info()['Content-Length']
  78                     except (compat_urllib_error.HTTPError, ) as err:
  79                         if err.code < 500 or err.code >= 600:
  80                             raise
  81                     else:
  82                         # Examine the reported length
  83                         if (content_length is not None and
  84                                 (resume_len - 100 < int(content_length) < resume_len + 100)):
  85                             # The file had already been fully downloaded.
  86                             # Explanation to the above condition: in issue #175 it was revealed that
  87                             # YouTube sometimes adds or removes a few bytes from the end of the file,
  88                             # changing the file size slightly and causing problems for some users. So
  89                             # I decided to implement a suggested change and consider the file
  90                             # completely downloaded if the file size differs less than 100 bytes from
  91                             # the one in the hard drive.
  92                             self.report_file_already_downloaded(filename)
  93                             self.try_rename(tmpfilename, filename)
  94                             self._hook_progress({
  95                                 'filename': filename,
  96                                 'status': 'finished',
  97                             })
  98                             return True
  99                         else:
 100                             # The length does not match, we start the download over
 101                             self.report_unable_to_resume()
 102                             resume_len = 0
 103                             open_mode = 'wb'
 104                             break
 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         data_len_str = format_bytes(data_len)
 136         byte_counter = 0 + resume_len
 137         block_size = self.params.get('buffersize', 1024)
 138         start = time.time()
 139         while True:
 140             # Download and write
 141             before = time.time()
 142             data_block = data.read(block_size if not is_test else min(block_size, data_len - byte_counter))
 143             after = time.time()
 144             if len(data_block) == 0:
 145                 break
 146             byte_counter += len(data_block)
 147
 148             # Open file just in time
 149             if stream is None:
 150                 try:
 151                     (stream, tmpfilename) = sanitize_open(tmpfilename, open_mode)
 152                     assert stream is not None
 153                     filename = self.undo_temp_name(tmpfilename)
 154                     self.report_destination(filename)
 155                 except (OSError, IOError) as err:
 156                     self.report_error('unable to open for writing: %s' % str(err))
 157                     return False
 158             try:
 159                 stream.write(data_block)
 160             except (IOError, OSError) as err:
 161                 self.to_stderr('\n')
 162                 self.report_error('unable to write data: %s' % str(err))
 163                 return False
 164             if not self.params.get('noresizebuffer', False):
 165                 block_size = self.best_block_size(after - before, len(data_block))
 166
 167             # Progress message
 168             speed = self.calc_speed(start, time.time(), byte_counter - resume_len)
 169             if data_len is None:
 170                 eta = percent = None
 171             else:
 172                 percent = self.calc_percent(byte_counter, data_len)
 173                 eta = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len)
 174             self.report_progress(percent, data_len_str, speed, eta)
 175
 176             self._hook_progress({
 177                 'downloaded_bytes': byte_counter,
 178                 'total_bytes': data_len,
 179                 'tmpfilename': tmpfilename,
 180                 'filename': filename,
 181                 'status': 'downloading',
 182                 'eta': eta,
 183                 'speed': speed,
 184             })
 185
 186             if is_test and byte_counter == data_len:
 187                 break
 188
 189             # Apply rate limit
 190             self.slow_down(start, byte_counter - resume_len)
 191
 192         if stream is None:
 193             self.to_stderr('\n')
 194             self.report_error('Did not get any data blocks')
 195             return False
 196         if tmpfilename != '-':
 197             stream.close()
 198         self.report_finish(data_len_str, (time.time() - start))
 199         if data_len is not None and byte_counter != data_len:
 200             raise ContentTooShortError(byte_counter, int(data_len))
 201         self.try_rename(tmpfilename, filename)
 202
 203         # Update file modification time
 204         if self.params.get('updatetime', True):
 205             info_dict['filetime'] = self.try_utime(filename, data.info().get('last-modified', None))
 206
 207         self._hook_progress({
 208             'downloaded_bytes': byte_counter,
 209             'total_bytes': byte_counter,
 210             'filename': filename,
 211             'status': 'finished',
 212         })
 213
 214         return True