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