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