1 from __future__ import unicode_literals
8 from .common import FileDownloader
9 from .dash import DashSegmentsFD
10 from ..compat import (
11 compat_urllib_request,
21 class HttpFD(FileDownloader):
22 def real_download(self, filename, info_dict):
23 if info_dict.get('initialization_url') and list(filter(None, info_dict.get('segment_urls', []))):
24 return DashSegmentsFD(self.ydl, self.params).real_download(filename, info_dict)
26 url = info_dict['url']
27 tmpfilename = self.temp_name(filename)
30 # Do not include the Accept-Encoding header
31 headers = {'Youtubedl-no-compression': 'True'}
32 add_headers = info_dict.get('http_headers')
34 headers.update(add_headers)
35 basic_request = compat_urllib_request.Request(url, None, headers)
36 request = compat_urllib_request.Request(url, None, headers)
38 is_test = self.params.get('test', False)
41 request.add_header('Range', 'bytes=0-%s' % str(self._TEST_FILE_SIZE - 1))
43 # Establish possible resume length
44 if os.path.isfile(encodeFilename(tmpfilename)):
45 resume_len = os.path.getsize(encodeFilename(tmpfilename))
51 if self.params.get('continuedl', True):
52 self.report_resuming_byte(resume_len)
53 request.add_header('Range', 'bytes=%d-' % resume_len)
59 retries = self.params.get('retries', 0)
60 while count <= retries:
61 # Establish connection
63 data = self.ydl.urlopen(request)
65 except (compat_urllib_error.HTTPError, ) as err:
66 if (err.code < 500 or err.code >= 600) and err.code != 416:
67 # Unexpected HTTP error
70 # Unable to resume (requested range not satisfiable)
72 # Open the connection again without the range header
73 data = self.ydl.urlopen(basic_request)
74 content_length = data.info()['Content-Length']
75 except (compat_urllib_error.HTTPError, ) as err:
76 if err.code < 500 or err.code >= 600:
79 # Examine the reported length
80 if (content_length is not None and
81 (resume_len - 100 < int(content_length) < resume_len + 100)):
82 # The file had already been fully downloaded.
83 # Explanation to the above condition: in issue #175 it was revealed that
84 # YouTube sometimes adds or removes a few bytes from the end of the file,
85 # changing the file size slightly and causing problems for some users. So
86 # I decided to implement a suggested change and consider the file
87 # completely downloaded if the file size differs less than 100 bytes from
88 # the one in the hard drive.
89 self.report_file_already_downloaded(filename)
90 self.try_rename(tmpfilename, filename)
94 'downloaded_bytes': resume_len,
95 'total_bytes': resume_len,
99 # The length does not match, we start the download over
100 self.report_unable_to_resume()
104 except socket.error as e:
105 if e.errno != errno.ECONNRESET:
106 # Connection reset is no problem, just retry
112 self.report_retry(count, retries)
115 self.report_error('giving up after %s retries' % retries)
118 data_len = data.info().get('Content-length', None)
120 # Range HTTP header may be ignored/unsupported by a webserver
121 # (e.g. extractor/scivee.py, extractor/bambuser.py).
122 # However, for a test we still would like to download just a piece of a file.
123 # To achieve this we limit data_len to _TEST_FILE_SIZE and manually control
124 # block size when downloading a file.
125 if is_test and (data_len is None or int(data_len) > self._TEST_FILE_SIZE):
126 data_len = self._TEST_FILE_SIZE
128 if data_len is not None:
129 data_len = int(data_len) + resume_len
130 min_data_len = self.params.get("min_filesize", None)
131 max_data_len = self.params.get("max_filesize", None)
132 if min_data_len is not None and data_len < min_data_len:
133 self.to_screen('\r[download] File is smaller than min-filesize (%s bytes < %s bytes). Aborting.' % (data_len, min_data_len))
135 if max_data_len is not None and data_len > max_data_len:
136 self.to_screen('\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len, max_data_len))
139 byte_counter = 0 + resume_len
140 block_size = self.params.get('buffersize', 1024)
143 # measure time over whole while-loop, so slow_down() and best_block_size() work together properly
144 now = None # needed for slow_down() in the first loop run
145 before = start # start measuring
149 data_block = data.read(block_size if not is_test else min(block_size, data_len - byte_counter))
150 byte_counter += len(data_block)
152 # exit loop when download is finished
153 if len(data_block) == 0:
156 # Open destination file just in time
159 (stream, tmpfilename) = sanitize_open(tmpfilename, open_mode)
160 assert stream is not None
161 filename = self.undo_temp_name(tmpfilename)
162 self.report_destination(filename)
163 except (OSError, IOError) as err:
164 self.report_error('unable to open for writing: %s' % str(err))
167 if self.params.get('xattr_set_filesize', False) and data_len is not None:
170 xattr.setxattr(tmpfilename, 'user.ytdl.filesize', str(data_len))
171 except(OSError, IOError, ImportError) as err:
172 self.report_error('unable to set filesize xattr: %s' % str(err))
175 stream.write(data_block)
176 except (IOError, OSError) as err:
178 self.report_error('unable to write data: %s' % str(err))
182 self.slow_down(start, now, byte_counter - resume_len)
184 # end measuring of one loop run
189 if not self.params.get('noresizebuffer', False):
190 block_size = self.best_block_size(after - before, len(data_block))
195 speed = self.calc_speed(start, now, byte_counter - resume_len)
199 eta = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len)
201 self._hook_progress({
202 'status': 'downloading',
203 'downloaded_bytes': byte_counter,
204 'total_bytes': data_len,
205 'tmpfilename': tmpfilename,
206 'filename': filename,
209 'elapsed': now - start,
212 if is_test and byte_counter == data_len:
217 self.report_error('Did not get any data blocks')
219 if tmpfilename != '-':
222 if data_len is not None and byte_counter != data_len:
223 raise ContentTooShortError(byte_counter, int(data_len))
224 self.try_rename(tmpfilename, filename)
226 # Update file modification time
227 if self.params.get('updatetime', True):
228 info_dict['filetime'] = self.try_utime(filename, data.info().get('last-modified', None))
230 self._hook_progress({
231 'downloaded_bytes': byte_counter,
232 'total_bytes': byte_counter,
233 'filename': filename,
234 'status': 'finished',
235 'elapsed': time.time() - start,