1 from __future__ import unicode_literals
8 from .common import FileDownloader
10 compat_urllib_request,
20 class HttpFD(FileDownloader):
21 def real_download(self, filename, info_dict):
22 url = info_dict['url']
23 tmpfilename = self.temp_name(filename)
26 # Do not include the Accept-Encoding header
27 headers = {'Youtubedl-no-compression': 'True'}
28 add_headers = info_dict.get('http_headers')
30 headers.update(add_headers)
31 basic_request = compat_urllib_request.Request(url, None, headers)
32 request = compat_urllib_request.Request(url, None, headers)
34 is_test = self.params.get('test', False)
37 request.add_header('Range', 'bytes=0-%s' % str(self._TEST_FILE_SIZE - 1))
39 # Establish possible resume length
40 if os.path.isfile(encodeFilename(tmpfilename)):
41 resume_len = os.path.getsize(encodeFilename(tmpfilename))
47 if self.params.get('continuedl', True):
48 self.report_resuming_byte(resume_len)
49 request.add_header('Range', 'bytes=%d-' % resume_len)
55 retries = self.params.get('retries', 0)
56 while count <= retries:
57 # Establish connection
59 data = self.ydl.urlopen(request)
62 content_range = data.headers.get('Content-Range')
64 content_range_m = re.search(r'bytes (\d+)-', content_range)
66 # Content-Range is correct - go on
67 if resume_len == int(content_range_m.group(1)):
70 # Content-Range is invalid - wipe the file and do entire redownload
75 except (compat_urllib_error.HTTPError, ) as err:
76 if (err.code < 500 or err.code >= 600) and err.code != 416:
77 # Unexpected HTTP error
80 # Unable to resume (requested range not satisfiable)
82 # Open the connection again without the range header
83 data = self.ydl.urlopen(basic_request)
84 content_length = data.info()['Content-Length']
85 except (compat_urllib_error.HTTPError, ) as err:
86 if err.code < 500 or err.code >= 600:
89 # Examine the reported length
90 if (content_length is not None and
91 (resume_len - 100 < int(content_length) < resume_len + 100)):
92 # The file had already been fully downloaded.
93 # Explanation to the above condition: in issue #175 it was revealed that
94 # YouTube sometimes adds or removes a few bytes from the end of the file,
95 # changing the file size slightly and causing problems for some users. So
96 # I decided to implement a suggested change and consider the file
97 # completely downloaded if the file size differs less than 100 bytes from
98 # the one in the hard drive.
99 self.report_file_already_downloaded(filename)
100 self.try_rename(tmpfilename, filename)
101 self._hook_progress({
102 'filename': filename,
103 'status': 'finished',
104 'downloaded_bytes': resume_len,
105 'total_bytes': resume_len,
109 # The length does not match, we start the download over
110 self.report_unable_to_resume()
114 except socket.error as e:
115 if e.errno != errno.ECONNRESET:
116 # Connection reset is no problem, just retry
122 self.report_retry(count, retries)
125 self.report_error('giving up after %s retries' % retries)
128 data_len = data.info().get('Content-length', None)
130 # Range HTTP header may be ignored/unsupported by a webserver
131 # (e.g. extractor/scivee.py, extractor/bambuser.py).
132 # However, for a test we still would like to download just a piece of a file.
133 # To achieve this we limit data_len to _TEST_FILE_SIZE and manually control
134 # block size when downloading a file.
135 if is_test and (data_len is None or int(data_len) > self._TEST_FILE_SIZE):
136 data_len = self._TEST_FILE_SIZE
138 if data_len is not None:
139 data_len = int(data_len) + resume_len
140 min_data_len = self.params.get("min_filesize", None)
141 max_data_len = self.params.get("max_filesize", None)
142 if min_data_len is not None and data_len < min_data_len:
143 self.to_screen('\r[download] File is smaller than min-filesize (%s bytes < %s bytes). Aborting.' % (data_len, min_data_len))
145 if max_data_len is not None and data_len > max_data_len:
146 self.to_screen('\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len, max_data_len))
149 byte_counter = 0 + resume_len
150 block_size = self.params.get('buffersize', 1024)
153 # measure time over whole while-loop, so slow_down() and best_block_size() work together properly
154 now = None # needed for slow_down() in the first loop run
155 before = start # start measuring
159 data_block = data.read(block_size if not is_test else min(block_size, data_len - byte_counter))
160 byte_counter += len(data_block)
162 # exit loop when download is finished
163 if len(data_block) == 0:
166 # Open destination file just in time
169 (stream, tmpfilename) = sanitize_open(tmpfilename, open_mode)
170 assert stream is not None
171 filename = self.undo_temp_name(tmpfilename)
172 self.report_destination(filename)
173 except (OSError, IOError) as err:
174 self.report_error('unable to open for writing: %s' % str(err))
177 if self.params.get('xattr_set_filesize', False) and data_len is not None:
180 xattr.setxattr(tmpfilename, 'user.ytdl.filesize', str(data_len))
181 except(OSError, IOError, ImportError) as err:
182 self.report_error('unable to set filesize xattr: %s' % str(err))
185 stream.write(data_block)
186 except (IOError, OSError) as err:
188 self.report_error('unable to write data: %s' % str(err))
192 self.slow_down(start, now, byte_counter - resume_len)
194 # end measuring of one loop run
199 if not self.params.get('noresizebuffer', False):
200 block_size = self.best_block_size(after - before, len(data_block))
205 speed = self.calc_speed(start, now, byte_counter - resume_len)
209 eta = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len)
211 self._hook_progress({
212 'status': 'downloading',
213 'downloaded_bytes': byte_counter,
214 'total_bytes': data_len,
215 'tmpfilename': tmpfilename,
216 'filename': filename,
219 'elapsed': now - start,
222 if is_test and byte_counter == data_len:
227 self.report_error('Did not get any data blocks')
229 if tmpfilename != '-':
232 if data_len is not None and byte_counter != data_len:
233 raise ContentTooShortError(byte_counter, int(data_len))
234 self.try_rename(tmpfilename, filename)
236 # Update file modification time
237 if self.params.get('updatetime', True):
238 info_dict['filetime'] = self.try_utime(filename, data.info().get('last-modified', None))
240 self._hook_progress({
241 'downloaded_bytes': byte_counter,
242 'total_bytes': byte_counter,
243 'filename': filename,
244 'status': 'finished',
245 'elapsed': time.time() - start,