4 from .common import FileDownloader
16 class HttpFD(FileDownloader):
17 _TEST_FILE_SIZE = 10241
19 def real_download(self, filename, info_dict):
20 url = info_dict['url']
21 tmpfilename = self.temp_name(filename)
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 basic_request = compat_urllib_request.Request(url, None, headers)
31 request = compat_urllib_request.Request(url, None, headers)
33 is_test = self.params.get('test', False)
36 request.add_header('Range', 'bytes=0-%s' % str(self._TEST_FILE_SIZE - 1))
38 # Establish possible resume length
39 if os.path.isfile(encodeFilename(tmpfilename)):
40 resume_len = os.path.getsize(encodeFilename(tmpfilename))
46 if self.params.get('continuedl', False):
47 self.report_resuming_byte(resume_len)
48 request.add_header('Range', 'bytes=%d-' % resume_len)
54 retries = self.params.get('retries', 0)
55 while count <= retries:
56 # Establish connection
58 data = self.ydl.urlopen(request)
60 except (compat_urllib_error.HTTPError, ) as err:
61 if (err.code < 500 or err.code >= 600) and err.code != 416:
62 # Unexpected HTTP error
65 # Unable to resume (requested range not satisfiable)
67 # Open the connection again without the range header
68 data = self.ydl.urlopen(basic_request)
69 content_length = data.info()['Content-Length']
70 except (compat_urllib_error.HTTPError, ) as err:
71 if err.code < 500 or err.code >= 600:
74 # Examine the reported length
75 if (content_length is not None and
76 (resume_len - 100 < int(content_length) < resume_len + 100)):
77 # The file had already been fully downloaded.
78 # Explanation to the above condition: in issue #175 it was revealed that
79 # YouTube sometimes adds or removes a few bytes from the end of the file,
80 # changing the file size slightly and causing problems for some users. So
81 # I decided to implement a suggested change and consider the file
82 # completely downloaded if the file size differs less than 100 bytes from
83 # the one in the hard drive.
84 self.report_file_already_downloaded(filename)
85 self.try_rename(tmpfilename, filename)
92 # The length does not match, we start the download over
93 self.report_unable_to_resume()
100 self.report_retry(count, retries)
103 self.report_error(u'giving up after %s retries' % retries)
106 data_len = data.info().get('Content-length', None)
108 # Range HTTP header may be ignored/unsupported by a webserver
109 # (e.g. extractor/scivee.py, extractor/bambuser.py).
110 # However, for a test we still would like to download just a piece of a file.
111 # To achieve this we limit data_len to _TEST_FILE_SIZE and manually control
112 # block size when downloading a file.
113 if is_test and (data_len is None or int(data_len) > self._TEST_FILE_SIZE):
114 data_len = self._TEST_FILE_SIZE
116 if data_len is not None:
117 data_len = int(data_len) + resume_len
118 min_data_len = self.params.get("min_filesize", None)
119 max_data_len = self.params.get("max_filesize", None)
120 if min_data_len is not None and data_len < min_data_len:
121 self.to_screen(u'\r[download] File is smaller than min-filesize (%s bytes < %s bytes). Aborting.' % (data_len, min_data_len))
123 if max_data_len is not None and data_len > max_data_len:
124 self.to_screen(u'\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len, max_data_len))
127 data_len_str = format_bytes(data_len)
128 byte_counter = 0 + resume_len
129 block_size = self.params.get('buffersize', 1024)
132 # measure time over whole while-loop, so slow_down() and best_block_size() work together properly
133 now = None # needed for slow_down() in the first loop run
134 before = start # start measuring
138 data_block = data.read(block_size if not is_test else min(block_size, data_len - byte_counter))
139 byte_counter += len(data_block)
141 # exit loop when download is finished
142 if len(data_block) == 0:
145 # Open destination file just in time
148 (stream, tmpfilename) = sanitize_open(tmpfilename, open_mode)
149 assert stream is not None
150 filename = self.undo_temp_name(tmpfilename)
151 self.report_destination(filename)
152 except (OSError, IOError) as err:
153 self.report_error(u'unable to open for writing: %s' % str(err))
156 stream.write(data_block)
157 except (IOError, OSError) as err:
158 self.to_stderr(u"\n")
159 self.report_error(u'unable to write data: %s' % str(err))
163 self.slow_down(start, now, byte_counter - resume_len)
165 # end measuring of one loop run
170 if not self.params.get('noresizebuffer', False):
171 block_size = self.best_block_size(after - before, len(data_block))
176 speed = self.calc_speed(start, now, byte_counter - resume_len)
180 percent = self.calc_percent(byte_counter, data_len)
181 eta = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len)
182 self.report_progress(percent, data_len_str, speed, eta)
184 self._hook_progress({
185 'downloaded_bytes': byte_counter,
186 'total_bytes': data_len,
187 'tmpfilename': tmpfilename,
188 'filename': filename,
189 'status': 'downloading',
194 if is_test and byte_counter == data_len:
198 self.to_stderr(u"\n")
199 self.report_error(u'Did not get any data blocks')
202 self.report_finish(data_len_str, (time.time() - start))
203 if data_len is not None and byte_counter != data_len:
204 raise ContentTooShortError(byte_counter, int(data_len))
205 self.try_rename(tmpfilename, filename)
207 # Update file modification time
208 if self.params.get('updatetime', True):
209 info_dict['filetime'] = self.try_utime(filename, data.info().get('last-modified', None))
211 self._hook_progress({
212 'downloaded_bytes': byte_counter,
213 'total_bytes': byte_counter,
214 'filename': filename,
215 'status': 'finished',