]> Raphaƫl G. Git Repositories - youtubedl/blob - youtube_dl/downloader/http.py
f8b69d186ac5ee93c8402f85bc66e7ed59570118
[youtubedl] / youtube_dl / downloader / http.py
1 from __future__ import unicode_literals
2
3 import errno
4 import os
5 import socket
6 import time
7 import re
8
9 from .common import FileDownloader
10 from ..compat import compat_urllib_error
11 from ..utils import (
12 ContentTooShortError,
13 encodeFilename,
14 sanitize_open,
15 sanitized_Request,
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 add_headers = info_dict.get('http_headers')
28 if add_headers:
29 headers.update(add_headers)
30 basic_request = sanitized_Request(url, None, headers)
31 request = sanitized_Request(url, None, headers)
32
33 is_test = self.params.get('test', False)
34
35 if is_test:
36 request.add_header('Range', 'bytes=0-%s' % str(self._TEST_FILE_SIZE - 1))
37
38 # Establish possible resume length
39 if os.path.isfile(encodeFilename(tmpfilename)):
40 resume_len = os.path.getsize(encodeFilename(tmpfilename))
41 else:
42 resume_len = 0
43
44 open_mode = 'wb'
45 if resume_len != 0:
46 if self.params.get('continuedl', True):
47 self.report_resuming_byte(resume_len)
48 request.add_header('Range', 'bytes=%d-' % resume_len)
49 open_mode = 'ab'
50 else:
51 resume_len = 0
52
53 count = 0
54 retries = self.params.get('retries', 0)
55 while count <= retries:
56 # Establish connection
57 try:
58 data = self.ydl.urlopen(request)
59 # When trying to resume, Content-Range HTTP header of response has to be checked
60 # to match the value of requested Range HTTP header. This is due to a webservers
61 # that don't support resuming and serve a whole file with no Content-Range
62 # set in response despite of requested Range (see
63 # https://github.com/rg3/youtube-dl/issues/6057#issuecomment-126129799)
64 if resume_len > 0:
65 content_range = data.headers.get('Content-Range')
66 if content_range:
67 content_range_m = re.search(r'bytes (\d+)-', content_range)
68 # Content-Range is present and matches requested Range, resume is possible
69 if content_range_m and resume_len == int(content_range_m.group(1)):
70 break
71 # Content-Range is either not present or invalid. Assuming remote webserver is
72 # trying to send the whole file, resume is not possible, so wiping the local file
73 # and performing entire redownload
74 self.report_unable_to_resume()
75 resume_len = 0
76 open_mode = 'wb'
77 break
78 except (compat_urllib_error.HTTPError, ) as err:
79 if (err.code < 500 or err.code >= 600) and err.code != 416:
80 # Unexpected HTTP error
81 raise
82 elif err.code == 416:
83 # Unable to resume (requested range not satisfiable)
84 try:
85 # Open the connection again without the range header
86 data = self.ydl.urlopen(basic_request)
87 content_length = data.info()['Content-Length']
88 except (compat_urllib_error.HTTPError, ) as err:
89 if err.code < 500 or err.code >= 600:
90 raise
91 else:
92 # Examine the reported length
93 if (content_length is not None and
94 (resume_len - 100 < int(content_length) < resume_len + 100)):
95 # The file had already been fully downloaded.
96 # Explanation to the above condition: in issue #175 it was revealed that
97 # YouTube sometimes adds or removes a few bytes from the end of the file,
98 # changing the file size slightly and causing problems for some users. So
99 # I decided to implement a suggested change and consider the file
100 # completely downloaded if the file size differs less than 100 bytes from
101 # the one in the hard drive.
102 self.report_file_already_downloaded(filename)
103 self.try_rename(tmpfilename, filename)
104 self._hook_progress({
105 'filename': filename,
106 'status': 'finished',
107 'downloaded_bytes': resume_len,
108 'total_bytes': resume_len,
109 })
110 return True
111 else:
112 # The length does not match, we start the download over
113 self.report_unable_to_resume()
114 resume_len = 0
115 open_mode = 'wb'
116 break
117 except socket.error as e:
118 if e.errno != errno.ECONNRESET:
119 # Connection reset is no problem, just retry
120 raise
121
122 # Retry
123 count += 1
124 if count <= retries:
125 self.report_retry(count, retries)
126
127 if count > retries:
128 self.report_error('giving up after %s retries' % retries)
129 return False
130
131 data_len = data.info().get('Content-length', None)
132
133 # Range HTTP header may be ignored/unsupported by a webserver
134 # (e.g. extractor/scivee.py, extractor/bambuser.py).
135 # However, for a test we still would like to download just a piece of a file.
136 # To achieve this we limit data_len to _TEST_FILE_SIZE and manually control
137 # block size when downloading a file.
138 if is_test and (data_len is None or int(data_len) > self._TEST_FILE_SIZE):
139 data_len = self._TEST_FILE_SIZE
140
141 if data_len is not None:
142 data_len = int(data_len) + resume_len
143 min_data_len = self.params.get('min_filesize')
144 max_data_len = self.params.get('max_filesize')
145 if min_data_len is not None and data_len < min_data_len:
146 self.to_screen('\r[download] File is smaller than min-filesize (%s bytes < %s bytes). Aborting.' % (data_len, min_data_len))
147 return False
148 if max_data_len is not None and data_len > max_data_len:
149 self.to_screen('\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len, max_data_len))
150 return False
151
152 byte_counter = 0 + resume_len
153 block_size = self.params.get('buffersize', 1024)
154 start = time.time()
155
156 # measure time over whole while-loop, so slow_down() and best_block_size() work together properly
157 now = None # needed for slow_down() in the first loop run
158 before = start # start measuring
159 while True:
160
161 # Download and write
162 data_block = data.read(block_size if not is_test else min(block_size, data_len - byte_counter))
163 byte_counter += len(data_block)
164
165 # exit loop when download is finished
166 if len(data_block) == 0:
167 break
168
169 # Open destination file just in time
170 if stream is None:
171 try:
172 (stream, tmpfilename) = sanitize_open(tmpfilename, open_mode)
173 assert stream is not None
174 filename = self.undo_temp_name(tmpfilename)
175 self.report_destination(filename)
176 except (OSError, IOError) as err:
177 self.report_error('unable to open for writing: %s' % str(err))
178 return False
179
180 if self.params.get('xattr_set_filesize', False) and data_len is not None:
181 try:
182 import xattr
183 xattr.setxattr(tmpfilename, 'user.ytdl.filesize', str(data_len))
184 except(OSError, IOError, ImportError) as err:
185 self.report_error('unable to set filesize xattr: %s' % str(err))
186
187 try:
188 stream.write(data_block)
189 except (IOError, OSError) as err:
190 self.to_stderr('\n')
191 self.report_error('unable to write data: %s' % str(err))
192 return False
193
194 # Apply rate limit
195 self.slow_down(start, now, byte_counter - resume_len)
196
197 # end measuring of one loop run
198 now = time.time()
199 after = now
200
201 # Adjust block size
202 if not self.params.get('noresizebuffer', False):
203 block_size = self.best_block_size(after - before, len(data_block))
204
205 before = after
206
207 # Progress message
208 speed = self.calc_speed(start, now, byte_counter - resume_len)
209 if data_len is None:
210 eta = None
211 else:
212 eta = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len)
213
214 self._hook_progress({
215 'status': 'downloading',
216 'downloaded_bytes': byte_counter,
217 'total_bytes': data_len,
218 'tmpfilename': tmpfilename,
219 'filename': filename,
220 'eta': eta,
221 'speed': speed,
222 'elapsed': now - start,
223 })
224
225 if is_test and byte_counter == data_len:
226 break
227
228 if stream is None:
229 self.to_stderr('\n')
230 self.report_error('Did not get any data blocks')
231 return False
232 if tmpfilename != '-':
233 stream.close()
234
235 if data_len is not None and byte_counter != data_len:
236 raise ContentTooShortError(byte_counter, int(data_len))
237 self.try_rename(tmpfilename, filename)
238
239 # Update file modification time
240 if self.params.get('updatetime', True):
241 info_dict['filetime'] = self.try_utime(filename, data.info().get('last-modified', None))
242
243 self._hook_progress({
244 'downloaded_bytes': byte_counter,
245 'total_bytes': byte_counter,
246 'filename': filename,
247 'status': 'finished',
248 'elapsed': time.time() - start,
249 })
250
251 return True