9 from .common
import FileDownloader
15 XAttrUnavailableError
,
24 RESPONSE_READ_EXCEPTIONS
= (
26 socket
.timeout
, # compat: py < 3.10
29 http
.client
.HTTPException
33 class HttpFD(FileDownloader
):
34 def real_download(self
, filename
, info_dict
):
35 url
= info_dict
['url']
36 request_data
= info_dict
.get('request_data', None)
38 class DownloadContext(dict):
39 __getattr__
= dict.get
40 __setattr__
= dict.__setitem
__
41 __delattr__
= dict.__delitem
__
43 ctx
= DownloadContext()
44 ctx
.filename
= filename
45 ctx
.tmpfilename
= self
.temp_name(filename
)
48 # Do not include the Accept-Encoding header
49 headers
= {'Youtubedl-no-compression': 'True'}
50 add_headers
= info_dict
.get('http_headers')
52 headers
.update(add_headers
)
54 is_test
= self
.params
.get('test', False)
55 chunk_size
= self
._TEST
_FILE
_SIZE
if is_test
else (
56 self
.params
.get('http_chunk_size')
57 or info_dict
.get('downloader_options', {}).get('http_chunk_size')
62 ctx
.block_size
= self
.params
.get('buffersize', 1024)
63 ctx
.start_time
= time
.time()
66 req_start
, req_end
, _
= parse_http_range(headers
.get('Range'))
68 if self
.params
.get('continuedl', True):
69 # Establish possible resume length
70 if os
.path
.isfile(encodeFilename(ctx
.tmpfilename
)):
71 ctx
.resume_len
= os
.path
.getsize(
72 encodeFilename(ctx
.tmpfilename
))
74 ctx
.is_resume
= ctx
.resume_len
> 0
76 class SucceedDownload(Exception):
79 class RetryDownload(Exception):
80 def __init__(self
, source_error
):
81 self
.source_error
= source_error
83 class NextFragment(Exception):
86 def establish_connection():
87 ctx
.chunk_size
= (random
.randint(int(chunk_size
* 0.95), chunk_size
)
88 if not is_test
and chunk_size
else chunk_size
)
89 if ctx
.resume_len
> 0:
90 range_start
= ctx
.resume_len
91 if req_start
is not None:
92 # offset the beginning of Range to be within request
93 range_start
+= req_start
95 self
.report_resuming_byte(ctx
.resume_len
)
97 elif req_start
is not None:
98 range_start
= req_start
99 elif ctx
.chunk_size
> 0:
103 ctx
.is_resume
= False
106 chunk_aware_end
= range_start
+ ctx
.chunk_size
- 1
107 # we're not allowed to download outside Range
108 range_end
= chunk_aware_end
if req_end
is None else min(chunk_aware_end
, req_end
)
109 elif req_end
is not None:
110 # there's no need for chunked downloads, so download until the end of Range
115 if try_call(lambda: range_start
> range_end
):
118 raise RetryDownload(Exception(f
'Conflicting range. (start={range_start} > end={range_end})'))
120 if try_call(lambda: range_end
>= ctx
.content_len
):
121 range_end
= ctx
.content_len
- 1
123 request
= sanitized_Request(url
, request_data
, headers
)
124 has_range
= range_start
is not None
126 request
.add_header('Range', f
'bytes={int(range_start)}-{int_or_none(range_end) or ""}')
127 # Establish connection
129 ctx
.data
= self
.ydl
.urlopen(request
)
130 # When trying to resume, Content-Range HTTP header of response has to be checked
131 # to match the value of requested Range HTTP header. This is due to a webservers
132 # that don't support resuming and serve a whole file with no Content-Range
133 # set in response despite of requested Range (see
134 # https://github.com/ytdl-org/youtube-dl/issues/6057#issuecomment-126129799)
136 content_range
= ctx
.data
.headers
.get('Content-Range')
137 content_range_start
, content_range_end
, content_len
= parse_http_range(content_range
)
138 # Content-Range is present and matches requested Range, resume is possible
139 if range_start
== content_range_start
and (
140 # Non-chunked download
142 # Chunked download and requested piece or
143 # its part is promised to be served
144 or content_range_end
== range_end
145 or content_len
< range_end
):
146 ctx
.content_len
= content_len
147 if content_len
or req_end
:
148 ctx
.data_len
= min(content_len
or req_end
, req_end
or content_len
) - (req_start
or 0)
150 # Content-Range is either not present or invalid. Assuming remote webserver is
151 # trying to send the whole file, resume is not possible, so wiping the local file
152 # and performing entire redownload
153 self
.report_unable_to_resume()
156 ctx
.data_len
= ctx
.content_len
= int_or_none(ctx
.data
.info().get('Content-length', None))
157 except urllib
.error
.HTTPError
as err
:
159 # Unable to resume (requested range not satisfiable)
161 # Open the connection again without the range header
162 ctx
.data
= self
.ydl
.urlopen(
163 sanitized_Request(url
, request_data
, headers
))
164 content_length
= ctx
.data
.info()['Content-Length']
165 except urllib
.error
.HTTPError
as err
:
166 if err
.code
< 500 or err
.code
>= 600:
169 # Examine the reported length
170 if (content_length
is not None
171 and (ctx
.resume_len
- 100 < int(content_length
) < ctx
.resume_len
+ 100)):
172 # The file had already been fully downloaded.
173 # Explanation to the above condition: in issue #175 it was revealed that
174 # YouTube sometimes adds or removes a few bytes from the end of the file,
175 # changing the file size slightly and causing problems for some users. So
176 # I decided to implement a suggested change and consider the file
177 # completely downloaded if the file size differs less than 100 bytes from
178 # the one in the hard drive.
179 self
.report_file_already_downloaded(ctx
.filename
)
180 self
.try_rename(ctx
.tmpfilename
, ctx
.filename
)
181 self
._hook
_progress
({
182 'filename': ctx
.filename
,
183 'status': 'finished',
184 'downloaded_bytes': ctx
.resume_len
,
185 'total_bytes': ctx
.resume_len
,
187 raise SucceedDownload()
189 # The length does not match, we start the download over
190 self
.report_unable_to_resume()
194 elif err
.code
< 500 or err
.code
>= 600:
195 # Unexpected HTTP error
197 raise RetryDownload(err
)
198 except urllib
.error
.URLError
as err
:
199 if isinstance(err
.reason
, ssl
.CertificateError
):
201 raise RetryDownload(err
)
202 # In urllib.request.AbstractHTTPHandler, the response is partially read on request.
203 # Any errors that occur during this will not be wrapped by URLError
204 except RESPONSE_READ_EXCEPTIONS
as err
:
205 raise RetryDownload(err
)
208 if ctx
.stream
is not None:
209 if not ctx
.tmpfilename
== '-':
214 data_len
= ctx
.data
.info().get('Content-length', None)
216 # Range HTTP header may be ignored/unsupported by a webserver
217 # (e.g. extractor/scivee.py, extractor/bambuser.py).
218 # However, for a test we still would like to download just a piece of a file.
219 # To achieve this we limit data_len to _TEST_FILE_SIZE and manually control
220 # block size when downloading a file.
221 if is_test
and (data_len
is None or int(data_len
) > self
._TEST
_FILE
_SIZE
):
222 data_len
= self
._TEST
_FILE
_SIZE
224 if data_len
is not None:
225 data_len
= int(data_len
) + ctx
.resume_len
226 min_data_len
= self
.params
.get('min_filesize')
227 max_data_len
= self
.params
.get('max_filesize')
228 if min_data_len
is not None and data_len
< min_data_len
:
230 f
'\r[download] File is smaller than min-filesize ({data_len} bytes < {min_data_len} bytes). Aborting.')
232 if max_data_len
is not None and data_len
> max_data_len
:
234 f
'\r[download] File is larger than max-filesize ({data_len} bytes > {max_data_len} bytes). Aborting.')
237 byte_counter
= 0 + ctx
.resume_len
238 block_size
= ctx
.block_size
241 # measure time over whole while-loop, so slow_down() and best_block_size() work together properly
242 now
= None # needed for slow_down() in the first loop run
243 before
= start
# start measuring
247 ctx
.resume_len
= (byte_counter
if ctx
.tmpfilename
== '-'
248 else os
.path
.getsize(encodeFilename(ctx
.tmpfilename
)))
249 raise RetryDownload(e
)
254 data_block
= ctx
.data
.read(block_size
if not is_test
else min(block_size
, data_len
- byte_counter
))
255 except RESPONSE_READ_EXCEPTIONS
as err
:
258 byte_counter
+= len(data_block
)
260 # exit loop when download is finished
261 if len(data_block
) == 0:
264 # Open destination file just in time
265 if ctx
.stream
is None:
267 ctx
.stream
, ctx
.tmpfilename
= self
.sanitize_open(
268 ctx
.tmpfilename
, ctx
.open_mode
)
269 assert ctx
.stream
is not None
270 ctx
.filename
= self
.undo_temp_name(ctx
.tmpfilename
)
271 self
.report_destination(ctx
.filename
)
272 except OSError as err
:
273 self
.report_error('unable to open for writing: %s' % str(err
))
276 if self
.params
.get('xattr_set_filesize', False) and data_len
is not None:
278 write_xattr(ctx
.tmpfilename
, 'user.ytdl.filesize', str(data_len
).encode())
279 except (XAttrUnavailableError
, XAttrMetadataError
) as err
:
280 self
.report_error('unable to set filesize xattr: %s' % str(err
))
283 ctx
.stream
.write(data_block
)
284 except OSError as err
:
286 self
.report_error('unable to write data: %s' % str(err
))
290 self
.slow_down(start
, now
, byte_counter
- ctx
.resume_len
)
292 # end measuring of one loop run
297 if not self
.params
.get('noresizebuffer', False):
298 block_size
= self
.best_block_size(after
- before
, len(data_block
))
303 speed
= self
.calc_speed(start
, now
, byte_counter
- ctx
.resume_len
)
304 if ctx
.data_len
is None:
307 eta
= self
.calc_eta(start
, time
.time(), ctx
.data_len
- ctx
.resume_len
, byte_counter
- ctx
.resume_len
)
309 self
._hook
_progress
({
310 'status': 'downloading',
311 'downloaded_bytes': byte_counter
,
312 'total_bytes': ctx
.data_len
,
313 'tmpfilename': ctx
.tmpfilename
,
314 'filename': ctx
.filename
,
317 'elapsed': now
- ctx
.start_time
,
318 'ctx_id': info_dict
.get('ctx_id'),
321 if data_len
is not None and byte_counter
== data_len
:
324 if speed
and speed
< (self
.params
.get('throttledratelimit') or 0):
325 # The speed must stay below the limit for 3 seconds
326 # This prevents raising error when the speed temporarily goes down
327 if ctx
.throttle_start
is None:
328 ctx
.throttle_start
= now
329 elif now
- ctx
.throttle_start
> 3:
330 if ctx
.stream
is not None and ctx
.tmpfilename
!= '-':
332 raise ThrottledDownload()
334 ctx
.throttle_start
= None
336 if not is_test
and ctx
.chunk_size
and ctx
.content_len
is not None and byte_counter
< ctx
.content_len
:
337 ctx
.resume_len
= byte_counter
338 # ctx.block_size = block_size
341 if ctx
.stream
is None:
343 self
.report_error('Did not get any data blocks')
345 if ctx
.tmpfilename
!= '-':
348 if data_len
is not None and byte_counter
!= data_len
:
349 err
= ContentTooShortError(byte_counter
, int(data_len
))
352 self
.try_rename(ctx
.tmpfilename
, ctx
.filename
)
354 # Update file modification time
355 if self
.params
.get('updatetime', True):
356 info_dict
['filetime'] = self
.try_utime(ctx
.filename
, ctx
.data
.info().get('last-modified', None))
358 self
._hook
_progress
({
359 'downloaded_bytes': byte_counter
,
360 'total_bytes': byte_counter
,
361 'filename': ctx
.filename
,
362 'status': 'finished',
363 'elapsed': time
.time() - ctx
.start_time
,
364 'ctx_id': info_dict
.get('ctx_id'),
369 for retry
in RetryManager(self
.params
.get('retries'), self
.report_retry
):
371 establish_connection()
373 except RetryDownload
as err
:
374 retry
.error
= err
.source_error
380 except SucceedDownload
: