]>
jfr.im git - yt-dlp.git/blob - youtube_dl/downloader/common.py
1 from __future__
import division
, unicode_literals
8 from ..compat
import compat_str
16 class FileDownloader(object):
17 """File Downloader class.
19 File downloader objects are the ones responsible of downloading the
20 actual video file and writing it to disk.
22 File downloaders accept a lot of parameters. In order not to saturate
23 the object constructor with arguments, it receives a dictionary of
28 verbose: Print additional info to stdout.
29 quiet: Do not print messages to stdout.
30 ratelimit: Download speed limit, in bytes/sec.
31 retries: Number of times to retry for HTTP error 5xx
32 buffersize: Size of download buffer in bytes.
33 noresizebuffer: Do not automatically resize the download buffer.
34 continuedl: Try to continue downloads if possible.
35 noprogress: Do not print the progress bar.
36 logtostderr: Log messages to stderr instead of stdout.
37 consoletitle: Display progress in console window's titlebar.
38 nopart: Do not use temporary .part files.
39 updatetime: Use the Last-modified header to set output file timestamps.
40 test: Download only first bytes to test the downloader.
41 min_filesize: Skip files smaller than this size
42 max_filesize: Skip files larger than this size
43 xattr_set_filesize: Set ytdl.filesize user xattribute with expected size.
45 external_downloader_args: A list of additional command-line arguments for the
48 Subclasses of this one must re-define the real_download method.
51 _TEST_FILE_SIZE
= 10241
54 def __init__(self
, ydl
, params
):
55 """Create a FileDownloader object with the given options."""
57 self
._progress
_hooks
= []
59 self
.add_progress_hook(self
.report_progress
)
62 def format_seconds(seconds
):
63 (mins
, secs
) = divmod(seconds
, 60)
64 (hours
, mins
) = divmod(mins
, 60)
68 return '%02d:%02d' % (mins
, secs
)
70 return '%02d:%02d:%02d' % (hours
, mins
, secs
)
73 def calc_percent(byte_counter
, data_len
):
76 return float(byte_counter
) / float(data_len
) * 100.0
79 def format_percent(percent
):
82 return '%6s' % ('%3.1f%%' % percent
)
85 def calc_eta(start
, now
, total
, current
):
91 if current
== 0 or dif
< 0.001: # One millisecond
93 rate
= float(current
) / dif
94 return int((float(total
) - float(current
)) / rate
)
100 return FileDownloader
.format_seconds(eta
)
103 def calc_speed(start
, now
, bytes):
105 if bytes == 0 or dif
< 0.001: # One millisecond
107 return float(bytes) / dif
110 def format_speed(speed
):
112 return '%10s' % '---b/s'
113 return '%10s' % ('%s/s' % format_bytes(speed
))
116 def best_block_size(elapsed_time
, bytes):
117 new_min
= max(bytes / 2.0, 1.0)
118 new_max
= min(max(bytes * 2.0, 1.0), 4194304) # Do not surpass 4 MB
119 if elapsed_time
< 0.001:
121 rate
= bytes / elapsed_time
129 def parse_bytes(bytestr
):
130 """Parse a string indicating a byte quantity into an integer."""
131 matchobj
= re
.match(r
'(?i)^(\d+(?:\.\d+)?)([kMGTPEZY]?)$', bytestr
)
134 number
= float(matchobj
.group(1))
135 multiplier
= 1024.0 ** 'bkmgtpezy'.index(matchobj
.group(2).lower())
136 return int(round(number
* multiplier
))
138 def to_screen(self
, *args
, **kargs
):
139 self
.ydl
.to_screen(*args
, **kargs
)
141 def to_stderr(self
, message
):
142 self
.ydl
.to_screen(message
)
144 def to_console_title(self
, message
):
145 self
.ydl
.to_console_title(message
)
147 def trouble(self
, *args
, **kargs
):
148 self
.ydl
.trouble(*args
, **kargs
)
150 def report_warning(self
, *args
, **kargs
):
151 self
.ydl
.report_warning(*args
, **kargs
)
153 def report_error(self
, *args
, **kargs
):
154 self
.ydl
.report_error(*args
, **kargs
)
156 def slow_down(self
, start_time
, now
, byte_counter
):
157 """Sleep if the download speed is over the rate limit."""
158 rate_limit
= self
.params
.get('ratelimit', None)
159 if rate_limit
is None or byte_counter
== 0:
163 elapsed
= now
- start_time
166 speed
= float(byte_counter
) / elapsed
167 if speed
> rate_limit
:
168 time
.sleep(max((byte_counter
// rate_limit
) - elapsed
, 0))
170 def temp_name(self
, filename
):
171 """Returns a temporary filename for the given filename."""
172 if self
.params
.get('nopart', False) or filename
== '-' or \
173 (os
.path
.exists(encodeFilename(filename
)) and not os
.path
.isfile(encodeFilename(filename
))):
175 return filename
+ '.part'
177 def undo_temp_name(self
, filename
):
178 if filename
.endswith('.part'):
179 return filename
[:-len('.part')]
182 def try_rename(self
, old_filename
, new_filename
):
184 if old_filename
== new_filename
:
186 os
.rename(encodeFilename(old_filename
), encodeFilename(new_filename
))
187 except (IOError, OSError) as err
:
188 self
.report_error('unable to rename file: %s' % compat_str(err
))
190 def try_utime(self
, filename
, last_modified_hdr
):
191 """Try to set the last-modified time of the given file."""
192 if last_modified_hdr
is None:
194 if not os
.path
.isfile(encodeFilename(filename
)):
196 timestr
= last_modified_hdr
199 filetime
= timeconvert(timestr
)
202 # Ignore obviously invalid dates
206 os
.utime(filename
, (time
.time(), filetime
))
211 def report_destination(self
, filename
):
212 """Report destination filename."""
213 self
.to_screen('[download] Destination: ' + filename
)
215 def _report_progress_status(self
, msg
, is_last_line
=False):
216 fullmsg
= '[download] ' + msg
217 if self
.params
.get('progress_with_newline', False):
218 self
.to_screen(fullmsg
)
221 prev_len
= getattr(self
, '_report_progress_prev_line_length',
223 if prev_len
> len(fullmsg
):
224 fullmsg
+= ' ' * (prev_len
- len(fullmsg
))
225 self
._report
_progress
_prev
_line
_length
= len(fullmsg
)
228 clear_line
= ('\r\x1b[K' if sys
.stderr
.isatty() else '\r')
229 self
.to_screen(clear_line
+ fullmsg
, skip_eol
=not is_last_line
)
230 self
.to_console_title('youtube-dl ' + msg
)
232 def report_progress(self
, s
):
233 if s
['status'] == 'finished':
234 if self
.params
.get('noprogress', False):
235 self
.to_screen('[download] Download completed')
237 s
['_total_bytes_str'] = format_bytes(s
['total_bytes'])
238 if s
.get('elapsed') is not None:
239 s
['_elapsed_str'] = self
.format_seconds(s
['elapsed'])
240 msg_template
= '100%% of %(_total_bytes_str)s in %(_elapsed_str)s'
242 msg_template
= '100%% of %(_total_bytes_str)s'
243 self
._report
_progress
_status
(
244 msg_template
% s
, is_last_line
=True)
246 if self
.params
.get('noprogress'):
249 if s
['status'] != 'downloading':
252 if s
.get('eta') is not None:
253 s
['_eta_str'] = self
.format_eta(s
['eta'])
255 s
['_eta_str'] = 'Unknown ETA'
257 if s
.get('total_bytes') and s
.get('downloaded_bytes') is not None:
258 s
['_percent_str'] = self
.format_percent(100 * s
['downloaded_bytes'] / s
['total_bytes'])
259 elif s
.get('total_bytes_estimate') and s
.get('downloaded_bytes') is not None:
260 s
['_percent_str'] = self
.format_percent(100 * s
['downloaded_bytes'] / s
['total_bytes_estimate'])
262 if s
.get('downloaded_bytes') == 0:
263 s
['_percent_str'] = self
.format_percent(0)
265 s
['_percent_str'] = 'Unknown %'
267 if s
.get('speed') is not None:
268 s
['_speed_str'] = self
.format_speed(s
['speed'])
270 s
['_speed_str'] = 'Unknown speed'
272 if s
.get('total_bytes') is not None:
273 s
['_total_bytes_str'] = format_bytes(s
['total_bytes'])
274 msg_template
= '%(_percent_str)s of %(_total_bytes_str)s at %(_speed_str)s ETA %(_eta_str)s'
275 elif s
.get('total_bytes_estimate') is not None:
276 s
['_total_bytes_estimate_str'] = format_bytes(s
['total_bytes_estimate'])
277 msg_template
= '%(_percent_str)s of ~%(_total_bytes_estimate_str)s at %(_speed_str)s ETA %(_eta_str)s'
279 if s
.get('downloaded_bytes') is not None:
280 s
['_downloaded_bytes_str'] = format_bytes(s
['downloaded_bytes'])
282 s
['_elapsed_str'] = self
.format_seconds(s
['elapsed'])
283 msg_template
= '%(_downloaded_bytes_str)s at %(_speed_str)s (%(_elapsed_str)s)'
285 msg_template
= '%(_downloaded_bytes_str)s at %(_speed_str)s'
287 msg_template
= '%(_percent_str)s % at %(_speed_str)s ETA %(_eta_str)s'
289 self
._report
_progress
_status
(msg_template
% s
)
291 def report_resuming_byte(self
, resume_len
):
292 """Report attempt to resume at given byte."""
293 self
.to_screen('[download] Resuming download at byte %s' % resume_len
)
295 def report_retry(self
, count
, retries
):
296 """Report retry in case of HTTP error 5xx"""
297 self
.to_screen('[download] Got server HTTP error. Retrying (attempt %d of %d)...' % (count
, retries
))
299 def report_file_already_downloaded(self
, file_name
):
300 """Report file has already been fully downloaded."""
302 self
.to_screen('[download] %s has already been downloaded' % file_name
)
303 except UnicodeEncodeError:
304 self
.to_screen('[download] The file has already been downloaded')
306 def report_unable_to_resume(self
):
307 """Report it was impossible to resume download."""
308 self
.to_screen('[download] Unable to resume')
310 def download(self
, filename
, info_dict
):
311 """Download to a filename using the info from info_dict
312 Return True on success and False otherwise
315 nooverwrites_and_exists
= (
316 self
.params
.get('nooverwrites', False) and
317 os
.path
.exists(encodeFilename(filename
))
320 continuedl_and_exists
= (
321 self
.params
.get('continuedl', False) and
322 os
.path
.isfile(encodeFilename(filename
)) and
323 not self
.params
.get('nopart', False)
326 # Check file already present
327 if filename
!= '-' and nooverwrites_and_exists
or continuedl_and_exists
:
328 self
.report_file_already_downloaded(filename
)
329 self
._hook
_progress
({
330 'filename': filename
,
331 'status': 'finished',
332 'total_bytes': os
.path
.getsize(encodeFilename(filename
)),
336 sleep_interval
= self
.params
.get('sleep_interval')
338 self
.to_screen('[download] Sleeping %s seconds...' % sleep_interval
)
339 time
.sleep(sleep_interval
)
341 return self
.real_download(filename
, info_dict
)
343 def real_download(self
, filename
, info_dict
):
344 """Real download process. Redefine in subclasses."""
345 raise NotImplementedError('This method must be implemented by subclasses')
347 def _hook_progress(self
, status
):
348 for ph
in self
._progress
_hooks
:
351 def add_progress_hook(self
, ph
):
352 # See YoutubeDl.py (search for progress_hooks) for a description of
354 self
._progress
_hooks
.append(ph
)
356 def _debug_cmd(self
, args
, subprocess_encoding
, exe
=None):
357 if not self
.params
.get('verbose', False):
361 exe
= os
.path
.basename(args
[0])
363 if subprocess_encoding
:
365 a
.decode(subprocess_encoding
) if isinstance(a
, bytes) else a
371 shell_quote
= lambda args
: ' '.join(map(pipes
.quote
, str_args
))
374 self
.to_screen('[debug] %s command line: %s' % (
375 exe
, shell_quote(str_args
)))