]> Raphaƫl G. Git Repositories - youtubedl/blob - youtube_dl/downloader/common.py
debian/control: Add recommends on aria2 | wget | curl to use external downloaders.
[youtubedl] / youtube_dl / downloader / common.py
1 from __future__ import unicode_literals
2
3 import os
4 import re
5 import sys
6 import time
7
8 from ..compat import compat_str
9 from ..utils import (
10 encodeFilename,
11 format_bytes,
12 timeconvert,
13 )
14
15
16 class FileDownloader(object):
17 """File Downloader class.
18
19 File downloader objects are the ones responsible of downloading the
20 actual video file and writing it to disk.
21
22 File downloaders accept a lot of parameters. In order not to saturate
23 the object constructor with arguments, it receives a dictionary of
24 options instead.
25
26 Available options:
27
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.
44 (experimenatal)
45
46 Subclasses of this one must re-define the real_download method.
47 """
48
49 _TEST_FILE_SIZE = 10241
50 params = None
51
52 def __init__(self, ydl, params):
53 """Create a FileDownloader object with the given options."""
54 self.ydl = ydl
55 self._progress_hooks = []
56 self.params = params
57
58 @staticmethod
59 def format_seconds(seconds):
60 (mins, secs) = divmod(seconds, 60)
61 (hours, mins) = divmod(mins, 60)
62 if hours > 99:
63 return '--:--:--'
64 if hours == 0:
65 return '%02d:%02d' % (mins, secs)
66 else:
67 return '%02d:%02d:%02d' % (hours, mins, secs)
68
69 @staticmethod
70 def calc_percent(byte_counter, data_len):
71 if data_len is None:
72 return None
73 return float(byte_counter) / float(data_len) * 100.0
74
75 @staticmethod
76 def format_percent(percent):
77 if percent is None:
78 return '---.-%'
79 return '%6s' % ('%3.1f%%' % percent)
80
81 @staticmethod
82 def calc_eta(start, now, total, current):
83 if total is None:
84 return None
85 if now is None:
86 now = time.time()
87 dif = now - start
88 if current == 0 or dif < 0.001: # One millisecond
89 return None
90 rate = float(current) / dif
91 return int((float(total) - float(current)) / rate)
92
93 @staticmethod
94 def format_eta(eta):
95 if eta is None:
96 return '--:--'
97 return FileDownloader.format_seconds(eta)
98
99 @staticmethod
100 def calc_speed(start, now, bytes):
101 dif = now - start
102 if bytes == 0 or dif < 0.001: # One millisecond
103 return None
104 return float(bytes) / dif
105
106 @staticmethod
107 def format_speed(speed):
108 if speed is None:
109 return '%10s' % '---b/s'
110 return '%10s' % ('%s/s' % format_bytes(speed))
111
112 @staticmethod
113 def best_block_size(elapsed_time, bytes):
114 new_min = max(bytes / 2.0, 1.0)
115 new_max = min(max(bytes * 2.0, 1.0), 4194304) # Do not surpass 4 MB
116 if elapsed_time < 0.001:
117 return int(new_max)
118 rate = bytes / elapsed_time
119 if rate > new_max:
120 return int(new_max)
121 if rate < new_min:
122 return int(new_min)
123 return int(rate)
124
125 @staticmethod
126 def parse_bytes(bytestr):
127 """Parse a string indicating a byte quantity into an integer."""
128 matchobj = re.match(r'(?i)^(\d+(?:\.\d+)?)([kMGTPEZY]?)$', bytestr)
129 if matchobj is None:
130 return None
131 number = float(matchobj.group(1))
132 multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower())
133 return int(round(number * multiplier))
134
135 def to_screen(self, *args, **kargs):
136 self.ydl.to_screen(*args, **kargs)
137
138 def to_stderr(self, message):
139 self.ydl.to_screen(message)
140
141 def to_console_title(self, message):
142 self.ydl.to_console_title(message)
143
144 def trouble(self, *args, **kargs):
145 self.ydl.trouble(*args, **kargs)
146
147 def report_warning(self, *args, **kargs):
148 self.ydl.report_warning(*args, **kargs)
149
150 def report_error(self, *args, **kargs):
151 self.ydl.report_error(*args, **kargs)
152
153 def slow_down(self, start_time, now, byte_counter):
154 """Sleep if the download speed is over the rate limit."""
155 rate_limit = self.params.get('ratelimit', None)
156 if rate_limit is None or byte_counter == 0:
157 return
158 if now is None:
159 now = time.time()
160 elapsed = now - start_time
161 if elapsed <= 0.0:
162 return
163 speed = float(byte_counter) / elapsed
164 if speed > rate_limit:
165 time.sleep(max((byte_counter // rate_limit) - elapsed, 0))
166
167 def temp_name(self, filename):
168 """Returns a temporary filename for the given filename."""
169 if self.params.get('nopart', False) or filename == '-' or \
170 (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))):
171 return filename
172 return filename + '.part'
173
174 def undo_temp_name(self, filename):
175 if filename.endswith('.part'):
176 return filename[:-len('.part')]
177 return filename
178
179 def try_rename(self, old_filename, new_filename):
180 try:
181 if old_filename == new_filename:
182 return
183 os.rename(encodeFilename(old_filename), encodeFilename(new_filename))
184 except (IOError, OSError) as err:
185 self.report_error('unable to rename file: %s' % compat_str(err))
186
187 def try_utime(self, filename, last_modified_hdr):
188 """Try to set the last-modified time of the given file."""
189 if last_modified_hdr is None:
190 return
191 if not os.path.isfile(encodeFilename(filename)):
192 return
193 timestr = last_modified_hdr
194 if timestr is None:
195 return
196 filetime = timeconvert(timestr)
197 if filetime is None:
198 return filetime
199 # Ignore obviously invalid dates
200 if filetime == 0:
201 return
202 try:
203 os.utime(filename, (time.time(), filetime))
204 except:
205 pass
206 return filetime
207
208 def report_destination(self, filename):
209 """Report destination filename."""
210 self.to_screen('[download] Destination: ' + filename)
211
212 def _report_progress_status(self, msg, is_last_line=False):
213 fullmsg = '[download] ' + msg
214 if self.params.get('progress_with_newline', False):
215 self.to_screen(fullmsg)
216 else:
217 if os.name == 'nt':
218 prev_len = getattr(self, '_report_progress_prev_line_length',
219 0)
220 if prev_len > len(fullmsg):
221 fullmsg += ' ' * (prev_len - len(fullmsg))
222 self._report_progress_prev_line_length = len(fullmsg)
223 clear_line = '\r'
224 else:
225 clear_line = ('\r\x1b[K' if sys.stderr.isatty() else '\r')
226 self.to_screen(clear_line + fullmsg, skip_eol=not is_last_line)
227 self.to_console_title('youtube-dl ' + msg)
228
229 def report_progress(self, percent, data_len_str, speed, eta):
230 """Report download progress."""
231 if self.params.get('noprogress', False):
232 return
233 if eta is not None:
234 eta_str = self.format_eta(eta)
235 else:
236 eta_str = 'Unknown ETA'
237 if percent is not None:
238 percent_str = self.format_percent(percent)
239 else:
240 percent_str = 'Unknown %'
241 speed_str = self.format_speed(speed)
242
243 msg = ('%s of %s at %s ETA %s' %
244 (percent_str, data_len_str, speed_str, eta_str))
245 self._report_progress_status(msg)
246
247 def report_progress_live_stream(self, downloaded_data_len, speed, elapsed):
248 if self.params.get('noprogress', False):
249 return
250 downloaded_str = format_bytes(downloaded_data_len)
251 speed_str = self.format_speed(speed)
252 elapsed_str = FileDownloader.format_seconds(elapsed)
253 msg = '%s at %s (%s)' % (downloaded_str, speed_str, elapsed_str)
254 self._report_progress_status(msg)
255
256 def report_finish(self, data_len_str, tot_time):
257 """Report download finished."""
258 if self.params.get('noprogress', False):
259 self.to_screen('[download] Download completed')
260 else:
261 self._report_progress_status(
262 ('100%% of %s in %s' %
263 (data_len_str, self.format_seconds(tot_time))),
264 is_last_line=True)
265
266 def report_resuming_byte(self, resume_len):
267 """Report attempt to resume at given byte."""
268 self.to_screen('[download] Resuming download at byte %s' % resume_len)
269
270 def report_retry(self, count, retries):
271 """Report retry in case of HTTP error 5xx"""
272 self.to_screen('[download] Got server HTTP error. Retrying (attempt %d of %d)...' % (count, retries))
273
274 def report_file_already_downloaded(self, file_name):
275 """Report file has already been fully downloaded."""
276 try:
277 self.to_screen('[download] %s has already been downloaded' % file_name)
278 except UnicodeEncodeError:
279 self.to_screen('[download] The file has already been downloaded')
280
281 def report_unable_to_resume(self):
282 """Report it was impossible to resume download."""
283 self.to_screen('[download] Unable to resume')
284
285 def download(self, filename, info_dict):
286 """Download to a filename using the info from info_dict
287 Return True on success and False otherwise
288 """
289
290 nooverwrites_and_exists = (
291 self.params.get('nooverwrites', False)
292 and os.path.exists(encodeFilename(filename))
293 )
294
295 continuedl_and_exists = (
296 self.params.get('continuedl', False)
297 and os.path.isfile(encodeFilename(filename))
298 and not self.params.get('nopart', False)
299 )
300
301 # Check file already present
302 if filename != '-' and nooverwrites_and_exists or continuedl_and_exists:
303 self.report_file_already_downloaded(filename)
304 self._hook_progress({
305 'filename': filename,
306 'status': 'finished',
307 'total_bytes': os.path.getsize(encodeFilename(filename)),
308 })
309 return True
310
311 sleep_interval = self.params.get('sleep_interval')
312 if sleep_interval:
313 self.to_screen('[download] Sleeping %s seconds...' % sleep_interval)
314 time.sleep(sleep_interval)
315
316 return self.real_download(filename, info_dict)
317
318 def real_download(self, filename, info_dict):
319 """Real download process. Redefine in subclasses."""
320 raise NotImplementedError('This method must be implemented by subclasses')
321
322 def _hook_progress(self, status):
323 for ph in self._progress_hooks:
324 ph(status)
325
326 def add_progress_hook(self, ph):
327 # See YoutubeDl.py (search for progress_hooks) for a description of
328 # this interface
329 self._progress_hooks.append(ph)
330
331 def _debug_cmd(self, args, subprocess_encoding, exe=None):
332 if not self.params.get('verbose', False):
333 return
334
335 if exe is None:
336 exe = os.path.basename(args[0])
337
338 if subprocess_encoding:
339 str_args = [
340 a.decode(subprocess_encoding) if isinstance(a, bytes) else a
341 for a in args]
342 else:
343 str_args = args
344 try:
345 import pipes
346 shell_quote = lambda args: ' '.join(map(pipes.quote, str_args))
347 except ImportError:
348 shell_quote = repr
349 self.to_screen('[debug] %s command line: %s' % (
350 exe, shell_quote(str_args)))