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