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