Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/downloader/common.py

   1 import os
   2 import re
   3 import sys
   4 import time
   5
   6 from ..utils import (
   7     encodeFilename,
   8     timeconvert,
   9     format_bytes,
  10 )
  11
  12
  13 class FileDownloader(object):
  14     """File Downloader class.
  15
  16     File downloader objects are the ones responsible of downloading the
  17     actual video file and writing it to disk.
  18
  19     File downloaders accept a lot of parameters. In order not to saturate
  20     the object constructor with arguments, it receives a dictionary of
  21     options instead.
  22
  23     Available options:
  24
  25     verbose:           Print additional info to stdout.
  26     quiet:             Do not print messages to stdout.
  27     ratelimit:         Download speed limit, in bytes/sec.
  28     retries:           Number of times to retry for HTTP error 5xx
  29     buffersize:        Size of download buffer in bytes.
  30     noresizebuffer:    Do not automatically resize the download buffer.
  31     continuedl:        Try to continue downloads if possible.
  32     noprogress:        Do not print the progress bar.
  33     logtostderr:       Log messages to stderr instead of stdout.
  34     consoletitle:      Display progress in console window's titlebar.
  35     nopart:            Do not use temporary .part files.
  36     updatetime:        Use the Last-modified header to set output file timestamps.
  37     test:              Download only first bytes to test the downloader.
  38     min_filesize:      Skip files smaller than this size
  39     max_filesize:      Skip files larger than this size
  40
  41     Subclasses of this one must re-define the real_download method.
  42     """
  43
  44     params = None
  45
  46     def __init__(self, ydl, params):
  47         """Create a FileDownloader object with the given options."""
  48         self.ydl = ydl
  49         self._progress_hooks = []
  50         self.params = params
  51
  52     @staticmethod
  53     def format_seconds(seconds):
  54         (mins, secs) = divmod(seconds, 60)
  55         (hours, mins) = divmod(mins, 60)
  56         if hours > 99:
  57             return '--:--:--'
  58         if hours == 0:
  59             return '%02d:%02d' % (mins, secs)
  60         else:
  61             return '%02d:%02d:%02d' % (hours, mins, secs)
  62
  63     @staticmethod
  64     def calc_percent(byte_counter, data_len):
  65         if data_len is None:
  66             return None
  67         return float(byte_counter) / float(data_len) * 100.0
  68
  69     @staticmethod
  70     def format_percent(percent):
  71         if percent is None:
  72             return '---.-%'
  73         return '%6s' % ('%3.1f%%' % percent)
  74
  75     @staticmethod
  76     def calc_eta(start, now, total, current):
  77         if total is None:
  78             return None
  79         dif = now - start
  80         if current == 0 or dif < 0.001: # One millisecond
  81             return None
  82         rate = float(current) / dif
  83         return int((float(total) - float(current)) / rate)
  84
  85     @staticmethod
  86     def format_eta(eta):
  87         if eta is None:
  88             return '--:--'
  89         return FileDownloader.format_seconds(eta)
  90
  91     @staticmethod
  92     def calc_speed(start, now, bytes):
  93         dif = now - start
  94         if bytes == 0 or dif < 0.001: # One millisecond
  95             return None
  96         return float(bytes) / dif
  97
  98     @staticmethod
  99     def format_speed(speed):
 100         if speed is None:
 101             return '%10s' % '---b/s'
 102         return '%10s' % ('%s/s' % format_bytes(speed))
 103
 104     @staticmethod
 105     def best_block_size(elapsed_time, bytes):
 106         new_min = max(bytes / 2.0, 1.0)
 107         new_max = min(max(bytes * 2.0, 1.0), 4194304) # Do not surpass 4 MB
 108         if elapsed_time < 0.001:
 109             return int(new_max)
 110         rate = bytes / elapsed_time
 111         if rate > new_max:
 112             return int(new_max)
 113         if rate < new_min:
 114             return int(new_min)
 115         return int(rate)
 116
 117     @staticmethod
 118     def parse_bytes(bytestr):
 119         """Parse a string indicating a byte quantity into an integer."""
 120         matchobj = re.match(r'(?i)^(\d+(?:\.\d+)?)([kMGTPEZY]?)$', bytestr)
 121         if matchobj is None:
 122             return None
 123         number = float(matchobj.group(1))
 124         multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower())
 125         return int(round(number * multiplier))
 126
 127     def to_screen(self, *args, **kargs):
 128         self.ydl.to_screen(*args, **kargs)
 129
 130     def to_stderr(self, message):
 131         self.ydl.to_screen(message)
 132
 133     def to_console_title(self, message):
 134         self.ydl.to_console_title(message)
 135
 136     def trouble(self, *args, **kargs):
 137         self.ydl.trouble(*args, **kargs)
 138
 139     def report_warning(self, *args, **kargs):
 140         self.ydl.report_warning(*args, **kargs)
 141
 142     def report_error(self, *args, **kargs):
 143         self.ydl.report_error(*args, **kargs)
 144
 145     def slow_down(self, start_time, byte_counter):
 146         """Sleep if the download speed is over the rate limit."""
 147         rate_limit = self.params.get('ratelimit', None)
 148         if rate_limit is None or byte_counter == 0:
 149             return
 150         now = time.time()
 151         elapsed = now - start_time
 152         if elapsed <= 0.0:
 153             return
 154         speed = float(byte_counter) / elapsed
 155         if speed > rate_limit:
 156             time.sleep((byte_counter - rate_limit * (now - start_time)) / rate_limit)
 157
 158     def temp_name(self, filename):
 159         """Returns a temporary filename for the given filename."""
 160         if self.params.get('nopart', False) or filename == u'-' or \
 161                 (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))):
 162             return filename
 163         return filename + u'.part'
 164
 165     def undo_temp_name(self, filename):
 166         if filename.endswith(u'.part'):
 167             return filename[:-len(u'.part')]
 168         return filename
 169
 170     def try_rename(self, old_filename, new_filename):
 171         try:
 172             if old_filename == new_filename:
 173                 return
 174             os.rename(encodeFilename(old_filename), encodeFilename(new_filename))
 175         except (IOError, OSError) as err:
 176             self.report_error(u'unable to rename file: %s' % str(err))
 177
 178     def try_utime(self, filename, last_modified_hdr):
 179         """Try to set the last-modified time of the given file."""
 180         if last_modified_hdr is None:
 181             return
 182         if not os.path.isfile(encodeFilename(filename)):
 183             return
 184         timestr = last_modified_hdr
 185         if timestr is None:
 186             return
 187         filetime = timeconvert(timestr)
 188         if filetime is None:
 189             return filetime
 190         # Ignore obviously invalid dates
 191         if filetime == 0:
 192             return
 193         try:
 194             os.utime(filename, (time.time(), filetime))
 195         except:
 196             pass
 197         return filetime
 198
 199     def report_destination(self, filename):
 200         """Report destination filename."""
 201         self.to_screen(u'[download] Destination: ' + filename)
 202
 203     def _report_progress_status(self, msg, is_last_line=False):
 204         fullmsg = u'[download] ' + msg
 205         if self.params.get('progress_with_newline', False):
 206             self.to_screen(fullmsg)
 207         else:
 208             if os.name == 'nt':
 209                 prev_len = getattr(self, '_report_progress_prev_line_length',
 210                                    0)
 211                 if prev_len > len(fullmsg):
 212                     fullmsg += u' ' * (prev_len - len(fullmsg))
 213                 self._report_progress_prev_line_length = len(fullmsg)
 214                 clear_line = u'\r'
 215             else:
 216                 clear_line = (u'\r\x1b[K' if sys.stderr.isatty() else u'\r')
 217             self.to_screen(clear_line + fullmsg, skip_eol=not is_last_line)
 218         self.to_console_title(u'youtube-dl ' + msg)
 219
 220     def report_progress(self, percent, data_len_str, speed, eta):
 221         """Report download progress."""
 222         if self.params.get('noprogress', False):
 223             return
 224         if eta is not None:
 225             eta_str = self.format_eta(eta)
 226         else:
 227             eta_str = 'Unknown ETA'
 228         if percent is not None:
 229             percent_str = self.format_percent(percent)
 230         else:
 231             percent_str = 'Unknown %'
 232         speed_str = self.format_speed(speed)
 233
 234         msg = (u'%s of %s at %s ETA %s' %
 235                (percent_str, data_len_str, speed_str, eta_str))
 236         self._report_progress_status(msg)
 237
 238     def report_progress_live_stream(self, downloaded_data_len, speed, elapsed):
 239         if self.params.get('noprogress', False):
 240             return
 241         downloaded_str = format_bytes(downloaded_data_len)
 242         speed_str = self.format_speed(speed)
 243         elapsed_str = FileDownloader.format_seconds(elapsed)
 244         msg = u'%s at %s (%s)' % (downloaded_str, speed_str, elapsed_str)
 245         self._report_progress_status(msg)
 246
 247     def report_finish(self, data_len_str, tot_time):
 248         """Report download finished."""
 249         if self.params.get('noprogress', False):
 250             self.to_screen(u'[download] Download completed')
 251         else:
 252             self._report_progress_status(
 253                 (u'100%% of %s in %s' %
 254                  (data_len_str, self.format_seconds(tot_time))),
 255                 is_last_line=True)
 256
 257     def report_resuming_byte(self, resume_len):
 258         """Report attempt to resume at given byte."""
 259         self.to_screen(u'[download] Resuming download at byte %s' % resume_len)
 260
 261     def report_retry(self, count, retries):
 262         """Report retry in case of HTTP error 5xx"""
 263         self.to_screen(u'[download] Got server HTTP error. Retrying (attempt %d of %d)...' % (count, retries))
 264
 265     def report_file_already_downloaded(self, file_name):
 266         """Report file has already been fully downloaded."""
 267         try:
 268             self.to_screen(u'[download] %s has already been downloaded' % file_name)
 269         except UnicodeEncodeError:
 270             self.to_screen(u'[download] The file has already been downloaded')
 271
 272     def report_unable_to_resume(self):
 273         """Report it was impossible to resume download."""
 274         self.to_screen(u'[download] Unable to resume')
 275
 276     def download(self, filename, info_dict):
 277         """Download to a filename using the info from info_dict
 278         Return True on success and False otherwise
 279         """
 280         # Check file already present
 281         if self.params.get('continuedl', False) and os.path.isfile(encodeFilename(filename)) and not self.params.get('nopart', False):
 282             self.report_file_already_downloaded(filename)
 283             self._hook_progress({
 284                 'filename': filename,
 285                 'status': 'finished',
 286                 'total_bytes': os.path.getsize(encodeFilename(filename)),
 287             })
 288             return True
 289
 290         return self.real_download(filename, info_dict)
 291
 292     def real_download(self, filename, info_dict):
 293         """Real download process. Redefine in subclasses."""
 294         raise NotImplementedError(u'This method must be implemented by sublcasses')
 295
 296     def _hook_progress(self, status):
 297         for ph in self._progress_hooks:
 298             ph(status)
 299
 300     def add_progress_hook(self, ph):
 301         """ ph gets called on download progress, with a dictionary with the entries
 302         * filename: The final filename
 303         * status: One of "downloading" and "finished"
 304
 305         It can also have some of the following entries:
 306
 307         * downloaded_bytes: Bytes on disks
 308         * total_bytes: Total bytes, None if unknown
 309         * tmpfilename: The filename we're currently writing to
 310         * eta: The estimated time in seconds, None if unknown
 311         * speed: The download speed in bytes/second, None if unknown
 312
 313         Hooks are guaranteed to be called at least once (with status "finished")
 314         if the download is successful.
 315         """
 316         self._progress_hooks.append(ph)