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