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