X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/467f1716f3b46ffc038848b9b08a833a7811be09..252064fa082f84e06d1251185428b44c75c59109:/youtube_dl/FileDownloader.py diff --git a/youtube_dl/FileDownloader.py b/youtube_dl/FileDownloader.py index ea6b9d6..5c8e676 100644 --- a/youtube_dl/FileDownloader.py +++ b/youtube_dl/FileDownloader.py @@ -1,577 +1,12 @@ -import math -import os -import re -import subprocess -import sys -import time -import traceback - -if os.name == 'nt': - import ctypes - -from .utils import * - - -class FileDownloader(object): - """File Downloader class. - - File downloader objects are the ones responsible of downloading the - actual video file and writing it to disk. - - File downloaders accept a lot of parameters. In order not to saturate - the object constructor with arguments, it receives a dictionary of - options instead. - - Available options: - - verbose: Print additional info to stdout. - quiet: Do not print messages to stdout. - ratelimit: Download speed limit, in bytes/sec. - retries: Number of times to retry for HTTP error 5xx - buffersize: Size of download buffer in bytes. - noresizebuffer: Do not automatically resize the download buffer. - continuedl: Try to continue downloads if possible. - noprogress: Do not print the progress bar. - logtostderr: Log messages to stderr instead of stdout. - consoletitle: Display progress in console window's titlebar. - nopart: Do not use temporary .part files. - updatetime: Use the Last-modified header to set output file timestamps. - test: Download only first bytes to test the downloader. - min_filesize: Skip files smaller than this size - max_filesize: Skip files larger than this size - """ - - params = None - - def __init__(self, ydl, params): - """Create a FileDownloader object with the given options.""" - self.ydl = ydl - self._progress_hooks = [] - self.params = params - - @staticmethod - def format_bytes(bytes): - if bytes is None: - return 'N/A' - if type(bytes) is str: - bytes = float(bytes) - if bytes == 0.0: - exponent = 0 - else: - exponent = int(math.log(bytes, 1024.0)) - suffix = ['B','KiB','MiB','GiB','TiB','PiB','EiB','ZiB','YiB'][exponent] - converted = float(bytes) / float(1024 ** exponent) - return '%.2f%s' % (converted, suffix) - - @staticmethod - def calc_percent(byte_counter, data_len): - if data_len is None: - return '---.-%' - return '%6s' % ('%3.1f%%' % (float(byte_counter) / float(data_len) * 100.0)) - - @staticmethod - def calc_eta(start, now, total, current): - if total is None: - return '--:--' - dif = now - start - if current == 0 or dif < 0.001: # One millisecond - return '--:--' - rate = float(current) / dif - eta = int((float(total) - float(current)) / rate) - (eta_mins, eta_secs) = divmod(eta, 60) - if eta_mins > 99: - return '--:--' - return '%02d:%02d' % (eta_mins, eta_secs) - - @staticmethod - def calc_speed(start, now, bytes): - dif = now - start - if bytes == 0 or dif < 0.001: # One millisecond - return '%10s' % '---b/s' - return '%10s' % ('%s/s' % FileDownloader.format_bytes(float(bytes) / dif)) - - @staticmethod - def best_block_size(elapsed_time, bytes): - new_min = max(bytes / 2.0, 1.0) - new_max = min(max(bytes * 2.0, 1.0), 4194304) # Do not surpass 4 MB - if elapsed_time < 0.001: - return int(new_max) - rate = bytes / elapsed_time - if rate > new_max: - return int(new_max) - if rate < new_min: - return int(new_min) - return int(rate) - - @staticmethod - def parse_bytes(bytestr): - """Parse a string indicating a byte quantity into an integer.""" - matchobj = re.match(r'(?i)^(\d+(?:\.\d+)?)([kMGTPEZY]?)$', bytestr) - if matchobj is None: - return None - number = float(matchobj.group(1)) - multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower()) - return int(round(number * multiplier)) - - def to_screen(self, *args, **kargs): - self.ydl.to_screen(*args, **kargs) - - def to_stderr(self, message): - self.ydl.to_screen(message) - - def to_cons_title(self, message): - """Set console/terminal window title to message.""" - if not self.params.get('consoletitle', False): - return - if os.name == 'nt' and ctypes.windll.kernel32.GetConsoleWindow(): - # c_wchar_p() might not be necessary if `message` is - # already of type unicode() - ctypes.windll.kernel32.SetConsoleTitleW(ctypes.c_wchar_p(message)) - elif 'TERM' in os.environ: - self.to_screen('\033]0;%s\007' % message, skip_eol=True) - - def trouble(self, *args, **kargs): - self.ydl.trouble(*args, **kargs) - - def report_warning(self, *args, **kargs): - self.ydl.report_warning(*args, **kargs) - - def report_error(self, *args, **kargs): - self.ydl.report_error(*args, **kargs) - - def slow_down(self, start_time, byte_counter): - """Sleep if the download speed is over the rate limit.""" - rate_limit = self.params.get('ratelimit', None) - if rate_limit is None or byte_counter == 0: - return - now = time.time() - elapsed = now - start_time - if elapsed <= 0.0: - return - speed = float(byte_counter) / elapsed - if speed > rate_limit: - time.sleep((byte_counter - rate_limit * (now - start_time)) / rate_limit) - - def temp_name(self, filename): - """Returns a temporary filename for the given filename.""" - if self.params.get('nopart', False) or filename == u'-' or \ - (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))): - return filename - return filename + u'.part' - - def undo_temp_name(self, filename): - if filename.endswith(u'.part'): - return filename[:-len(u'.part')] - return filename - - def try_rename(self, old_filename, new_filename): - try: - if old_filename == new_filename: - return - os.rename(encodeFilename(old_filename), encodeFilename(new_filename)) - except (IOError, OSError) as err: - self.report_error(u'unable to rename file') - - def try_utime(self, filename, last_modified_hdr): - """Try to set the last-modified time of the given file.""" - if last_modified_hdr is None: - return - if not os.path.isfile(encodeFilename(filename)): - return - timestr = last_modified_hdr - if timestr is None: - return - filetime = timeconvert(timestr) - if filetime is None: - return filetime - # Ignore obviously invalid dates - if filetime == 0: - return - try: - os.utime(filename, (time.time(), filetime)) - except: - pass - return filetime - - def report_destination(self, filename): - """Report destination filename.""" - self.to_screen(u'[download] Destination: ' + filename) - - def report_progress(self, percent_str, data_len_str, speed_str, eta_str): - """Report download progress.""" - if self.params.get('noprogress', False): - return - clear_line = (u'\x1b[K' if sys.stderr.isatty() and os.name != 'nt' else u'') - if self.params.get('progress_with_newline', False): - self.to_screen(u'[download] %s of %s at %s ETA %s' % - (percent_str, data_len_str, speed_str, eta_str)) - else: - self.to_screen(u'\r%s[download] %s of %s at %s ETA %s' % - (clear_line, percent_str, data_len_str, speed_str, eta_str), skip_eol=True) - self.to_cons_title(u'youtube-dl - %s of %s at %s ETA %s' % - (percent_str.strip(), data_len_str.strip(), speed_str.strip(), eta_str.strip())) - - def report_resuming_byte(self, resume_len): - """Report attempt to resume at given byte.""" - self.to_screen(u'[download] Resuming download at byte %s' % resume_len) - - def report_retry(self, count, retries): - """Report retry in case of HTTP error 5xx""" - self.to_screen(u'[download] Got server HTTP error. Retrying (attempt %d of %d)...' % (count, retries)) - - def report_file_already_downloaded(self, file_name): - """Report file has already been fully downloaded.""" - try: - self.to_screen(u'[download] %s has already been downloaded' % file_name) - except (UnicodeEncodeError) as err: - self.to_screen(u'[download] The file has already been downloaded') - - def report_unable_to_resume(self): - """Report it was impossible to resume download.""" - self.to_screen(u'[download] Unable to resume') - - def report_finish(self): - """Report download finished.""" - if self.params.get('noprogress', False): - self.to_screen(u'[download] Download completed') - else: - self.to_screen(u'') - - def _download_with_rtmpdump(self, filename, url, player_url, page_url, play_path, tc_url): - self.report_destination(filename) - tmpfilename = self.temp_name(filename) - - # Check for rtmpdump first - try: - subprocess.call(['rtmpdump', '-h'], stdout=(open(os.path.devnull, 'w')), stderr=subprocess.STDOUT) - except (OSError, IOError): - self.report_error(u'RTMP download detected but "rtmpdump" could not be run') - return False - verbosity_option = '--verbose' if self.params.get('verbose', False) else '--quiet' - - # Download using rtmpdump. rtmpdump returns exit code 2 when - # the connection was interrumpted and resuming appears to be - # possible. This is part of rtmpdump's normal usage, AFAIK. - basic_args = ['rtmpdump', verbosity_option, '-r', url, '-o', tmpfilename] - if player_url is not None: - basic_args += ['--swfVfy', player_url] - if page_url is not None: - basic_args += ['--pageUrl', page_url] - if play_path is not None: - basic_args += ['--playpath', play_path] - if tc_url is not None: - basic_args += ['--tcUrl', url] - args = basic_args + [[], ['--resume', '--skip', '1']][self.params.get('continuedl', False)] - if self.params.get('verbose', False): - try: - import pipes - shell_quote = lambda args: ' '.join(map(pipes.quote, args)) - except ImportError: - shell_quote = repr - self.to_screen(u'[debug] rtmpdump command line: ' + shell_quote(args)) - retval = subprocess.call(args) - while retval == 2 or retval == 1: - prevsize = os.path.getsize(encodeFilename(tmpfilename)) - self.to_screen(u'\r[rtmpdump] %s bytes' % prevsize, skip_eol=True) - time.sleep(5.0) # This seems to be needed - retval = subprocess.call(basic_args + ['-e'] + [[], ['-k', '1']][retval == 1]) - cursize = os.path.getsize(encodeFilename(tmpfilename)) - if prevsize == cursize and retval == 1: - break - # Some rtmp streams seem abort after ~ 99.8%. Don't complain for those - if prevsize == cursize and retval == 2 and cursize > 1024: - self.to_screen(u'\r[rtmpdump] Could not download the whole video. This can happen for some advertisements.') - retval = 0 - break - if retval == 0: - fsize = os.path.getsize(encodeFilename(tmpfilename)) - self.to_screen(u'\r[rtmpdump] %s bytes' % fsize) - self.try_rename(tmpfilename, filename) - self._hook_progress({ - 'downloaded_bytes': fsize, - 'total_bytes': fsize, - 'filename': filename, - 'status': 'finished', - }) - return True - else: - self.to_stderr(u"\n") - self.report_error(u'rtmpdump exited with code %d' % retval) - return False - - def _download_with_mplayer(self, filename, url): - self.report_destination(filename) - tmpfilename = self.temp_name(filename) - - args = ['mplayer', '-really-quiet', '-vo', 'null', '-vc', 'dummy', '-dumpstream', '-dumpfile', tmpfilename, url] - # Check for mplayer first - try: - subprocess.call(['mplayer', '-h'], stdout=(open(os.path.devnull, 'w')), stderr=subprocess.STDOUT) - except (OSError, IOError): - self.report_error(u'MMS or RTSP download detected but "%s" could not be run' % args[0] ) - return False - - # Download using mplayer. - retval = subprocess.call(args) - if retval == 0: - fsize = os.path.getsize(encodeFilename(tmpfilename)) - self.to_screen(u'\r[%s] %s bytes' % (args[0], fsize)) - self.try_rename(tmpfilename, filename) - self._hook_progress({ - 'downloaded_bytes': fsize, - 'total_bytes': fsize, - 'filename': filename, - 'status': 'finished', - }) - return True - else: - self.to_stderr(u"\n") - self.report_error(u'mplayer exited with code %d' % retval) - return False - - def _download_m3u8_with_ffmpeg(self, filename, url): - self.report_destination(filename) - tmpfilename = self.temp_name(filename) - - args = ['ffmpeg', '-y', '-i', url, '-f', 'mp4', tmpfilename] - # Check for ffmpeg first - try: - subprocess.call(['ffmpeg', '-h'], stdout=(open(os.path.devnull, 'w')), stderr=subprocess.STDOUT) - except (OSError, IOError): - self.report_error(u'm3u8 download detected but "%s" could not be run' % args[0] ) - return False - - retval = subprocess.call(args) - if retval == 0: - fsize = os.path.getsize(encodeFilename(tmpfilename)) - self.to_screen(u'\r[%s] %s bytes' % (args[0], fsize)) - self.try_rename(tmpfilename, filename) - self._hook_progress({ - 'downloaded_bytes': fsize, - 'total_bytes': fsize, - 'filename': filename, - 'status': 'finished', - }) - return True - else: - self.to_stderr(u"\n") - self.report_error(u'ffmpeg exited with code %d' % retval) - return False +# Legacy file for backwards compatibility, use youtube_dl.downloader instead! +from .downloader import FileDownloader as RealFileDownloader +from .downloader import get_suitable_downloader +# This class reproduces the old behaviour of FileDownloader +class FileDownloader(RealFileDownloader): def _do_download(self, filename, info_dict): - url = info_dict['url'] - - # Check file already present - if self.params.get('continuedl', False) and os.path.isfile(encodeFilename(filename)) and not self.params.get('nopart', False): - self.report_file_already_downloaded(filename) - self._hook_progress({ - 'filename': filename, - 'status': 'finished', - }) - return True - - # Attempt to download using rtmpdump - if url.startswith('rtmp'): - return self._download_with_rtmpdump(filename, url, - info_dict.get('player_url', None), - info_dict.get('page_url', None), - info_dict.get('play_path', None), - info_dict.get('tc_url', None)) - - # Attempt to download using mplayer - if url.startswith('mms') or url.startswith('rtsp'): - return self._download_with_mplayer(filename, url) - - # m3u8 manifest are downloaded with ffmpeg - if determine_ext(url) == u'm3u8': - return self._download_m3u8_with_ffmpeg(filename, url) - - tmpfilename = self.temp_name(filename) - stream = None - - # Do not include the Accept-Encoding header - headers = {'Youtubedl-no-compression': 'True'} - if 'user_agent' in info_dict: - headers['Youtubedl-user-agent'] = info_dict['user_agent'] - basic_request = compat_urllib_request.Request(url, None, headers) - request = compat_urllib_request.Request(url, None, headers) - - if self.params.get('test', False): - request.add_header('Range','bytes=0-10240') - - # Establish possible resume length - if os.path.isfile(encodeFilename(tmpfilename)): - resume_len = os.path.getsize(encodeFilename(tmpfilename)) - else: - resume_len = 0 - - open_mode = 'wb' - if resume_len != 0: - if self.params.get('continuedl', False): - self.report_resuming_byte(resume_len) - request.add_header('Range','bytes=%d-' % resume_len) - open_mode = 'ab' - else: - resume_len = 0 - - count = 0 - retries = self.params.get('retries', 0) - while count <= retries: - # Establish connection - try: - if count == 0 and 'urlhandle' in info_dict: - data = info_dict['urlhandle'] - data = compat_urllib_request.urlopen(request) - break - except (compat_urllib_error.HTTPError, ) as err: - if (err.code < 500 or err.code >= 600) and err.code != 416: - # Unexpected HTTP error - raise - elif err.code == 416: - # Unable to resume (requested range not satisfiable) - try: - # Open the connection again without the range header - data = compat_urllib_request.urlopen(basic_request) - content_length = data.info()['Content-Length'] - except (compat_urllib_error.HTTPError, ) as err: - if err.code < 500 or err.code >= 600: - raise - else: - # Examine the reported length - if (content_length is not None and - (resume_len - 100 < int(content_length) < resume_len + 100)): - # The file had already been fully downloaded. - # Explanation to the above condition: in issue #175 it was revealed that - # YouTube sometimes adds or removes a few bytes from the end of the file, - # changing the file size slightly and causing problems for some users. So - # I decided to implement a suggested change and consider the file - # completely downloaded if the file size differs less than 100 bytes from - # the one in the hard drive. - self.report_file_already_downloaded(filename) - self.try_rename(tmpfilename, filename) - self._hook_progress({ - 'filename': filename, - 'status': 'finished', - }) - return True - else: - # The length does not match, we start the download over - self.report_unable_to_resume() - open_mode = 'wb' - break - # Retry - count += 1 - if count <= retries: - self.report_retry(count, retries) - - if count > retries: - self.report_error(u'giving up after %s retries' % retries) - return False - - data_len = data.info().get('Content-length', None) - if data_len is not None: - data_len = int(data_len) + resume_len - min_data_len = self.params.get("min_filesize", None) - max_data_len = self.params.get("max_filesize", None) - if min_data_len is not None and data_len < min_data_len: - self.to_screen(u'\r[download] File is smaller than min-filesize (%s bytes < %s bytes). Aborting.' % (data_len, min_data_len)) - return False - if max_data_len is not None and data_len > max_data_len: - self.to_screen(u'\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len, max_data_len)) - return False - - data_len_str = self.format_bytes(data_len) - byte_counter = 0 + resume_len - block_size = self.params.get('buffersize', 1024) - start = time.time() - while True: - # Download and write - before = time.time() - data_block = data.read(block_size) - after = time.time() - if len(data_block) == 0: - break - byte_counter += len(data_block) - - # Open file just in time - if stream is None: - try: - (stream, tmpfilename) = sanitize_open(tmpfilename, open_mode) - assert stream is not None - filename = self.undo_temp_name(tmpfilename) - self.report_destination(filename) - except (OSError, IOError) as err: - self.report_error(u'unable to open for writing: %s' % str(err)) - return False - try: - stream.write(data_block) - except (IOError, OSError) as err: - self.to_stderr(u"\n") - self.report_error(u'unable to write data: %s' % str(err)) - return False - if not self.params.get('noresizebuffer', False): - block_size = self.best_block_size(after - before, len(data_block)) - - # Progress message - speed_str = self.calc_speed(start, time.time(), byte_counter - resume_len) - if data_len is None: - self.report_progress('Unknown %', data_len_str, speed_str, 'Unknown ETA') - else: - percent_str = self.calc_percent(byte_counter, data_len) - eta_str = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len) - self.report_progress(percent_str, data_len_str, speed_str, eta_str) - - self._hook_progress({ - 'downloaded_bytes': byte_counter, - 'total_bytes': data_len, - 'tmpfilename': tmpfilename, - 'filename': filename, - 'status': 'downloading', - }) - - # Apply rate limit - self.slow_down(start, byte_counter - resume_len) - - if stream is None: - self.to_stderr(u"\n") - self.report_error(u'Did not get any data blocks') - return False - stream.close() - self.report_finish() - if data_len is not None and byte_counter != data_len: - raise ContentTooShortError(byte_counter, int(data_len)) - self.try_rename(tmpfilename, filename) - - # Update file modification time - if self.params.get('updatetime', True): - info_dict['filetime'] = self.try_utime(filename, data.info().get('last-modified', None)) - - self._hook_progress({ - 'downloaded_bytes': byte_counter, - 'total_bytes': byte_counter, - 'filename': filename, - 'status': 'finished', - }) - - return True - - def _hook_progress(self, status): + real_fd = get_suitable_downloader(info_dict)(self.ydl, self.params) for ph in self._progress_hooks: - ph(status) - - def add_progress_hook(self, ph): - """ ph gets called on download progress, with a dictionary with the entries - * filename: The final filename - * status: One of "downloading" and "finished" - - It can also have some of the following entries: - - * downloaded_bytes: Bytes on disks - * total_bytes: Total bytes, None if unknown - * tmpfilename: The filename we're currently writing to - - Hooks are guaranteed to be called at least once (with status "finished") - if the download is successful. - """ - self._progress_hooks.append(ph) + real_fd.add_progress_hook(ph) + return real_fd.download(filename, info_dict)