diff options
Diffstat (limited to 'youtube_dl/downloader')
| -rw-r--r-- | youtube_dl/downloader/__init__.py | 55 | ||||
| -rw-r--r-- | youtube_dl/downloader/common.py | 139 | ||||
| -rw-r--r-- | youtube_dl/downloader/dash.py | 113 | ||||
| -rw-r--r-- | youtube_dl/downloader/external.py | 435 | ||||
| -rw-r--r-- | youtube_dl/downloader/f4m.py | 155 | ||||
| -rw-r--r-- | youtube_dl/downloader/fragment.py | 242 | ||||
| -rw-r--r-- | youtube_dl/downloader/hls.py | 268 | ||||
| -rw-r--r-- | youtube_dl/downloader/http.py | 443 | ||||
| -rw-r--r-- | youtube_dl/downloader/ism.py | 259 | ||||
| -rw-r--r-- | youtube_dl/downloader/niconico.py | 66 | ||||
| -rw-r--r-- | youtube_dl/downloader/rtmp.py | 141 | ||||
| -rw-r--r-- | youtube_dl/downloader/rtsp.py | 2 |
12 files changed, 1787 insertions, 531 deletions
diff --git a/youtube_dl/downloader/__init__.py b/youtube_dl/downloader/__init__.py index dccc59212..d701d6292 100644 --- a/youtube_dl/downloader/__init__.py +++ b/youtube_dl/downloader/__init__.py @@ -1,43 +1,68 @@ from __future__ import unicode_literals +from ..utils import ( + determine_protocol, +) + + +def get_suitable_downloader(info_dict, params={}): + info_dict['protocol'] = determine_protocol(info_dict) + info_copy = info_dict.copy() + return _get_suitable_downloader(info_copy, params) + + +# Some of these require get_suitable_downloader from .common import FileDownloader -from .external import get_external_downloader +from .dash import DashSegmentsFD from .f4m import F4mFD from .hls import HlsFD -from .hls import NativeHlsFD from .http import HttpFD -from .rtsp import RtspFD from .rtmp import RtmpFD -from .dash import DashSegmentsFD - -from ..utils import ( - determine_protocol, +from .rtsp import RtspFD +from .ism import IsmFD +from .niconico import NiconicoDmcFD +from .external import ( + get_external_downloader, + FFmpegFD, ) PROTOCOL_MAP = { 'rtmp': RtmpFD, - 'm3u8_native': NativeHlsFD, - 'm3u8': HlsFD, + 'm3u8_native': HlsFD, + 'm3u8': FFmpegFD, 'mms': RtspFD, 'rtsp': RtspFD, 'f4m': F4mFD, 'http_dash_segments': DashSegmentsFD, + 'ism': IsmFD, + 'niconico_dmc': NiconicoDmcFD, } -def get_suitable_downloader(info_dict, params={}): +def _get_suitable_downloader(info_dict, params={}): """Get the downloader class that can handle the info dict.""" - protocol = determine_protocol(info_dict) - info_dict['protocol'] = protocol + + # if (info_dict.get('start_time') or info_dict.get('end_time')) and not info_dict.get('requested_formats') and FFmpegFD.can_download(info_dict): + # return FFmpegFD external_downloader = params.get('external_downloader') if external_downloader is not None: ed = get_external_downloader(external_downloader) - if ed.supports(info_dict): + if ed.can_download(info_dict): return ed + # Avoid using unwanted args since external_downloader was rejected + if params.get('external_downloader_args'): + params['external_downloader_args'] = None + + protocol = info_dict['protocol'] + if protocol.startswith('m3u8') and info_dict.get('is_live'): + return FFmpegFD + + if protocol == 'm3u8' and params.get('hls_prefer_native') is True: + return HlsFD - if protocol == 'm3u8' and params.get('hls_prefer_native'): - return NativeHlsFD + if protocol == 'm3u8_native' and params.get('hls_prefer_native') is False: + return FFmpegFD return PROTOCOL_MAP.get(protocol, HttpFD) diff --git a/youtube_dl/downloader/common.py b/youtube_dl/downloader/common.py index b8bf8daf8..8354030a9 100644 --- a/youtube_dl/downloader/common.py +++ b/youtube_dl/downloader/common.py @@ -4,12 +4,16 @@ import os import re import sys import time +import random -from ..compat import compat_str +from ..compat import compat_os_name from ..utils import ( - encodeFilename, decodeArgument, + encodeFilename, + error_to_compat_str, + float_or_none, format_bytes, + shell_quote, timeconvert, ) @@ -42,9 +46,12 @@ class FileDownloader(object): min_filesize: Skip files smaller than this size max_filesize: Skip files larger than this size xattr_set_filesize: Set ytdl.filesize user xattribute with expected size. - (experimental) external_downloader_args: A list of additional command-line arguments for the external downloader. + hls_use_mpegts: Use the mpegts container for HLS videos. + http_chunk_size: Size of a chunk for chunk-based HTTP downloading. May be + useful for bypassing bandwidth throttling imposed by + a webserver (experimental) Subclasses of this one must re-define the real_download method. """ @@ -82,17 +89,21 @@ class FileDownloader(object): return '---.-%' return '%6s' % ('%3.1f%%' % percent) - @staticmethod - def calc_eta(start, now, total, current): + @classmethod + def calc_eta(cls, start_or_rate, now_or_remaining, *args): + if len(args) < 2: + rate, remaining = (start_or_rate, now_or_remaining) + if None in (rate, remaining): + return None + return int(float(remaining) / rate) + start, now = (start_or_rate, now_or_remaining) + total, current = args[:2] if total is None: return None if now is None: now = time.time() - dif = now - start - if current == 0 or dif < 0.001: # One millisecond - return None - rate = float(current) / dif - return int((float(total) - float(current)) / rate) + rate = cls.calc_speed(start, now, current) + return rate and int((float(total) - float(current)) / rate) @staticmethod def format_eta(eta): @@ -114,6 +125,16 @@ class FileDownloader(object): return '%10s' % ('%s/s' % format_bytes(speed)) @staticmethod + def format_retries(retries): + return 'inf' if retries == float('inf') else '%.0f' % retries + + @staticmethod + def filesize_or_none(unencoded_filename): + fn = encodeFilename(unencoded_filename) + if os.path.isfile(fn): + return os.path.getsize(fn) + + @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 @@ -156,7 +177,7 @@ class FileDownloader(object): def slow_down(self, start_time, now, byte_counter): """Sleep if the download speed is over the rate limit.""" - rate_limit = self.params.get('ratelimit', None) + rate_limit = self.params.get('ratelimit') if rate_limit is None or byte_counter == 0: return if now is None: @@ -166,7 +187,9 @@ class FileDownloader(object): return speed = float(byte_counter) / elapsed if speed > rate_limit: - time.sleep(max((byte_counter // rate_limit) - elapsed, 0)) + sleep_time = float(byte_counter) / rate_limit - elapsed + if sleep_time > 0: + time.sleep(sleep_time) def temp_name(self, filename): """Returns a temporary filename for the given filename.""" @@ -180,13 +203,16 @@ class FileDownloader(object): return filename[:-len('.part')] return filename + def ytdl_filename(self, filename): + return filename + '.ytdl' + 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('unable to rename file: %s' % compat_str(err)) + self.report_error('unable to rename file: %s' % error_to_compat_str(err)) def try_utime(self, filename, last_modified_hdr): """Try to set the last-modified time of the given file.""" @@ -218,7 +244,7 @@ class FileDownloader(object): if self.params.get('progress_with_newline', False): self.to_screen(fullmsg) else: - if os.name == 'nt': + if compat_os_name == 'nt': prev_len = getattr(self, '_report_progress_prev_line_length', 0) if prev_len > len(fullmsg): @@ -235,12 +261,13 @@ class FileDownloader(object): if self.params.get('noprogress', False): self.to_screen('[download] Download completed') else: - s['_total_bytes_str'] = format_bytes(s['total_bytes']) + msg_template = '100%%' + if s.get('total_bytes') is not None: + s['_total_bytes_str'] = format_bytes(s['total_bytes']) + msg_template += ' of %(_total_bytes_str)s' if s.get('elapsed') is not None: s['_elapsed_str'] = self.format_seconds(s['elapsed']) - msg_template = '100%% of %(_total_bytes_str)s in %(_elapsed_str)s' - else: - msg_template = '100%% of %(_total_bytes_str)s' + msg_template += ' in %(_elapsed_str)s' self._report_progress_status( msg_template % s, is_last_line=True) @@ -293,9 +320,11 @@ class FileDownloader(object): """Report attempt to resume at given byte.""" self.to_screen('[download] Resuming download at byte %s' % resume_len) - def report_retry(self, count, retries): + def report_retry(self, err, count, retries): """Report retry in case of HTTP error 5xx""" - self.to_screen('[download] Got server HTTP error. Retrying (attempt %d of %d)...' % (count, retries)) + self.to_screen( + '[download] Got server HTTP error: %s. Retrying (attempt %d of %s)...' + % (error_to_compat_str(err), count, self.format_retries(retries))) def report_file_already_downloaded(self, file_name): """Report file has already been fully downloaded.""" @@ -311,32 +340,55 @@ class FileDownloader(object): def download(self, filename, info_dict): """Download to a filename using the info from info_dict Return True on success and False otherwise + + This method filters the `Cookie` header from the info_dict to prevent leaks. + Downloaders have their own way of handling cookies. + See: https://github.com/yt-dlp/yt-dlp/security/advisories/GHSA-v8mc-9377-rwjj """ nooverwrites_and_exists = ( - self.params.get('nooverwrites', False) and - os.path.exists(encodeFilename(filename)) + self.params.get('nooverwrites', False) + and os.path.exists(encodeFilename(filename)) ) - continuedl_and_exists = ( - self.params.get('continuedl', True) and - os.path.isfile(encodeFilename(filename)) and - not self.params.get('nopart', False) - ) - - # Check file already present - if filename != '-' and (nooverwrites_and_exists or continuedl_and_exists): - self.report_file_already_downloaded(filename) - self._hook_progress({ - 'filename': filename, - 'status': 'finished', - 'total_bytes': os.path.getsize(encodeFilename(filename)), - }) - return True - - sleep_interval = self.params.get('sleep_interval') - if sleep_interval: - self.to_screen('[download] Sleeping %s seconds...' % sleep_interval) + if not hasattr(filename, 'write'): + continuedl_and_exists = ( + self.params.get('continuedl', True) + and os.path.isfile(encodeFilename(filename)) + and not self.params.get('nopart', False) + ) + + # Check file already present + if filename != '-' and (nooverwrites_and_exists or continuedl_and_exists): + self.report_file_already_downloaded(filename) + self._hook_progress({ + 'filename': filename, + 'status': 'finished', + 'total_bytes': os.path.getsize(encodeFilename(filename)), + }) + return True + + min_sleep_interval, max_sleep_interval = ( + float_or_none(self.params.get(interval), default=0) + for interval in ('sleep_interval', 'max_sleep_interval')) + + sleep_note = '' + available_at = info_dict.get('available_at') + if available_at: + forced_sleep_interval = available_at - int(time.time()) + if forced_sleep_interval > min_sleep_interval: + sleep_note = 'as required by the site' + min_sleep_interval = forced_sleep_interval + if forced_sleep_interval > max_sleep_interval: + max_sleep_interval = forced_sleep_interval + + sleep_interval = random.uniform( + min_sleep_interval, max_sleep_interval or min_sleep_interval) + + if sleep_interval > 0: + self.to_screen( + '[download] Sleeping %.2f seconds %s...' % ( + sleep_interval, sleep_note)) time.sleep(sleep_interval) return self.real_download(filename, info_dict) @@ -363,10 +415,5 @@ class FileDownloader(object): if exe is None: exe = os.path.basename(str_args[0]) - try: - import pipes - shell_quote = lambda args: ' '.join(map(pipes.quote, str_args)) - except ImportError: - shell_quote = repr self.to_screen('[debug] %s command line: %s' % ( exe, shell_quote(str_args))) diff --git a/youtube_dl/downloader/dash.py b/youtube_dl/downloader/dash.py index 535f2a7fc..f3c058879 100644 --- a/youtube_dl/downloader/dash.py +++ b/youtube_dl/downloader/dash.py @@ -1,66 +1,83 @@ from __future__ import unicode_literals -import re +import itertools -from .common import FileDownloader -from ..utils import sanitized_Request +from .fragment import FragmentFD +from ..compat import compat_urllib_error +from ..utils import ( + DownloadError, + urljoin, +) -class DashSegmentsFD(FileDownloader): +class DashSegmentsFD(FragmentFD): """ Download segments in a DASH manifest """ - def real_download(self, filename, info_dict): - self.report_destination(filename) - tmpfilename = self.temp_name(filename) - base_url = info_dict['url'] - segment_urls = info_dict['segment_urls'] - - is_test = self.params.get('test', False) - remaining_bytes = self._TEST_FILE_SIZE if is_test else None - byte_counter = 0 - def append_url_to_file(outf, target_url, target_name, remaining_bytes=None): - self.to_screen('[DashSegments] %s: Downloading %s' % (info_dict['id'], target_name)) - req = sanitized_Request(target_url) - if remaining_bytes is not None: - req.add_header('Range', 'bytes=0-%d' % (remaining_bytes - 1)) + FD_NAME = 'dashsegments' - data = self.ydl.urlopen(req).read() + def real_download(self, filename, info_dict): + fragment_base_url = info_dict.get('fragment_base_url') + fragments = info_dict['fragments'][:1] if self.params.get( + 'test', False) else info_dict['fragments'] - if remaining_bytes is not None: - data = data[:remaining_bytes] + ctx = { + 'filename': filename, + 'total_frags': len(fragments), + } - outf.write(data) - return len(data) + self._prepare_and_start_frag_download(ctx) - def combine_url(base_url, target_url): - if re.match(r'^https?://', target_url): - return target_url - return '%s%s%s' % (base_url, '' if base_url.endswith('/') else '/', target_url) + fragment_retries = self.params.get('fragment_retries', 0) + skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True) - with open(tmpfilename, 'wb') as outf: - append_url_to_file( - outf, combine_url(base_url, info_dict['initialization_url']), - 'initialization segment') - for i, segment_url in enumerate(segment_urls): - segment_len = append_url_to_file( - outf, combine_url(base_url, segment_url), - 'segment %d / %d' % (i + 1, len(segment_urls)), - remaining_bytes) - byte_counter += segment_len - if remaining_bytes is not None: - remaining_bytes -= segment_len - if remaining_bytes <= 0: - break + for frag_index, fragment in enumerate(fragments, 1): + if frag_index <= ctx['fragment_index']: + continue + success = False + # In DASH, the first segment contains necessary headers to + # generate a valid MP4 file, so always abort for the first segment + fatal = frag_index == 1 or not skip_unavailable_fragments + fragment_url = fragment.get('url') + if not fragment_url: + assert fragment_base_url + fragment_url = urljoin(fragment_base_url, fragment['path']) + headers = info_dict.get('http_headers') + fragment_range = fragment.get('range') + if fragment_range: + headers = headers.copy() if headers else {} + headers['Range'] = 'bytes=%s' % (fragment_range,) + for count in itertools.count(): + try: + success, frag_content = self._download_fragment(ctx, fragment_url, info_dict, headers) + if not success: + return False + self._append_fragment(ctx, frag_content) + except compat_urllib_error.HTTPError as err: + # YouTube may often return 404 HTTP error for a fragment causing the + # whole download to fail. However if the same fragment is immediately + # retried with the same request data this usually succeeds (1-2 attempts + # is usually enough) thus allowing to download the whole file successfully. + # To be future-proof we will retry all fragments that fail with any + # HTTP error. + if count < fragment_retries: + self.report_retry_fragment(err, frag_index, count + 1, fragment_retries) + continue + except DownloadError: + # Don't retry fragment if error occurred during HTTP downloading + # itself since it has its own retry settings + if fatal: + raise + break - self.try_rename(tmpfilename, filename) + if not success: + if not fatal: + self.report_skip_fragment(frag_index) + continue + self.report_error('giving up after %s fragment retries' % count) + return False - self._hook_progress({ - 'downloaded_bytes': byte_counter, - 'total_bytes': byte_counter, - 'filename': filename, - 'status': 'finished', - }) + self._finish_frag_download(ctx) return True diff --git a/youtube_dl/downloader/external.py b/youtube_dl/downloader/external.py index 2bc011266..4fbc0f520 100644 --- a/youtube_dl/downloader/external.py +++ b/youtube_dl/downloader/external.py @@ -1,9 +1,24 @@ from __future__ import unicode_literals -import os.path +import os +import re import subprocess +import sys +import tempfile +import time from .common import FileDownloader +from ..compat import ( + compat_setenv, + compat_str, + compat_subprocess_Popen, +) + +try: + from ..postprocessor.ffmpeg import FFmpegPostProcessor, EXT_TO_OUT_FORMATS +except ImportError: + FFmpegPostProcessor = None + from ..utils import ( cli_option, cli_valueless_option, @@ -11,6 +26,12 @@ from ..utils import ( cli_configuration_args, encodeFilename, encodeArgument, + handle_youtubedl_headers, + check_executable, + is_outdated_version, + process_communicate_or_kill, + T, + traverse_obj, ) @@ -18,18 +39,42 @@ class ExternalFD(FileDownloader): def real_download(self, filename, info_dict): self.report_destination(filename) tmpfilename = self.temp_name(filename) + self._cookies_tempfile = None + + try: + started = time.time() + retval = self._call_downloader(tmpfilename, info_dict) + except KeyboardInterrupt: + if not info_dict.get('is_live'): + raise + # Live stream downloading cancellation should be considered as + # correct and expected termination thus all postprocessing + # should take place + retval = 0 + self.to_screen('[%s] Interrupted by user' % self.get_basename()) + finally: + if self._cookies_tempfile and os.path.isfile(self._cookies_tempfile): + try: + os.remove(self._cookies_tempfile) + except OSError: + self.report_warning( + 'Unable to delete temporary cookies file "{0}"'.format(self._cookies_tempfile)) - retval = self._call_downloader(tmpfilename, info_dict) if retval == 0: - fsize = os.path.getsize(encodeFilename(tmpfilename)) - self.to_screen('\r[%s] Downloaded %s bytes' % (self.get_basename(), fsize)) - self.try_rename(tmpfilename, filename) - self._hook_progress({ - 'downloaded_bytes': fsize, - 'total_bytes': fsize, + status = { 'filename': filename, 'status': 'finished', - }) + 'elapsed': time.time() - started, + } + if filename != '-': + fsize = os.path.getsize(encodeFilename(tmpfilename)) + self.to_screen('\r[%s] Downloaded %s bytes' % (self.get_basename(), fsize)) + self.try_rename(tmpfilename, filename) + status.update({ + 'downloaded_bytes': fsize, + 'total_bytes': fsize, + }) + self._hook_progress(status) return True else: self.to_stderr('\n') @@ -46,9 +91,17 @@ class ExternalFD(FileDownloader): return self.params.get('external_downloader') @classmethod + def available(cls): + return check_executable(cls.get_basename(), [cls.AVAILABLE_OPT]) + + @classmethod def supports(cls, info_dict): return info_dict['protocol'] in ('http', 'https', 'ftp', 'ftps') + @classmethod + def can_download(cls, info_dict): + return cls.available() and cls.supports(info_dict) + def _option(self, command_option, param): return cli_option(self.params, command_option, param) @@ -61,6 +114,16 @@ class ExternalFD(FileDownloader): def _configuration_args(self, default=[]): return cli_configuration_args(self.params, 'external_downloader_args', default) + def _write_cookies(self): + if not self.ydl.cookiejar.filename: + tmp_cookies = tempfile.NamedTemporaryFile(suffix='.cookies', delete=False) + tmp_cookies.close() + self._cookies_tempfile = tmp_cookies.name + self.to_screen('[download] Writing temporary cookies file to "{0}"'.format(self._cookies_tempfile)) + # real_download resets _cookies_tempfile; if it's None, save() will write to cookiejar.filename + self.ydl.cookiejar.save(self._cookies_tempfile, ignore_discard=True, ignore_expires=True) + return self.ydl.cookiejar.filename or self._cookies_tempfile + def _call_downloader(self, tmpfilename, info_dict): """ Either overwrite this or implement _make_cmd """ cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)] @@ -69,17 +132,37 @@ class ExternalFD(FileDownloader): p = subprocess.Popen( cmd, stderr=subprocess.PIPE) - _, stderr = p.communicate() + _, stderr = process_communicate_or_kill(p) if p.returncode != 0: - self.to_stderr(stderr) + self.to_stderr(stderr.decode('utf-8', 'replace')) return p.returncode + @staticmethod + def _header_items(info_dict): + return traverse_obj( + info_dict, ('http_headers', T(dict.items), Ellipsis)) + class CurlFD(ExternalFD): + AVAILABLE_OPT = '-V' + def _make_cmd(self, tmpfilename, info_dict): - cmd = [self.exe, '--location', '-o', tmpfilename] - for key, val in info_dict['http_headers'].items(): + cmd = [self.exe, '--location', '-o', tmpfilename, '--compressed'] + cookie_header = self.ydl.cookiejar.get_cookie_header(info_dict['url']) + if cookie_header: + cmd += ['--cookie', cookie_header] + for key, val in self._header_items(info_dict): cmd += ['--header', '%s: %s' % (key, val)] + cmd += self._bool_option('--continue-at', 'continuedl', '-', '0') + cmd += self._valueless_option('--silent', 'noprogress') + cmd += self._valueless_option('--verbose', 'verbose') + cmd += self._option('--limit-rate', 'ratelimit') + retry = self._option('--retry', 'retries') + if len(retry) == 2: + if retry[1] in ('inf', 'infinite'): + retry[1] = '2147483647' + cmd += retry + cmd += self._option('--max-filesize', 'max_filesize') cmd += self._option('--interface', 'source_address') cmd += self._option('--proxy', 'proxy') cmd += self._valueless_option('--insecure', 'nocheckcertificate') @@ -87,24 +170,52 @@ class CurlFD(ExternalFD): cmd += ['--', info_dict['url']] return cmd + def _call_downloader(self, tmpfilename, info_dict): + cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)] + + self._debug_cmd(cmd) + + # curl writes the progress to stderr so don't capture it. + p = subprocess.Popen(cmd) + process_communicate_or_kill(p) + return p.returncode + class AxelFD(ExternalFD): + AVAILABLE_OPT = '-V' + def _make_cmd(self, tmpfilename, info_dict): cmd = [self.exe, '-o', tmpfilename] - for key, val in info_dict['http_headers'].items(): + for key, val in self._header_items(info_dict): cmd += ['-H', '%s: %s' % (key, val)] + cookie_header = self.ydl.cookiejar.get_cookie_header(info_dict['url']) + if cookie_header: + cmd += ['-H', 'Cookie: {0}'.format(cookie_header), '--max-redirect=0'] cmd += self._configuration_args() cmd += ['--', info_dict['url']] return cmd class WgetFD(ExternalFD): + AVAILABLE_OPT = '--version' + def _make_cmd(self, tmpfilename, info_dict): - cmd = [self.exe, '-O', tmpfilename, '-nv', '--no-cookies'] - for key, val in info_dict['http_headers'].items(): + cmd = [self.exe, '-O', tmpfilename, '-nv', '--compression=auto'] + if self.ydl.cookiejar.get_cookie_header(info_dict['url']): + cmd += ['--load-cookies', self._write_cookies()] + for key, val in self._header_items(info_dict): cmd += ['--header', '%s: %s' % (key, val)] + cmd += self._option('--limit-rate', 'ratelimit') + retry = self._option('--tries', 'retries') + if len(retry) == 2: + if retry[1] in ('inf', 'infinite'): + retry[1] = '0' + cmd += retry cmd += self._option('--bind-address', 'source_address') - cmd += self._option('--proxy', 'proxy') + proxy = self.params.get('proxy') + if proxy: + for var in ('http_proxy', 'https_proxy'): + cmd += ['--execute', '%s=%s' % (var, proxy)] cmd += self._valueless_option('--no-check-certificate', 'nocheckcertificate') cmd += self._configuration_args() cmd += ['--', info_dict['url']] @@ -112,30 +223,300 @@ class WgetFD(ExternalFD): class Aria2cFD(ExternalFD): + AVAILABLE_OPT = '-v' + + @staticmethod + def _aria2c_filename(fn): + return fn if os.path.isabs(fn) else os.path.join('.', fn) + def _make_cmd(self, tmpfilename, info_dict): - cmd = [self.exe, '-c'] - cmd += self._configuration_args([ - '--min-split-size', '1M', '--max-connection-per-server', '4']) - dn = os.path.dirname(tmpfilename) - if dn: - cmd += ['--dir', dn] - cmd += ['--out', os.path.basename(tmpfilename)] - for key, val in info_dict['http_headers'].items(): + cmd = [self.exe, '-c', + '--console-log-level=warn', '--summary-interval=0', '--download-result=hide', + '--http-accept-gzip=true', '--file-allocation=none', '-x16', '-j16', '-s16'] + if 'fragments' in info_dict: + cmd += ['--allow-overwrite=true', '--allow-piece-length-change=true'] + else: + cmd += ['--min-split-size', '1M'] + + if self.ydl.cookiejar.get_cookie_header(info_dict['url']): + cmd += ['--load-cookies={0}'.format(self._write_cookies())] + for key, val in self._header_items(info_dict): cmd += ['--header', '%s: %s' % (key, val)] + cmd += self._configuration_args(['--max-connection-per-server', '4']) + cmd += ['--out', os.path.basename(tmpfilename)] + cmd += self._option('--max-overall-download-limit', 'ratelimit') cmd += self._option('--interface', 'source_address') cmd += self._option('--all-proxy', 'proxy') cmd += self._bool_option('--check-certificate', 'nocheckcertificate', 'false', 'true', '=') - cmd += ['--', info_dict['url']] + cmd += self._bool_option('--remote-time', 'updatetime', 'true', 'false', '=') + cmd += self._bool_option('--show-console-readout', 'noprogress', 'false', 'true', '=') + cmd += self._configuration_args() + + # aria2c strips out spaces from the beginning/end of filenames and paths. + # We work around this issue by adding a "./" to the beginning of the + # filename and relative path, and adding a "/" at the end of the path. + # See: https://github.com/yt-dlp/yt-dlp/issues/276 + # https://github.com/ytdl-org/youtube-dl/issues/20312 + # https://github.com/aria2/aria2/issues/1373 + dn = os.path.dirname(tmpfilename) + if dn: + cmd += ['--dir', self._aria2c_filename(dn) + os.path.sep] + if 'fragments' not in info_dict: + cmd += ['--out', self._aria2c_filename(os.path.basename(tmpfilename))] + cmd += ['--auto-file-renaming=false'] + if 'fragments' in info_dict: + cmd += ['--file-allocation=none', '--uri-selector=inorder'] + url_list_file = '%s.frag.urls' % (tmpfilename, ) + url_list = [] + for frag_index, fragment in enumerate(info_dict['fragments']): + fragment_filename = '%s-Frag%d' % (os.path.basename(tmpfilename), frag_index) + url_list.append('%s\n\tout=%s' % (fragment['url'], self._aria2c_filename(fragment_filename))) + stream, _ = self.sanitize_open(url_list_file, 'wb') + stream.write('\n'.join(url_list).encode()) + stream.close() + cmd += ['-i', self._aria2c_filename(url_list_file)] + else: + cmd += ['--', info_dict['url']] return cmd +class Aria2pFD(ExternalFD): + ''' Aria2pFD class + This class support to use aria2p as downloader. + (Aria2p, a command-line tool and Python library to interact with an aria2c daemon process + through JSON-RPC.) + It can help you to get download progress more easily. + To use aria2p as downloader, you need to install aria2c and aria2p, aria2p can download with pip. + Then run aria2c in the background and enable with the --enable-rpc option. + ''' + try: + import aria2p + __avail = True + except ImportError: + __avail = False + + @classmethod + def available(cls): + return cls.__avail + + def _call_downloader(self, tmpfilename, info_dict): + aria2 = self.aria2p.API( + self.aria2p.Client( + host='http://localhost', + port=6800, + secret='' + ) + ) + + options = { + 'min-split-size': '1M', + 'max-connection-per-server': 4, + 'auto-file-renaming': 'false', + } + options['dir'] = os.path.dirname(tmpfilename) or os.path.abspath('.') + options['out'] = os.path.basename(tmpfilename) + if self.ydl.cookiejar.get_cookie_header(info_dict['url']): + options['load-cookies'] = self._write_cookies() + options['header'] = [] + for key, val in self._header_items(info_dict): + options['header'].append('{0}: {1}'.format(key, val)) + download = aria2.add_uris([info_dict['url']], options) + status = { + 'status': 'downloading', + 'tmpfilename': tmpfilename, + } + started = time.time() + while download.status in ['active', 'waiting']: + download = aria2.get_download(download.gid) + status.update({ + 'downloaded_bytes': download.completed_length, + 'total_bytes': download.total_length, + 'elapsed': time.time() - started, + 'eta': download.eta.total_seconds(), + 'speed': download.download_speed, + }) + self._hook_progress(status) + time.sleep(.5) + return download.status != 'complete' + + class HttpieFD(ExternalFD): + @classmethod + def available(cls): + return check_executable('http', ['--version']) + def _make_cmd(self, tmpfilename, info_dict): cmd = ['http', '--download', '--output', tmpfilename, info_dict['url']] - for key, val in info_dict['http_headers'].items(): + for key, val in self._header_items(info_dict): cmd += ['%s:%s' % (key, val)] + + # httpie 3.1.0+ removes the Cookie header on redirect, so this should be safe for now. [1] + # If we ever need cookie handling for redirects, we can export the cookiejar into a session. [2] + # 1: https://github.com/httpie/httpie/security/advisories/GHSA-9w4w-cpc8-h2fq + # 2: https://httpie.io/docs/cli/sessions + cookie_header = self.ydl.cookiejar.get_cookie_header(info_dict['url']) + if cookie_header: + cmd += ['Cookie:%s' % cookie_header] return cmd + +class FFmpegFD(ExternalFD): + @classmethod + def supports(cls, info_dict): + return info_dict['protocol'] in ('http', 'https', 'ftp', 'ftps', 'm3u8', 'rtsp', 'rtmp', 'mms', 'http_dash_segments') + + @classmethod + def available(cls): + # actual availability can only be confirmed for an instance + return bool(FFmpegPostProcessor) + + def _call_downloader(self, tmpfilename, info_dict): + # `downloader` means the parent `YoutubeDL` + ffpp = FFmpegPostProcessor(downloader=self.ydl) + if not ffpp.available: + self.report_error('ffmpeg required for download but no ffmpeg (nor avconv) executable could be found. Please install one.') + return False + ffpp.check_version() + + args = [ffpp.executable, '-y'] + + for log_level in ('quiet', 'verbose'): + if self.params.get(log_level, False): + args += ['-loglevel', log_level] + break + + seekable = info_dict.get('_seekable') + if seekable is not None: + # setting -seekable prevents ffmpeg from guessing if the server + # supports seeking(by adding the header `Range: bytes=0-`), which + # can cause problems in some cases + # https://github.com/ytdl-org/youtube-dl/issues/11800#issuecomment-275037127 + # http://trac.ffmpeg.org/ticket/6125#comment:10 + args += ['-seekable', '1' if seekable else '0'] + + args += self._configuration_args() + + # start_time = info_dict.get('start_time') or 0 + # if start_time: + # args += ['-ss', compat_str(start_time)] + # end_time = info_dict.get('end_time') + # if end_time: + # args += ['-t', compat_str(end_time - start_time)] + + url = info_dict['url'] + cookies = self.ydl.cookiejar.get_cookies_for_url(url) + if cookies: + args.extend(['-cookies', ''.join( + '{0}={1}; path={2}; domain={3};\r\n'.format( + cookie.name, cookie.value, cookie.path, cookie.domain) + for cookie in cookies)]) + + if info_dict.get('http_headers') and re.match(r'^https?://', url): + # Trailing \r\n after each HTTP header is important to prevent warning from ffmpeg/avconv: + # [http @ 00000000003d2fa0] No trailing CRLF found in HTTP header. + headers = handle_youtubedl_headers(info_dict['http_headers']) + args += [ + '-headers', + ''.join('%s: %s\r\n' % (key, val) for key, val in headers.items())] + + env = None + proxy = self.params.get('proxy') + if proxy: + if not re.match(r'^[\da-zA-Z]+://', proxy): + proxy = 'http://%s' % proxy + + if proxy.startswith('socks'): + self.report_warning( + '%s does not support SOCKS proxies. Downloading is likely to fail. ' + 'Consider adding --hls-prefer-native to your command.' % self.get_basename()) + + # Since December 2015 ffmpeg supports -http_proxy option (see + # http://git.videolan.org/?p=ffmpeg.git;a=commit;h=b4eb1f29ebddd60c41a2eb39f5af701e38e0d3fd) + # We could switch to the following code if we are able to detect version properly + # args += ['-http_proxy', proxy] + env = os.environ.copy() + compat_setenv('HTTP_PROXY', proxy, env=env) + compat_setenv('http_proxy', proxy, env=env) + + protocol = info_dict.get('protocol') + + if protocol == 'rtmp': + player_url = info_dict.get('player_url') + page_url = info_dict.get('page_url') + app = info_dict.get('app') + play_path = info_dict.get('play_path') + tc_url = info_dict.get('tc_url') + flash_version = info_dict.get('flash_version') + live = info_dict.get('rtmp_live', False) + conn = info_dict.get('rtmp_conn') + if player_url is not None: + args += ['-rtmp_swfverify', player_url] + if page_url is not None: + args += ['-rtmp_pageurl', page_url] + if app is not None: + args += ['-rtmp_app', app] + if play_path is not None: + args += ['-rtmp_playpath', play_path] + if tc_url is not None: + args += ['-rtmp_tcurl', tc_url] + if flash_version is not None: + args += ['-rtmp_flashver', flash_version] + if live: + args += ['-rtmp_live', 'live'] + if isinstance(conn, list): + for entry in conn: + args += ['-rtmp_conn', entry] + elif isinstance(conn, compat_str): + args += ['-rtmp_conn', conn] + + args += ['-i', url, '-c', 'copy'] + + if self.params.get('test', False): + args += ['-fs', compat_str(self._TEST_FILE_SIZE)] + + if protocol in ('m3u8', 'm3u8_native'): + if self.params.get('hls_use_mpegts', False) or tmpfilename == '-': + args += ['-f', 'mpegts'] + else: + args += ['-f', 'mp4'] + if (ffpp.basename == 'ffmpeg' and is_outdated_version(ffpp._versions['ffmpeg'], '3.2', False)) and (not info_dict.get('acodec') or info_dict['acodec'].split('.')[0] in ('aac', 'mp4a')): + args += ['-bsf:a', 'aac_adtstoasc'] + elif protocol == 'rtmp': + args += ['-f', 'flv'] + else: + args += ['-f', EXT_TO_OUT_FORMATS.get(info_dict['ext'], info_dict['ext'])] + + args = [encodeArgument(opt) for opt in args] + args.append(encodeFilename(ffpp._ffmpeg_filename_argument(tmpfilename), True)) + + self._debug_cmd(args) + + # From [1], a PIPE opened in Popen() should be closed, unless + # .communicate() is called. Avoid leaking any PIPEs by using Popen + # as a context manager (newer Python 3.x and compat) + # Fixes "Resource Warning" in test/test_downloader_external.py + # [1] https://devpress.csdn.net/python/62fde12d7e66823466192e48.html + with compat_subprocess_Popen(args, stdin=subprocess.PIPE, env=env) as proc: + try: + retval = proc.wait() + except BaseException as e: + # subprocess.run would send the SIGKILL signal to ffmpeg and the + # mp4 file couldn't be played, but if we ask ffmpeg to quit it + # produces a file that is playable (this is mostly useful for live + # streams). Note that Windows is not affected and produces playable + # files (see https://github.com/ytdl-org/youtube-dl/issues/8300). + if isinstance(e, KeyboardInterrupt) and sys.platform != 'win32': + process_communicate_or_kill(proc, b'q') + else: + proc.kill() + raise + return retval + + +class AVconvFD(FFmpegFD): + pass + + _BY_NAME = dict( (klass.get_basename(), klass) for name, klass in globals().items() diff --git a/youtube_dl/downloader/f4m.py b/youtube_dl/downloader/f4m.py index 6170cc155..8dd3c2eeb 100644 --- a/youtube_dl/downloader/f4m.py +++ b/youtube_dl/downloader/f4m.py @@ -1,47 +1,57 @@ from __future__ import division, unicode_literals -import base64 import io import itertools -import os import time from .fragment import FragmentFD from ..compat import ( + compat_b64decode, compat_etree_fromstring, compat_urlparse, compat_urllib_error, compat_urllib_parse_urlparse, + compat_struct_pack, + compat_struct_unpack, ) from ..utils import ( - encodeFilename, - sanitize_open, - struct_pack, - struct_unpack, + fix_xml_ampersands, xpath_text, ) +class DataTruncatedError(Exception): + pass + + class FlvReader(io.BytesIO): """ Reader for Flv files The file format is documented in https://www.adobe.com/devnet/f4v.html """ + def read_bytes(self, n): + data = self.read(n) + if len(data) < n: + raise DataTruncatedError( + 'FlvReader error: need %d bytes while only %d bytes got' % ( + n, len(data))) + return data + # Utility functions for reading numbers and strings def read_unsigned_long_long(self): - return struct_unpack('!Q', self.read(8))[0] + return compat_struct_unpack('!Q', self.read_bytes(8))[0] def read_unsigned_int(self): - return struct_unpack('!I', self.read(4))[0] + return compat_struct_unpack('!I', self.read_bytes(4))[0] def read_unsigned_char(self): - return struct_unpack('!B', self.read(1))[0] + return compat_struct_unpack('!B', self.read_bytes(1))[0] def read_string(self): res = b'' while True: - char = self.read(1) + char = self.read_bytes(1) if char == b'\x00': break res += char @@ -52,18 +62,18 @@ class FlvReader(io.BytesIO): Read a box and return the info as a tuple: (box_size, box_type, box_data) """ real_size = size = self.read_unsigned_int() - box_type = self.read(4) + box_type = self.read_bytes(4) header_end = 8 if size == 1: real_size = self.read_unsigned_long_long() header_end = 16 - return real_size, box_type, self.read(real_size - header_end) + return real_size, box_type, self.read_bytes(real_size - header_end) def read_asrt(self): # version self.read_unsigned_char() # flags - self.read(3) + self.read_bytes(3) quality_entry_count = self.read_unsigned_char() # QualityEntryCount for i in range(quality_entry_count): @@ -84,7 +94,7 @@ class FlvReader(io.BytesIO): # version self.read_unsigned_char() # flags - self.read(3) + self.read_bytes(3) # time scale self.read_unsigned_int() @@ -118,7 +128,7 @@ class FlvReader(io.BytesIO): # version self.read_unsigned_char() # flags - self.read(3) + self.read_bytes(3) self.read_unsigned_int() # BootstrapinfoVersion # Profile,Live,Update,Reserved @@ -183,6 +193,11 @@ def build_fragments_list(boot_info): first_frag_number = fragment_run_entry_table[0]['first'] fragments_counter = itertools.count(first_frag_number) for segment, fragments_count in segment_run_table['segment_run']: + # In some live HDS streams (for example Rai), `fragments_count` is + # abnormal and causing out-of-memory errors. It's OK to change the + # number of fragments for live streams as they are updated periodically + if fragments_count == 4294967295 and boot_info['live']: + fragments_count = 2 for _ in range(fragments_count): res.append((segment, next(fragments_counter))) @@ -193,11 +208,11 @@ def build_fragments_list(boot_info): def write_unsigned_int(stream, val): - stream.write(struct_pack('!I', val)) + stream.write(compat_struct_pack('!I', val)) def write_unsigned_int_24(stream, val): - stream.write(struct_pack('!I', val)[1:]) + stream.write(compat_struct_pack('!I', val)[1:]) def write_flv_header(stream): @@ -222,8 +237,23 @@ def write_metadata_tag(stream, metadata): write_unsigned_int(stream, FLV_TAG_HEADER_LEN + len(metadata)) -def _add_ns(prop): - return '{http://ns.adobe.com/f4m/1.0}%s' % prop +def remove_encrypted_media(media): + return list(filter(lambda e: 'drmAdditionalHeaderId' not in e.attrib + and 'drmAdditionalHeaderSetId' not in e.attrib, + media)) + + +def _add_ns(prop, ver=1): + return '{http://ns.adobe.com/f4m/%d.0}%s' % (ver, prop) + + +def get_base_url(manifest): + base_url = xpath_text( + manifest, [_add_ns('baseURL'), _add_ns('baseURL', 2)], + 'base URL', default=None) + if base_url: + base_url = base_url.strip() + return base_url class F4mFD(FragmentFD): @@ -237,15 +267,13 @@ class F4mFD(FragmentFD): media = doc.findall(_add_ns('media')) if not media: self.report_error('No media found') - for e in (doc.findall(_add_ns('drmAdditionalHeader')) + - doc.findall(_add_ns('drmAdditionalHeaderSet'))): + for e in (doc.findall(_add_ns('drmAdditionalHeader')) + + doc.findall(_add_ns('drmAdditionalHeaderSet'))): # If id attribute is missing it's valid for all media nodes # without drmAdditionalHeaderId or drmAdditionalHeaderSetId attribute if 'id' not in e.attrib: self.report_error('Missing ID in f4m DRM') - media = list(filter(lambda e: 'drmAdditionalHeaderId' not in e.attrib and - 'drmAdditionalHeaderSetId' not in e.attrib, - media)) + media = remove_encrypted_media(media) if not media: self.report_error('Unsupported DRM') return media @@ -272,28 +300,38 @@ class F4mFD(FragmentFD): return fragments_list def _parse_bootstrap_node(self, node, base_url): - if node.text is None: + # Sometimes non empty inline bootstrap info can be specified along + # with bootstrap url attribute (e.g. dummy inline bootstrap info + # contains whitespace characters in [1]). We will prefer bootstrap + # url over inline bootstrap info when present. + # 1. http://live-1-1.rutube.ru/stream/1024/HDS/SD/C2NKsS85HQNckgn5HdEmOQ/1454167650/S-s604419906/move/four/dirs/upper/1024-576p.f4m + bootstrap_url = node.get('url') + if bootstrap_url: bootstrap_url = compat_urlparse.urljoin( - base_url, node.attrib['url']) + base_url, bootstrap_url) boot_info = self._get_bootstrap_from_url(bootstrap_url) else: bootstrap_url = None - bootstrap = base64.b64decode(node.text.encode('ascii')) + bootstrap = compat_b64decode(node.text) boot_info = read_bootstrap_info(bootstrap) - return (boot_info, bootstrap_url) + return boot_info, bootstrap_url def real_download(self, filename, info_dict): man_url = info_dict['url'] requested_bitrate = info_dict.get('tbr') self.to_screen('[%s] Downloading f4m manifest' % self.FD_NAME) - urlh = self.ydl.urlopen(man_url) + + urlh = self.ydl.urlopen(self._prepare_url(info_dict, man_url)) man_url = urlh.geturl() - manifest = urlh.read() + # Some manifests may be malformed, e.g. prosiebensat1 generated manifests + # (see https://github.com/ytdl-org/youtube-dl/issues/6215#issuecomment-121704244 + # and https://github.com/ytdl-org/youtube-dl/issues/7823) + manifest = fix_xml_ampersands(urlh.read().decode('utf-8', 'ignore')).strip() doc = compat_etree_fromstring(manifest) formats = [(int(f.attrib.get('bitrate', -1)), f) for f in self._get_unencrypted_media(doc)] - if requested_bitrate is None: + if requested_bitrate is None or len(formats) == 1: # get the best format formats = sorted(formats, key=lambda f: f[0]) rate, media = formats[-1] @@ -301,18 +339,23 @@ class F4mFD(FragmentFD): rate, media = list(filter( lambda f: int(f[0]) == requested_bitrate, formats))[0] - base_url = compat_urlparse.urljoin(man_url, media.attrib['url']) + # Prefer baseURL for relative URLs as per 11.2 of F4M 3.0 spec. + man_base_url = get_base_url(doc) or man_url + + base_url = compat_urlparse.urljoin(man_base_url, media.attrib['url']) bootstrap_node = doc.find(_add_ns('bootstrapInfo')) - boot_info, bootstrap_url = self._parse_bootstrap_node(bootstrap_node, base_url) + boot_info, bootstrap_url = self._parse_bootstrap_node( + bootstrap_node, man_base_url) live = boot_info['live'] metadata_node = media.find(_add_ns('metadata')) if metadata_node is not None: - metadata = base64.b64decode(metadata_node.text.encode('ascii')) + metadata = compat_b64decode(metadata_node.text) else: metadata = None fragments_list = build_fragments_list(boot_info) - if self.params.get('test', False): + test = self.params.get('test', False) + if test: # We only download the first fragment fragments_list = fragments_list[:1] total_frags = len(fragments_list) @@ -322,23 +365,28 @@ class F4mFD(FragmentFD): ctx = { 'filename': filename, 'total_frags': total_frags, + 'live': live, } self._prepare_frag_download(ctx) dest_stream = ctx['dest_stream'] - write_flv_header(dest_stream) - if not live: - write_metadata_tag(dest_stream, metadata) + if ctx['complete_frags_downloaded_bytes'] == 0: + write_flv_header(dest_stream) + if not live: + write_metadata_tag(dest_stream, metadata) base_url_parsed = compat_urllib_parse_urlparse(base_url) self._start_frag_download(ctx) - frags_filenames = [] + frag_index = 0 while fragments_list: seg_i, frag_i = fragments_list.pop(0) + frag_index += 1 + if frag_index <= ctx['fragment_index']: + continue name = 'Seg%d-Frag%d' % (seg_i, frag_i) query = [] if base_url_parsed.query: @@ -348,24 +396,26 @@ class F4mFD(FragmentFD): if info_dict.get('extra_param_to_segment_url'): query.append(info_dict['extra_param_to_segment_url']) url_parsed = base_url_parsed._replace(path=base_url_parsed.path + name, query='&'.join(query)) - frag_filename = '%s-%s' % (ctx['tmpfilename'], name) try: - success = ctx['dl'].download(frag_filename, {'url': url_parsed.geturl()}) + success, down_data = self._download_fragment(ctx, url_parsed.geturl(), info_dict) if not success: return False - (down, frag_sanitized) = sanitize_open(frag_filename, 'rb') - down_data = down.read() - down.close() reader = FlvReader(down_data) while True: - _, box_type, box_data = reader.read_box_info() + try: + _, box_type, box_data = reader.read_box_info() + except DataTruncatedError: + if test: + # In tests, segments may be truncated, and thus + # FlvReader may not be able to parse the whole + # chunk. If so, write the segment as is + # See https://github.com/ytdl-org/youtube-dl/issues/9214 + dest_stream.write(down_data) + break + raise if box_type == b'mdat': - dest_stream.write(box_data) + self._append_fragment(ctx, box_data) break - if live: - os.remove(encodeFilename(frag_sanitized)) - else: - frags_filenames.append(frag_sanitized) except (compat_urllib_error.HTTPError, ) as err: if live and (err.code == 404 or err.code == 410): # We didn't keep up with the live window. Continue @@ -376,7 +426,7 @@ class F4mFD(FragmentFD): else: raise - if not fragments_list and live and bootstrap_url: + if not fragments_list and not test and live and bootstrap_url: fragments_list = self._update_live_fragments(bootstrap_url, frag_i) total_frags += len(fragments_list) if fragments_list and (fragments_list[0][1] > frag_i + 1): @@ -385,7 +435,4 @@ class F4mFD(FragmentFD): self._finish_frag_download(ctx) - for frag_file in frags_filenames: - os.remove(encodeFilename(frag_file)) - return True diff --git a/youtube_dl/downloader/fragment.py b/youtube_dl/downloader/fragment.py index 5a64b29ee..913e91b64 100644 --- a/youtube_dl/downloader/fragment.py +++ b/youtube_dl/downloader/fragment.py @@ -2,12 +2,15 @@ from __future__ import division, unicode_literals import os import time +import json from .common import FileDownloader from .http import HttpFD from ..utils import ( + error_to_compat_str, encodeFilename, sanitize_open, + sanitized_Request, ) @@ -19,77 +22,243 @@ class HttpQuietDownloader(HttpFD): class FragmentFD(FileDownloader): """ A base file downloader class for fragmented media (e.g. f4m/m3u8 manifests). + + Available options: + + fragment_retries: Number of times to retry a fragment for HTTP error (DASH + and hlsnative only) + skip_unavailable_fragments: + Skip unavailable fragments (DASH and hlsnative only) + keep_fragments: Keep downloaded fragments on disk after downloading is + finished + + For each incomplete fragment download youtube-dl keeps on disk a special + bookkeeping file with download state and metadata (in future such files will + be used for any incomplete download handled by youtube-dl). This file is + used to properly handle resuming, check download file consistency and detect + potential errors. The file has a .ytdl extension and represents a standard + JSON file of the following format: + + extractor: + Dictionary of extractor related data. TBD. + + downloader: + Dictionary of downloader related data. May contain following data: + current_fragment: + Dictionary with current (being downloaded) fragment data: + index: 0-based index of current fragment among all fragments + fragment_count: + Total count of fragments + + This feature is experimental and file format may change in future. """ + def report_retry_fragment(self, err, frag_index, count, retries): + self.to_screen( + '[download] Got server HTTP error: %s. Retrying fragment %d (attempt %d of %s)...' + % (error_to_compat_str(err), frag_index, count, self.format_retries(retries))) + + def report_skip_fragment(self, frag_index): + self.to_screen('[download] Skipping fragment %d...' % frag_index) + + def _prepare_url(self, info_dict, url): + headers = info_dict.get('http_headers') + return sanitized_Request(url, None, headers) if headers else url + def _prepare_and_start_frag_download(self, ctx): self._prepare_frag_download(ctx) self._start_frag_download(ctx) + @staticmethod + def __do_ytdl_file(ctx): + return ctx['live'] is not True and ctx['tmpfilename'] != '-' + + def _read_ytdl_file(self, ctx): + assert 'ytdl_corrupt' not in ctx + stream, _ = sanitize_open(self.ytdl_filename(ctx['filename']), 'r') + try: + ctx['fragment_index'] = json.loads(stream.read())['downloader']['current_fragment']['index'] + except Exception: + ctx['ytdl_corrupt'] = True + finally: + stream.close() + + def _write_ytdl_file(self, ctx): + frag_index_stream, _ = sanitize_open(self.ytdl_filename(ctx['filename']), 'w') + downloader = { + 'current_fragment': { + 'index': ctx['fragment_index'], + }, + } + if ctx.get('fragment_count') is not None: + downloader['fragment_count'] = ctx['fragment_count'] + frag_index_stream.write(json.dumps({'downloader': downloader})) + frag_index_stream.close() + + def _download_fragment(self, ctx, frag_url, info_dict, headers=None): + fragment_filename = '%s-Frag%d' % (ctx['tmpfilename'], ctx['fragment_index']) + fragment_info_dict = { + 'url': frag_url, + 'http_headers': headers or info_dict.get('http_headers'), + } + frag_resume_len = 0 + if ctx['dl'].params.get('continuedl', True): + frag_resume_len = self.filesize_or_none( + self.temp_name(fragment_filename)) + fragment_info_dict['frag_resume_len'] = frag_resume_len + ctx['frag_resume_len'] = frag_resume_len or 0 + + success = ctx['dl'].download(fragment_filename, fragment_info_dict) + if not success: + return False, None + if fragment_info_dict.get('filetime'): + ctx['fragment_filetime'] = fragment_info_dict.get('filetime') + down, frag_sanitized = sanitize_open(fragment_filename, 'rb') + ctx['fragment_filename_sanitized'] = frag_sanitized + frag_content = down.read() + down.close() + return True, frag_content + + def _append_fragment(self, ctx, frag_content): + try: + ctx['dest_stream'].write(frag_content) + ctx['dest_stream'].flush() + finally: + if self.__do_ytdl_file(ctx): + self._write_ytdl_file(ctx) + if not self.params.get('keep_fragments', False): + os.remove(encodeFilename(ctx['fragment_filename_sanitized'])) + del ctx['fragment_filename_sanitized'] + def _prepare_frag_download(self, ctx): - self.to_screen('[%s] Total fragments: %d' % (self.FD_NAME, ctx['total_frags'])) + if not ctx.setdefault('live', False): + total_frags_str = '%d' % ctx['total_frags'] + ad_frags = ctx.get('ad_frags', 0) + if ad_frags: + total_frags_str += ' (not including %d ad)' % ad_frags + else: + total_frags_str = 'unknown (live)' + self.to_screen( + '[%s] Total fragments: %s' % (self.FD_NAME, total_frags_str)) self.report_destination(ctx['filename']) + continuedl = self.params.get('continuedl', True) dl = HttpQuietDownloader( self.ydl, { - 'continuedl': True, + 'continuedl': continuedl, 'quiet': True, 'noprogress': True, - 'ratelimit': self.params.get('ratelimit', None), + 'ratelimit': self.params.get('ratelimit'), 'retries': self.params.get('retries', 0), + 'nopart': self.params.get('nopart', False), 'test': self.params.get('test', False), } ) tmpfilename = self.temp_name(ctx['filename']) - dest_stream, tmpfilename = sanitize_open(tmpfilename, 'wb') + open_mode = 'wb' + + # Establish possible resume length + resume_len = self.filesize_or_none(tmpfilename) or 0 + if resume_len > 0: + open_mode = 'ab' + + # Should be initialized before ytdl file check + ctx.update({ + 'tmpfilename': tmpfilename, + 'fragment_index': 0, + }) + + if self.__do_ytdl_file(ctx): + ytdl_file_exists = os.path.isfile(encodeFilename(self.ytdl_filename(ctx['filename']))) + if continuedl and ytdl_file_exists: + self._read_ytdl_file(ctx) + is_corrupt = ctx.get('ytdl_corrupt') is True + is_inconsistent = ctx['fragment_index'] > 0 and resume_len == 0 + if is_corrupt or is_inconsistent: + message = ( + '.ytdl file is corrupt' if is_corrupt else + 'Inconsistent state of incomplete fragment download') + self.report_warning( + '%s. Restarting from the beginning...' % message) + ctx['fragment_index'] = resume_len = 0 + if 'ytdl_corrupt' in ctx: + del ctx['ytdl_corrupt'] + self._write_ytdl_file(ctx) + + else: + if not continuedl: + if ytdl_file_exists: + self._read_ytdl_file(ctx) + ctx['fragment_index'] = resume_len = 0 + self._write_ytdl_file(ctx) + assert ctx['fragment_index'] == 0 + + dest_stream, tmpfilename = sanitize_open(tmpfilename, open_mode) + ctx.update({ 'dl': dl, 'dest_stream': dest_stream, 'tmpfilename': tmpfilename, + # Total complete fragments downloaded so far in bytes + 'complete_frags_downloaded_bytes': resume_len, }) def _start_frag_download(self, ctx): + resume_len = ctx['complete_frags_downloaded_bytes'] total_frags = ctx['total_frags'] # This dict stores the download progress, it's updated by the progress # hook state = { 'status': 'downloading', - 'downloaded_bytes': 0, - 'frag_index': 0, - 'frag_count': total_frags, + 'downloaded_bytes': resume_len, + 'fragment_index': ctx['fragment_index'], + 'fragment_count': total_frags, 'filename': ctx['filename'], 'tmpfilename': ctx['tmpfilename'], } + start = time.time() - ctx['started'] = start + ctx.update({ + 'started': start, + 'fragment_started': start, + # Amount of fragment's bytes downloaded by the time of the previous + # frag progress hook invocation + 'prev_frag_downloaded_bytes': 0, + }) def frag_progress_hook(s): if s['status'] not in ('downloading', 'finished'): return - frag_total_bytes = s.get('total_bytes', 0) - if s['status'] == 'finished': - state['downloaded_bytes'] += frag_total_bytes - state['frag_index'] += 1 + if not total_frags and ctx.get('fragment_count'): + state['fragment_count'] = ctx['fragment_count'] - estimated_size = ( - (state['downloaded_bytes'] + frag_total_bytes) / - (state['frag_index'] + 1) * total_frags) time_now = time.time() - state['total_bytes_estimate'] = estimated_size state['elapsed'] = time_now - start + frag_total_bytes = s.get('total_bytes') or 0 + if not ctx['live']: + estimated_size = ( + (ctx['complete_frags_downloaded_bytes'] + frag_total_bytes) + / (state['fragment_index'] + 1) * total_frags) + state['total_bytes_estimate'] = estimated_size if s['status'] == 'finished': - progress = self.calc_percent(state['frag_index'], total_frags) + state['fragment_index'] += 1 + ctx['fragment_index'] = state['fragment_index'] + state['downloaded_bytes'] += frag_total_bytes - ctx['prev_frag_downloaded_bytes'] + ctx['complete_frags_downloaded_bytes'] = state['downloaded_bytes'] + ctx['speed'] = state['speed'] = self.calc_speed( + ctx['fragment_started'], time_now, frag_total_bytes) + ctx['fragment_started'] = time.time() + ctx['prev_frag_downloaded_bytes'] = 0 else: frag_downloaded_bytes = s['downloaded_bytes'] - frag_progress = self.calc_percent(frag_downloaded_bytes, - frag_total_bytes) - progress = self.calc_percent(state['frag_index'], total_frags) - progress += frag_progress / float(total_frags) - - state['eta'] = self.calc_eta( - start, time_now, estimated_size, state['downloaded_bytes'] + frag_downloaded_bytes) - state['speed'] = s.get('speed') + state['downloaded_bytes'] += frag_downloaded_bytes - ctx['prev_frag_downloaded_bytes'] + ctx['speed'] = state['speed'] = self.calc_speed( + ctx['fragment_started'], time_now, frag_downloaded_bytes - ctx['frag_resume_len']) + if not ctx['live']: + state['eta'] = self.calc_eta(state['speed'], estimated_size - state['downloaded_bytes']) + ctx['prev_frag_downloaded_bytes'] = frag_downloaded_bytes self._hook_progress(state) ctx['dl'].add_progress_hook(frag_progress_hook) @@ -98,13 +267,28 @@ class FragmentFD(FileDownloader): def _finish_frag_download(self, ctx): ctx['dest_stream'].close() + if self.__do_ytdl_file(ctx): + ytdl_filename = encodeFilename(self.ytdl_filename(ctx['filename'])) + if os.path.isfile(ytdl_filename): + os.remove(ytdl_filename) elapsed = time.time() - ctx['started'] - self.try_rename(ctx['tmpfilename'], ctx['filename']) - fsize = os.path.getsize(encodeFilename(ctx['filename'])) + + if ctx['tmpfilename'] == '-': + downloaded_bytes = ctx['complete_frags_downloaded_bytes'] + else: + self.try_rename(ctx['tmpfilename'], ctx['filename']) + if self.params.get('updatetime', True): + filetime = ctx.get('fragment_filetime') + if filetime: + try: + os.utime(ctx['filename'], (time.time(), filetime)) + except Exception: + pass + downloaded_bytes = self.filesize_or_none(ctx['filename']) or 0 self._hook_progress({ - 'downloaded_bytes': fsize, - 'total_bytes': fsize, + 'downloaded_bytes': downloaded_bytes, + 'total_bytes': downloaded_bytes, 'filename': ctx['filename'], 'status': 'finished', 'elapsed': elapsed, diff --git a/youtube_dl/downloader/hls.py b/youtube_dl/downloader/hls.py index 92765a3f9..7aaebc940 100644 --- a/youtube_dl/downloader/hls.py +++ b/youtube_dl/downloader/hls.py @@ -1,112 +1,216 @@ from __future__ import unicode_literals -import os import re -import subprocess +import binascii +try: + from Crypto.Cipher import AES + can_decrypt_frag = True +except ImportError: + can_decrypt_frag = False -from .common import FileDownloader from .fragment import FragmentFD +from .external import FFmpegFD -from ..compat import compat_urlparse -from ..postprocessor.ffmpeg import FFmpegPostProcessor +from ..compat import ( + compat_urllib_error, + compat_urlparse, + compat_struct_pack, +) from ..utils import ( - encodeArgument, - encodeFilename, - sanitize_open, + parse_m3u8_attributes, + update_url_query, ) -class HlsFD(FileDownloader): - def real_download(self, filename, info_dict): - url = info_dict['url'] - self.report_destination(filename) - tmpfilename = self.temp_name(filename) - - ffpp = FFmpegPostProcessor(downloader=self) - if not ffpp.available: - self.report_error('m3u8 download detected but ffmpeg or avconv could not be found. Please install one.') - return False - ffpp.check_version() - - args = [ffpp.executable, '-y'] - - if info_dict['http_headers'] and re.match(r'^https?://', url): - # Trailing \r\n after each HTTP header is important to prevent warning from ffmpeg/avconv: - # [http @ 00000000003d2fa0] No trailing CRLF found in HTTP header. - args += [ - '-headers', - ''.join('%s: %s\r\n' % (key, val) for key, val in info_dict['http_headers'].items() if key.lower() != 'accept-encoding')] - - args += ['-i', url, '-f', 'mp4', '-c', 'copy', '-bsf:a', 'aac_adtstoasc'] - - args = [encodeArgument(opt) for opt in args] - args.append(encodeFilename(ffpp._ffmpeg_filename_argument(tmpfilename), True)) - - self._debug_cmd(args) - - retval = subprocess.call(args) - if retval == 0: - fsize = os.path.getsize(encodeFilename(tmpfilename)) - self.to_screen('\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('\n') - self.report_error('%s exited with code %d' % (ffpp.basename, retval)) - return False - - -class NativeHlsFD(FragmentFD): - """ A more limited implementation that does not require ffmpeg """ +class HlsFD(FragmentFD): + """ A limited implementation that does not require ffmpeg """ FD_NAME = 'hlsnative' + @staticmethod + def can_download(manifest, info_dict): + UNSUPPORTED_FEATURES = ( + r'#EXT-X-KEY:METHOD=(?!NONE|AES-128)', # encrypted streams [1] + # r'#EXT-X-BYTERANGE', # playlists composed of byte ranges of media files [2] + + # Live streams heuristic does not always work (e.g. geo restricted to Germany + # http://hls-geo.daserste.de/i/videoportal/Film/c_620000/622873/format,716451,716457,716450,716458,716459,.mp4.csmil/index_4_av.m3u8?null=0) + # r'#EXT-X-MEDIA-SEQUENCE:(?!0$)', # live streams [3] + + # This heuristic also is not correct since segments may not be appended as well. + # Twitch vods of finished streams have EXT-X-PLAYLIST-TYPE:EVENT despite + # no segments will definitely be appended to the end of the playlist. + # r'#EXT-X-PLAYLIST-TYPE:EVENT', # media segments may be appended to the end of + # # event media playlists [4] + r'#EXT-X-MAP:', # media initialization [5] + + # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.4 + # 2. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.2 + # 3. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.2 + # 4. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.5 + # 5. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.5 + ) + check_results = [not re.search(feature, manifest) for feature in UNSUPPORTED_FEATURES] + is_aes128_enc = '#EXT-X-KEY:METHOD=AES-128' in manifest + check_results.append(can_decrypt_frag or not is_aes128_enc) + check_results.append(not (is_aes128_enc and r'#EXT-X-BYTERANGE' in manifest)) + check_results.append(not info_dict.get('is_live')) + return all(check_results) + def real_download(self, filename, info_dict): man_url = info_dict['url'] self.to_screen('[%s] Downloading m3u8 manifest' % self.FD_NAME) - manifest = self.ydl.urlopen(man_url).read() - s = manifest.decode('utf-8', 'ignore') - fragment_urls = [] + urlh = self.ydl.urlopen(self._prepare_url(info_dict, man_url)) + man_url = urlh.geturl() + s = urlh.read().decode('utf-8', 'ignore') + + if not self.can_download(s, info_dict): + if info_dict.get('extra_param_to_segment_url') or info_dict.get('_decryption_key_url'): + self.report_error('pycrypto not found. Please install it.') + return False + self.report_warning( + 'hlsnative has detected features it does not support, ' + 'extraction will be delegated to ffmpeg') + fd = FFmpegFD(self.ydl, self.params) + for ph in self._progress_hooks: + fd.add_progress_hook(ph) + return fd.real_download(filename, info_dict) + + def is_ad_fragment_start(s): + return (s.startswith('#ANVATO-SEGMENT-INFO') and 'type=ad' in s + or s.startswith('#UPLYNK-SEGMENT') and s.endswith(',ad')) + + def is_ad_fragment_end(s): + return (s.startswith('#ANVATO-SEGMENT-INFO') and 'type=master' in s + or s.startswith('#UPLYNK-SEGMENT') and s.endswith(',segment')) + + media_frags = 0 + ad_frags = 0 + ad_frag_next = False for line in s.splitlines(): line = line.strip() - if line and not line.startswith('#'): - segment_url = ( - line - if re.match(r'^https?://', line) - else compat_urlparse.urljoin(man_url, line)) - fragment_urls.append(segment_url) - # We only download the first fragment during the test - if self.params.get('test', False): - break + if not line: + continue + if line.startswith('#'): + if is_ad_fragment_start(line): + ad_frag_next = True + elif is_ad_fragment_end(line): + ad_frag_next = False + continue + if ad_frag_next: + ad_frags += 1 + continue + media_frags += 1 ctx = { 'filename': filename, - 'total_frags': len(fragment_urls), + 'total_frags': media_frags, + 'ad_frags': ad_frags, } self._prepare_and_start_frag_download(ctx) - frags_filenames = [] - for i, frag_url in enumerate(fragment_urls): - frag_filename = '%s-Frag%d' % (ctx['tmpfilename'], i) - success = ctx['dl'].download(frag_filename, {'url': frag_url}) - if not success: - return False - down, frag_sanitized = sanitize_open(frag_filename, 'rb') - ctx['dest_stream'].write(down.read()) - down.close() - frags_filenames.append(frag_sanitized) + fragment_retries = self.params.get('fragment_retries', 0) + skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True) + test = self.params.get('test', False) + + extra_query = None + extra_param_to_segment_url = info_dict.get('extra_param_to_segment_url') + if extra_param_to_segment_url: + extra_query = compat_urlparse.parse_qs(extra_param_to_segment_url) + i = 0 + media_sequence = 0 + decrypt_info = {'METHOD': 'NONE'} + byte_range = {} + frag_index = 0 + ad_frag_next = False + for line in s.splitlines(): + line = line.strip() + if line: + if not line.startswith('#'): + if ad_frag_next: + continue + frag_index += 1 + if frag_index <= ctx['fragment_index']: + continue + frag_url = ( + line + if re.match(r'^https?://', line) + else compat_urlparse.urljoin(man_url, line)) + if extra_query: + frag_url = update_url_query(frag_url, extra_query) + count = 0 + headers = info_dict.get('http_headers', {}) + if byte_range: + headers['Range'] = 'bytes=%d-%d' % (byte_range['start'], byte_range['end'] - 1) + while count <= fragment_retries: + try: + success, frag_content = self._download_fragment( + ctx, frag_url, info_dict, headers) + if not success: + return False + break + except compat_urllib_error.HTTPError as err: + # Unavailable (possibly temporary) fragments may be served. + # First we try to retry then either skip or abort. + # See https://github.com/ytdl-org/youtube-dl/issues/10165, + # https://github.com/ytdl-org/youtube-dl/issues/10448). + count += 1 + if count <= fragment_retries: + self.report_retry_fragment(err, frag_index, count, fragment_retries) + if count > fragment_retries: + if skip_unavailable_fragments: + i += 1 + media_sequence += 1 + self.report_skip_fragment(frag_index) + continue + self.report_error( + 'giving up after %s fragment retries' % fragment_retries) + return False + if decrypt_info['METHOD'] == 'AES-128': + iv = decrypt_info.get('IV') or compat_struct_pack('>8xq', media_sequence) + decrypt_info['KEY'] = decrypt_info.get('KEY') or self.ydl.urlopen( + self._prepare_url(info_dict, info_dict.get('_decryption_key_url') or decrypt_info['URI'])).read() + # Don't decrypt the content in tests since the data is explicitly truncated and it's not to a valid block + # size (see https://github.com/ytdl-org/youtube-dl/pull/27660). Tests only care that the correct data downloaded, + # not what it decrypts to. + if not test: + frag_content = AES.new( + decrypt_info['KEY'], AES.MODE_CBC, iv).decrypt(frag_content) + self._append_fragment(ctx, frag_content) + # We only download the first fragment during the test + if test: + break + i += 1 + media_sequence += 1 + elif line.startswith('#EXT-X-KEY'): + decrypt_url = decrypt_info.get('URI') + decrypt_info = parse_m3u8_attributes(line[11:]) + if decrypt_info['METHOD'] == 'AES-128': + if 'IV' in decrypt_info: + decrypt_info['IV'] = binascii.unhexlify(decrypt_info['IV'][2:].zfill(32)) + if not re.match(r'^https?://', decrypt_info['URI']): + decrypt_info['URI'] = compat_urlparse.urljoin( + man_url, decrypt_info['URI']) + if extra_query: + decrypt_info['URI'] = update_url_query(decrypt_info['URI'], extra_query) + if decrypt_url != decrypt_info['URI']: + decrypt_info['KEY'] = None + elif line.startswith('#EXT-X-MEDIA-SEQUENCE'): + media_sequence = int(line[22:]) + elif line.startswith('#EXT-X-BYTERANGE'): + splitted_byte_range = line[17:].split('@') + sub_range_start = int(splitted_byte_range[1]) if len(splitted_byte_range) == 2 else byte_range['end'] + byte_range = { + 'start': sub_range_start, + 'end': sub_range_start + int(splitted_byte_range[0]), + } + elif is_ad_fragment_start(line): + ad_frag_next = True + elif is_ad_fragment_end(line): + ad_frag_next = False self._finish_frag_download(ctx) - for frag_file in frags_filenames: - os.remove(encodeFilename(frag_file)) - return True diff --git a/youtube_dl/downloader/http.py b/youtube_dl/downloader/http.py index 56840e026..3cad87420 100644 --- a/youtube_dl/downloader/http.py +++ b/youtube_dl/downloader/http.py @@ -4,94 +4,164 @@ import errno import os import socket import time +import random import re from .common import FileDownloader -from ..compat import compat_urllib_error +from ..compat import ( + compat_str, + compat_urllib_error, +) from ..utils import ( ContentTooShortError, encodeFilename, + int_or_none, sanitize_open, sanitized_Request, + write_xattr, + XAttrMetadataError, + XAttrUnavailableError, ) class HttpFD(FileDownloader): def real_download(self, filename, info_dict): url = info_dict['url'] - tmpfilename = self.temp_name(filename) - stream = None + + class DownloadContext(dict): + __getattr__ = dict.get + __setattr__ = dict.__setitem__ + __delattr__ = dict.__delitem__ + + ctx = DownloadContext() + ctx.filename = filename + ctx.tmpfilename = self.temp_name(filename) + ctx.stream = None # Do not include the Accept-Encoding header headers = {'Youtubedl-no-compression': 'True'} add_headers = info_dict.get('http_headers') if add_headers: headers.update(add_headers) - basic_request = sanitized_Request(url, None, headers) - request = sanitized_Request(url, None, headers) is_test = self.params.get('test', False) + chunk_size = self._TEST_FILE_SIZE if is_test else ( + info_dict.get('downloader_options', {}).get('http_chunk_size') + or self.params.get('http_chunk_size') or 0) - if is_test: - request.add_header('Range', 'bytes=0-%s' % str(self._TEST_FILE_SIZE - 1)) - - # 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', True): - self.report_resuming_byte(resume_len) - request.add_header('Range', 'bytes=%d-' % resume_len) - open_mode = 'ab' - else: - resume_len = 0 + ctx.open_mode = 'wb' + ctx.resume_len = 0 + ctx.data_len = None + ctx.block_size = self.params.get('buffersize', 1024) + ctx.start_time = time.time() + ctx.chunk_size = None + + if self.params.get('continuedl', True): + # Establish possible resume length + ctx.resume_len = info_dict.get('frag_resume_len') + if ctx.resume_len is None: + ctx.resume_len = self.filesize_or_none(ctx.tmpfilename) or 0 + + ctx.is_resume = ctx.resume_len > 0 count = 0 retries = self.params.get('retries', 0) - while count <= retries: + + class SucceedDownload(Exception): + pass + + class RetryDownload(Exception): + def __init__(self, source_error): + self.source_error = source_error + + class NextFragment(Exception): + pass + + def set_range(req, start, end): + range_header = 'bytes=%d-' % start + if end: + range_header += compat_str(end) + req.add_header('Range', range_header) + + def establish_connection(): + ctx.chunk_size = (random.randint(int(chunk_size * 0.95), chunk_size) + if not is_test and chunk_size else chunk_size) + if ctx.resume_len > 0: + range_start = ctx.resume_len + if ctx.is_resume: + self.report_resuming_byte(ctx.resume_len) + ctx.open_mode = 'ab' + elif ctx.chunk_size > 0: + range_start = 0 + else: + range_start = None + ctx.is_resume = False + range_end = range_start + ctx.chunk_size - 1 if ctx.chunk_size else None + if range_end and ctx.data_len is not None and range_end >= ctx.data_len: + range_end = ctx.data_len - 1 + has_range = range_start is not None + ctx.has_range = has_range + request = sanitized_Request(url, None, headers) + if has_range: + set_range(request, range_start, range_end) # Establish connection try: - data = self.ydl.urlopen(request) + try: + ctx.data = self.ydl.urlopen(request) + except (compat_urllib_error.URLError, ) as err: + # reason may not be available, e.g. for urllib2.HTTPError on python 2.6 + reason = getattr(err, 'reason', None) + if isinstance(reason, socket.timeout): + raise RetryDownload(err) + raise err # When trying to resume, Content-Range HTTP header of response has to be checked - # to match the value of requested Range HTTP header. This is due to a webservers + # to match the value of requested Range HTTP header. This is due to webservers # that don't support resuming and serve a whole file with no Content-Range - # set in response despite of requested Range (see - # https://github.com/rg3/youtube-dl/issues/6057#issuecomment-126129799) - if resume_len > 0: - content_range = data.headers.get('Content-Range') + # set in response despite requested Range (see + # https://github.com/ytdl-org/youtube-dl/issues/6057#issuecomment-126129799) + if has_range: + content_range = ctx.data.headers.get('Content-Range') if content_range: - content_range_m = re.search(r'bytes (\d+)-', content_range) + content_range_m = re.search(r'bytes (\d+)-(\d+)?(?:/(\d+))?', content_range) # Content-Range is present and matches requested Range, resume is possible - if content_range_m and resume_len == int(content_range_m.group(1)): - break + if content_range_m: + if range_start == int(content_range_m.group(1)): + content_range_end = int_or_none(content_range_m.group(2)) + content_len = int_or_none(content_range_m.group(3)) + accept_content_len = ( + # Non-chunked download + not ctx.chunk_size + # Chunked download and requested piece or + # its part is promised to be served + or content_range_end == range_end + or content_len < range_end) + if accept_content_len: + ctx.data_len = content_len + return # Content-Range is either not present or invalid. Assuming remote webserver is # trying to send the whole file, resume is not possible, so wiping the local file # and performing entire redownload - self.report_unable_to_resume() - resume_len = 0 - open_mode = 'wb' - break + if range_start > 0: + self.report_unable_to_resume() + ctx.resume_len = 0 + ctx.open_mode = 'wb' + ctx.data_len = int_or_none(ctx.data.info().get('Content-length', None)) + return 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: + if err.code == 416: # Unable to resume (requested range not satisfiable) try: # Open the connection again without the range header - data = self.ydl.urlopen(basic_request) - content_length = data.info()['Content-Length'] + ctx.data = self.ydl.urlopen( + sanitized_Request(url, None, headers)) + content_length = ctx.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)): + if (content_length is not None + and (ctx.resume_len - 100 < int(content_length) < ctx.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, @@ -99,153 +169,194 @@ class HttpFD(FileDownloader): # 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.report_file_already_downloaded(ctx.filename) + self.try_rename(ctx.tmpfilename, ctx.filename) self._hook_progress({ - 'filename': filename, + 'filename': ctx.filename, 'status': 'finished', - 'downloaded_bytes': resume_len, - 'total_bytes': resume_len, + 'downloaded_bytes': ctx.resume_len, + 'total_bytes': ctx.resume_len, }) - return True + raise SucceedDownload() else: # The length does not match, we start the download over self.report_unable_to_resume() - resume_len = 0 - open_mode = 'wb' - break - except socket.error as e: - if e.errno != errno.ECONNRESET: + ctx.resume_len = 0 + ctx.open_mode = 'wb' + return + elif err.code < 500 or err.code >= 600: + # Unexpected HTTP error + raise + raise RetryDownload(err) + except socket.error as err: + if err.errno != errno.ECONNRESET: # Connection reset is no problem, just retry raise + raise RetryDownload(err) + + def download(): + data_len = ctx.data.info().get('Content-length', None) + + # Range HTTP header may be ignored/unsupported by a webserver + # (e.g. extractor/scivee.py, extractor/bambuser.py). + # However, for a test we still would like to download just a piece of a file. + # To achieve this we limit data_len to _TEST_FILE_SIZE and manually control + # block size when downloading a file. + if is_test and (data_len is None or int(data_len) > self._TEST_FILE_SIZE): + data_len = self._TEST_FILE_SIZE + + if data_len is not None: + data_len = int(data_len) + ctx.resume_len + min_data_len = self.params.get('min_filesize') + max_data_len = self.params.get('max_filesize') + if min_data_len is not None and data_len < min_data_len: + self.to_screen('\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('\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len, max_data_len)) + return False - # Retry - count += 1 - if count <= retries: - self.report_retry(count, retries) - - if count > retries: - self.report_error('giving up after %s retries' % retries) - return False - - data_len = data.info().get('Content-length', None) - - # Range HTTP header may be ignored/unsupported by a webserver - # (e.g. extractor/scivee.py, extractor/bambuser.py). - # However, for a test we still would like to download just a piece of a file. - # To achieve this we limit data_len to _TEST_FILE_SIZE and manually control - # block size when downloading a file. - if is_test and (data_len is None or int(data_len) > self._TEST_FILE_SIZE): - data_len = self._TEST_FILE_SIZE - - 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('\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('\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len, max_data_len)) - return False + byte_counter = 0 + ctx.resume_len + block_size = ctx.block_size + start = time.time() + + # measure time over whole while-loop, so slow_down() and best_block_size() work together properly + now = None # needed for slow_down() in the first loop run + before = start # start measuring - byte_counter = 0 + resume_len - block_size = self.params.get('buffersize', 1024) - start = time.time() + def retry(e): + to_stdout = ctx.tmpfilename == '-' + if ctx.stream is not None: + if not to_stdout: + ctx.stream.close() + ctx.stream = None + ctx.resume_len = byte_counter if to_stdout else os.path.getsize(encodeFilename(ctx.tmpfilename)) + raise RetryDownload(e) - # measure time over whole while-loop, so slow_down() and best_block_size() work together properly - now = None # needed for slow_down() in the first loop run - before = start # start measuring - while True: + while True: + try: + # Download and write + data_block = ctx.data.read(block_size if data_len is None else min(block_size, data_len - byte_counter)) + # socket.timeout is a subclass of socket.error but may not have + # errno set + except socket.timeout as e: + retry(e) + except socket.error as e: + # SSLError on python 2 (inherits socket.error) may have + # no errno set but this error message + if e.errno in (errno.ECONNRESET, errno.ETIMEDOUT) or getattr(e, 'message', None) == 'The read operation timed out': + retry(e) + raise - # Download and write - data_block = data.read(block_size if not is_test else min(block_size, data_len - byte_counter)) - byte_counter += len(data_block) + byte_counter += len(data_block) - # exit loop when download is finished - if len(data_block) == 0: - break + # exit loop when download is finished + if len(data_block) == 0: + break + + # Open destination file just in time + if ctx.stream is None: + try: + ctx.stream, ctx.tmpfilename = sanitize_open( + ctx.tmpfilename, ctx.open_mode) + assert ctx.stream is not None + ctx.filename = self.undo_temp_name(ctx.tmpfilename) + self.report_destination(ctx.filename) + except (OSError, IOError) as err: + self.report_error('unable to open for writing: %s' % str(err)) + return False + + if self.params.get('xattr_set_filesize', False) and data_len is not None: + try: + write_xattr(ctx.tmpfilename, 'user.ytdl.filesize', str(data_len).encode('utf-8')) + except (XAttrUnavailableError, XAttrMetadataError) as err: + self.report_error('unable to set filesize xattr: %s' % str(err)) - # Open destination 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('unable to open for writing: %s' % str(err)) + ctx.stream.write(data_block) + except (IOError, OSError) as err: + self.to_stderr('\n') + self.report_error('unable to write data: %s' % str(err)) return False - if self.params.get('xattr_set_filesize', False) and data_len is not None: - try: - import xattr - xattr.setxattr(tmpfilename, 'user.ytdl.filesize', str(data_len)) - except(OSError, IOError, ImportError) as err: - self.report_error('unable to set filesize xattr: %s' % str(err)) + # Apply rate limit + self.slow_down(start, now, byte_counter - ctx.resume_len) - try: - stream.write(data_block) - except (IOError, OSError) as err: - self.to_stderr('\n') - self.report_error('unable to write data: %s' % str(err)) - return False + # end measuring of one loop run + now = time.time() + after = now - # Apply rate limit - self.slow_down(start, now, byte_counter - resume_len) + # Adjust block size + if not self.params.get('noresizebuffer', False): + block_size = self.best_block_size(after - before, len(data_block)) - # end measuring of one loop run - now = time.time() - after = now + before = after - # Adjust block size - if not self.params.get('noresizebuffer', False): - block_size = self.best_block_size(after - before, len(data_block)) + # Progress message + speed = self.calc_speed(start, now, byte_counter - ctx.resume_len) + eta = self.calc_eta(speed, ctx.data_len and (ctx.data_len - byte_counter)) - before = after + self._hook_progress({ + 'status': 'downloading', + 'downloaded_bytes': byte_counter, + 'total_bytes': ctx.data_len, + 'tmpfilename': ctx.tmpfilename, + 'filename': ctx.filename, + 'eta': eta, + 'speed': speed, + 'elapsed': now - ctx.start_time, + }) - # Progress message - speed = self.calc_speed(start, now, byte_counter - resume_len) - if data_len is None: - eta = None - else: - eta = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len) + if data_len is not None and byte_counter == data_len: + break + + if not is_test and ctx.chunk_size and ctx.data_len is not None and byte_counter < ctx.data_len: + ctx.resume_len = byte_counter + # ctx.block_size = block_size + raise NextFragment() + + if ctx.stream is None: + self.to_stderr('\n') + self.report_error('Did not get any data blocks') + return False + if ctx.tmpfilename != '-': + ctx.stream.close() + + if data_len is not None and byte_counter != data_len: + err = ContentTooShortError(byte_counter, int(data_len)) + if count <= retries: + retry(err) + raise err + + self.try_rename(ctx.tmpfilename, ctx.filename) + + # Update file modification time + if self.params.get('updatetime', True): + info_dict['filetime'] = self.try_utime(ctx.filename, ctx.data.info().get('last-modified', None)) self._hook_progress({ - 'status': 'downloading', 'downloaded_bytes': byte_counter, - 'total_bytes': data_len, - 'tmpfilename': tmpfilename, - 'filename': filename, - 'eta': eta, - 'speed': speed, - 'elapsed': now - start, + 'total_bytes': byte_counter, + 'filename': ctx.filename, + 'status': 'finished', + 'elapsed': time.time() - ctx.start_time, }) - if is_test and byte_counter == data_len: - break - - if stream is None: - self.to_stderr('\n') - self.report_error('Did not get any data blocks') - return False - if tmpfilename != '-': - stream.close() - - 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', - 'elapsed': time.time() - start, - }) - - return True + return True + + while count <= retries: + try: + establish_connection() + return download() + except RetryDownload as e: + count += 1 + if count <= retries: + self.report_retry(e.source_error, count, retries) + continue + except NextFragment: + continue + except SucceedDownload: + return True + + self.report_error('giving up after %s retries' % retries) + return False diff --git a/youtube_dl/downloader/ism.py b/youtube_dl/downloader/ism.py new file mode 100644 index 000000000..1ca666b4a --- /dev/null +++ b/youtube_dl/downloader/ism.py @@ -0,0 +1,259 @@ +from __future__ import unicode_literals + +import time +import binascii +import io + +from .fragment import FragmentFD +from ..compat import ( + compat_Struct, + compat_urllib_error, +) + + +u8 = compat_Struct('>B') +u88 = compat_Struct('>Bx') +u16 = compat_Struct('>H') +u1616 = compat_Struct('>Hxx') +u32 = compat_Struct('>I') +u64 = compat_Struct('>Q') + +s88 = compat_Struct('>bx') +s16 = compat_Struct('>h') +s1616 = compat_Struct('>hxx') +s32 = compat_Struct('>i') + +unity_matrix = (s32.pack(0x10000) + s32.pack(0) * 3) * 2 + s32.pack(0x40000000) + +TRACK_ENABLED = 0x1 +TRACK_IN_MOVIE = 0x2 +TRACK_IN_PREVIEW = 0x4 + +SELF_CONTAINED = 0x1 + + +def box(box_type, payload): + return u32.pack(8 + len(payload)) + box_type + payload + + +def full_box(box_type, version, flags, payload): + return box(box_type, u8.pack(version) + u32.pack(flags)[1:] + payload) + + +def write_piff_header(stream, params): + track_id = params['track_id'] + fourcc = params['fourcc'] + duration = params['duration'] + timescale = params.get('timescale', 10000000) + language = params.get('language', 'und') + height = params.get('height', 0) + width = params.get('width', 0) + is_audio = width == 0 and height == 0 + creation_time = modification_time = int(time.time()) + + ftyp_payload = b'isml' # major brand + ftyp_payload += u32.pack(1) # minor version + ftyp_payload += b'piff' + b'iso2' # compatible brands + stream.write(box(b'ftyp', ftyp_payload)) # File Type Box + + mvhd_payload = u64.pack(creation_time) + mvhd_payload += u64.pack(modification_time) + mvhd_payload += u32.pack(timescale) + mvhd_payload += u64.pack(duration) + mvhd_payload += s1616.pack(1) # rate + mvhd_payload += s88.pack(1) # volume + mvhd_payload += u16.pack(0) # reserved + mvhd_payload += u32.pack(0) * 2 # reserved + mvhd_payload += unity_matrix + mvhd_payload += u32.pack(0) * 6 # pre defined + mvhd_payload += u32.pack(0xffffffff) # next track id + moov_payload = full_box(b'mvhd', 1, 0, mvhd_payload) # Movie Header Box + + tkhd_payload = u64.pack(creation_time) + tkhd_payload += u64.pack(modification_time) + tkhd_payload += u32.pack(track_id) # track id + tkhd_payload += u32.pack(0) # reserved + tkhd_payload += u64.pack(duration) + tkhd_payload += u32.pack(0) * 2 # reserved + tkhd_payload += s16.pack(0) # layer + tkhd_payload += s16.pack(0) # alternate group + tkhd_payload += s88.pack(1 if is_audio else 0) # volume + tkhd_payload += u16.pack(0) # reserved + tkhd_payload += unity_matrix + tkhd_payload += u1616.pack(width) + tkhd_payload += u1616.pack(height) + trak_payload = full_box(b'tkhd', 1, TRACK_ENABLED | TRACK_IN_MOVIE | TRACK_IN_PREVIEW, tkhd_payload) # Track Header Box + + mdhd_payload = u64.pack(creation_time) + mdhd_payload += u64.pack(modification_time) + mdhd_payload += u32.pack(timescale) + mdhd_payload += u64.pack(duration) + mdhd_payload += u16.pack(((ord(language[0]) - 0x60) << 10) | ((ord(language[1]) - 0x60) << 5) | (ord(language[2]) - 0x60)) + mdhd_payload += u16.pack(0) # pre defined + mdia_payload = full_box(b'mdhd', 1, 0, mdhd_payload) # Media Header Box + + hdlr_payload = u32.pack(0) # pre defined + hdlr_payload += b'soun' if is_audio else b'vide' # handler type + hdlr_payload += u32.pack(0) * 3 # reserved + hdlr_payload += (b'Sound' if is_audio else b'Video') + b'Handler\0' # name + mdia_payload += full_box(b'hdlr', 0, 0, hdlr_payload) # Handler Reference Box + + if is_audio: + smhd_payload = s88.pack(0) # balance + smhd_payload += u16.pack(0) # reserved + media_header_box = full_box(b'smhd', 0, 0, smhd_payload) # Sound Media Header + else: + vmhd_payload = u16.pack(0) # graphics mode + vmhd_payload += u16.pack(0) * 3 # opcolor + media_header_box = full_box(b'vmhd', 0, 1, vmhd_payload) # Video Media Header + minf_payload = media_header_box + + dref_payload = u32.pack(1) # entry count + dref_payload += full_box(b'url ', 0, SELF_CONTAINED, b'') # Data Entry URL Box + dinf_payload = full_box(b'dref', 0, 0, dref_payload) # Data Reference Box + minf_payload += box(b'dinf', dinf_payload) # Data Information Box + + stsd_payload = u32.pack(1) # entry count + + sample_entry_payload = u8.pack(0) * 6 # reserved + sample_entry_payload += u16.pack(1) # data reference index + if is_audio: + sample_entry_payload += u32.pack(0) * 2 # reserved + sample_entry_payload += u16.pack(params.get('channels', 2)) + sample_entry_payload += u16.pack(params.get('bits_per_sample', 16)) + sample_entry_payload += u16.pack(0) # pre defined + sample_entry_payload += u16.pack(0) # reserved + sample_entry_payload += u1616.pack(params['sampling_rate']) + + if fourcc == 'AACL': + sample_entry_box = box(b'mp4a', sample_entry_payload) + else: + sample_entry_payload += u16.pack(0) # pre defined + sample_entry_payload += u16.pack(0) # reserved + sample_entry_payload += u32.pack(0) * 3 # pre defined + sample_entry_payload += u16.pack(width) + sample_entry_payload += u16.pack(height) + sample_entry_payload += u1616.pack(0x48) # horiz resolution 72 dpi + sample_entry_payload += u1616.pack(0x48) # vert resolution 72 dpi + sample_entry_payload += u32.pack(0) # reserved + sample_entry_payload += u16.pack(1) # frame count + sample_entry_payload += u8.pack(0) * 32 # compressor name + sample_entry_payload += u16.pack(0x18) # depth + sample_entry_payload += s16.pack(-1) # pre defined + + codec_private_data = binascii.unhexlify(params['codec_private_data'].encode('utf-8')) + if fourcc in ('H264', 'AVC1'): + sps, pps = codec_private_data.split(u32.pack(1))[1:] + avcc_payload = u8.pack(1) # configuration version + avcc_payload += sps[1:4] # avc profile indication + profile compatibility + avc level indication + avcc_payload += u8.pack(0xfc | (params.get('nal_unit_length_field', 4) - 1)) # complete representation (1) + reserved (11111) + length size minus one + avcc_payload += u8.pack(1) # reserved (0) + number of sps (0000001) + avcc_payload += u16.pack(len(sps)) + avcc_payload += sps + avcc_payload += u8.pack(1) # number of pps + avcc_payload += u16.pack(len(pps)) + avcc_payload += pps + sample_entry_payload += box(b'avcC', avcc_payload) # AVC Decoder Configuration Record + sample_entry_box = box(b'avc1', sample_entry_payload) # AVC Simple Entry + stsd_payload += sample_entry_box + + stbl_payload = full_box(b'stsd', 0, 0, stsd_payload) # Sample Description Box + + stts_payload = u32.pack(0) # entry count + stbl_payload += full_box(b'stts', 0, 0, stts_payload) # Decoding Time to Sample Box + + stsc_payload = u32.pack(0) # entry count + stbl_payload += full_box(b'stsc', 0, 0, stsc_payload) # Sample To Chunk Box + + stco_payload = u32.pack(0) # entry count + stbl_payload += full_box(b'stco', 0, 0, stco_payload) # Chunk Offset Box + + minf_payload += box(b'stbl', stbl_payload) # Sample Table Box + + mdia_payload += box(b'minf', minf_payload) # Media Information Box + + trak_payload += box(b'mdia', mdia_payload) # Media Box + + moov_payload += box(b'trak', trak_payload) # Track Box + + mehd_payload = u64.pack(duration) + mvex_payload = full_box(b'mehd', 1, 0, mehd_payload) # Movie Extends Header Box + + trex_payload = u32.pack(track_id) # track id + trex_payload += u32.pack(1) # default sample description index + trex_payload += u32.pack(0) # default sample duration + trex_payload += u32.pack(0) # default sample size + trex_payload += u32.pack(0) # default sample flags + mvex_payload += full_box(b'trex', 0, 0, trex_payload) # Track Extends Box + + moov_payload += box(b'mvex', mvex_payload) # Movie Extends Box + stream.write(box(b'moov', moov_payload)) # Movie Box + + +def extract_box_data(data, box_sequence): + data_reader = io.BytesIO(data) + while True: + box_size = u32.unpack(data_reader.read(4))[0] + box_type = data_reader.read(4) + if box_type == box_sequence[0]: + box_data = data_reader.read(box_size - 8) + if len(box_sequence) == 1: + return box_data + return extract_box_data(box_data, box_sequence[1:]) + data_reader.seek(box_size - 8, 1) + + +class IsmFD(FragmentFD): + """ + Download segments in a ISM manifest + """ + + FD_NAME = 'ism' + + def real_download(self, filename, info_dict): + segments = info_dict['fragments'][:1] if self.params.get( + 'test', False) else info_dict['fragments'] + + ctx = { + 'filename': filename, + 'total_frags': len(segments), + } + + self._prepare_and_start_frag_download(ctx) + + fragment_retries = self.params.get('fragment_retries', 0) + skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True) + + track_written = False + frag_index = 0 + for i, segment in enumerate(segments): + frag_index += 1 + if frag_index <= ctx['fragment_index']: + continue + count = 0 + while count <= fragment_retries: + try: + success, frag_content = self._download_fragment(ctx, segment['url'], info_dict) + if not success: + return False + if not track_written: + tfhd_data = extract_box_data(frag_content, [b'moof', b'traf', b'tfhd']) + info_dict['_download_params']['track_id'] = u32.unpack(tfhd_data[4:8])[0] + write_piff_header(ctx['dest_stream'], info_dict['_download_params']) + track_written = True + self._append_fragment(ctx, frag_content) + break + except compat_urllib_error.HTTPError as err: + count += 1 + if count <= fragment_retries: + self.report_retry_fragment(err, frag_index, count, fragment_retries) + if count > fragment_retries: + if skip_unavailable_fragments: + self.report_skip_fragment(frag_index) + continue + self.report_error('giving up after %s fragment retries' % fragment_retries) + return False + + self._finish_frag_download(ctx) + + return True diff --git a/youtube_dl/downloader/niconico.py b/youtube_dl/downloader/niconico.py new file mode 100644 index 000000000..6392c9989 --- /dev/null +++ b/youtube_dl/downloader/niconico.py @@ -0,0 +1,66 @@ +# coding: utf-8 +from __future__ import unicode_literals + +try: + import threading +except ImportError: + threading = None + +from .common import FileDownloader +from ..downloader import get_suitable_downloader +from ..extractor.niconico import NiconicoIE +from ..utils import sanitized_Request + + +class NiconicoDmcFD(FileDownloader): + """ Downloading niconico douga from DMC with heartbeat """ + + FD_NAME = 'niconico_dmc' + + def real_download(self, filename, info_dict): + self.to_screen('[%s] Downloading from DMC' % self.FD_NAME) + + ie = NiconicoIE(self.ydl) + info_dict, heartbeat_info_dict = ie._get_heartbeat_info(info_dict) + + fd = get_suitable_downloader(info_dict, params=self.params)(self.ydl, self.params) + for ph in self._progress_hooks: + fd.add_progress_hook(ph) + + if not threading: + self.to_screen('[%s] Threading for Heartbeat not available' % self.FD_NAME) + return fd.real_download(filename, info_dict) + + success = download_complete = False + timer = [None] + heartbeat_lock = threading.Lock() + heartbeat_url = heartbeat_info_dict['url'] + heartbeat_data = heartbeat_info_dict['data'].encode() + heartbeat_interval = heartbeat_info_dict.get('interval', 30) + + request = sanitized_Request(heartbeat_url, heartbeat_data) + + def heartbeat(): + try: + self.ydl.urlopen(request).read() + except Exception: + self.to_screen('[%s] Heartbeat failed' % self.FD_NAME) + + with heartbeat_lock: + if not download_complete: + timer[0] = threading.Timer(heartbeat_interval, heartbeat) + timer[0].start() + + heartbeat_info_dict['ping']() + self.to_screen('[%s] Heartbeat with %d second interval ...' % (self.FD_NAME, heartbeat_interval)) + try: + heartbeat() + if type(fd).__name__ == 'HlsFD': + info_dict.update(ie._extract_m3u8_formats(info_dict['url'], info_dict['id'])[0]) + success = fd.real_download(filename, info_dict) + finally: + if heartbeat_lock: + with heartbeat_lock: + timer[0].cancel() + download_complete = True + return success diff --git a/youtube_dl/downloader/rtmp.py b/youtube_dl/downloader/rtmp.py index 14d56db47..8a25dbc8d 100644 --- a/youtube_dl/downloader/rtmp.py +++ b/youtube_dl/downloader/rtmp.py @@ -29,80 +29,84 @@ class RtmpFD(FileDownloader): proc = subprocess.Popen(args, stderr=subprocess.PIPE) cursor_in_new_line = True proc_stderr_closed = False - while not proc_stderr_closed: - # read line from stderr - line = '' - while True: - char = proc.stderr.read(1) - if not char: - proc_stderr_closed = True - break - if char in [b'\r', b'\n']: - break - line += char.decode('ascii', 'replace') - if not line: - # proc_stderr_closed is True - continue - mobj = re.search(r'([0-9]+\.[0-9]{3}) kB / [0-9]+\.[0-9]{2} sec \(([0-9]{1,2}\.[0-9])%\)', line) - if mobj: - downloaded_data_len = int(float(mobj.group(1)) * 1024) - percent = float(mobj.group(2)) - if not resume_percent: - resume_percent = percent - resume_downloaded_data_len = downloaded_data_len - time_now = time.time() - eta = self.calc_eta(start, time_now, 100 - resume_percent, percent - resume_percent) - speed = self.calc_speed(start, time_now, downloaded_data_len - resume_downloaded_data_len) - data_len = None - if percent > 0: - data_len = int(downloaded_data_len * 100 / percent) - self._hook_progress({ - 'status': 'downloading', - 'downloaded_bytes': downloaded_data_len, - 'total_bytes_estimate': data_len, - 'tmpfilename': tmpfilename, - 'filename': filename, - 'eta': eta, - 'elapsed': time_now - start, - 'speed': speed, - }) - cursor_in_new_line = False - else: - # no percent for live streams - mobj = re.search(r'([0-9]+\.[0-9]{3}) kB / [0-9]+\.[0-9]{2} sec', line) + try: + while not proc_stderr_closed: + # read line from stderr + line = '' + while True: + char = proc.stderr.read(1) + if not char: + proc_stderr_closed = True + break + if char in [b'\r', b'\n']: + break + line += char.decode('ascii', 'replace') + if not line: + # proc_stderr_closed is True + continue + mobj = re.search(r'([0-9]+\.[0-9]{3}) kB / [0-9]+\.[0-9]{2} sec \(([0-9]{1,2}\.[0-9])%\)', line) if mobj: downloaded_data_len = int(float(mobj.group(1)) * 1024) + percent = float(mobj.group(2)) + if not resume_percent: + resume_percent = percent + resume_downloaded_data_len = downloaded_data_len time_now = time.time() - speed = self.calc_speed(start, time_now, downloaded_data_len) + eta = self.calc_eta(start, time_now, 100 - resume_percent, percent - resume_percent) + speed = self.calc_speed(start, time_now, downloaded_data_len - resume_downloaded_data_len) + data_len = None + if percent > 0: + data_len = int(downloaded_data_len * 100 / percent) self._hook_progress({ + 'status': 'downloading', 'downloaded_bytes': downloaded_data_len, + 'total_bytes_estimate': data_len, 'tmpfilename': tmpfilename, 'filename': filename, - 'status': 'downloading', + 'eta': eta, 'elapsed': time_now - start, 'speed': speed, }) cursor_in_new_line = False - elif self.params.get('verbose', False): - if not cursor_in_new_line: - self.to_screen('') - cursor_in_new_line = True - self.to_screen('[rtmpdump] ' + line) - proc.wait() - if not cursor_in_new_line: - self.to_screen('') - return proc.returncode + else: + # no percent for live streams + mobj = re.search(r'([0-9]+\.[0-9]{3}) kB / [0-9]+\.[0-9]{2} sec', line) + if mobj: + downloaded_data_len = int(float(mobj.group(1)) * 1024) + time_now = time.time() + speed = self.calc_speed(start, time_now, downloaded_data_len) + self._hook_progress({ + 'downloaded_bytes': downloaded_data_len, + 'tmpfilename': tmpfilename, + 'filename': filename, + 'status': 'downloading', + 'elapsed': time_now - start, + 'speed': speed, + }) + cursor_in_new_line = False + elif self.params.get('verbose', False): + if not cursor_in_new_line: + self.to_screen('') + cursor_in_new_line = True + self.to_screen('[rtmpdump] ' + line) + if not cursor_in_new_line: + self.to_screen('') + return proc.wait() + except BaseException: # Including KeyboardInterrupt + proc.kill() + proc.wait() + raise url = info_dict['url'] - player_url = info_dict.get('player_url', None) - page_url = info_dict.get('page_url', None) - app = info_dict.get('app', None) - play_path = info_dict.get('play_path', None) - tc_url = info_dict.get('tc_url', None) - flash_version = info_dict.get('flash_version', None) + player_url = info_dict.get('player_url') + page_url = info_dict.get('page_url') + app = info_dict.get('app') + play_path = info_dict.get('play_path') + tc_url = info_dict.get('tc_url') + flash_version = info_dict.get('flash_version') live = info_dict.get('rtmp_live', False) - conn = info_dict.get('rtmp_conn', None) - protocol = info_dict.get('rtmp_protocol', None) + conn = info_dict.get('rtmp_conn') + protocol = info_dict.get('rtmp_protocol') real_time = info_dict.get('rtmp_real_time', False) no_resume = info_dict.get('no_resume', False) continue_dl = self.params.get('continuedl', True) @@ -163,15 +167,23 @@ class RtmpFD(FileDownloader): RD_INCOMPLETE = 2 RD_NO_CONNECT = 3 - retval = run_rtmpdump(args) + started = time.time() + + try: + retval = run_rtmpdump(args) + except KeyboardInterrupt: + if not info_dict.get('is_live'): + raise + retval = RD_SUCCESS + self.to_screen('\n[rtmpdump] Interrupted by user') if retval == RD_NO_CONNECT: self.report_error('[rtmpdump] Could not connect to RTMP server.') return False - while (retval == RD_INCOMPLETE or retval == RD_FAILED) and not test and not live: + while retval in (RD_INCOMPLETE, RD_FAILED) and not test and not live: prevsize = os.path.getsize(encodeFilename(tmpfilename)) - self.to_screen('[rtmpdump] %s bytes' % prevsize) + self.to_screen('[rtmpdump] Downloaded %s bytes' % prevsize) time.sleep(5.0) # This seems to be needed args = basic_args + ['--resume'] if retval == RD_FAILED: @@ -188,13 +200,14 @@ class RtmpFD(FileDownloader): break if retval == RD_SUCCESS or (test and retval == RD_INCOMPLETE): fsize = os.path.getsize(encodeFilename(tmpfilename)) - self.to_screen('[rtmpdump] %s bytes' % fsize) + self.to_screen('[rtmpdump] Downloaded %s bytes' % fsize) self.try_rename(tmpfilename, filename) self._hook_progress({ 'downloaded_bytes': fsize, 'total_bytes': fsize, 'filename': filename, 'status': 'finished', + 'elapsed': time.time() - started, }) return True else: diff --git a/youtube_dl/downloader/rtsp.py b/youtube_dl/downloader/rtsp.py index 3eb29526c..939358b2a 100644 --- a/youtube_dl/downloader/rtsp.py +++ b/youtube_dl/downloader/rtsp.py @@ -27,6 +27,8 @@ class RtspFD(FileDownloader): self.report_error('MMS or RTSP download detected but neither "mplayer" nor "mpv" could be run. Please install any.') return False + self._debug_cmd(args) + retval = subprocess.call(args) if retval == 0: fsize = os.path.getsize(encodeFilename(tmpfilename)) |