X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/540fd68c40df72763aee5d75598675c45cfa9aba..7537b0e3b51e7491205fd2198e8b8b9ae334e50d:/youtube_dl/YoutubeDL.py diff --git a/youtube_dl/YoutubeDL.py b/youtube_dl/YoutubeDL.py index 691f3e0..c3d1407 100755 --- a/youtube_dl/YoutubeDL.py +++ b/youtube_dl/YoutubeDL.py @@ -1,10 +1,11 @@ #!/usr/bin/env python -# -*- coding: utf-8 -*- +# coding: utf-8 from __future__ import absolute_import, unicode_literals import collections import contextlib +import copy import datetime import errno import fileinput @@ -21,65 +22,84 @@ import subprocess import socket import sys import time +import tokenize import traceback +import random -if os.name == 'nt': - import ctypes +from string import ascii_letters from .compat import ( compat_basestring, compat_cookiejar, - compat_expanduser, compat_get_terminal_size, compat_http_client, compat_kwargs, + compat_numeric_types, + compat_os_name, compat_str, + compat_tokenize_tokenize, compat_urllib_error, compat_urllib_request, + compat_urllib_request_DataHandler, ) from .utils import ( - escape_url, + age_restricted, + args_to_str, ContentTooShortError, date_from_str, DateRange, DEFAULT_OUTTMPL, determine_ext, + determine_protocol, DownloadError, + encode_compat_str, encodeFilename, + error_to_compat_str, + expand_path, ExtractorError, format_bytes, formatSeconds, + GeoRestrictedError, + int_or_none, + ISO3166Utils, locked_file, make_HTTPS_handler, MaxDownloadsReached, + orderedSet, PagedList, parse_filesize, PerRequestProxyHandler, - PostProcessingError, platform_name, + PostProcessingError, preferredencoding, + prepend_extension, + register_socks_protocols, render_table, + replace_extension, SameFileError, sanitize_filename, sanitize_path, + sanitize_url, + sanitized_Request, std_headers, + str_or_none, subtitles_filename, UnavailableVideoError, url_basename, version_tuple, write_json_file, write_string, + YoutubeDLCookieJar, + YoutubeDLCookieProcessor, YoutubeDLHandler, - prepend_extension, - replace_extension, - args_to_str, - age_restricted, ) from .cache import Cache -from .extractor import get_info_extractor, gen_extractors +from .extractor import get_info_extractor, gen_extractor_classes, _LAZY_LOADER +from .extractor.openload import PhantomJSwrapper from .downloader import get_suitable_downloader from .downloader.rtmp import rtmpdump_version from .postprocessor import ( + FFmpegFixupM3u8PP, FFmpegFixupM4aPP, FFmpegFixupStretchedPP, FFmpegMergerPP, @@ -88,6 +108,9 @@ from .postprocessor import ( ) from .version import __version__ +if compat_os_name == 'nt': + import ctypes + class YoutubeDL(object): """YoutubeDL class. @@ -118,7 +141,10 @@ class YoutubeDL(object): username: Username for authentication purposes. password: Password for authentication purposes. - videopassword: Password for acces a video. + videopassword: Password for accessing a video. + ap_mso: Adobe Pass multiple-system operator identifier. + ap_username: Multiple-system operator account username. + ap_password: Multiple-system operator account password. usenetrc: Use netrc for authentication instead. verbose: Print additional info to stdout. quiet: Do not print messages to stdout. @@ -138,11 +164,13 @@ class YoutubeDL(object): outtmpl: Template for output names. restrictfilenames: Do not allow "&" and spaces in file names ignoreerrors: Do not stop on download errors. + force_generic_extractor: Force downloader to use the generic extractor nooverwrites: Prevent overwriting files. playliststart: Playlist item to start at. playlistend: Playlist item to end at. playlist_items: Specific indices of playlist to download. playlistreverse: Download playlist items in reverse order. + playlistrandom: Download playlist items in random order. matchtitle: Download only matching titles. rejecttitle: Reject downloads for matching titles. logger: Log messages to a logging.Logger instance. @@ -153,7 +181,7 @@ class YoutubeDL(object): writethumbnail: Write the thumbnail image to a file write_all_thumbnails: Write all thumbnail formats to files writesubtitles: Write the video subtitles to a file - writeautomaticsub: Write the automatic subtitles to a file + writeautomaticsub: Write the automatically generated subtitles to a file allsubtitles: Downloads all the subtitles of the video (requires writesubtitles or writeautomaticsub) listsubtitles: Lists all available subtitles for the video @@ -184,8 +212,8 @@ class YoutubeDL(object): prefer_insecure: Use HTTP instead of HTTPS to retrieve information. At the moment, this is only supported by YouTube. proxy: URL of the proxy server to use - cn_verification_proxy: URL of the proxy to use for IP address verification - on Chinese sites. (Experimental) + geo_verification_proxy: URL of the proxy to use for IP address verification + on geo-restricted sites. socket_timeout: Time to wait for unresponsive hosts, in seconds bidi_workaround: Work around buggy terminals without bidirectional text support, using fridibi @@ -233,10 +261,19 @@ class YoutubeDL(object): - "warn": only emit a warning - "detect_or_warn": check whether we can do anything about it, warn otherwise (default) - source_address: (Experimental) Client-side IP address to bind to. + source_address: Client-side IP address to bind to. call_home: Boolean, true iff we are allowed to contact the youtube-dl servers for debugging. - sleep_interval: Number of seconds to sleep before each download. + sleep_interval: Number of seconds to sleep before each download when + used alone or a lower bound of a range for randomized + sleep before each download (minimum possible number + of seconds to sleep) when used along with + max_sleep_interval. + max_sleep_interval:Upper bound of a range for randomized sleep before each + download (maximum possible number of seconds to sleep). + Must only be used along with sleep_interval. + Actual sleep time will be a random float from range + [sleep_interval; max_sleep_interval]. listformats: Print an overview of available video formats and exit. list_thumbnails: Print a table of all thumbnails and exit. match_filter: A function that gets called with the info_dict of @@ -245,23 +282,56 @@ class YoutubeDL(object): If it returns None, the video is downloaded. match_filter_func in utils.py is one example for this. no_color: Do not emit color codes in output. + geo_bypass: Bypass geographic restriction via faking X-Forwarded-For + HTTP header + geo_bypass_country: + Two-letter ISO 3166-2 country code that will be used for + explicit geographic restriction bypassing via faking + X-Forwarded-For HTTP header + geo_bypass_ip_block: + IP range in CIDR notation that will be used similarly to + geo_bypass_country The following options determine which downloader is picked: external_downloader: Executable of the external downloader to call. None or unset for standard (built-in) downloader. - hls_prefer_native: Use the native HLS downloader instead of ffmpeg/avconv. + hls_prefer_native: Use the native HLS downloader instead of ffmpeg/avconv + if True, otherwise use ffmpeg/avconv if False, otherwise + use downloader suggested by extractor if None. The following parameters are not used by YoutubeDL itself, they are used by the downloader (see youtube_dl/downloader/common.py): nopart, updatetime, buffersize, ratelimit, min_filesize, max_filesize, test, noresizebuffer, retries, continuedl, noprogress, consoletitle, - xattr_set_filesize, external_downloader_args. + xattr_set_filesize, external_downloader_args, hls_use_mpegts, + http_chunk_size. The following options are used by the post processors: - prefer_ffmpeg: If True, use ffmpeg instead of avconv if both are available, - otherwise prefer avconv. + prefer_ffmpeg: If False, use avconv instead of ffmpeg if both are available, + otherwise prefer ffmpeg. + ffmpeg_location: Location of the ffmpeg/avconv binary; either the path + to the binary or its containing directory. + postprocessor_args: A list of additional command-line arguments for the + postprocessor. + + The following options are used by the Youtube extractor: + youtube_include_dash_manifest: If True (default), DASH manifests and related + data will be downloaded and processed by extractor. + You can reduce network I/O by disabling it if you don't + care about DASH. """ + _NUMERIC_FIELDS = set(( + 'width', 'height', 'tbr', 'abr', 'asr', 'vbr', 'fps', 'filesize', 'filesize_approx', + 'timestamp', 'upload_year', 'upload_month', 'upload_day', + 'duration', 'view_count', 'like_count', 'dislike_count', 'repost_count', + 'average_rating', 'comment_count', 'age_limit', + 'start_time', 'end_time', + 'chapter_number', 'season_number', 'episode_number', + 'track_number', 'disc_number', 'release_year', + 'playlist_index', + )) + params = None _ies = [] _pps = [] @@ -281,9 +351,28 @@ class YoutubeDL(object): self._num_downloads = 0 self._screen_file = [sys.stdout, sys.stderr][params.get('logtostderr', False)] self._err_file = sys.stderr - self.params = params + self.params = { + # Default parameters + 'nocheckcertificate': False, + } + self.params.update(params) self.cache = Cache(self) + def check_deprecated(param, option, suggestion): + if self.params.get(param) is not None: + self.report_warning( + '%s is deprecated. Use %s instead.' % (option, suggestion)) + return True + return False + + if check_deprecated('cn_verification_proxy', '--cn-verification-proxy', '--geo-verification-proxy'): + if self.params.get('geo_verification_proxy') is None: + self.params['geo_verification_proxy'] = self.params['cn_verification_proxy'] + + check_deprecated('autonumber_size', '--autonumber-size', 'output template with %(autonumber)0Nd, where N in the number of digits') + check_deprecated('autonumber', '--auto-number', '-o "%(autonumber)s-%(title)s.%(ext)s"') + check_deprecated('usetitle', '--title', '-o "%(title)s-%(id)s.%(ext)s"') + if params.get('bidi_workaround', False): try: import pty @@ -306,15 +395,15 @@ class YoutubeDL(object): ['fribidi', '-c', 'UTF-8'] + width_args, **sp_kwargs) self._output_channel = os.fdopen(master, 'rb') except OSError as ose: - if ose.errno == 2: + if ose.errno == errno.ENOENT: self.report_warning('Could not find fribidi executable, ignoring --bidi-workaround . Make sure that fribidi is an executable file in one of the directories in your $PATH.') else: raise - if (sys.version_info >= (3,) and sys.platform != 'win32' and - sys.getfilesystemencoding() in ['ascii', 'ANSI_X3.4-1968'] and - not params.get('restrictfilenames', False)): - # On Python 3, the Unicode filesystem API will throw errors (#1474) + if (sys.platform != 'win32' + and sys.getfilesystemencoding() in ['ascii', 'ANSI_X3.4-1968'] + and not params.get('restrictfilenames', False)): + # Unicode filesystem API will throw errors (#1474, #13027) self.report_warning( 'Assuming --restrict-filenames since file system encoding ' 'cannot encode all characters. ' @@ -342,6 +431,8 @@ class YoutubeDL(object): for ph in self.params.get('progress_hooks', []): self.add_progress_hook(ph) + register_socks_protocols() + def warn_if_short_id(self, argv): # short YouTube ID starting with dash? idxs = [ @@ -349,9 +440,9 @@ class YoutubeDL(object): if re.match(r'^-[0-9A-Za-z_-]{10}$', a)] if idxs: correct_argv = ( - ['youtube-dl'] + - [a for i, a in enumerate(argv) if i not in idxs] + - ['--'] + [argv[i] for i in idxs] + ['youtube-dl'] + + [a for i, a in enumerate(argv) if i not in idxs] + + ['--'] + [argv[i] for i in idxs] ) self.report_warning( 'Long argument string detected. ' @@ -361,8 +452,9 @@ class YoutubeDL(object): def add_info_extractor(self, ie): """Add an InfoExtractor object to the end of the list.""" self._ies.append(ie) - self._ies_instances[ie.ie_key()] = ie - ie.set_downloader(self) + if not isinstance(ie, type): + self._ies_instances[ie.ie_key()] = ie + ie.set_downloader(self) def get_info_extractor(self, ie_key): """ @@ -380,7 +472,7 @@ class YoutubeDL(object): """ Add the InfoExtractors returned by gen_extractors to the end of the list """ - for ie in gen_extractors(): + for ie in gen_extractor_classes(): self.add_info_extractor(ie) def add_post_processor(self, pp): @@ -436,24 +528,29 @@ class YoutubeDL(object): def to_console_title(self, message): if not self.params.get('consoletitle', False): return - if os.name == 'nt' and ctypes.windll.kernel32.GetConsoleWindow(): - # c_wchar_p() might not be necessary if `message` is - # already of type unicode() - ctypes.windll.kernel32.SetConsoleTitleW(ctypes.c_wchar_p(message)) + if compat_os_name == 'nt': + if ctypes.windll.kernel32.GetConsoleWindow(): + # c_wchar_p() might not be necessary if `message` is + # already of type unicode() + ctypes.windll.kernel32.SetConsoleTitleW(ctypes.c_wchar_p(message)) elif 'TERM' in os.environ: self._write_string('\033]0;%s\007' % message, self._screen_file) def save_console_title(self): if not self.params.get('consoletitle', False): return - if 'TERM' in os.environ: + if self.params.get('simulate', False): + return + if compat_os_name != 'nt' and 'TERM' in os.environ: # Save the title on stack self._write_string('\033[22;0t', self._screen_file) def restore_console_title(self): if not self.params.get('consoletitle', False): return - if 'TERM' in os.environ: + if self.params.get('simulate', False): + return + if compat_os_name != 'nt' and 'TERM' in os.environ: # Restore the title from stack self._write_string('\033[23;0t', self._screen_file) @@ -465,7 +562,7 @@ class YoutubeDL(object): self.restore_console_title() if self.params.get('cookiefile') is not None: - self.cookiejar.save() + self.cookiejar.save(ignore_discard=True, ignore_expires=True) def trouble(self, message=None, tb=None): """Determine action to take when a download problem appears. @@ -484,7 +581,7 @@ class YoutubeDL(object): tb = '' if hasattr(sys.exc_info()[1], 'exc_info') and sys.exc_info()[1].exc_info[0]: tb += ''.join(traceback.format_exception(*sys.exc_info()[1].exc_info)) - tb += compat_str(traceback.format_exc()) + tb += encode_compat_str(traceback.format_exc()) else: tb_data = traceback.format_list(traceback.extract_stack()) tb = ''.join(tb_data) @@ -507,7 +604,7 @@ class YoutubeDL(object): else: if self.params.get('no_warnings'): return - if not self.params.get('no_color') and self._err_file.isatty() and os.name != 'nt': + if not self.params.get('no_color') and self._err_file.isatty() and compat_os_name != 'nt': _msg_header = '\033[0;33mWARNING:\033[0m' else: _msg_header = 'WARNING:' @@ -519,7 +616,7 @@ class YoutubeDL(object): Do the same as trouble, but prefixes the message with 'ERROR:', colored in red if stderr is a tty file. ''' - if not self.params.get('no_color') and self._err_file.isatty() and os.name != 'nt': + if not self.params.get('no_color') and self._err_file.isatty() and compat_os_name != 'nt': _msg_header = '\033[0;31mERROR:\033[0m' else: _msg_header = 'ERROR:' @@ -542,36 +639,82 @@ class YoutubeDL(object): autonumber_size = self.params.get('autonumber_size') if autonumber_size is None: autonumber_size = 5 - autonumber_templ = '%0' + str(autonumber_size) + 'd' - template_dict['autonumber'] = autonumber_templ % self._num_downloads - if template_dict.get('playlist_index') is not None: - template_dict['playlist_index'] = '%0*d' % (len(str(template_dict['n_entries'])), template_dict['playlist_index']) + template_dict['autonumber'] = self.params.get('autonumber_start', 1) - 1 + self._num_downloads if template_dict.get('resolution') is None: if template_dict.get('width') and template_dict.get('height'): template_dict['resolution'] = '%dx%d' % (template_dict['width'], template_dict['height']) elif template_dict.get('height'): template_dict['resolution'] = '%sp' % template_dict['height'] elif template_dict.get('width'): - template_dict['resolution'] = '?x%d' % template_dict['width'] + template_dict['resolution'] = '%dx?' % template_dict['width'] sanitize = lambda k, v: sanitize_filename( compat_str(v), restricted=self.params.get('restrictfilenames'), - is_id=(k == 'id')) - template_dict = dict((k, sanitize(k, v)) + is_id=(k == 'id' or k.endswith('_id'))) + template_dict = dict((k, v if isinstance(v, compat_numeric_types) else sanitize(k, v)) for k, v in template_dict.items() - if v is not None) + if v is not None and not isinstance(v, (list, tuple, dict))) template_dict = collections.defaultdict(lambda: 'NA', template_dict) - outtmpl = sanitize_path(self.params.get('outtmpl', DEFAULT_OUTTMPL)) - tmpl = compat_expanduser(outtmpl) - filename = tmpl % template_dict + outtmpl = self.params.get('outtmpl', DEFAULT_OUTTMPL) + + # For fields playlist_index and autonumber convert all occurrences + # of %(field)s to %(field)0Nd for backward compatibility + field_size_compat_map = { + 'playlist_index': len(str(template_dict['n_entries'])), + 'autonumber': autonumber_size, + } + FIELD_SIZE_COMPAT_RE = r'(?autonumber|playlist_index)\)s' + mobj = re.search(FIELD_SIZE_COMPAT_RE, outtmpl) + if mobj: + outtmpl = re.sub( + FIELD_SIZE_COMPAT_RE, + r'%%(\1)0%dd' % field_size_compat_map[mobj.group('field')], + outtmpl) + + # Missing numeric fields used together with integer presentation types + # in format specification will break the argument substitution since + # string 'NA' is returned for missing fields. We will patch output + # template for missing fields to meet string presentation type. + for numeric_field in self._NUMERIC_FIELDS: + if numeric_field not in template_dict: + # As of [1] format syntax is: + # %[mapping_key][conversion_flags][minimum_width][.precision][length_modifier]type + # 1. https://docs.python.org/2/library/stdtypes.html#string-formatting + FORMAT_RE = r'''(?x) + (?width|height|tbr|abr|vbr|asr|filesize|fps) + operator_rex = re.compile(r'''(?x)\s* + (?Pwidth|height|tbr|abr|vbr|asr|filesize|filesize_approx|fps) \s*(?P%s)(?P\s*\?)?\s* (?P[0-9.]+(?:[kKmMgGtTpPeEzZyY]i?[Bb]?)?) - \]$ + $ ''' % '|'.join(map(re.escape, OPERATORS.keys()))) - m = operator_rex.search(format_spec) + m = operator_rex.search(filter_spec) if m: try: comparison_value = int(m.group('value')) @@ -872,92 +1061,329 @@ class YoutubeDL(object): if comparison_value is None: raise ValueError( 'Invalid value %r in format specification %r' % ( - m.group('value'), format_spec)) + m.group('value'), filter_spec)) op = OPERATORS[m.group('op')] if not m: STR_OPERATORS = { '=': operator.eq, - '!=': operator.ne, + '^=': lambda attr, value: attr.startswith(value), + '$=': lambda attr, value: attr.endswith(value), + '*=': lambda attr, value: value in attr, } - str_operator_rex = re.compile(r'''(?x)\s*\[ - \s*(?Pext|acodec|vcodec|container|protocol) - \s*(?P%s)(?P\s*\?)? - \s*(?P[a-zA-Z0-9_-]+) - \s*\]$ + str_operator_rex = re.compile(r'''(?x) + \s*(?Pext|acodec|vcodec|container|protocol|format_id) + \s*(?P!\s*)?(?P%s)(?P\s*\?)? + \s*(?P[a-zA-Z0-9._-]+) + \s*$ ''' % '|'.join(map(re.escape, STR_OPERATORS.keys()))) - m = str_operator_rex.search(format_spec) + m = str_operator_rex.search(filter_spec) if m: comparison_value = m.group('value') - op = STR_OPERATORS[m.group('op')] + str_op = STR_OPERATORS[m.group('op')] + if m.group('negation'): + op = lambda attr, value: not str_op(attr, value) + else: + op = str_op if not m: - raise ValueError('Invalid format specification %r' % format_spec) + raise ValueError('Invalid filter specification %r' % filter_spec) def _filter(f): actual_value = f.get(m.group('key')) if actual_value is None: return m.group('none_inclusive') return op(actual_value, comparison_value) - new_formats = [f for f in available_formats if _filter(f)] + return _filter + + def _default_format_spec(self, info_dict, download=True): + + def can_merge(): + merger = FFmpegMergerPP(self) + return merger.available and merger.can_merge() + + def prefer_best(): + if self.params.get('simulate', False): + return False + if not download: + return False + if self.params.get('outtmpl', DEFAULT_OUTTMPL) == '-': + return True + if info_dict.get('is_live'): + return True + if not can_merge(): + return True + return False - new_format_spec = format_spec[:-len(m.group(0))] - if not new_format_spec: - new_format_spec = 'best' + req_format_list = ['bestvideo+bestaudio', 'best'] + if prefer_best(): + req_format_list.reverse() + return '/'.join(req_format_list) + + def build_format_selector(self, format_spec): + def syntax_error(note, start): + message = ( + 'Invalid format specification: ' + '{0}\n\t{1}\n\t{2}^'.format(note, format_spec, ' ' * start[1])) + return SyntaxError(message) + + PICKFIRST = 'PICKFIRST' + MERGE = 'MERGE' + SINGLE = 'SINGLE' + GROUP = 'GROUP' + FormatSelector = collections.namedtuple('FormatSelector', ['type', 'selector', 'filters']) + + def _parse_filter(tokens): + filter_parts = [] + for type, string, start, _, _ in tokens: + if type == tokenize.OP and string == ']': + return ''.join(filter_parts) + else: + filter_parts.append(string) + + def _remove_unused_ops(tokens): + # Remove operators that we don't use and join them with the surrounding strings + # for example: 'mp4' '-' 'baseline' '-' '16x9' is converted to 'mp4-baseline-16x9' + ALLOWED_OPS = ('/', '+', ',', '(', ')') + last_string, last_start, last_end, last_line = None, None, None, None + for type, string, start, end, line in tokens: + if type == tokenize.OP and string == '[': + if last_string: + yield tokenize.NAME, last_string, last_start, last_end, last_line + last_string = None + yield type, string, start, end, line + # everything inside brackets will be handled by _parse_filter + for type, string, start, end, line in tokens: + yield type, string, start, end, line + if type == tokenize.OP and string == ']': + break + elif type == tokenize.OP and string in ALLOWED_OPS: + if last_string: + yield tokenize.NAME, last_string, last_start, last_end, last_line + last_string = None + yield type, string, start, end, line + elif type in [tokenize.NAME, tokenize.NUMBER, tokenize.OP]: + if not last_string: + last_string = string + last_start = start + last_end = end + else: + last_string += string + if last_string: + yield tokenize.NAME, last_string, last_start, last_end, last_line + + def _parse_format_selection(tokens, inside_merge=False, inside_choice=False, inside_group=False): + selectors = [] + current_selector = None + for type, string, start, _, _ in tokens: + # ENCODING is only defined in python 3.x + if type == getattr(tokenize, 'ENCODING', None): + continue + elif type in [tokenize.NAME, tokenize.NUMBER]: + current_selector = FormatSelector(SINGLE, string, []) + elif type == tokenize.OP: + if string == ')': + if not inside_group: + # ')' will be handled by the parentheses group + tokens.restore_last_token() + break + elif inside_merge and string in ['/', ',']: + tokens.restore_last_token() + break + elif inside_choice and string == ',': + tokens.restore_last_token() + break + elif string == ',': + if not current_selector: + raise syntax_error('"," must follow a format selector', start) + selectors.append(current_selector) + current_selector = None + elif string == '/': + if not current_selector: + raise syntax_error('"/" must follow a format selector', start) + first_choice = current_selector + second_choice = _parse_format_selection(tokens, inside_choice=True) + current_selector = FormatSelector(PICKFIRST, (first_choice, second_choice), []) + elif string == '[': + if not current_selector: + current_selector = FormatSelector(SINGLE, 'best', []) + format_filter = _parse_filter(tokens) + current_selector.filters.append(format_filter) + elif string == '(': + if current_selector: + raise syntax_error('Unexpected "("', start) + group = _parse_format_selection(tokens, inside_group=True) + current_selector = FormatSelector(GROUP, group, []) + elif string == '+': + video_selector = current_selector + audio_selector = _parse_format_selection(tokens, inside_merge=True) + if not video_selector or not audio_selector: + raise syntax_error('"+" must be between two format selectors', start) + current_selector = FormatSelector(MERGE, (video_selector, audio_selector), []) + else: + raise syntax_error('Operator not recognized: "{0}"'.format(string), start) + elif type == tokenize.ENDMARKER: + break + if current_selector: + selectors.append(current_selector) + return selectors + + def _build_selector_function(selector): + if isinstance(selector, list): + fs = [_build_selector_function(s) for s in selector] + + def selector_function(ctx): + for f in fs: + for format in f(ctx): + yield format + return selector_function + elif selector.type == GROUP: + selector_function = _build_selector_function(selector.selector) + elif selector.type == PICKFIRST: + fs = [_build_selector_function(s) for s in selector.selector] + + def selector_function(ctx): + for f in fs: + picked_formats = list(f(ctx)) + if picked_formats: + return picked_formats + return [] + elif selector.type == SINGLE: + format_spec = selector.selector + + def selector_function(ctx): + formats = list(ctx['formats']) + if not formats: + return + if format_spec == 'all': + for f in formats: + yield f + elif format_spec in ['best', 'worst', None]: + format_idx = 0 if format_spec == 'worst' else -1 + audiovideo_formats = [ + f for f in formats + if f.get('vcodec') != 'none' and f.get('acodec') != 'none'] + if audiovideo_formats: + yield audiovideo_formats[format_idx] + # for extractors with incomplete formats (audio only (soundcloud) + # or video only (imgur)) we will fallback to best/worst + # {video,audio}-only format + elif ctx['incomplete_formats']: + yield formats[format_idx] + elif format_spec == 'bestaudio': + audio_formats = [ + f for f in formats + if f.get('vcodec') == 'none'] + if audio_formats: + yield audio_formats[-1] + elif format_spec == 'worstaudio': + audio_formats = [ + f for f in formats + if f.get('vcodec') == 'none'] + if audio_formats: + yield audio_formats[0] + elif format_spec == 'bestvideo': + video_formats = [ + f for f in formats + if f.get('acodec') == 'none'] + if video_formats: + yield video_formats[-1] + elif format_spec == 'worstvideo': + video_formats = [ + f for f in formats + if f.get('acodec') == 'none'] + if video_formats: + yield video_formats[0] + else: + extensions = ['mp4', 'flv', 'webm', '3gp', 'm4a', 'mp3', 'ogg', 'aac', 'wav'] + if format_spec in extensions: + filter_f = lambda f: f['ext'] == format_spec + else: + filter_f = lambda f: f['format_id'] == format_spec + matches = list(filter(filter_f, formats)) + if matches: + yield matches[-1] + elif selector.type == MERGE: + def _merge(formats_info): + format_1, format_2 = [f['format_id'] for f in formats_info] + # The first format must contain the video and the + # second the audio + if formats_info[0].get('vcodec') == 'none': + self.report_error('The first format must ' + 'contain the video, try using ' + '"-f %s+%s"' % (format_2, format_1)) + return + # Formats must be opposite (video+audio) + if formats_info[0].get('acodec') == 'none' and formats_info[1].get('acodec') == 'none': + self.report_error( + 'Both formats %s and %s are video-only, you must specify "-f video+audio"' + % (format_1, format_2)) + return + output_ext = ( + formats_info[0]['ext'] + if self.params.get('merge_output_format') is None + else self.params['merge_output_format']) + return { + 'requested_formats': formats_info, + 'format': '%s+%s' % (formats_info[0].get('format'), + formats_info[1].get('format')), + 'format_id': '%s+%s' % (formats_info[0].get('format_id'), + formats_info[1].get('format_id')), + 'width': formats_info[0].get('width'), + 'height': formats_info[0].get('height'), + 'resolution': formats_info[0].get('resolution'), + 'fps': formats_info[0].get('fps'), + 'vcodec': formats_info[0].get('vcodec'), + 'vbr': formats_info[0].get('vbr'), + 'stretched_ratio': formats_info[0].get('stretched_ratio'), + 'acodec': formats_info[1].get('acodec'), + 'abr': formats_info[1].get('abr'), + 'ext': output_ext, + } + video_selector, audio_selector = map(_build_selector_function, selector.selector) - return (new_format_spec, new_formats) + def selector_function(ctx): + for pair in itertools.product( + video_selector(copy.deepcopy(ctx)), audio_selector(copy.deepcopy(ctx))): + yield _merge(pair) - def select_format(self, format_spec, available_formats): - while format_spec.endswith(']'): - format_spec, available_formats = self._apply_format_filter( - format_spec, available_formats) - if not available_formats: - return None + filters = [self._build_format_filter(f) for f in selector.filters] - if format_spec in ['best', 'worst', None]: - format_idx = 0 if format_spec == 'worst' else -1 - audiovideo_formats = [ - f for f in available_formats - if f.get('vcodec') != 'none' and f.get('acodec') != 'none'] - if audiovideo_formats: - return audiovideo_formats[format_idx] - # for audio only urls, select the best/worst audio format - elif all(f.get('acodec') != 'none' for f in available_formats): - return available_formats[format_idx] - elif format_spec == 'bestaudio': - audio_formats = [ - f for f in available_formats - if f.get('vcodec') == 'none'] - if audio_formats: - return audio_formats[-1] - elif format_spec == 'worstaudio': - audio_formats = [ - f for f in available_formats - if f.get('vcodec') == 'none'] - if audio_formats: - return audio_formats[0] - elif format_spec == 'bestvideo': - video_formats = [ - f for f in available_formats - if f.get('acodec') == 'none'] - if video_formats: - return video_formats[-1] - elif format_spec == 'worstvideo': - video_formats = [ - f for f in available_formats - if f.get('acodec') == 'none'] - if video_formats: - return video_formats[0] - else: - extensions = ['mp4', 'flv', 'webm', '3gp', 'm4a', 'mp3', 'ogg', 'aac', 'wav'] - if format_spec in extensions: - filter_f = lambda f: f['ext'] == format_spec - else: - filter_f = lambda f: f['format_id'] == format_spec - matches = list(filter(filter_f, available_formats)) - if matches: - return matches[-1] - return None + def final_selector(ctx): + ctx_copy = copy.deepcopy(ctx) + for _filter in filters: + ctx_copy['formats'] = list(filter(_filter, ctx_copy['formats'])) + return selector_function(ctx_copy) + return final_selector + + stream = io.BytesIO(format_spec.encode('utf-8')) + try: + tokens = list(_remove_unused_ops(compat_tokenize_tokenize(stream.readline))) + except tokenize.TokenError: + raise syntax_error('Missing closing/opening brackets or parenthesis', (0, len(format_spec))) + + class TokenIterator(object): + def __init__(self, tokens): + self.tokens = tokens + self.counter = 0 + + def __iter__(self): + return self + + def __next__(self): + if self.counter >= len(self.tokens): + raise StopIteration() + value = self.tokens[self.counter] + self.counter += 1 + return value + + next = __next__ + + def restore_last_token(self): + self.counter -= 1 + + parsed_selector = _parse_format_selection(iter(TokenIterator(tokens))) + return _build_selector_function(parsed_selector) def _calc_headers(self, info_dict): res = std_headers.copy() @@ -970,10 +1396,15 @@ class YoutubeDL(object): if cookies: res['Cookie'] = cookies + if 'X-Forwarded-For' not in res: + x_forwarded_for_ip = info_dict.get('__x_forwarded_for_ip') + if x_forwarded_for_ip: + res['X-Forwarded-For'] = x_forwarded_for_ip + return res def _calc_cookies(self, info_dict): - pr = compat_urllib_request.Request(info_dict['url']) + pr = sanitized_Request(info_dict['url']) self.cookiejar.add_cookie_header(pr) return pr.get_header('Cookie') @@ -985,6 +1416,29 @@ class YoutubeDL(object): if 'title' not in info_dict: raise ExtractorError('Missing "title" field in extractor result') + def report_force_conversion(field, field_not, conversion): + self.report_warning( + '"%s" field is not %s - forcing %s conversion, there is an error in extractor' + % (field, field_not, conversion)) + + def sanitize_string_field(info, string_field): + field = info.get(string_field) + if field is None or isinstance(field, compat_str): + return + report_force_conversion(string_field, 'a string', 'string') + info[string_field] = compat_str(field) + + def sanitize_numeric_fields(info): + for numeric_field in self._NUMERIC_FIELDS: + field = info.get(numeric_field) + if field is None or isinstance(field, compat_numeric_types): + continue + report_force_conversion(numeric_field, 'numeric', 'int') + info[numeric_field] = int_or_none(field) + + sanitize_string_field(info_dict, 'id') + sanitize_numeric_fields(info_dict) + if 'playlist' not in info_dict: # It isn't part of a playlist info_dict['playlist'] = None @@ -997,43 +1451,67 @@ class YoutubeDL(object): info_dict['thumbnails'] = thumbnails = [{'url': thumbnail}] if thumbnails: thumbnails.sort(key=lambda t: ( - t.get('preference'), t.get('width'), t.get('height'), - t.get('id'), t.get('url'))) + t.get('preference') if t.get('preference') is not None else -1, + t.get('width') if t.get('width') is not None else -1, + t.get('height') if t.get('height') is not None else -1, + t.get('id') if t.get('id') is not None else '', t.get('url'))) for i, t in enumerate(thumbnails): - if 'width' in t and 'height' in t: + t['url'] = sanitize_url(t['url']) + if t.get('width') and t.get('height'): t['resolution'] = '%dx%d' % (t['width'], t['height']) if t.get('id') is None: t['id'] = '%d' % i - if thumbnails and 'thumbnail' not in info_dict: + if self.params.get('list_thumbnails'): + self.list_thumbnails(info_dict) + return + + thumbnail = info_dict.get('thumbnail') + if thumbnail: + info_dict['thumbnail'] = sanitize_url(thumbnail) + elif thumbnails: info_dict['thumbnail'] = thumbnails[-1]['url'] if 'display_id' not in info_dict and 'id' in info_dict: info_dict['display_id'] = info_dict['id'] if info_dict.get('upload_date') is None and info_dict.get('timestamp') is not None: - # Working around negative timestamps in Windows - # (see http://bugs.python.org/issue1646728) - if info_dict['timestamp'] < 0 and os.name == 'nt': - info_dict['timestamp'] = 0 - upload_date = datetime.datetime.utcfromtimestamp( - info_dict['timestamp']) - info_dict['upload_date'] = upload_date.strftime('%Y%m%d') + # Working around out-of-range timestamp values (e.g. negative ones on Windows, + # see http://bugs.python.org/issue1646728) + try: + upload_date = datetime.datetime.utcfromtimestamp(info_dict['timestamp']) + info_dict['upload_date'] = upload_date.strftime('%Y%m%d') + except (ValueError, OverflowError, OSError): + pass + + # Auto generate title fields corresponding to the *_number fields when missing + # in order to always have clean titles. This is very common for TV series. + for field in ('chapter', 'season', 'episode'): + if info_dict.get('%s_number' % field) is not None and not info_dict.get(field): + info_dict[field] = '%s %d' % (field.capitalize(), info_dict['%s_number' % field]) + + for cc_kind in ('subtitles', 'automatic_captions'): + cc = info_dict.get(cc_kind) + if cc: + for _, subtitle in cc.items(): + for subtitle_format in subtitle: + if subtitle_format.get('url'): + subtitle_format['url'] = sanitize_url(subtitle_format['url']) + if subtitle_format.get('ext') is None: + subtitle_format['ext'] = determine_ext(subtitle_format['url']).lower() + + automatic_captions = info_dict.get('automatic_captions') + subtitles = info_dict.get('subtitles') if self.params.get('listsubtitles', False): if 'automatic_captions' in info_dict: - self.list_subtitles(info_dict['id'], info_dict.get('automatic_captions'), 'automatic captions') - self.list_subtitles(info_dict['id'], info_dict.get('subtitles'), 'subtitles') + self.list_subtitles( + info_dict['id'], automatic_captions, 'automatic captions') + self.list_subtitles(info_dict['id'], subtitles, 'subtitles') return - info_dict['requested_subtitles'] = self.process_subtitles( - info_dict['id'], info_dict.get('subtitles'), - info_dict.get('automatic_captions')) - # This extractors handle format selection themselves - if info_dict['extractor'] in ['Youku']: - if download: - self.process_info(info_dict) - return info_dict + info_dict['requested_subtitles'] = self.process_subtitles( + info_dict['id'], subtitles, automatic_captions) # We now pick which formats have to be downloaded if info_dict.get('formats') is None: @@ -1045,13 +1523,44 @@ class YoutubeDL(object): if not formats: raise ExtractorError('No video formats found!') + def is_wellformed(f): + url = f.get('url') + if not url: + self.report_warning( + '"url" field is missing or empty - skipping format, ' + 'there is an error in extractor') + return False + if isinstance(url, bytes): + sanitize_string_field(f, 'url') + return True + + # Filter out malformed formats for better extraction robustness + formats = list(filter(is_wellformed, formats)) + + formats_dict = {} + # We check that all the formats have the format and format_id fields for i, format in enumerate(formats): - if 'url' not in format: - raise ExtractorError('Missing "url" key in result (index %d)' % i) - - if format.get('format_id') is None: + sanitize_string_field(format, 'format_id') + sanitize_numeric_fields(format) + format['url'] = sanitize_url(format['url']) + if not format.get('format_id'): format['format_id'] = compat_str(i) + else: + # Sanitize format_id from characters used in format selector expression + format['format_id'] = re.sub(r'[\s,/+\[\]()]', '_', format['format_id']) + format_id = format['format_id'] + if format_id not in formats_dict: + formats_dict[format_id] = [] + formats_dict[format_id].append(format) + + # Make sure all formats have unique format_id + for format_id, ambiguous_formats in formats_dict.items(): + if len(ambiguous_formats) > 1: + for i, format in enumerate(ambiguous_formats): + format['format_id'] = '%s-%d' % (format_id, i) + + for i, format in enumerate(formats): if format.get('format') is None: format['format'] = '{id} - {res}{note}'.format( id=format['format_id'], @@ -1059,13 +1568,20 @@ class YoutubeDL(object): note=' ({0})'.format(format['format_note']) if format.get('format_note') is not None else '', ) # Automatically determine file extension if missing - if 'ext' not in format: + if format.get('ext') is None: format['ext'] = determine_ext(format['url']).lower() + # Automatically determine protocol if missing (useful for format + # selection purposes) + if format.get('protocol') is None: + format['protocol'] = determine_protocol(format) # Add HTTP headers, so that external programs can use them from the # json output full_format_info = info_dict.copy() full_format_info.update(format) format['http_headers'] = self._calc_headers(full_format_info) + # Remove private housekeeping stuff + if '__x_forwarded_for_ip' in info_dict: + del info_dict['__x_forwarded_for_ip'] # TODO Central sorting goes here @@ -1073,75 +1589,47 @@ class YoutubeDL(object): # only set the 'formats' fields if the original info_dict list them # otherwise we end up with a circular reference, the first (and unique) # element in the 'formats' field in info_dict is info_dict itself, - # wich can't be exported to json + # which can't be exported to json info_dict['formats'] = formats if self.params.get('listformats'): self.list_formats(info_dict) return - if self.params.get('list_thumbnails'): - self.list_thumbnails(info_dict) - return req_format = self.params.get('format') if req_format is None: - req_format_list = [] - if (self.params.get('outtmpl', DEFAULT_OUTTMPL) != '-' and - info_dict['extractor'] in ['youtube', 'ted']): - merger = FFmpegMergerPP(self) - if merger.available and merger.can_merge(): - req_format_list.append('bestvideo+bestaudio') - req_format_list.append('best') - req_format = '/'.join(req_format_list) - formats_to_download = [] - if req_format == 'all': - formats_to_download = formats - else: - for rfstr in req_format.split(','): - # We can accept formats requested in the format: 34/5/best, we pick - # the first that is available, starting from left - req_formats = rfstr.split('/') - for rf in req_formats: - if re.match(r'.+?\+.+?', rf) is not None: - # Two formats have been requested like '137+139' - format_1, format_2 = rf.split('+') - formats_info = (self.select_format(format_1, formats), - self.select_format(format_2, formats)) - if all(formats_info): - # The first format must contain the video and the - # second the audio - if formats_info[0].get('vcodec') == 'none': - self.report_error('The first format must ' - 'contain the video, try using ' - '"-f %s+%s"' % (format_2, format_1)) - return - output_ext = ( - formats_info[0]['ext'] - if self.params.get('merge_output_format') is None - else self.params['merge_output_format']) - selected_format = { - 'requested_formats': formats_info, - 'format': '%s+%s' % (formats_info[0].get('format'), - formats_info[1].get('format')), - 'format_id': '%s+%s' % (formats_info[0].get('format_id'), - formats_info[1].get('format_id')), - 'width': formats_info[0].get('width'), - 'height': formats_info[0].get('height'), - 'resolution': formats_info[0].get('resolution'), - 'fps': formats_info[0].get('fps'), - 'vcodec': formats_info[0].get('vcodec'), - 'vbr': formats_info[0].get('vbr'), - 'stretched_ratio': formats_info[0].get('stretched_ratio'), - 'acodec': formats_info[1].get('acodec'), - 'abr': formats_info[1].get('abr'), - 'ext': output_ext, - } - else: - selected_format = None - else: - selected_format = self.select_format(rf, formats) - if selected_format is not None: - formats_to_download.append(selected_format) - break + req_format = self._default_format_spec(info_dict, download=download) + if self.params.get('verbose'): + self.to_stdout('[debug] Default format spec: %s' % req_format) + + format_selector = self.build_format_selector(req_format) + + # While in format selection we may need to have an access to the original + # format set in order to calculate some metrics or do some processing. + # For now we need to be able to guess whether original formats provided + # by extractor are incomplete or not (i.e. whether extractor provides only + # video-only or audio-only formats) for proper formats selection for + # extractors with such incomplete formats (see + # https://github.com/ytdl-org/youtube-dl/pull/5556). + # Since formats may be filtered during format selection and may not match + # the original formats the results may be incorrect. Thus original formats + # or pre-calculated metrics should be passed to format selection routines + # as well. + # We will pass a context object containing all necessary additional data + # instead of just formats. + # This fixes incorrect format selection issue (see + # https://github.com/ytdl-org/youtube-dl/issues/10083). + incomplete_formats = ( + # All formats are video-only or + all(f.get('vcodec') != 'none' and f.get('acodec') == 'none' for f in formats) + # all formats are audio-only + or all(f.get('vcodec') == 'none' and f.get('acodec') != 'none' for f in formats)) + + ctx = { + 'formats': formats, + 'incomplete_formats': incomplete_formats, + } + + formats_to_download = list(format_selector(ctx)) if not formats_to_download: raise ExtractorError('requested format not available', expected=True) @@ -1206,6 +1694,36 @@ class YoutubeDL(object): subs[lang] = f return subs + def __forced_printings(self, info_dict, filename, incomplete): + def print_mandatory(field): + if (self.params.get('force%s' % field, False) + and (not incomplete or info_dict.get(field) is not None)): + self.to_stdout(info_dict[field]) + + def print_optional(field): + if (self.params.get('force%s' % field, False) + and info_dict.get(field) is not None): + self.to_stdout(info_dict[field]) + + print_mandatory('title') + print_mandatory('id') + if self.params.get('forceurl', False) and not incomplete: + if info_dict.get('requested_formats') is not None: + for f in info_dict['requested_formats']: + self.to_stdout(f['url'] + f.get('play_path', '')) + else: + # For RTMP URLs, also include the playpath + self.to_stdout(info_dict['url'] + info_dict.get('play_path', '')) + print_optional('thumbnail') + print_optional('description') + if self.params.get('forcefilename', False) and filename is not None: + self.to_stdout(filename) + if self.params.get('forceduration', False) and info_dict.get('duration') is not None: + self.to_stdout(formatSeconds(info_dict['duration'])) + print_mandatory('format') + if self.params.get('forcejson', False): + self.to_stdout(json.dumps(info_dict)) + def process_info(self, info_dict): """Process a single resolved IE result.""" @@ -1216,9 +1734,8 @@ class YoutubeDL(object): if self._num_downloads >= int(max_downloads): raise MaxDownloadsReached() + # TODO: backward compatibility, to be removed info_dict['fulltitle'] = info_dict['title'] - if len(info_dict['title']) > 200: - info_dict['title'] = info_dict['title'][:197] + '...' if 'format' not in info_dict: info_dict['format'] = info_dict['ext'] @@ -1233,29 +1750,7 @@ class YoutubeDL(object): info_dict['_filename'] = filename = self.prepare_filename(info_dict) # Forced printings - if self.params.get('forcetitle', False): - self.to_stdout(info_dict['fulltitle']) - if self.params.get('forceid', False): - self.to_stdout(info_dict['id']) - if self.params.get('forceurl', False): - if info_dict.get('requested_formats') is not None: - for f in info_dict['requested_formats']: - self.to_stdout(f['url'] + f.get('play_path', '')) - else: - # For RTMP URLs, also include the playpath - self.to_stdout(info_dict['url'] + info_dict.get('play_path', '')) - if self.params.get('forcethumbnail', False) and info_dict.get('thumbnail') is not None: - self.to_stdout(info_dict['thumbnail']) - if self.params.get('forcedescription', False) and info_dict.get('description') is not None: - self.to_stdout(info_dict['description']) - if self.params.get('forcefilename', False) and filename is not None: - self.to_stdout(filename) - if self.params.get('forceduration', False) and info_dict.get('duration') is not None: - self.to_stdout(formatSeconds(info_dict['duration'])) - if self.params.get('forceformat', False): - self.to_stdout(info_dict['format']) - if self.params.get('forcejson', False): - self.to_stdout(json.dumps(info_dict)) + self.__forced_printings(info_dict, filename, incomplete=False) # Do nothing else if in simulate mode if self.params.get('simulate', False): @@ -1264,12 +1759,17 @@ class YoutubeDL(object): if filename is None: return - try: - dn = os.path.dirname(sanitize_path(encodeFilename(filename))) - if dn and not os.path.exists(dn): - os.makedirs(dn) - except (OSError, IOError) as err: - self.report_error('unable to create directory ' + compat_str(err)) + def ensure_dir_exists(path): + try: + dn = os.path.dirname(path) + if dn and not os.path.exists(dn): + os.makedirs(dn) + return True + except (OSError, IOError) as err: + self.report_error('unable to create directory ' + error_to_compat_str(err)) + return False + + if not ensure_dir_exists(sanitize_path(encodeFilename(filename))): return if self.params.get('writedescription', False): @@ -1291,6 +1791,8 @@ class YoutubeDL(object): annofn = replace_extension(filename, 'annotations.xml', info_dict.get('ext')) if self.params.get('nooverwrites', False) and os.path.exists(encodeFilename(annofn)): self.to_screen('[info] Video annotations are already present') + elif not info_dict.get('annotations'): + self.report_warning('There are no annotations to write.') else: try: self.to_screen('[info] Writing video annotations to: ' + annofn) @@ -1312,27 +1814,30 @@ class YoutubeDL(object): ie = self.get_info_extractor(info_dict['extractor_key']) for sub_lang, sub_info in subtitles.items(): sub_format = sub_info['ext'] - if sub_info.get('data') is not None: - sub_data = sub_info['data'] + sub_filename = subtitles_filename(filename, sub_lang, sub_format) + if self.params.get('nooverwrites', False) and os.path.exists(encodeFilename(sub_filename)): + self.to_screen('[info] Video subtitle %s.%s is already present' % (sub_lang, sub_format)) else: - try: - sub_data = ie._download_webpage( - sub_info['url'], info_dict['id'], note=False) - except ExtractorError as err: - self.report_warning('Unable to download subtitle for "%s": %s' % - (sub_lang, compat_str(err.cause))) - continue - try: - sub_filename = subtitles_filename(filename, sub_lang, sub_format) - if self.params.get('nooverwrites', False) and os.path.exists(encodeFilename(sub_filename)): - self.to_screen('[info] Video subtitle %s.%s is already_present' % (sub_lang, sub_format)) + self.to_screen('[info] Writing video subtitles to: ' + sub_filename) + if sub_info.get('data') is not None: + try: + # Use newline='' to prevent conversion of newline characters + # See https://github.com/ytdl-org/youtube-dl/issues/10268 + with io.open(encodeFilename(sub_filename), 'w', encoding='utf-8', newline='') as subfile: + subfile.write(sub_info['data']) + except (OSError, IOError): + self.report_error('Cannot write subtitles file ' + sub_filename) + return else: - self.to_screen('[info] Writing video subtitles to: ' + sub_filename) - with io.open(encodeFilename(sub_filename), 'w', encoding='utf-8') as subfile: - subfile.write(sub_data) - except (OSError, IOError): - self.report_error('Cannot write subtitles file ' + sub_filename) - return + try: + sub_data = ie._request_webpage( + sub_info['url'], info_dict['id'], note=False).read() + with io.open(encodeFilename(sub_filename), 'wb') as subfile: + subfile.write(sub_data) + except (ExtractorError, IOError, OSError, ValueError) as err: + self.report_warning('Unable to download subtitle for "%s": %s' % + (sub_lang, error_to_compat_str(err))) + continue if self.params.get('writeinfojson', False): infofn = replace_extension(filename, 'info.json', info_dict.get('ext')) @@ -1366,17 +1871,17 @@ class YoutubeDL(object): postprocessors = [] self.report_warning('You have requested multiple ' 'formats but ffmpeg or avconv are not installed.' - ' The formats won\'t be merged') + ' The formats won\'t be merged.') else: postprocessors = [merger] def compatible_formats(formats): video, audio = formats # Check extension - video_ext, audio_ext = audio.get('ext'), video.get('ext') + video_ext, audio_ext = video.get('ext'), audio.get('ext') if video_ext and audio_ext: COMPATIBLE_EXTS = ( - ('mp3', 'mp4', 'm4a', 'm4p', 'm4b', 'm4r', 'm4v'), + ('mp3', 'mp4', 'm4a', 'm4p', 'm4b', 'm4r', 'm4v', 'ismv', 'isma'), ('webm') ) for exts in COMPATIBLE_EXTS: @@ -1393,8 +1898,8 @@ class YoutubeDL(object): requested_formats = info_dict['requested_formats'] if self.params.get('merge_output_format') is None and not compatible_formats(requested_formats): info_dict['ext'] = 'mkv' - self.report_warning('You have requested formats incompatible for merge. ' - 'The formats will be merged into mkv') + self.report_warning( + 'Requested formats are incompatible for merge and will be merged into mkv.') # Ensure filename always has a correct extension for successful merge filename = '%s.%s' % (filename_wo_ext, info_dict['ext']) if os.path.exists(encodeFilename(filename)): @@ -1405,8 +1910,11 @@ class YoutubeDL(object): for f in requested_formats: new_info = dict(info_dict) new_info.update(f) - fname = self.prepare_filename(new_info) - fname = prepend_extension(fname, 'f%s' % f['format_id'], new_info['ext']) + fname = prepend_extension( + self.prepare_filename(new_info), + 'f%s' % f['format_id'], new_info['ext']) + if not ensure_dir_exists(fname): + return downloaded.append(fname) partial_success = dl(fname, new_info) success = success and partial_success @@ -1416,7 +1924,7 @@ class YoutubeDL(object): # Just a single file success = dl(filename, info_dict) except (compat_urllib_error.URLError, compat_http_client.HTTPException, socket.error) as err: - self.report_error('unable to download video data: %s' % str(err)) + self.report_error('unable to download video data: %s' % error_to_compat_str(err)) return except (OSError, IOError) as err: raise UnavailableVideoError(err) @@ -1424,12 +1932,14 @@ class YoutubeDL(object): self.report_error('content too short (expected %s bytes and served %s)' % (err.expected, err.downloaded)) return - if success: + if success and filename != '-': # Fixup content fixup_policy = self.params.get('fixup') if fixup_policy is None: fixup_policy = 'detect_or_warn' + INSTALL_FFMPEG_MESSAGE = 'Install ffmpeg or avconv to fix this automatically.' + stretched_ratio = info_dict.get('stretched_ratio') if stretched_ratio is not None and stretched_ratio != 1: if fixup_policy == 'warn': @@ -1442,15 +1952,18 @@ class YoutubeDL(object): info_dict['__postprocessors'].append(stretched_pp) else: self.report_warning( - '%s: Non-uniform pixel ratio (%s). Install ffmpeg or avconv to fix this automatically.' % ( - info_dict['id'], stretched_ratio)) + '%s: Non-uniform pixel ratio (%s). %s' + % (info_dict['id'], stretched_ratio, INSTALL_FFMPEG_MESSAGE)) else: assert fixup_policy in ('ignore', 'never') - if info_dict.get('requested_formats') is None and info_dict.get('container') == 'm4a_dash': + if (info_dict.get('requested_formats') is None + and info_dict.get('container') == 'm4a_dash'): if fixup_policy == 'warn': - self.report_warning('%s: writing DASH m4a. Only some players support this container.' % ( - info_dict['id'])) + self.report_warning( + '%s: writing DASH m4a. ' + 'Only some players support this container.' + % info_dict['id']) elif fixup_policy == 'detect_or_warn': fixup_pp = FFmpegFixupM4aPP(self) if fixup_pp.available: @@ -1458,8 +1971,27 @@ class YoutubeDL(object): info_dict['__postprocessors'].append(fixup_pp) else: self.report_warning( - '%s: writing DASH m4a. Only some players support this container. Install ffmpeg or avconv to fix this automatically.' % ( - info_dict['id'])) + '%s: writing DASH m4a. ' + 'Only some players support this container. %s' + % (info_dict['id'], INSTALL_FFMPEG_MESSAGE)) + else: + assert fixup_policy in ('ignore', 'never') + + if (info_dict.get('protocol') == 'm3u8_native' + or info_dict.get('protocol') == 'm3u8' + and self.params.get('hls_prefer_native')): + if fixup_policy == 'warn': + self.report_warning('%s: malformed AAC bitstream detected.' % ( + info_dict['id'])) + elif fixup_policy == 'detect_or_warn': + fixup_pp = FFmpegFixupM3u8PP(self) + if fixup_pp.available: + info_dict.setdefault('__postprocessors', []) + info_dict['__postprocessors'].append(fixup_pp) + else: + self.report_warning( + '%s: malformed AAC bitstream detected. %s' + % (info_dict['id'], INSTALL_FFMPEG_MESSAGE)) else: assert fixup_policy in ('ignore', 'never') @@ -1473,15 +2005,17 @@ class YoutubeDL(object): def download(self, url_list): """Download a given list of URLs.""" outtmpl = self.params.get('outtmpl', DEFAULT_OUTTMPL) - if (len(url_list) > 1 and - '%' not in outtmpl and - self.params.get('max_downloads') != 1): + if (len(url_list) > 1 + and outtmpl != '-' + and '%' not in outtmpl + and self.params.get('max_downloads') != 1): raise SameFileError(outtmpl) for url in url_list: try: # It also downloads the videos - res = self.extract_info(url) + res = self.extract_info( + url, force_generic_extractor=self.params.get('force_generic_extractor', False)) except UnavailableVideoError: self.report_error('unable to download video') except MaxDownloadsReached: @@ -1525,6 +2059,7 @@ class YoutubeDL(object): pps_chain.extend(ie_info['__postprocessors']) pps_chain.extend(self._pps) for pp in pps_chain: + files_to_delete = [] try: files_to_delete, info = pp.run(info) except PostProcessingError as e: @@ -1538,15 +2073,24 @@ class YoutubeDL(object): self.report_warning('Unable to remove downloaded original file') def _make_archive_id(self, info_dict): + video_id = info_dict.get('id') + if not video_id: + return # Future-proof against any change in case # and backwards compatibility with prior versions - extractor = info_dict.get('extractor_key') - if extractor is None: - if 'id' in info_dict: - extractor = info_dict.get('ie_key') # key in a playlist + extractor = info_dict.get('extractor_key') or info_dict.get('ie_key') # key in a playlist if extractor is None: - return None # Incomplete video information - return extractor.lower() + ' ' + info_dict['id'] + url = str_or_none(info_dict.get('url')) + if not url: + return + # Try to find matching extractor for the URL and take its ie_key + for ie in self._ies: + if ie.suitable(url): + extractor = ie.ie_key() + break + else: + return + return extractor.lower() + ' ' + video_id def in_download_archive(self, info_dict): fn = self.params.get('download_archive') @@ -1554,7 +2098,7 @@ class YoutubeDL(object): return False vid_id = self._make_archive_id(info_dict) - if vid_id is None: + if not vid_id: return False # Incomplete video information try: @@ -1588,7 +2132,7 @@ class YoutubeDL(object): else: res = '%sp' % format['height'] elif format.get('width') is not None: - res = '?x%d' % format['width'] + res = '%dx?' % format['width'] else: res = default return res @@ -1597,6 +2141,10 @@ class YoutubeDL(object): res = '' if fdict.get('ext') in ['f4f', 'f4m']: res += '(unsupported) ' + if fdict.get('language'): + if res: + res += ' ' + res += '[%s] ' % fdict['language'] if fdict.get('format_note') is not None: res += fdict['format_note'] + ' ' if fdict.get('tbr') is not None: @@ -1605,8 +2153,8 @@ class YoutubeDL(object): if res: res += ', ' res += '%s container' % fdict['container'] - if (fdict.get('vcodec') is not None and - fdict.get('vcodec') != 'none'): + if (fdict.get('vcodec') is not None + and fdict.get('vcodec') != 'none'): if res: res += ', ' res += fdict['vcodec'] @@ -1617,7 +2165,9 @@ class YoutubeDL(object): if fdict.get('vbr') is not None: res += '%4dk' % fdict['vbr'] if fdict.get('fps') is not None: - res += ', %sfps' % fdict['fps'] + if res: + res += ', ' + res += '%sfps' % fdict['fps'] if fdict.get('acodec') is not None: if res: res += ', ' @@ -1660,13 +2210,8 @@ class YoutubeDL(object): def list_thumbnails(self, info_dict): thumbnails = info_dict.get('thumbnails') if not thumbnails: - tn_url = info_dict.get('thumbnail') - if tn_url: - thumbnails = [{'id': '0', 'url': tn_url}] - else: - self.to_screen( - '[info] No thumbnails present for %s' % info_dict['id']) - return + self.to_screen('[info] No thumbnails present for %s' % info_dict['id']) + return self.to_screen( '[info] Thumbnails for %s:' % info_dict['id']) @@ -1687,26 +2232,8 @@ class YoutubeDL(object): def urlopen(self, req): """ Start an HTTP download """ - - # According to RFC 3986, URLs can not contain non-ASCII characters, however this is not - # always respected by websites, some tend to give out URLs with non percent-encoded - # non-ASCII characters (see telemb.py, ard.py [#3412]) - # urllib chokes on URLs with non-ASCII characters (see http://bugs.python.org/issue3991) - # To work around aforementioned issue we will replace request's original URL with - # percent-encoded one - req_is_string = isinstance(req, compat_basestring) - url = req if req_is_string else req.get_full_url() - url_escaped = escape_url(url) - - # Substitute URL if any change after escaping - if url != url_escaped: - if req_is_string: - req = url_escaped - else: - req = compat_urllib_request.Request( - url_escaped, data=req.data, headers=req.headers, - origin_req_host=req.origin_req_host, unverifiable=req.unverifiable) - + if isinstance(req, compat_basestring): + req = sanitized_Request(req) return self._opener.open(req, timeout=self._socket_timeout) def print_debug_header(self): @@ -1714,7 +2241,7 @@ class YoutubeDL(object): return if type('') is not compat_str: - # Python 2.6 on SLES11 SP1 (https://github.com/rg3/youtube-dl/issues/3326) + # Python 2.6 on SLES11 SP1 (https://github.com/ytdl-org/youtube-dl/issues/3326) self.report_warning( 'Your Python is broken! Update to a newer and supported version') @@ -1729,6 +2256,8 @@ class YoutubeDL(object): write_string(encoding_str, encoding=None) self._write_string('[debug] youtube-dl version ' + __version__ + '\n') + if _LAZY_LOADER: + self._write_string('[debug] Lazy loading extractors enabled' + '\n') try: sp = subprocess.Popen( ['git', 'rev-parse', '--short', 'HEAD'], @@ -1743,11 +2272,20 @@ class YoutubeDL(object): sys.exc_clear() except Exception: pass - self._write_string('[debug] Python version %s - %s\n' % ( - platform.python_version(), platform_name())) + + def python_implementation(): + impl_name = platform.python_implementation() + if impl_name == 'PyPy' and hasattr(sys, 'pypy_version_info'): + return impl_name + ' version %d.%d.%d' % sys.pypy_version_info[:3] + return impl_name + + self._write_string('[debug] Python version %s (%s) - %s\n' % ( + platform.python_version(), python_implementation(), + platform_name())) exe_versions = FFmpegPostProcessor.get_versions(self) exe_versions['rtmpdump'] = rtmpdump_version() + exe_versions['phantomjs'] = PhantomJSwrapper._version() exe_str = ', '.join( '%s %s' % (exe, v) for exe, v in sorted(exe_versions.items()) @@ -1784,13 +2322,12 @@ class YoutubeDL(object): if opts_cookiefile is None: self.cookiejar = compat_cookiejar.CookieJar() else: - self.cookiejar = compat_cookiejar.MozillaCookieJar( - opts_cookiefile) + opts_cookiefile = expand_path(opts_cookiefile) + self.cookiejar = YoutubeDLCookieJar(opts_cookiefile) if os.access(opts_cookiefile, os.R_OK): - self.cookiejar.load() + self.cookiejar.load(ignore_discard=True, ignore_expires=True) - cookie_processor = compat_urllib_request.HTTPCookieProcessor( - self.cookiejar) + cookie_processor = YoutubeDLCookieProcessor(self.cookiejar) if opts_proxy is not None: if opts_proxy == '': proxies = {} @@ -1798,7 +2335,7 @@ class YoutubeDL(object): proxies = {'http': opts_proxy, 'https': opts_proxy} else: proxies = compat_urllib_request.getproxies() - # Set HTTPS proxy to HTTP one if given (https://github.com/rg3/youtube-dl/issues/805) + # Set HTTPS proxy to HTTP one if given (https://github.com/ytdl-org/youtube-dl/issues/805) if 'http' in proxies and 'https' not in proxies: proxies['https'] = proxies['http'] proxy_handler = PerRequestProxyHandler(proxies) @@ -1806,12 +2343,24 @@ class YoutubeDL(object): debuglevel = 1 if self.params.get('debug_printtraffic') else 0 https_handler = make_HTTPS_handler(self.params, debuglevel=debuglevel) ydlh = YoutubeDLHandler(self.params, debuglevel=debuglevel) + data_handler = compat_urllib_request_DataHandler() + + # When passing our own FileHandler instance, build_opener won't add the + # default FileHandler and allows us to disable the file protocol, which + # can be used for malicious purposes (see + # https://github.com/ytdl-org/youtube-dl/issues/8227) + file_handler = compat_urllib_request.FileHandler() + + def file_open(*args, **kwargs): + raise compat_urllib_error.URLError('file:// scheme is explicitly disabled in youtube-dl for security reasons') + file_handler.file_open = file_open + opener = compat_urllib_request.build_opener( - proxy_handler, https_handler, cookie_processor, ydlh) + proxy_handler, https_handler, cookie_processor, ydlh, data_handler, file_handler) # Delete the default user-agent header, which would otherwise apply in # cases where our custom HTTP handler doesn't come into play - # (See https://github.com/rg3/youtube-dl/issues/1309 for details) + # (See https://github.com/ytdl-org/youtube-dl/issues/1309 for details) opener.addheaders = [] self._opener = opener @@ -1859,10 +2408,10 @@ class YoutubeDL(object): (info_dict['extractor'], info_dict['id'], thumb_display_id)) try: uf = self.urlopen(t['url']) - with open(thumb_filename, 'wb') as thumbf: + with open(encodeFilename(thumb_filename), 'wb') as thumbf: shutil.copyfileobj(uf, thumbf) self.to_screen('[%s] %s: Writing thumbnail %sto: %s' % (info_dict['extractor'], info_dict['id'], thumb_display_id, thumb_filename)) except (compat_urllib_error.URLError, compat_http_client.HTTPException, socket.error) as err: self.report_warning('Unable to download thumbnail "%s": %s' % - (t['url'], compat_str(err))) + (t['url'], error_to_compat_str(err)))