X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/d9d7cd0e85dc712461d9185db9df9d6c900a573b..b6db0e9853c4704fd25a056749ab237cc74bcd00:/youtube_dl/YoutubeDL.py diff --git a/youtube_dl/YoutubeDL.py b/youtube_dl/YoutubeDL.py index 4c33d49..2a405c5 100755 --- a/youtube_dl/YoutubeDL.py +++ b/youtube_dl/YoutubeDL.py @@ -26,6 +26,8 @@ import tokenize import traceback import random +from string import ascii_letters + from .compat import ( compat_basestring, compat_cookiejar, @@ -58,10 +60,12 @@ from .utils import ( format_bytes, formatSeconds, GeoRestrictedError, + int_or_none, ISO3166Utils, locked_file, make_HTTPS_handler, MaxDownloadsReached, + orderedSet, PagedList, parse_filesize, PerRequestProxyHandler, @@ -89,6 +93,7 @@ from .utils import ( ) from .cache import Cache 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 ( @@ -206,7 +211,7 @@ class YoutubeDL(object): At the moment, this is only supported by YouTube. proxy: URL of the proxy server to use geo_verification_proxy: URL of the proxy to use for IP address verification - on geo-restricted sites. (Experimental) + 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 @@ -254,7 +259,7 @@ 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 when @@ -276,11 +281,14 @@ class YoutubeDL(object): 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 (experimental) + 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 (experimental) + 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. @@ -293,15 +301,33 @@ class YoutubeDL(object): 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, hls_use_mpegts. + 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. 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 = [] @@ -498,24 +524,29 @@ class YoutubeDL(object): def to_console_title(self, message): if not self.params.get('consoletitle', False): return - if compat_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) @@ -638,22 +669,11 @@ class YoutubeDL(object): r'%%(\1)0%dd' % field_size_compat_map[mobj.group('field')], outtmpl) - 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', - )) - # 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 NUMERIC_FIELDS: + 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 @@ -672,7 +692,19 @@ class YoutubeDL(object): FORMAT_RE.format(numeric_field), r'%({0})s'.format(numeric_field), outtmpl) - filename = expand_path(outtmpl % template_dict) + # expand_path translates '%%' into '%' and '$$' into '$' + # correspondingly that is not what we want since we need to keep + # '%%' intact for template dict substitution step. Working around + # with boundary-alike separator hack. + sep = ''.join([random.choice(ascii_letters) for _ in range(32)]) + outtmpl = outtmpl.replace('%%', '%{0}%'.format(sep)).replace('$$', '${0}$'.format(sep)) + + # outtmpl should be expand_path'ed before template dict substitution + # because meta fields may contain env variables we don't want to + # be expanded. For example, for outtmpl "%(title)s.%(ext)s" and + # title "Hello $PATH", we don't want `$PATH` to be expanded. + filename = expand_path(outtmpl).replace(sep, '') % template_dict + # Temporary fix for #4787 # 'Treat' all problem characters by passing filename through preferredencoding # to workaround encoding issues with subprocess on python2 @ Windows @@ -844,7 +876,7 @@ class YoutubeDL(object): force_properties = dict( (k, v) for k, v in ie_result.items() if v is not None) - for f in ('_type', 'url', 'ie_key'): + for f in ('_type', 'url', 'id', 'extractor', 'extractor_key', 'ie_key'): if f in force_properties: del force_properties[f] new_result = info.copy() @@ -885,15 +917,25 @@ class YoutubeDL(object): yield int(item) else: yield int(string_segment) - playlistitems = iter_playlistitems(playlistitems_str) + playlistitems = orderedSet(iter_playlistitems(playlistitems_str)) ie_entries = ie_result['entries'] + + def make_playlistitems_entries(list_ie_entries): + num_entries = len(list_ie_entries) + return [ + list_ie_entries[i - 1] for i in playlistitems + if -num_entries <= i - 1 < num_entries] + + def report_download(num_entries): + self.to_screen( + '[%s] playlist %s: Downloading %d videos' % + (ie_result['extractor'], playlist, num_entries)) + if isinstance(ie_entries, list): n_all_entries = len(ie_entries) if playlistitems: - entries = [ - ie_entries[i - 1] for i in playlistitems - if -n_all_entries <= i - 1 < n_all_entries] + entries = make_playlistitems_entries(ie_entries) else: entries = ie_entries[playliststart:playlistend] n_entries = len(entries) @@ -911,20 +953,16 @@ class YoutubeDL(object): entries = ie_entries.getslice( playliststart, playlistend) n_entries = len(entries) - self.to_screen( - '[%s] playlist %s: Downloading %d videos' % - (ie_result['extractor'], playlist, n_entries)) + report_download(n_entries) else: # iterable if playlistitems: - entry_list = list(ie_entries) - entries = [entry_list[i - 1] for i in playlistitems] + entries = make_playlistitems_entries(list(itertools.islice( + ie_entries, 0, max(playlistitems)))) else: entries = list(itertools.islice( ie_entries, playliststart, playlistend)) n_entries = len(entries) - self.to_screen( - '[%s] playlist %s: Downloading %d videos' % - (ie_result['extractor'], playlist, n_entries)) + report_download(n_entries) if self.params.get('playlistreverse', False): entries = entries[::-1] @@ -945,6 +983,8 @@ class YoutubeDL(object): 'playlist': playlist, 'playlist_id': ie_result.get('id'), 'playlist_title': ie_result.get('title'), + 'playlist_uploader': ie_result.get('uploader'), + 'playlist_uploader_id': ie_result.get('uploader_id'), 'playlist_index': i + playliststart, 'extractor': ie_result['extractor'], 'webpage_url': ie_result['webpage_url'], @@ -1000,7 +1040,7 @@ class YoutubeDL(object): '!=': operator.ne, } operator_rex = re.compile(r'''(?x)\s* - (?Pwidth|height|tbr|abr|vbr|asr|filesize|fps) + (?Pwidth|height|tbr|abr|vbr|asr|filesize|filesize_approx|fps) \s*(?P%s)(?P\s*\?)?\s* (?P[0-9.]+(?:[kKmMgGtTpPeEzZyY]i?[Bb]?)?) $ @@ -1048,6 +1088,30 @@ class YoutubeDL(object): return op(actual_value, comparison_value) 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 + + 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 = ( @@ -1344,9 +1408,28 @@ class YoutubeDL(object): if 'title' not in info_dict: raise ExtractorError('Missing "title" field in extractor result') - if not isinstance(info_dict['id'], compat_str): - self.report_warning('"id" field is not a string - forcing string conversion') - info_dict['id'] = compat_str(info_dict['id']) + 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 @@ -1399,23 +1482,28 @@ class YoutubeDL(object): 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 subtitles: - for _, subtitle in subtitles.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() 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'], automatic_captions, 'automatic captions') self.list_subtitles(info_dict['id'], subtitles, 'subtitles') return + info_dict['requested_subtitles'] = self.process_subtitles( - info_dict['id'], subtitles, - info_dict.get('automatic_captions')) + info_dict['id'], subtitles, automatic_captions) # We now pick which formats have to be downloaded if info_dict.get('formats') is None: @@ -1427,16 +1515,28 @@ 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) - + sanitize_string_field(format, 'format_id') + sanitize_numeric_fields(format) format['url'] = sanitize_url(format['url']) - - if format.get('format_id') is None: + if not format.get('format_id'): format['format_id'] = compat_str(i) else: # Sanitize format_id from characters used in format selector expression @@ -1489,14 +1589,10 @@ class YoutubeDL(object): req_format = self.params.get('format') if req_format is None: - req_format_list = [] - if (self.params.get('outtmpl', DEFAULT_OUTTMPL) != '-' and - not info_dict.get('is_live')): - 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) + 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 @@ -1648,12 +1744,17 @@ 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 ' + error_to_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): @@ -1696,29 +1797,30 @@ 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, error_to_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/rg3/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) - # Use newline='' to prevent conversion of newline characters - # See https://github.com/rg3/youtube-dl/issues/10268 - with io.open(encodeFilename(sub_filename), 'w', encoding='utf-8', newline='') 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')) @@ -1759,7 +1861,7 @@ class YoutubeDL(object): 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', 'ismv', 'isma'), @@ -1791,8 +1893,11 @@ 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 @@ -1859,7 +1964,7 @@ class YoutubeDL(object): info_dict.get('protocol') == 'm3u8' and self.params.get('hls_prefer_native')): if fixup_policy == 'warn': - self.report_warning('%s: malformated aac bitstream.' % ( + self.report_warning('%s: malformed AAC bitstream detected.' % ( info_dict['id'])) elif fixup_policy == 'detect_or_warn': fixup_pp = FFmpegFixupM3u8PP(self) @@ -1868,7 +1973,7 @@ class YoutubeDL(object): info_dict['__postprocessors'].append(fixup_pp) else: self.report_warning( - '%s: malformated aac bitstream. %s' + '%s: malformed AAC bitstream detected. %s' % (info_dict['id'], INSTALL_FFMPEG_MESSAGE)) else: assert fixup_policy in ('ignore', 'never') @@ -2141,11 +2246,20 @@ 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())