X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/99b46abf79266c2bf5eabc291dc4f366e61b5dde..d2632ebbe0759622d4ab7aff134421194974b394:/youtube_dl/extractor/common.py diff --git a/youtube_dl/extractor/common.py b/youtube_dl/extractor/common.py index 9427ff4..0b4e2ac 100644 --- a/youtube_dl/extractor/common.py +++ b/youtube_dl/extractor/common.py @@ -21,6 +21,7 @@ from ..compat import ( compat_os_name, compat_str, compat_urllib_error, + compat_urllib_parse_unquote, compat_urllib_parse_urlencode, compat_urllib_request, compat_urlparse, @@ -29,6 +30,7 @@ from ..downloader.f4m import remove_encrypted_media from ..utils import ( NO_DEFAULT, age_restricted, + base_url, bug_reports_message, clean_html, compiled_regex_type, @@ -57,6 +59,7 @@ from ..utils import ( parse_m3u8_attributes, extract_attributes, parse_codecs, + urljoin, ) @@ -87,6 +90,9 @@ class InfoExtractor(object): Potential fields: * url Mandatory. The URL of the video file + * manifest_url + The URL of the manifest file in case of + fragmented media (DASH, hls, hds) * ext Will be calculated from URL if missing * format A human-readable description of the format ("mp4 container with h264/opus"). @@ -115,6 +121,21 @@ class InfoExtractor(object): download, lower-case. "http", "https", "rtsp", "rtmp", "rtmpe", "m3u8", "m3u8_native" or "http_dash_segments". + * fragment_base_url + Base URL for fragments. Each fragment's path + value (if present) will be relative to + this URL. + * fragments A list of fragments of a fragmented media. + Each fragment entry must contain either an url + or a path. If an url is present it should be + considered by a client. Otherwise both path and + fragment_base_url must be present. Here is + the list of all potential fields: + * "url" - fragment's URL + * "path" - fragment's path relative to + fragment_base_url + * "duration" (optional, int or float) + * "filesize" (optional, int) * preference Order number of this format. If this field is present and not None, the formats get sorted by this field, regardless of all other values. @@ -178,9 +199,10 @@ class InfoExtractor(object): uploader_url: Full URL to a personal webpage of the video uploader. location: Physical location where the video was filmed. subtitles: The available subtitles as a dictionary in the format - {language: subformats}. "subformats" is a list sorted from - lower to higher preference, each element is a dictionary - with the "ext" entry and one of: + {tag: subformats}. "tag" is usually a language code, and + "subformats" is a list sorted from lower to higher + preference, each element is a dictionary with the "ext" + entry and one of: * "data": The subtitles file contents * "url": A URL pointing to the subtitles file "ext" will be calculated from URL if missing @@ -226,7 +248,7 @@ class InfoExtractor(object): chapter_id: Id of the chapter the video belongs to, as a unicode string. The following fields should only be used when the video is an episode of some - series or programme: + series, programme or podcast: series: Title of the series or programme the video episode belongs to. season: Title of the season the video episode belongs to. @@ -674,33 +696,36 @@ class InfoExtractor(object): username = info[0] password = info[2] else: - raise netrc.NetrcParseError('No authenticators for %s' % netrc_machine) + raise netrc.NetrcParseError( + 'No authenticators for %s' % netrc_machine) except (IOError, netrc.NetrcParseError) as err: - self._downloader.report_warning('parsing .netrc: %s' % error_to_compat_str(err)) + self._downloader.report_warning( + 'parsing .netrc: %s' % error_to_compat_str(err)) - return (username, password) + return username, password - def _get_login_info(self): + def _get_login_info(self, username_option='username', password_option='password', netrc_machine=None): """ Get the login info as (username, password) - It will look in the netrc file using the _NETRC_MACHINE value + First look for the manually specified credentials using username_option + and password_option as keys in params dictionary. If no such credentials + available look in the netrc file using the netrc_machine or _NETRC_MACHINE + value. If there's no info available, return (None, None) """ if self._downloader is None: return (None, None) - username = None - password = None downloader_params = self._downloader.params # Attempt to use provided username and password or .netrc data - if downloader_params.get('username') is not None: - username = downloader_params['username'] - password = downloader_params['password'] + if downloader_params.get(username_option) is not None: + username = downloader_params[username_option] + password = downloader_params[password_option] else: - username, password = self._get_netrc_login_info() + username, password = self._get_netrc_login_info(netrc_machine) - return (username, password) + return username, password def _get_tfa_info(self, note='two-factor verification code'): """ @@ -873,7 +898,7 @@ class InfoExtractor(object): 'url': e.get('contentUrl'), 'title': unescapeHTML(e.get('name')), 'description': unescapeHTML(e.get('description')), - 'thumbnail': e.get('thumbnailUrl'), + 'thumbnail': e.get('thumbnailUrl') or e.get('thumbnailURL'), 'duration': parse_duration(e.get('duration')), 'timestamp': unified_timestamp(e.get('uploadDate')), 'filesize': float_or_none(e.get('contentSize')), @@ -888,16 +913,16 @@ class InfoExtractor(object): def _hidden_inputs(html): html = re.sub(r'', '', html) hidden_inputs = {} - for input in re.findall(r'(?i)]+)>', html): - if not re.search(r'type=(["\'])(?:hidden|submit)\1', input): - continue - name = re.search(r'(?:name|id)=(["\'])(?P.+?)\1', input) - if not name: + for input in re.findall(r'(?i)(]+>)', html): + attrs = extract_attributes(input) + if not input: continue - value = re.search(r'value=(["\'])(?P.*?)\1', input) - if not value: + if attrs.get('type') not in ('hidden', 'submit'): continue - hidden_inputs[name.group('value')] = value.group('value') + name = attrs.get('name') or attrs.get('id') + value = attrs.get('value') + if name and value is not None: + hidden_inputs[name] = value return hidden_inputs def _form_hidden_inputs(self, form_id, html): @@ -1000,13 +1025,13 @@ class InfoExtractor(object): unique_formats.append(f) formats[:] = unique_formats - def _is_valid_url(self, url, video_id, item='video'): + def _is_valid_url(self, url, video_id, item='video', headers={}): url = self._proto_relative_url(url, scheme='http:') # For now assume non HTTP(S) URLs always valid if not (url.startswith('http://') or url.startswith('https://')): return True try: - self._request_webpage(url, video_id, 'Checking %s URL' % item) + self._request_webpage(url, video_id, 'Checking %s URL' % item, headers=headers) return True except ExtractorError as e: if isinstance(e.cause, compat_urllib_error.URLError): @@ -1088,6 +1113,13 @@ class InfoExtractor(object): manifest, ['{http://ns.adobe.com/f4m/1.0}bootstrapInfo', '{http://ns.adobe.com/f4m/2.0}bootstrapInfo'], 'bootstrap info', default=None) + vcodec = None + mime_type = xpath_text( + manifest, ['{http://ns.adobe.com/f4m/1.0}mimeType', '{http://ns.adobe.com/f4m/2.0}mimeType'], + 'base URL', default=None) + if mime_type and mime_type.startswith('audio/'): + vcodec = 'none' + for i, media_el in enumerate(media_nodes): tbr = int_or_none(media_el.attrib.get('bitrate')) width = int_or_none(media_el.attrib.get('width')) @@ -1128,6 +1160,7 @@ class InfoExtractor(object): 'width': f.get('width') or width, 'height': f.get('height') or height, 'format_id': f.get('format_id') if not tbr else format_id, + 'vcodec': vcodec, }) formats.extend(f4m_formats) continue @@ -1139,10 +1172,12 @@ class InfoExtractor(object): formats.append({ 'format_id': format_id, 'url': manifest_url, + 'manifest_url': manifest_url, 'ext': 'flv' if bootstrap_info is not None else None, 'tbr': tbr, 'width': width, 'height': height, + 'vcodec': vcodec, 'preference': preference, }) return formats @@ -1163,13 +1198,6 @@ class InfoExtractor(object): m3u8_id=None, note=None, errnote=None, fatal=True, live=False): - formats = [self._m3u8_meta_format(m3u8_url, ext, preference, m3u8_id)] - - format_url = lambda u: ( - u - if re.match(r'^https?://', u) - else compat_urlparse.urljoin(m3u8_url, u)) - res = self._download_webpage_handle( m3u8_url, video_id, note=note or 'Downloading m3u8 information', @@ -1180,6 +1208,13 @@ class InfoExtractor(object): m3u8_doc, urlh = res m3u8_url = urlh.geturl() + formats = [self._m3u8_meta_format(m3u8_url, ext, preference, m3u8_id)] + + format_url = lambda u: ( + u + if re.match(r'^https?://', u) + else compat_urlparse.urljoin(m3u8_url, u)) + # We should try extracting formats only from master playlists [1], i.e. # playlists that describe available qualities. On the other hand media # playlists [2] should be returned as is since they contain just the media @@ -1201,35 +1236,62 @@ class InfoExtractor(object): 'protocol': entry_protocol, 'preference': preference, }] - last_info = None - last_media = None + audio_in_video_stream = {} + last_info = {} + last_media = {} for line in m3u8_doc.splitlines(): if line.startswith('#EXT-X-STREAM-INF:'): last_info = parse_m3u8_attributes(line) elif line.startswith('#EXT-X-MEDIA:'): - last_media = parse_m3u8_attributes(line) + media = parse_m3u8_attributes(line) + media_type = media.get('TYPE') + if media_type in ('VIDEO', 'AUDIO'): + group_id = media.get('GROUP-ID') + media_url = media.get('URI') + if media_url: + format_id = [] + for v in (group_id, media.get('NAME')): + if v: + format_id.append(v) + f = { + 'format_id': '-'.join(format_id), + 'url': format_url(media_url), + 'language': media.get('LANGUAGE'), + 'ext': ext, + 'protocol': entry_protocol, + 'preference': preference, + } + if media_type == 'AUDIO': + f['vcodec'] = 'none' + if group_id and not audio_in_video_stream.get(group_id): + audio_in_video_stream[group_id] = False + formats.append(f) + else: + # When there is no URI in EXT-X-MEDIA let this tag's + # data be used by regular URI lines below + last_media = media + if media_type == 'AUDIO' and group_id: + audio_in_video_stream[group_id] = True elif line.startswith('#') or not line.strip(): continue else: - if last_info is None: - formats.append({'url': format_url(line)}) - continue - tbr = int_or_none(last_info.get('BANDWIDTH'), scale=1000) + tbr = int_or_none(last_info.get('AVERAGE-BANDWIDTH') or last_info.get('BANDWIDTH'), scale=1000) format_id = [] if m3u8_id: format_id.append(m3u8_id) - last_media_name = last_media.get('NAME') if last_media and last_media.get('TYPE') not in ('SUBTITLES', 'CLOSED-CAPTIONS') else None # Despite specification does not mention NAME attribute for # EXT-X-STREAM-INF it still sometimes may be present - stream_name = last_info.get('NAME') or last_media_name + stream_name = last_info.get('NAME') or last_media.get('NAME') # Bandwidth of live streams may differ over time thus making # format_id unpredictable. So it's better to keep provided # format_id intact. if not live: format_id.append(stream_name if stream_name else '%d' % (tbr if tbr else len(formats))) + manifest_url = format_url(line.strip()) f = { 'format_id': '-'.join(format_id), - 'url': format_url(line.strip()), + 'url': manifest_url, + 'manifest_url': manifest_url, 'tbr': tbr, 'ext': ext, 'fps': float_or_none(last_info.get('FRAME-RATE')), @@ -1238,9 +1300,10 @@ class InfoExtractor(object): } resolution = last_info.get('RESOLUTION') if resolution: - width_str, height_str = resolution.split('x') - f['width'] = int(width_str) - f['height'] = int(height_str) + mobj = re.search(r'(?P\d+)[xX](?P\d+)', resolution) + if mobj: + f['width'] = int(mobj.group('width')) + f['height'] = int(mobj.group('height')) # Unified Streaming Platform mobj = re.search( r'audio.*?(?:%3D|=)(\d+)(?:-video.*?(?:%3D|=)(\d+))?', f['url']) @@ -1252,11 +1315,12 @@ class InfoExtractor(object): 'abr': abr, }) f.update(parse_codecs(last_info.get('CODECS'))) - if last_media is not None: - f['m3u8_media'] = last_media - last_media = None + if audio_in_video_stream.get(last_info.get('AUDIO')) is False and f['vcodec'] != 'none': + # TODO: update acodec for audio only formats with the same GROUP-ID + f['acodec'] = 'none' formats.append(f) last_info = {} + last_media = {} return formats @staticmethod @@ -1500,12 +1564,13 @@ class InfoExtractor(object): if res is False: return [] mpd, urlh = res - mpd_base_url = re.match(r'https?://.+/', urlh.geturl()).group() + mpd_base_url = base_url(urlh.geturl()) return self._parse_mpd_formats( - compat_etree_fromstring(mpd.encode('utf-8')), mpd_id, mpd_base_url, formats_dict=formats_dict) + compat_etree_fromstring(mpd.encode('utf-8')), mpd_id, mpd_base_url, + formats_dict=formats_dict, mpd_url=mpd_url) - def _parse_mpd_formats(self, mpd_doc, mpd_id=None, mpd_base_url='', formats_dict={}): + def _parse_mpd_formats(self, mpd_doc, mpd_id=None, mpd_base_url='', formats_dict={}, mpd_url=None): """ Parse formats from MPD manifest. References: @@ -1526,52 +1591,60 @@ class InfoExtractor(object): def extract_multisegment_info(element, ms_parent_info): ms_info = ms_parent_info.copy() + + # As per [1, 5.3.9.2.2] SegmentList and SegmentTemplate share some + # common attributes and elements. We will only extract relevant + # for us. + def extract_common(source): + segment_timeline = source.find(_add_ns('SegmentTimeline')) + if segment_timeline is not None: + s_e = segment_timeline.findall(_add_ns('S')) + if s_e: + ms_info['total_number'] = 0 + ms_info['s'] = [] + for s in s_e: + r = int(s.get('r', 0)) + ms_info['total_number'] += 1 + r + ms_info['s'].append({ + 't': int(s.get('t', 0)), + # @d is mandatory (see [1, 5.3.9.6.2, Table 17, page 60]) + 'd': int(s.attrib['d']), + 'r': r, + }) + start_number = source.get('startNumber') + if start_number: + ms_info['start_number'] = int(start_number) + timescale = source.get('timescale') + if timescale: + ms_info['timescale'] = int(timescale) + segment_duration = source.get('duration') + if segment_duration: + ms_info['segment_duration'] = int(segment_duration) + + def extract_Initialization(source): + initialization = source.find(_add_ns('Initialization')) + if initialization is not None: + ms_info['initialization_url'] = initialization.attrib['sourceURL'] + segment_list = element.find(_add_ns('SegmentList')) if segment_list is not None: + extract_common(segment_list) + extract_Initialization(segment_list) segment_urls_e = segment_list.findall(_add_ns('SegmentURL')) if segment_urls_e: ms_info['segment_urls'] = [segment.attrib['media'] for segment in segment_urls_e] - initialization = segment_list.find(_add_ns('Initialization')) - if initialization is not None: - ms_info['initialization_url'] = initialization.attrib['sourceURL'] else: segment_template = element.find(_add_ns('SegmentTemplate')) if segment_template is not None: - start_number = segment_template.get('startNumber') - if start_number: - ms_info['start_number'] = int(start_number) - segment_timeline = segment_template.find(_add_ns('SegmentTimeline')) - if segment_timeline is not None: - s_e = segment_timeline.findall(_add_ns('S')) - if s_e: - ms_info['total_number'] = 0 - ms_info['s'] = [] - for s in s_e: - r = int(s.get('r', 0)) - ms_info['total_number'] += 1 + r - ms_info['s'].append({ - 't': int(s.get('t', 0)), - # @d is mandatory (see [1, 5.3.9.6.2, Table 17, page 60]) - 'd': int(s.attrib['d']), - 'r': r, - }) - else: - timescale = segment_template.get('timescale') - if timescale: - ms_info['timescale'] = int(timescale) - segment_duration = segment_template.get('duration') - if segment_duration: - ms_info['segment_duration'] = int(segment_duration) - media_template = segment_template.get('media') - if media_template: - ms_info['media_template'] = media_template + extract_common(segment_template) + media = segment_template.get('media') + if media: + ms_info['media'] = media initialization = segment_template.get('initialization') if initialization: - ms_info['initialization_url'] = initialization + ms_info['initialization'] = initialization else: - initialization = segment_template.find(_add_ns('Initialization')) - if initialization is not None: - ms_info['initialization_url'] = initialization.attrib['sourceURL'] + extract_Initialization(segment_template) return ms_info mpd_duration = parse_duration(mpd_doc.get('mediaPresentationDuration')) @@ -1613,74 +1686,127 @@ class InfoExtractor(object): lang = representation_attrib.get('lang') url_el = representation.find(_add_ns('BaseURL')) filesize = int_or_none(url_el.attrib.get('{http://youtube.com/yt/2012/10/10}contentLength') if url_el is not None else None) + bandwidth = int_or_none(representation_attrib.get('bandwidth')) f = { 'format_id': '%s-%s' % (mpd_id, representation_id) if mpd_id else representation_id, 'url': base_url, + 'manifest_url': mpd_url, 'ext': mimetype2ext(mime_type), 'width': int_or_none(representation_attrib.get('width')), 'height': int_or_none(representation_attrib.get('height')), - 'tbr': int_or_none(representation_attrib.get('bandwidth'), 1000), + 'tbr': int_or_none(bandwidth, 1000), 'asr': int_or_none(representation_attrib.get('audioSamplingRate')), 'fps': int_or_none(representation_attrib.get('frameRate')), - 'vcodec': 'none' if content_type == 'audio' else representation_attrib.get('codecs'), - 'acodec': 'none' if content_type == 'video' else representation_attrib.get('codecs'), 'language': lang if lang not in ('mul', 'und', 'zxx', 'mis') else None, 'format_note': 'DASH %s' % content_type, 'filesize': filesize, } + f.update(parse_codecs(representation_attrib.get('codecs'))) representation_ms_info = extract_multisegment_info(representation, adaption_set_ms_info) - if 'segment_urls' not in representation_ms_info and 'media_template' in representation_ms_info: - if 'total_number' not in representation_ms_info and 'segment_duration': - segment_duration = float(representation_ms_info['segment_duration']) / float(representation_ms_info['timescale']) - representation_ms_info['total_number'] = int(math.ceil(float(period_duration) / segment_duration)) - media_template = representation_ms_info['media_template'] - media_template = media_template.replace('$RepresentationID$', representation_id) - media_template = re.sub(r'\$(Number|Bandwidth|Time)\$', r'%(\1)d', media_template) - media_template = re.sub(r'\$(Number|Bandwidth|Time)%([^$]+)\$', r'%(\1)\2', media_template) - media_template.replace('$$', '$') + + def prepare_template(template_name, identifiers): + t = representation_ms_info[template_name] + t = t.replace('$RepresentationID$', representation_id) + t = re.sub(r'\$(%s)\$' % '|'.join(identifiers), r'%(\1)d', t) + t = re.sub(r'\$(%s)%%([^$]+)\$' % '|'.join(identifiers), r'%(\1)\2', t) + t.replace('$$', '$') + return t + + # @initialization is a regular template like @media one + # so it should be handled just the same way (see + # https://github.com/rg3/youtube-dl/issues/11605) + if 'initialization' in representation_ms_info: + initialization_template = prepare_template( + 'initialization', + # As per [1, 5.3.9.4.2, Table 15, page 54] $Number$ and + # $Time$ shall not be included for @initialization thus + # only $Bandwidth$ remains + ('Bandwidth', )) + representation_ms_info['initialization_url'] = initialization_template % { + 'Bandwidth': bandwidth, + } + + if 'segment_urls' not in representation_ms_info and 'media' in representation_ms_info: + + media_template = prepare_template('media', ('Number', 'Bandwidth', 'Time')) # As per [1, 5.3.9.4.4, Table 16, page 55] $Number$ and $Time$ # can't be used at the same time - if '%(Number' in media_template: - representation_ms_info['segment_urls'] = [ - media_template % { + if '%(Number' in media_template and 's' not in representation_ms_info: + segment_duration = None + if 'total_number' not in representation_ms_info and 'segment_duration': + segment_duration = float_or_none(representation_ms_info['segment_duration'], representation_ms_info['timescale']) + representation_ms_info['total_number'] = int(math.ceil(float(period_duration) / segment_duration)) + representation_ms_info['fragments'] = [{ + 'url': media_template % { 'Number': segment_number, - 'Bandwidth': representation_attrib.get('bandwidth'), - } - for segment_number in range( - representation_ms_info['start_number'], - representation_ms_info['total_number'] + representation_ms_info['start_number'])] + 'Bandwidth': bandwidth, + }, + 'duration': segment_duration, + } for segment_number in range( + representation_ms_info['start_number'], + representation_ms_info['total_number'] + representation_ms_info['start_number'])] else: - representation_ms_info['segment_urls'] = [] + # $Number*$ or $Time$ in media template with S list available + # Example $Number*$: http://www.svtplay.se/klipp/9023742/stopptid-om-bjorn-borg + # Example $Time$: https://play.arkena.com/embed/avp/v2/player/media/b41dda37-d8e7-4d3f-b1b5-9a9db578bdfe/1/129411 + representation_ms_info['fragments'] = [] segment_time = 0 + segment_d = None + segment_number = representation_ms_info['start_number'] def add_segment_url(): - representation_ms_info['segment_urls'].append( - media_template % { - 'Time': segment_time, - 'Bandwidth': representation_attrib.get('bandwidth'), - } - ) + segment_url = media_template % { + 'Time': segment_time, + 'Bandwidth': bandwidth, + 'Number': segment_number, + } + representation_ms_info['fragments'].append({ + 'url': segment_url, + 'duration': float_or_none(segment_d, representation_ms_info['timescale']), + }) for num, s in enumerate(representation_ms_info['s']): segment_time = s.get('t') or segment_time + segment_d = s['d'] add_segment_url() + segment_number += 1 for r in range(s.get('r', 0)): - segment_time += s['d'] + segment_time += segment_d add_segment_url() - segment_time += s['d'] - if 'segment_urls' in representation_ms_info: + segment_number += 1 + segment_time += segment_d + elif 'segment_urls' in representation_ms_info and 's' in representation_ms_info: + # No media template + # Example: https://www.youtube.com/watch?v=iXZV5uAYMJI + # or any YouTube dashsegments video + fragments = [] + segment_index = 0 + timescale = representation_ms_info['timescale'] + for s in representation_ms_info['s']: + duration = float_or_none(s['d'], timescale) + for r in range(s.get('r', 0) + 1): + fragments.append({ + 'url': representation_ms_info['segment_urls'][segment_index], + 'duration': duration, + }) + segment_index += 1 + representation_ms_info['fragments'] = fragments + # NB: MPD manifest may contain direct URLs to unfragmented media. + # No fragments key is present in this case. + if 'fragments' in representation_ms_info: f.update({ - 'segment_urls': representation_ms_info['segment_urls'], + 'fragments': [], 'protocol': 'http_dash_segments', }) if 'initialization_url' in representation_ms_info: - initialization_url = representation_ms_info['initialization_url'].replace('$RepresentationID$', representation_id) - f.update({ - 'initialization_url': initialization_url, - }) + initialization_url = representation_ms_info['initialization_url'] if not f.get('url'): f['url'] = initialization_url + f['fragments'].append({'url': initialization_url}) + f['fragments'].extend(representation_ms_info['fragments']) + for fragment in f['fragments']: + fragment['url'] = urljoin(base_url, fragment['url']) try: existing_format = next( fo for fo in formats @@ -1695,7 +1821,106 @@ class InfoExtractor(object): self.report_warning('Unknown MIME type %s in DASH manifest' % mime_type) return formats - def _parse_html5_media_entries(self, base_url, webpage): + def _extract_ism_formats(self, ism_url, video_id, ism_id=None, note=None, errnote=None, fatal=True): + res = self._download_webpage_handle( + ism_url, video_id, + note=note or 'Downloading ISM manifest', + errnote=errnote or 'Failed to download ISM manifest', + fatal=fatal) + if res is False: + return [] + ism, urlh = res + + return self._parse_ism_formats( + compat_etree_fromstring(ism.encode('utf-8')), urlh.geturl(), ism_id) + + def _parse_ism_formats(self, ism_doc, ism_url, ism_id=None): + if ism_doc.get('IsLive') == 'TRUE' or ism_doc.find('Protection') is not None: + return [] + + duration = int(ism_doc.attrib['Duration']) + timescale = int_or_none(ism_doc.get('TimeScale')) or 10000000 + + formats = [] + for stream in ism_doc.findall('StreamIndex'): + stream_type = stream.get('Type') + if stream_type not in ('video', 'audio'): + continue + url_pattern = stream.attrib['Url'] + stream_timescale = int_or_none(stream.get('TimeScale')) or timescale + stream_name = stream.get('Name') + for track in stream.findall('QualityLevel'): + fourcc = track.get('FourCC') + # TODO: add support for WVC1 and WMAP + if fourcc not in ('H264', 'AVC1', 'AACL'): + self.report_warning('%s is not a supported codec' % fourcc) + continue + tbr = int(track.attrib['Bitrate']) // 1000 + width = int_or_none(track.get('MaxWidth')) + height = int_or_none(track.get('MaxHeight')) + sampling_rate = int_or_none(track.get('SamplingRate')) + + track_url_pattern = re.sub(r'{[Bb]itrate}', track.attrib['Bitrate'], url_pattern) + track_url_pattern = compat_urlparse.urljoin(ism_url, track_url_pattern) + + fragments = [] + fragment_ctx = { + 'time': 0, + } + stream_fragments = stream.findall('c') + for stream_fragment_index, stream_fragment in enumerate(stream_fragments): + fragment_ctx['time'] = int_or_none(stream_fragment.get('t')) or fragment_ctx['time'] + fragment_repeat = int_or_none(stream_fragment.get('r')) or 1 + fragment_ctx['duration'] = int_or_none(stream_fragment.get('d')) + if not fragment_ctx['duration']: + try: + next_fragment_time = int(stream_fragment[stream_fragment_index + 1].attrib['t']) + except IndexError: + next_fragment_time = duration + fragment_ctx['duration'] = (next_fragment_time - fragment_ctx['time']) / fragment_repeat + for _ in range(fragment_repeat): + fragments.append({ + 'url': re.sub(r'{start[ _]time}', compat_str(fragment_ctx['time']), track_url_pattern), + 'duration': fragment_ctx['duration'] / stream_timescale, + }) + fragment_ctx['time'] += fragment_ctx['duration'] + + format_id = [] + if ism_id: + format_id.append(ism_id) + if stream_name: + format_id.append(stream_name) + format_id.append(compat_str(tbr)) + + formats.append({ + 'format_id': '-'.join(format_id), + 'url': ism_url, + 'manifest_url': ism_url, + 'ext': 'ismv' if stream_type == 'video' else 'isma', + 'width': width, + 'height': height, + 'tbr': tbr, + 'asr': sampling_rate, + 'vcodec': 'none' if stream_type == 'audio' else fourcc, + 'acodec': 'none' if stream_type == 'video' else fourcc, + 'protocol': 'ism', + 'fragments': fragments, + '_download_params': { + 'duration': duration, + 'timescale': stream_timescale, + 'width': width or 0, + 'height': height or 0, + 'fourcc': fourcc, + 'codec_private_data': track.get('CodecPrivateData'), + 'sampling_rate': sampling_rate, + 'channels': int_or_none(track.get('Channels', 2)), + 'bits_per_sample': int_or_none(track.get('BitsPerSample', 16)), + 'nal_unit_length_field': int_or_none(track.get('NALUnitLengthField', 4)), + }, + }) + return formats + + def _parse_html5_media_entries(self, base_url, webpage, video_id, m3u8_id=None, m3u8_entry_protocol='m3u8', mpd_id=None): def absolute_url(video_url): return compat_urlparse.urljoin(base_url, video_url) @@ -1710,8 +1935,37 @@ class InfoExtractor(object): return f return {} + def _media_formats(src, cur_media_type): + full_url = absolute_url(src) + ext = determine_ext(full_url) + if ext == 'm3u8': + is_plain_url = False + formats = self._extract_m3u8_formats( + full_url, video_id, ext='mp4', + entry_protocol=m3u8_entry_protocol, m3u8_id=m3u8_id) + elif ext == 'mpd': + is_plain_url = False + formats = self._extract_mpd_formats( + full_url, video_id, mpd_id=mpd_id) + else: + is_plain_url = True + formats = [{ + 'url': full_url, + 'vcodec': 'none' if cur_media_type == 'audio' else None, + }] + return is_plain_url, formats + entries = [] - for media_tag, media_type, media_content in re.findall(r'(?s)(<(?Pvideo|audio)[^>]*>)(.*?)', webpage): + media_tags = [(media_tag, media_type, '') + for media_tag, media_type + in re.findall(r'(?s)(<(video|audio)[^>]*/>)', webpage)] + media_tags.extend(re.findall( + # We only allow video|audio followed by a whitespace or '>'. + # Allowing more characters may end up in significant slow down (see + # https://github.com/rg3/youtube-dl/issues/11979, example URL: + # http://www.porntrex.com/maps/videositemap.xml). + r'(?s)(<(?Pvideo|audio)(?:\s+[^>]*)?>)(.*?)', webpage)) + for media_tag, media_type, media_content in media_tags: media_info = { 'formats': [], 'subtitles': {}, @@ -1719,10 +1973,8 @@ class InfoExtractor(object): media_attributes = extract_attributes(media_tag) src = media_attributes.get('src') if src: - media_info['formats'].append({ - 'url': absolute_url(src), - 'vcodec': 'none' if media_type == 'audio' else None, - }) + _, formats = _media_formats(src, media_type) + media_info['formats'].extend(formats) media_info['thumbnail'] = media_attributes.get('poster') if media_content: for source_tag in re.findall(r']+>', media_content): @@ -1730,16 +1982,17 @@ class InfoExtractor(object): src = source_attributes.get('src') if not src: continue - f = parse_content_type(source_attributes.get('type')) - f.update({ - 'url': absolute_url(src), - 'vcodec': 'none' if media_type == 'audio' else None, - }) - media_info['formats'].append(f) + is_plain_url, formats = _media_formats(src, media_type) + if is_plain_url: + f = parse_content_type(source_attributes.get('type')) + f.update(formats[0]) + media_info['formats'].append(f) + else: + media_info['formats'].extend(formats) for track_tag in re.findall(r']+>', media_content): track_attributes = extract_attributes(track_tag) kind = track_attributes.get('kind') - if not kind or kind == 'subtitles': + if not kind or kind in ('subtitles', 'captions'): src = track_attributes.get('src') if not src: continue @@ -1747,10 +2000,76 @@ class InfoExtractor(object): media_info['subtitles'].setdefault(lang, []).append({ 'url': absolute_url(src), }) - if media_info['formats']: + if media_info['formats'] or media_info['subtitles']: entries.append(media_info) return entries + def _extract_akamai_formats(self, manifest_url, video_id, hosts={}): + formats = [] + hdcore_sign = 'hdcore=3.7.0' + f4m_url = re.sub(r'(https?://[^/+])/i/', r'\1/z/', manifest_url).replace('/master.m3u8', '/manifest.f4m') + hds_host = hosts.get('hds') + if hds_host: + f4m_url = re.sub(r'(https?://)[^/]+', r'\1' + hds_host, f4m_url) + if 'hdcore=' not in f4m_url: + f4m_url += ('&' if '?' in f4m_url else '?') + hdcore_sign + f4m_formats = self._extract_f4m_formats( + f4m_url, video_id, f4m_id='hds', fatal=False) + for entry in f4m_formats: + entry.update({'extra_param_to_segment_url': hdcore_sign}) + formats.extend(f4m_formats) + m3u8_url = re.sub(r'(https?://[^/]+)/z/', r'\1/i/', manifest_url).replace('/manifest.f4m', '/master.m3u8') + hls_host = hosts.get('hls') + if hls_host: + m3u8_url = re.sub(r'(https?://)[^/]+', r'\1' + hls_host, m3u8_url) + formats.extend(self._extract_m3u8_formats( + m3u8_url, video_id, 'mp4', 'm3u8_native', + m3u8_id='hls', fatal=False)) + return formats + + def _extract_wowza_formats(self, url, video_id, m3u8_entry_protocol='m3u8_native', skip_protocols=[]): + url = re.sub(r'/(?:manifest|playlist|jwplayer)\.(?:m3u8|f4m|mpd|smil)', '', url) + url_base = self._search_regex(r'(?:https?|rtmp|rtsp)(://[^?]+)', url, 'format url') + http_base_url = 'http' + url_base + formats = [] + if 'm3u8' not in skip_protocols: + formats.extend(self._extract_m3u8_formats( + http_base_url + '/playlist.m3u8', video_id, 'mp4', + m3u8_entry_protocol, m3u8_id='hls', fatal=False)) + if 'f4m' not in skip_protocols: + formats.extend(self._extract_f4m_formats( + http_base_url + '/manifest.f4m', + video_id, f4m_id='hds', fatal=False)) + if 'dash' not in skip_protocols: + formats.extend(self._extract_mpd_formats( + http_base_url + '/manifest.mpd', + video_id, mpd_id='dash', fatal=False)) + if re.search(r'(?:/smil:|\.smil)', url_base): + if 'smil' not in skip_protocols: + rtmp_formats = self._extract_smil_formats( + http_base_url + '/jwplayer.smil', + video_id, fatal=False) + for rtmp_format in rtmp_formats: + rtsp_format = rtmp_format.copy() + rtsp_format['url'] = '%s/%s' % (rtmp_format['url'], rtmp_format['play_path']) + del rtsp_format['play_path'] + del rtsp_format['ext'] + rtsp_format.update({ + 'url': rtsp_format['url'].replace('rtmp://', 'rtsp://'), + 'format_id': rtmp_format['format_id'].replace('rtmp', 'rtsp'), + 'protocol': 'rtsp', + }) + formats.extend([rtmp_format, rtsp_format]) + else: + for protocol in ('rtmp', 'rtsp'): + if protocol not in skip_protocols: + formats.append({ + 'url': protocol + url_base, + 'format_id': protocol, + 'protocol': protocol, + }) + return formats + def _live_title(self, name): """ Generate the title for a live video """ now = datetime.datetime.now() @@ -1871,6 +2190,12 @@ class InfoExtractor(object): headers['Ytdl-request-proxy'] = geo_verification_proxy return headers + def _generic_id(self, url): + return compat_urllib_parse_unquote(os.path.splitext(url.rstrip('/').split('/')[-1])[0]) + + def _generic_title(self, url): + return compat_urllib_parse_unquote(os.path.splitext(url_basename(url))[0]) + class SearchInfoExtractor(InfoExtractor): """