X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fcommon.py;h=5cb4479ec5271256cef27836ecbe1cdc5b8df3f3;hb=5968d7d2fe619e85eb424d6e47d000f0b295d4a2;hp=e47770c1d1717db7970fe46080ef410655b00911;hpb=321b5e082a7340225091ea5ad852e12b2dc499e9;p=youtube-dl diff --git a/youtube_dl/extractor/common.py b/youtube_dl/extractor/common.py index e47770c1d..5cb4479ec 100644 --- a/youtube_dl/extractor/common.py +++ b/youtube_dl/extractor/common.py @@ -87,6 +87,9 @@ 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 +118,11 @@ class InfoExtractor(object): download, lower-case. "http", "https", "rtsp", "rtmp", "rtmpe", "m3u8", "m3u8_native" or "http_dash_segments". + * fragments A list of fragments of the fragmented media, + with the following entries: + * "url" (mandatory) - fragment's 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. @@ -662,35 +670,48 @@ class InfoExtractor(object): else: return res - def _get_login_info(self): + def _get_netrc_login_info(self, netrc_machine=None): + username = None + password = None + netrc_machine = netrc_machine or self._NETRC_MACHINE + + if self._downloader.params.get('usenetrc', False): + try: + info = netrc.netrc().authenticators(netrc_machine) + if info is not None: + username = info[0] + password = info[2] + else: + 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)) + + return username, password + + 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'] - elif downloader_params.get('usenetrc', False): - try: - info = netrc.netrc().authenticators(self._NETRC_MACHINE) - if info is not None: - username = info[0] - password = info[2] - else: - raise netrc.NetrcParseError('No authenticators for %s' % self._NETRC_MACHINE) - except (IOError, netrc.NetrcParseError) as err: - self._downloader.report_warning('parsing .netrc: %s' % error_to_compat_str(err)) + 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(netrc_machine) - return (username, password) + return username, password def _get_tfa_info(self, note='two-factor verification code'): """ @@ -878,16 +899,16 @@ 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): + for input in re.findall(r'(?i)(]+>)', html): + attrs = extract_attributes(input) + if not input: continue - name = re.search(r'(?:name|id)=(["\'])(?P.+?)\1', input) - if not name: + if attrs.get('type') not in ('hidden', 'submit'): continue - value = re.search(r'value=(["\'])(?P.*?)\1', input) - if not value: - 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): @@ -1129,6 +1150,7 @@ 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, @@ -1153,13 +1175,6 @@ 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', @@ -1170,6 +1185,13 @@ 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 @@ -1191,35 +1213,54 @@ class InfoExtractor(object): 'protocol': entry_protocol, 'preference': preference, }] - last_info = None - last_media = None + 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'): + media_url = media.get('URI') + if media_url: + format_id = [] + for v in (media.get('GROUP-ID'), media.get('NAME')): + if v: + format_id.append(v) + formats.append({ + 'format_id': '-'.join(format_id), + 'url': format_url(media_url), + 'language': media.get('LANGUAGE'), + 'vcodec': 'none' if media_type == 'AUDIO' else None, + 'ext': ext, + 'protocol': entry_protocol, + 'preference': preference, + }) + 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 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')), @@ -1242,11 +1283,9 @@ 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 formats.append(f) last_info = {} + last_media = {} return formats @staticmethod @@ -1493,9 +1532,10 @@ class InfoExtractor(object): mpd_base_url = re.match(r'https?://.+/', urlh.geturl()).group() 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: @@ -1516,42 +1556,52 @@ 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) + extract_common(segment_template) media_template = segment_template.get('media') if media_template: ms_info['media_template'] = media_template @@ -1559,11 +1609,14 @@ class InfoExtractor(object): if initialization: ms_info['initialization_url'] = 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 + def combine_url(base_url, target_url): + if re.match(r'^https?://', target_url): + return target_url + return '%s%s%s' % (base_url, '' if base_url.endswith('/') else '/', target_url) + mpd_duration = parse_duration(mpd_doc.get('mediaPresentationDuration')) formats = [] for period in mpd_doc.findall(_add_ns('Period')): @@ -1606,6 +1659,7 @@ class InfoExtractor(object): 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')), @@ -1620,9 +1674,7 @@ class InfoExtractor(object): } 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) @@ -1631,46 +1683,79 @@ class InfoExtractor(object): # 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'])] + }, + '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': representation_attrib.get('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 = [] + s_num = 0 + for segment_url in representation_ms_info['segment_urls']: + s = representation_ms_info['s'][s_num] + for r in range(s.get('r', 0) + 1): + fragments.append({ + 'url': segment_url, + 'duration': float_or_none(s['d'], representation_ms_info['timescale']), + }) + 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, - }) 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'] = combine_url(base_url, fragment['url']) try: existing_format = next( fo for fo in formats @@ -1685,7 +1770,7 @@ 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 _parse_html5_media_entries(self, base_url, webpage, video_id, m3u8_id=None, m3u8_entry_protocol='m3u8'): def absolute_url(video_url): return compat_urlparse.urljoin(base_url, video_url) @@ -1700,6 +1785,21 @@ class InfoExtractor(object): return f return {} + def _media_formats(src, cur_media_type): + full_url = absolute_url(src) + if determine_ext(full_url) == '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) + 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_info = { @@ -1709,10 +1809,8 @@ 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): @@ -1720,16 +1818,17 @@ 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 @@ -1737,10 +1836,65 @@ 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): + formats = [] + f4m_url = re.sub(r'(https?://.+?)/i/', r'\1/z/', manifest_url).replace('/master.m3u8', '/manifest.f4m') + formats.extend(self._extract_f4m_formats( + update_url_query(f4m_url, {'hdcore': '3.7.0'}), + video_id, f4m_id='hds', fatal=False)) + m3u8_url = re.sub(r'(https?://.+?)/z/', r'\1/i/', manifest_url).replace('/manifest.f4m', '/master.m3u8') + 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 re.search(r'(?:/smil:|\.smil)', url_base): + 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 '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()