X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/47d80ec0b18245caeb97018d4c1af18d0b5b972b..9fb40a4e5f59ba243dee2edad23a2c5555338113:/youtube_dl/extractor/common.py?ds=sidebyside diff --git a/youtube_dl/extractor/common.py b/youtube_dl/extractor/common.py index 05c51fa..4252d68 100644 --- a/youtube_dl/extractor/common.py +++ b/youtube_dl/extractor/common.py @@ -6,6 +6,7 @@ import hashlib import json import netrc import os +import random import re import socket import sys @@ -39,7 +40,10 @@ from ..utils import ( ExtractorError, fix_xml_ampersands, float_or_none, + GeoRestrictedError, + GeoUtils, int_or_none, + js_to_json, parse_iso8601, RegexNotFoundError, sanitize_filename, @@ -59,6 +63,7 @@ from ..utils import ( parse_m3u8_attributes, extract_attributes, parse_codecs, + urljoin, ) @@ -120,9 +125,19 @@ 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 + * 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 @@ -188,9 +203,10 @@ 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 @@ -307,17 +323,34 @@ class InfoExtractor(object): _real_extract() methods and define a _VALID_URL regexp. Probably, they should also be added to the list of extractors. + _GEO_BYPASS attribute may be set to False in order to disable + geo restriction bypass mechanisms for a particular extractor. + Though it won't disable explicit geo restriction bypass based on + country code provided with geo_bypass_country. (experimental) + + _GEO_COUNTRIES attribute may contain a list of presumably geo unrestricted + countries for this extractor. One of these countries will be used by + geo restriction bypass mechanism right away in order to bypass + geo restriction, of course, if the mechanism is not disabled. (experimental) + + NB: both these geo attributes are experimental and may change in future + or be completely removed. + Finally, the _WORKING attribute should be set to False for broken IEs in order to warn the users and skip the tests. """ _ready = False _downloader = None + _x_forwarded_for_ip = None + _GEO_BYPASS = True + _GEO_COUNTRIES = None _WORKING = True def __init__(self, downloader=None): """Constructor. Receives an optional downloader.""" self._ready = False + self._x_forwarded_for_ip = None self.set_downloader(downloader) @classmethod @@ -346,15 +379,59 @@ class InfoExtractor(object): def initialize(self): """Initializes an instance (authentication, etc).""" + self._initialize_geo_bypass(self._GEO_COUNTRIES) if not self._ready: self._real_initialize() self._ready = True + def _initialize_geo_bypass(self, countries): + """ + Initialize geo restriction bypass mechanism. + + This method is used to initialize geo bypass mechanism based on faking + X-Forwarded-For HTTP header. A random country from provided country list + is selected and a random IP belonging to this country is generated. This + IP will be passed as X-Forwarded-For HTTP header in all subsequent + HTTP requests. + + This method will be used for initial geo bypass mechanism initialization + during the instance initialization with _GEO_COUNTRIES. + + You may also manually call it from extractor's code if geo countries + information is not available beforehand (e.g. obtained during + extraction) or due to some another reason. + """ + if not self._x_forwarded_for_ip: + country_code = self._downloader.params.get('geo_bypass_country', None) + # If there is no explicit country for geo bypass specified and + # the extractor is known to be geo restricted let's fake IP + # as X-Forwarded-For right away. + if (not country_code and + self._GEO_BYPASS and + self._downloader.params.get('geo_bypass', True) and + countries): + country_code = random.choice(countries) + if country_code: + self._x_forwarded_for_ip = GeoUtils.random_ipv4(country_code) + if self._downloader.params.get('verbose', False): + self._downloader.to_stdout( + '[debug] Using fake IP %s (%s) as X-Forwarded-For.' + % (self._x_forwarded_for_ip, country_code.upper())) + def extract(self, url): """Extracts URL information and returns it in list of dicts.""" try: - self.initialize() - return self._real_extract(url) + for _ in range(2): + try: + self.initialize() + ie_result = self._real_extract(url) + if self._x_forwarded_for_ip: + ie_result['__x_forwarded_for_ip'] = self._x_forwarded_for_ip + return ie_result + except GeoRestrictedError as e: + if self.__maybe_fake_ip_and_retry(e.countries): + continue + raise except ExtractorError: raise except compat_http_client.IncompleteRead as e: @@ -362,6 +439,21 @@ class InfoExtractor(object): except (KeyError, StopIteration) as e: raise ExtractorError('An extractor error has occurred.', cause=e) + def __maybe_fake_ip_and_retry(self, countries): + if (not self._downloader.params.get('geo_bypass_country', None) and + self._GEO_BYPASS and + self._downloader.params.get('geo_bypass', True) and + not self._x_forwarded_for_ip and + countries): + country_code = random.choice(countries) + self._x_forwarded_for_ip = GeoUtils.random_ipv4(country_code) + if self._x_forwarded_for_ip: + self.report_warning( + 'Video is geo restricted. Retrying extraction with fake IP %s (%s) as X-Forwarded-For.' + % (self._x_forwarded_for_ip, country_code.upper())) + return True + return False + def set_downloader(self, downloader): """Sets the downloader for this IE.""" self._downloader = downloader @@ -421,6 +513,15 @@ class InfoExtractor(object): if isinstance(url_or_request, (compat_str, str)): url_or_request = url_or_request.partition('#')[0] + # Some sites check X-Forwarded-For HTTP header in order to figure out + # the origin of the client behind proxy. This allows bypassing geo + # restriction by faking this header's value to IP that belongs to some + # geo unrestricted country. We will do so once we encounter any + # geo restriction error. + if self._x_forwarded_for_ip: + if 'X-Forwarded-For' not in headers: + headers['X-Forwarded-For'] = self._x_forwarded_for_ip + urlh = self._request_webpage(url_or_request, video_id, note, errnote, fatal, data=data, headers=headers, query=query) if urlh is False: assert not fatal @@ -596,10 +697,8 @@ class InfoExtractor(object): expected=True) @staticmethod - def raise_geo_restricted(msg='This video is not available from your location due to geo restriction'): - raise ExtractorError( - '%s. You might want to use --proxy to workaround.' % msg, - expected=True) + def raise_geo_restricted(msg='This video is not available from your location due to geo restriction', countries=None): + raise GeoRestrictedError(msg, countries=countries) # Methods for following #608 @staticmethod @@ -1013,13 +1112,13 @@ 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): @@ -1196,6 +1295,9 @@ class InfoExtractor(object): m3u8_doc, urlh = res m3u8_url = urlh.geturl() + if '#EXT-X-FAXS-CM:' in m3u8_doc: # Adobe Flash Access + return [] + formats = [self._m3u8_meta_format(m3u8_url, ext, preference, m3u8_id)] format_url = lambda u: ( @@ -1224,6 +1326,7 @@ class InfoExtractor(object): 'protocol': entry_protocol, 'preference': preference, }] + audio_in_video_stream = {} last_info = {} last_media = {} for line in m3u8_doc.splitlines(): @@ -1233,25 +1336,32 @@ class InfoExtractor(object): 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 (media.get('GROUP-ID'), media.get('NAME')): + for v in (group_id, media.get('NAME')): if v: format_id.append(v) - formats.append({ + f = { '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, - }) + } + 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: @@ -1295,6 +1405,9 @@ class InfoExtractor(object): 'abr': abr, }) f.update(parse_codecs(last_info.get('CODECS'))) + 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 = {} @@ -1614,21 +1727,16 @@ class InfoExtractor(object): segment_template = element.find(_add_ns('SegmentTemplate')) if segment_template is not None: extract_common(segment_template) - media_template = segment_template.get('media') - if media_template: - ms_info['media_template'] = media_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: 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')): @@ -1668,6 +1776,7 @@ 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, @@ -1675,23 +1784,41 @@ class InfoExtractor(object): '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: - 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 @@ -1703,7 +1830,7 @@ class InfoExtractor(object): representation_ms_info['fragments'] = [{ 'url': media_template % { 'Number': segment_number, - 'Bandwidth': int_or_none(representation_attrib.get('bandwidth')), + 'Bandwidth': bandwidth, }, 'duration': segment_duration, } for segment_number in range( @@ -1721,7 +1848,7 @@ class InfoExtractor(object): def add_segment_url(): segment_url = media_template % { 'Time': segment_time, - 'Bandwidth': int_or_none(representation_attrib.get('bandwidth')), + 'Bandwidth': bandwidth, 'Number': segment_number, } representation_ms_info['fragments'].append({ @@ -1744,14 +1871,16 @@ class InfoExtractor(object): # 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] + 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': segment_url, - 'duration': float_or_none(s['d'], representation_ms_info['timescale']), + '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. @@ -1761,13 +1890,13 @@ class InfoExtractor(object): 'protocol': 'http_dash_segments', }) if 'initialization_url' in representation_ms_info: - initialization_url = representation_ms_info['initialization_url'].replace('$RepresentationID$', representation_id) + 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'] = combine_url(base_url, fragment['url']) + fragment['url'] = urljoin(base_url, fragment['url']) try: existing_format = next( fo for fo in formats @@ -1881,7 +2010,7 @@ class InfoExtractor(object): }) return formats - def _parse_html5_media_entries(self, base_url, webpage, video_id, m3u8_id=None, m3u8_entry_protocol='m3u8'): + 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) @@ -1898,11 +2027,16 @@ class InfoExtractor(object): def _media_formats(src, cur_media_type): full_url = absolute_url(src) - if determine_ext(full_url) == 'm3u8': + 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 = [{ @@ -1915,7 +2049,12 @@ class InfoExtractor(object): media_tags = [(media_tag, media_type, '') for media_tag, media_type in re.findall(r'(?s)(<(video|audio)[^>]*/>)', webpage)] - media_tags.extend(re.findall(r'(?s)(<(?Pvideo|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': [], @@ -1955,10 +2094,13 @@ class InfoExtractor(object): entries.append(media_info) return entries - def _extract_akamai_formats(self, manifest_url, video_id): + 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') + 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( @@ -1966,7 +2108,10 @@ class InfoExtractor(object): 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') + 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)) @@ -2015,6 +2160,123 @@ class InfoExtractor(object): }) return formats + @staticmethod + def _find_jwplayer_data(webpage): + mobj = re.search( + r'jwplayer\((?P[\'"])[^\'" ]+(?P=quote)\)\.setup\s*\((?P[^)]+)\)', + webpage) + if mobj: + return mobj.group('options') + + def _extract_jwplayer_data(self, webpage, video_id, *args, **kwargs): + jwplayer_data = self._parse_json( + self._find_jwplayer_data(webpage), video_id, + transform_source=js_to_json) + return self._parse_jwplayer_data( + jwplayer_data, video_id, *args, **kwargs) + + def _parse_jwplayer_data(self, jwplayer_data, video_id=None, require_title=True, + m3u8_id=None, mpd_id=None, rtmp_params=None, base_url=None): + # JWPlayer backward compatibility: flattened playlists + # https://github.com/jwplayer/jwplayer/blob/v7.4.3/src/js/api/config.js#L81-L96 + if 'playlist' not in jwplayer_data: + jwplayer_data = {'playlist': [jwplayer_data]} + + entries = [] + + # JWPlayer backward compatibility: single playlist item + # https://github.com/jwplayer/jwplayer/blob/v7.7.0/src/js/playlist/playlist.js#L10 + if not isinstance(jwplayer_data['playlist'], list): + jwplayer_data['playlist'] = [jwplayer_data['playlist']] + + for video_data in jwplayer_data['playlist']: + # JWPlayer backward compatibility: flattened sources + # https://github.com/jwplayer/jwplayer/blob/v7.4.3/src/js/playlist/item.js#L29-L35 + if 'sources' not in video_data: + video_data['sources'] = [video_data] + + this_video_id = video_id or video_data['mediaid'] + + formats = [] + for source in video_data['sources']: + source_url = self._proto_relative_url(source['file']) + if base_url: + source_url = compat_urlparse.urljoin(base_url, source_url) + source_type = source.get('type') or '' + ext = mimetype2ext(source_type) or determine_ext(source_url) + if source_type == 'hls' or ext == 'm3u8': + formats.extend(self._extract_m3u8_formats( + source_url, this_video_id, 'mp4', 'm3u8_native', m3u8_id=m3u8_id, fatal=False)) + elif ext == 'mpd': + formats.extend(self._extract_mpd_formats( + source_url, this_video_id, mpd_id=mpd_id, fatal=False)) + # https://github.com/jwplayer/jwplayer/blob/master/src/js/providers/default.js#L67 + elif source_type.startswith('audio') or ext in ('oga', 'aac', 'mp3', 'mpeg', 'vorbis'): + formats.append({ + 'url': source_url, + 'vcodec': 'none', + 'ext': ext, + }) + else: + height = int_or_none(source.get('height')) + if height is None: + # Often no height is provided but there is a label in + # format like 1080p. + height = int_or_none(self._search_regex( + r'^(\d{3,})[pP]$', source.get('label') or '', + 'height', default=None)) + a_format = { + 'url': source_url, + 'width': int_or_none(source.get('width')), + 'height': height, + 'ext': ext, + } + if source_url.startswith('rtmp'): + a_format['ext'] = 'flv' + + # See com/longtailvideo/jwplayer/media/RTMPMediaProvider.as + # of jwplayer.flash.swf + rtmp_url_parts = re.split( + r'((?:mp4|mp3|flv):)', source_url, 1) + if len(rtmp_url_parts) == 3: + rtmp_url, prefix, play_path = rtmp_url_parts + a_format.update({ + 'url': rtmp_url, + 'play_path': prefix + play_path, + }) + if rtmp_params: + a_format.update(rtmp_params) + formats.append(a_format) + self._sort_formats(formats) + + subtitles = {} + tracks = video_data.get('tracks') + if tracks and isinstance(tracks, list): + for track in tracks: + if track.get('kind') != 'captions': + continue + track_url = urljoin(base_url, track.get('file')) + if not track_url: + continue + subtitles.setdefault(track.get('label') or 'en', []).append({ + 'url': self._proto_relative_url(track_url) + }) + + entries.append({ + 'id': this_video_id, + 'title': video_data['title'] if require_title else video_data.get('title'), + 'description': video_data.get('description'), + 'thumbnail': self._proto_relative_url(video_data.get('image')), + 'timestamp': int_or_none(video_data.get('pubdate')), + 'duration': float_or_none(jwplayer_data.get('duration') or video_data.get('duration')), + 'subtitles': subtitles, + 'formats': formats, + }) + if len(entries) == 1: + return entries[0] + else: + return self.playlist_result(entries) + def _live_title(self, name): """ Generate the title for a live video """ now = datetime.datetime.now()