diff options
Diffstat (limited to 'youtube_dl/extractor/common.py')
| -rw-r--r-- | youtube_dl/extractor/common.py | 1943 |
1 files changed, 1431 insertions, 512 deletions
diff --git a/youtube_dl/extractor/common.py b/youtube_dl/extractor/common.py index dcc9d628a..a64fcfccc 100644 --- a/youtube_dl/extractor/common.py +++ b/youtube_dl/extractor/common.py @@ -2,7 +2,9 @@ from __future__ import unicode_literals import base64 +import collections import datetime +import functools import hashlib import json import netrc @@ -10,16 +12,22 @@ import os import random import re import socket +import ssl import sys import time import math from ..compat import ( - compat_cookiejar, - compat_cookies, + compat_cookiejar_Cookie, + compat_cookies_SimpleCookie, + compat_etree_Element, compat_etree_fromstring, compat_getpass, + compat_integer_types, compat_http_client, + compat_kwargs, + compat_map as map, + compat_open as open, compat_os_name, compat_str, compat_urllib_error, @@ -27,8 +35,13 @@ from ..compat import ( compat_urllib_parse_urlencode, compat_urllib_request, compat_urlparse, + compat_xml_parse_error, + compat_zip as zip, +) +from ..downloader.f4m import ( + get_base_url, + remove_encrypted_media, ) -from ..downloader.f4m import remove_encrypted_media from ..utils import ( NO_DEFAULT, age_restricted, @@ -38,6 +51,7 @@ from ..utils import ( compiled_regex_type, determine_ext, determine_protocol, + dict_get, error_to_compat_str, ExtractorError, extract_attributes, @@ -46,16 +60,26 @@ from ..utils import ( GeoRestrictedError, GeoUtils, int_or_none, + join_nonempty, js_to_json, + JSON_LD_RE, mimetype2ext, orderedSet, + parse_bitrate, parse_codecs, parse_duration, parse_iso8601, parse_m3u8_attributes, + parse_resolution, RegexNotFoundError, sanitized_Request, sanitize_filename, + str_or_none, + str_to_int, + strip_or_none, + T, + traverse_obj, + try_get, unescapeHTML, unified_strdate, unified_timestamp, @@ -63,6 +87,8 @@ from ..utils import ( update_url_query, urljoin, url_basename, + url_or_none, + variadic, xpath_element, xpath_text, xpath_with_ns, @@ -95,10 +121,26 @@ class InfoExtractor(object): from worst to best quality. Potential fields: - * url Mandatory. The URL of the video file + * url The mandatory URL representing the media: + for plain file media - HTTP URL of this file, + for RTMP - RTMP URL, + for HLS - URL of the M3U8 media playlist, + for HDS - URL of the F4M manifest, + for DASH + - HTTP URL to plain file media (in case of + unfragmented media) + - URL of the MPD manifest or base URL + representing the media if MPD manifest + is parsed from a string (in case of + fragmented media) + for MSS - URL of the ISM manifest. * manifest_url The URL of the manifest file in case of - fragmented media (DASH, hls, hds) + fragmented media: + for HLS - URL of the M3U8 master playlist, + for HDS - URL of the F4M manifest, + for DASH - URL of the MPD manifest, + for MSS - URL of the ISM manifest. * ext Will be calculated from URL if missing * format A human-readable description of the format ("mp4 container with h264/opus"). @@ -142,6 +184,8 @@ class InfoExtractor(object): fragment_base_url * "duration" (optional, int or float) * "filesize" (optional, int) + * "range" (optional, str of the form "start-end" + to use in HTTP Range header) * 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. @@ -170,6 +214,8 @@ class InfoExtractor(object): width : height ratio as float. * no_resume The server does not support resuming the (HTTP or RTMP) download. Boolean. + * downloader_options A dictionary of downloader options as + described in FileDownloader url: Final video URL. ext: Video filename extension. @@ -189,7 +235,7 @@ class InfoExtractor(object): * "preference" (optional, int) - quality of the image * "width" (optional, int) * "height" (optional, int) - * "resolution" (optional, string "{width}x{height"}, + * "resolution" (optional, string "{width}x{height}", deprecated) * "filesize" (optional, int) thumbnail: Full URL to a video thumbnail image. @@ -197,12 +243,19 @@ class InfoExtractor(object): uploader: Full name of the video uploader. license: License name the video is licensed under. creator: The creator of the video. + release_timestamp: UNIX timestamp of the moment the video was released. release_date: The date (YYYYMMDD) when the video was released. - timestamp: UNIX timestamp of the moment the video became available. + timestamp: UNIX timestamp of the moment the video became available + (uploaded). upload_date: Video upload date (YYYYMMDD). If not explicitly set, calculated from timestamp. uploader_id: Nickname or id of the video uploader. uploader_url: Full URL to a personal webpage of the video uploader. + channel: Full name of the channel the video is uploaded on. + Note that channel fields may or may not repeat uploader + fields. This depends on a particular extractor. + channel_id: Id of the channel. + channel_url: Full URL to a channel webpage. location: Physical location where the video was filmed. subtitles: The available subtitles as a dictionary in the format {tag: subformats}. "tag" is usually a language code, and @@ -245,6 +298,10 @@ class InfoExtractor(object): specified in the URL. end_time: Time in seconds where the reproduction should end, as specified in the URL. + chapters: A list of dictionaries, with the following entries: + * "start_time" - The start time of the chapter in seconds + * "end_time" - The end time of the chapter in seconds + * "title" (optional, string) The following fields should only be used when the video belongs to some logical chapter or section: @@ -293,8 +350,9 @@ class InfoExtractor(object): There must be a key "entries", which is a list, an iterable, or a PagedList object, each element of which is a valid dictionary by this specification. - Additionally, playlists can have "title", "description" and "id" attributes - with the same semantics as videos (see above). + Additionally, playlists can have "id", "title", "description", "uploader", + "uploader_id", "uploader_url", "duration" attributes with the same semantics + as videos (see above). _type "multi_video" indicates that there are multiple videos that @@ -321,22 +379,37 @@ class InfoExtractor(object): title, description etc. - Subclasses of this one should re-define the _real_initialize() and - _real_extract() methods and define a _VALID_URL regexp. - Probably, they should also be added to the list of extractors. + A subclass of InfoExtractor must be defined to handle each specific site (or + several sites). Such a concrete subclass should be added to the list of + extractors. It should also: + * define its _VALID_URL attribute as a regexp, or a Sequence of alternative + regexps (but see below) + * re-define the _real_extract() method + * optionally re-define the _real_initialize() method. + + An extractor subclass may also override suitable() if necessary, but the + function signature must be preserved and the function must import everything + it needs (except other extractors), so that lazy_extractors works correctly. + If the subclass's suitable() and _real_extract() functions avoid using + _VALID_URL, the subclass need not set that class attribute. + + An abstract subclass of InfoExtractor may be used to simplify implementation + within an extractor module; it should not 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) + country code provided with geo_bypass_country. _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) + geo restriction, of course, if the mechanism is not disabled. - NB: both these geo attributes are experimental and may change in future - or be completely removed. + _GEO_IP_BLOCKS attribute may contain a list of presumably geo unrestricted + IP blocks in CIDR notation for this extractor. One of these IP blocks + will be used by geo restriction bypass mechanism similarly + to _GEO_COUNTRIES. Finally, the _WORKING attribute should be set to False for broken IEs in order to warn the users and skip the tests. @@ -347,7 +420,10 @@ class InfoExtractor(object): _x_forwarded_for_ip = None _GEO_BYPASS = True _GEO_COUNTRIES = None + _GEO_IP_BLOCKS = None _WORKING = True + # supply this in public subclasses: used in supported sites list, etc + # IE_DESC = 'short description of IE' def __init__(self, downloader=None): """Constructor. Receives an optional downloader.""" @@ -356,23 +432,34 @@ class InfoExtractor(object): self.set_downloader(downloader) @classmethod - def suitable(cls, url): - """Receives a URL and returns True if suitable for this IE.""" - + def __match_valid_url(cls, url): # This does not use has/getattr intentionally - we want to know whether - # we have cached the regexp for *this* class, whereas getattr would also - # match the superclass + # we have cached the regexp for cls, whereas getattr would also + # match its superclass if '_VALID_URL_RE' not in cls.__dict__: - cls._VALID_URL_RE = re.compile(cls._VALID_URL) - return cls._VALID_URL_RE.match(url) is not None + # _VALID_URL can now be a list/tuple of patterns + cls._VALID_URL_RE = tuple(map(re.compile, variadic(cls._VALID_URL))) + # 20% faster than next(filter(None, (p.match(url) for p in cls._VALID_URL_RE)), None) in 2.7 + for p in cls._VALID_URL_RE: + p = p.match(url) + if p: + return p + + # The public alias can safely be overridden, as in some back-ports + _match_valid_url = __match_valid_url + + @classmethod + def suitable(cls, url): + """Receives a URL and returns True if suitable for this IE.""" + # This function must import everything it needs (except other extractors), + # so that lazy_extractors works correctly + return cls.__match_valid_url(url) is not None @classmethod def _match_id(cls, url): - if '_VALID_URL_RE' not in cls.__dict__: - cls._VALID_URL_RE = re.compile(cls._VALID_URL) - m = cls._VALID_URL_RE.match(url) + m = cls.__match_valid_url(url) assert m - return m.group('id') + return compat_str(m.group('id')) @classmethod def working(cls): @@ -381,12 +468,15 @@ class InfoExtractor(object): def initialize(self): """Initializes an instance (authentication, etc).""" - self._initialize_geo_bypass(self._GEO_COUNTRIES) + self._initialize_geo_bypass({ + 'countries': self._GEO_COUNTRIES, + 'ip_blocks': self._GEO_IP_BLOCKS, + }) if not self._ready: self._real_initialize() self._ready = True - def _initialize_geo_bypass(self, countries): + def _initialize_geo_bypass(self, geo_bypass_context): """ Initialize geo restriction bypass mechanism. @@ -397,28 +487,82 @@ class InfoExtractor(object): HTTP requests. This method will be used for initial geo bypass mechanism initialization - during the instance initialization with _GEO_COUNTRIES. + during the instance initialization with _GEO_COUNTRIES and + _GEO_IP_BLOCKS. - You may also manually call it from extractor's code if geo countries + You may also manually call it from extractor's code if geo bypass information is not available beforehand (e.g. obtained during - extraction) or due to some another reason. + extraction) or due to some other reason. In this case you should pass + this information in geo bypass context passed as first argument. It may + contain following fields: + + countries: List of geo unrestricted countries (similar + to _GEO_COUNTRIES) + ip_blocks: List of geo unrestricted IP blocks in CIDR notation + (similar to _GEO_IP_BLOCKS) + """ 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( + + # Geo bypass mechanism is explicitly disabled by user + if not self.get_param('geo_bypass', True): + return + + if not geo_bypass_context: + geo_bypass_context = {} + + # Backward compatibility: previously _initialize_geo_bypass + # expected a list of countries, some 3rd party code may still use + # it this way + if isinstance(geo_bypass_context, (list, tuple)): + geo_bypass_context = { + 'countries': geo_bypass_context, + } + + # The whole point of geo bypass mechanism is to fake IP + # as X-Forwarded-For HTTP header based on some IP block or + # country code. + + # Path 1: bypassing based on IP block in CIDR notation + + # Explicit IP block specified by user, use it right away + # regardless of whether extractor is geo bypassable or not + ip_block = self.get_param('geo_bypass_ip_block', None) + + # Otherwise use random IP block from geo bypass context but only + # if extractor is known as geo bypassable + if not ip_block: + ip_blocks = geo_bypass_context.get('ip_blocks') + if self._GEO_BYPASS and ip_blocks: + ip_block = random.choice(ip_blocks) + + if ip_block: + self._x_forwarded_for_ip = GeoUtils.random_ipv4(ip_block) + if self.get_param('verbose', False): + self.to_screen( + '[debug] Using fake IP %s as X-Forwarded-For.' + % self._x_forwarded_for_ip) + return + + # Path 2: bypassing based on country code + + # Explicit country code specified by user, use it right away + # regardless of whether extractor is geo bypassable or not + country = self.get_param('geo_bypass_country', None) + + # Otherwise use random country code from geo bypass context but + # only if extractor is known as geo bypassable + if not country: + countries = geo_bypass_context.get('countries') + if self._GEO_BYPASS and countries: + country = random.choice(countries) + + if country: + self._x_forwarded_for_ip = GeoUtils.random_ipv4(country) + if self.get_param('verbose', False): + self.to_screen( '[debug] Using fake IP %s (%s) as X-Forwarded-For.' - % (self._x_forwarded_for_ip, country_code.upper())) + % (self._x_forwarded_for_ip, country.upper())) def extract(self, url): """Extracts URL information and returns it in list of dicts.""" @@ -442,11 +586,11 @@ class InfoExtractor(object): 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): + if (not self.get_param('geo_bypass_country', None) + and self._GEO_BYPASS + and self.get_param('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: @@ -460,6 +604,14 @@ class InfoExtractor(object): """Sets the downloader for this IE.""" self._downloader = downloader + @property + def cache(self): + return self._downloader.cache + + @property + def cookiejar(self): + return self._downloader.cookiejar + def _real_initialize(self): """Real initialization process. Redefine in subclasses.""" pass @@ -477,8 +629,26 @@ class InfoExtractor(object): def IE_NAME(self): return compat_str(type(self).__name__[:-2]) - def _request_webpage(self, url_or_request, video_id, note=None, errnote=None, fatal=True, data=None, headers={}, query={}): - """ Returns the response handle """ + @staticmethod + def __can_accept_status_code(err, expected_status): + assert isinstance(err, compat_urllib_error.HTTPError) + if expected_status is None: + return False + if isinstance(expected_status, compat_integer_types): + return err.code == expected_status + elif isinstance(expected_status, (list, tuple)): + return err.code in expected_status + elif callable(expected_status): + return expected_status(err.code) is True + else: + assert False + + def _request_webpage(self, url_or_request, video_id, note=None, errnote=None, fatal=True, data=None, headers={}, query={}, expected_status=None): + """ + Return the response handle. + + See _download_webpage docstring for arguments specification. + """ if note is None: self.report_download_webpage(video_id) elif note is not False: @@ -486,6 +656,16 @@ class InfoExtractor(object): self.to_screen('%s' % (note,)) else: self.to_screen('%s: %s' % (video_id, note)) + + # 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 + if isinstance(url_or_request, compat_urllib_request.Request): url_or_request = update_Request( url_or_request, data=data, headers=headers, query=query) @@ -494,9 +674,21 @@ class InfoExtractor(object): url_or_request = update_url_query(url_or_request, query) if data is not None or headers: url_or_request = sanitized_Request(url_or_request, data, headers) + exceptions = [compat_urllib_error.URLError, compat_http_client.HTTPException, socket.error] + if hasattr(ssl, 'CertificateError'): + exceptions.append(ssl.CertificateError) try: return self._downloader.urlopen(url_or_request) - except (compat_urllib_error.URLError, compat_http_client.HTTPException, socket.error) as err: + except tuple(exceptions) as err: + if isinstance(err, compat_urllib_error.HTTPError): + if self.__can_accept_status_code(err, expected_status): + # Retain reference to error to prevent file object from + # being closed before it can be read. Works around the + # effects of <https://bugs.python.org/issue15002> + # introduced in Python 3.4.1. + err.fp._error = err + return err.fp + if errnote is False: return False if errnote is None: @@ -506,25 +698,20 @@ class InfoExtractor(object): if fatal: raise ExtractorError(errmsg, sys.exc_info()[2], cause=err) else: - self._downloader.report_warning(errmsg) + self.report_warning(errmsg) return False - def _download_webpage_handle(self, url_or_request, video_id, note=None, errnote=None, fatal=True, encoding=None, data=None, headers={}, query={}): - """ Returns a tuple (page content as string, URL handle) """ + def _download_webpage_handle(self, url_or_request, video_id, note=None, errnote=None, fatal=True, encoding=None, data=None, headers={}, query={}, expected_status=None): + """ + Return a tuple (page content as string, URL handle). + + See _download_webpage docstring for arguments specification. + """ # Strip hashes from the URL (#1038) 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) + urlh = self._request_webpage(url_or_request, video_id, note, errnote, fatal, data=data, headers=headers, query=query, expected_status=expected_status) if urlh is False: assert not fatal return False @@ -550,8 +737,8 @@ class InfoExtractor(object): def __check_blocked(self, content): first_block = content[:512] - if ('<title>Access to this site is blocked</title>' in content and - 'Websense' in first_block): + if ('<title>Access to this site is blocked</title>' in content + and 'Websense' in first_block): msg = 'Access to this webpage has been blocked by Websense filtering software in your network.' blocked_iframe = self._html_search_regex( r'<iframe src="([^"]+)"', content, @@ -569,8 +756,8 @@ class InfoExtractor(object): if block_msg: msg += ' (Message: "%s")' % block_msg.replace('\n', ' ') raise ExtractorError(msg, expected=True) - if ('<title>TTK :: Доступ к ресурсу ограничен</title>' in content and - 'blocklist.rkn.gov.ru' in content): + if ('<title>TTK :: Доступ к ресурсу ограничен</title>' in content + and 'blocklist.rkn.gov.ru' in content): raise ExtractorError( 'Access to this webpage has been blocked by decision of the Russian government. ' 'Visit http://blocklist.rkn.gov.ru/ for a block reason.', @@ -583,20 +770,12 @@ class InfoExtractor(object): webpage_bytes = prefix + webpage_bytes if not encoding: encoding = self._guess_encoding_from_content(content_type, webpage_bytes) - if self._downloader.params.get('dump_intermediate_pages', False): - try: - url = url_or_request.get_full_url() - except AttributeError: - url = url_or_request - self.to_screen('Dumping request to ' + url) + if self.get_param('dump_intermediate_pages', False): + self.to_screen('Dumping request to ' + urlh.geturl()) dump = base64.b64encode(webpage_bytes).decode('ascii') - self._downloader.to_screen(dump) - if self._downloader.params.get('write_pages', False): - try: - url = url_or_request.get_full_url() - except AttributeError: - url = url_or_request - basen = '%s_%s' % (video_id, url) + self.to_screen(dump) + if self.get_param('write_pages', False): + basen = '%s_%s' % (video_id, urlh.geturl()) if len(basen) > 240: h = '___' + hashlib.md5(basen.encode('utf-8')).hexdigest() basen = basen[:240 - len(h)] + h @@ -621,13 +800,52 @@ class InfoExtractor(object): return content - def _download_webpage(self, url_or_request, video_id, note=None, errnote=None, fatal=True, tries=1, timeout=5, encoding=None, data=None, headers={}, query={}): - """ Returns the data of the page as a string """ + def _download_webpage( + self, url_or_request, video_id, note=None, errnote=None, + fatal=True, tries=1, timeout=5, encoding=None, data=None, + headers={}, query={}, expected_status=None): + """ + Return the data of the page as a string. + + Arguments: + url_or_request -- plain text URL as a string or + a compat_urllib_request.Requestobject + video_id -- Video/playlist/item identifier (string) + + Keyword arguments: + note -- note printed before downloading (string) + errnote -- note printed in case of an error (string) + fatal -- flag denoting whether error should be considered fatal, + i.e. whether it should cause ExtractionError to be raised, + otherwise a warning will be reported and extraction continued + tries -- number of tries + timeout -- sleep interval between tries + encoding -- encoding for a page content decoding, guessed automatically + when not explicitly specified + data -- POST data (bytes) + headers -- HTTP headers (dict) + query -- URL query (dict) + expected_status -- allows to accept failed HTTP requests (non 2xx + status code) by explicitly specifying a set of accepted status + codes. Can be any of the following entities: + - an integer type specifying an exact failed status code to + accept + - a list or a tuple of integer types specifying a list of + failed status codes to accept + - a callable accepting an actual failed status code and + returning True if it should be accepted + Note that this argument does not affect success status codes (2xx) + which are always accepted. + """ + success = False try_count = 0 while success is False: try: - res = self._download_webpage_handle(url_or_request, video_id, note, errnote, fatal, encoding=encoding, data=data, headers=headers, query=query) + res = self._download_webpage_handle( + url_or_request, video_id, note, errnote, fatal, + encoding=encoding, data=data, headers=headers, query=query, + expected_status=expected_status) success = True except compat_http_client.IncompleteRead as e: try_count += 1 @@ -640,30 +858,93 @@ class InfoExtractor(object): content, _ = res return content - def _download_xml(self, url_or_request, video_id, - note='Downloading XML', errnote='Unable to download XML', - transform_source=None, fatal=True, encoding=None, data=None, headers={}, query={}): - """Return the xml as an xml.etree.ElementTree.Element""" - xml_string = self._download_webpage( - url_or_request, video_id, note, errnote, fatal=fatal, encoding=encoding, data=data, headers=headers, query=query) - if xml_string is False: - return xml_string + def _download_xml_handle( + self, url_or_request, video_id, note='Downloading XML', + errnote='Unable to download XML', transform_source=None, + fatal=True, encoding=None, data=None, headers={}, query={}, + expected_status=None): + """ + Return a tuple (xml as an compat_etree_Element, URL handle). + + See _download_webpage docstring for arguments specification. + """ + res = self._download_webpage_handle( + url_or_request, video_id, note, errnote, fatal=fatal, + encoding=encoding, data=data, headers=headers, query=query, + expected_status=expected_status) + if res is False: + return res + xml_string, urlh = res + return self._parse_xml( + xml_string, video_id, transform_source=transform_source, + fatal=fatal), urlh + + def _download_xml( + self, url_or_request, video_id, + note='Downloading XML', errnote='Unable to download XML', + transform_source=None, fatal=True, encoding=None, + data=None, headers={}, query={}, expected_status=None): + """ + Return the xml as an compat_etree_Element. + + See _download_webpage docstring for arguments specification. + """ + res = self._download_xml_handle( + url_or_request, video_id, note=note, errnote=errnote, + transform_source=transform_source, fatal=fatal, encoding=encoding, + data=data, headers=headers, query=query, + expected_status=expected_status) + return res if res is False else res[0] + + def _parse_xml(self, xml_string, video_id, transform_source=None, fatal=True): if transform_source: xml_string = transform_source(xml_string) - return compat_etree_fromstring(xml_string.encode('utf-8')) - - def _download_json(self, url_or_request, video_id, - note='Downloading JSON metadata', - errnote='Unable to download JSON metadata', - transform_source=None, - fatal=True, encoding=None, data=None, headers={}, query={}): - json_string = self._download_webpage( + try: + return compat_etree_fromstring(xml_string.encode('utf-8')) + except compat_xml_parse_error as ve: + errmsg = '%s: Failed to parse XML ' % video_id + if fatal: + raise ExtractorError(errmsg, cause=ve) + else: + self.report_warning(errmsg + str(ve)) + + def _download_json_handle( + self, url_or_request, video_id, note='Downloading JSON metadata', + errnote='Unable to download JSON metadata', transform_source=None, + fatal=True, encoding=None, data=None, headers={}, query={}, + expected_status=None): + """ + Return a tuple (JSON object, URL handle). + + See _download_webpage docstring for arguments specification. + """ + res = self._download_webpage_handle( url_or_request, video_id, note, errnote, fatal=fatal, - encoding=encoding, data=data, headers=headers, query=query) - if (not fatal) and json_string is False: - return None + encoding=encoding, data=data, headers=headers, query=query, + expected_status=expected_status) + if res is False: + return res + json_string, urlh = res return self._parse_json( - json_string, video_id, transform_source=transform_source, fatal=fatal) + json_string, video_id, transform_source=transform_source, + fatal=fatal), urlh + + def _download_json( + self, url_or_request, video_id, note='Downloading JSON metadata', + errnote='Unable to download JSON metadata', transform_source=None, + fatal=True, encoding=None, data=None, headers={}, query={}, + expected_status=None): + """ + Return the JSON object as a dict. + + See _download_webpage docstring for arguments specification. + """ + res = self._download_json_handle( + url_or_request, video_id, note=note, errnote=errnote, + transform_source=transform_source, fatal=fatal, encoding=encoding, + data=data, headers=headers, query=query, + expected_status=expected_status) + return res if res is False else res[0] def _parse_json(self, json_string, video_id, transform_source=None, fatal=True): if transform_source: @@ -677,14 +958,37 @@ class InfoExtractor(object): else: self.report_warning(errmsg + str(ve)) - def report_warning(self, msg, video_id=None): + def __ie_msg(self, *msg): + return '[{0}] {1}'.format(self.IE_NAME, ''.join(msg)) + + # msg, video_id=None, *args, only_once=False, **kwargs + def report_warning(self, msg, *args, **kwargs): + if len(args) > 0: + video_id = args[0] + args = args[1:] + else: + video_id = kwargs.pop('video_id', None) idstr = '' if video_id is None else '%s: ' % video_id self._downloader.report_warning( - '[%s] %s%s' % (self.IE_NAME, idstr, msg)) + self.__ie_msg(idstr, msg), *args, **kwargs) def to_screen(self, msg): """Print msg to screen, prefixing it with '[ie_name]'""" - self._downloader.to_screen('[%s] %s' % (self.IE_NAME, msg)) + self._downloader.to_screen(self.__ie_msg(msg)) + + def write_debug(self, msg, only_once=False): + '''Log debug message or Print message to stderr''' + self._downloader.write_debug(self.__ie_msg(msg), only_once=only_once) + + # name, default=None, *args, **kwargs + def get_param(self, name, *args, **kwargs): + default, args = (args[0], args[1:]) if len(args) > 0 else (kwargs.pop('default', None), args) + if self._downloader: + return self._downloader.params.get(name, default, *args, **kwargs) + return default + + def report_drm(self, video_id): + self.raise_no_formats('This video is DRM protected', expected=True, video_id=video_id) def report_extraction(self, id_or_name): """Report information extraction.""" @@ -712,6 +1016,15 @@ class InfoExtractor(object): def raise_geo_restricted(msg='This video is not available from your location due to geo restriction', countries=None): raise GeoRestrictedError(msg, countries=countries) + def raise_no_formats(self, msg, expected=False, video_id=None): + if expected and ( + self.get_param('ignore_no_formats_error') or self.get_param('wait_for_video')): + self.report_warning(msg, video_id) + elif isinstance(msg, ExtractorError): + raise msg + else: + raise ExtractorError(msg, expected=expected, video_id=video_id) + # Methods for following #608 @staticmethod def url_result(url, ie=None, video_id=None, video_title=None): @@ -726,12 +1039,12 @@ class InfoExtractor(object): video_info['title'] = video_title return video_info - def playlist_from_matches(self, matches, video_id, video_title, getter=None, ie=None): - urlrs = orderedSet( + def playlist_from_matches(self, matches, playlist_id=None, playlist_title=None, getter=None, ie=None): + urls = orderedSet( self.url_result(self._proto_relative_url(getter(m) if getter else m), ie) for m in matches) return self.playlist_result( - urlrs, playlist_id=video_id, playlist_title=video_title) + urls, playlist_id=playlist_id, playlist_title=playlist_title) @staticmethod def playlist_result(entries, playlist_id=None, playlist_title=None, playlist_description=None): @@ -761,7 +1074,7 @@ class InfoExtractor(object): if mobj: break - if not self._downloader.params.get('no_color') and compat_os_name != 'nt' and sys.stderr.isatty(): + if not self.get_param('no_color') and compat_os_name != 'nt' and sys.stderr.isatty(): _name = '\033[0;34m%s\033[0m' % name else: _name = name @@ -770,6 +1083,8 @@ class InfoExtractor(object): if group is None: # return the first matching group return next(g for g in mobj.groups() if g is not None) + elif isinstance(group, (list, tuple)): + return tuple(mobj.group(g) for g in group) else: return mobj.group(group) elif default is not NO_DEFAULT: @@ -777,26 +1092,79 @@ class InfoExtractor(object): elif fatal: raise RegexNotFoundError('Unable to extract %s' % _name) else: - self._downloader.report_warning('unable to extract %s' % _name + bug_reports_message()) + self.report_warning('unable to extract %s' % _name + bug_reports_message()) return None + def _search_json(self, start_pattern, string, name, video_id, **kwargs): + """Searches string for the JSON object specified by start_pattern""" + + # self, start_pattern, string, name, video_id, *, end_pattern='', + # contains_pattern=r'{(?s:.+)}', fatal=True, default=NO_DEFAULT + # NB: end_pattern is only used to reduce the size of the initial match + end_pattern = kwargs.pop('end_pattern', '') + # (?:[\s\S]) simulates (?(s):.) (eg) + contains_pattern = kwargs.pop('contains_pattern', r'{[\s\S]+}') + fatal = kwargs.pop('fatal', True) + default = kwargs.pop('default', NO_DEFAULT) + + if default is NO_DEFAULT: + default, has_default = {}, False + else: + fatal, has_default = False, True + + json_string = self._search_regex( + r'(?:{0})\s*(?P<json>{1})\s*(?:{2})'.format( + start_pattern, contains_pattern, end_pattern), + string, name, group='json', fatal=fatal, default=None if has_default else NO_DEFAULT) + if not json_string: + return default + + # yt-dlp has a special JSON parser that allows trailing text. + # Until that arrives here, the diagnostic from the exception + # raised by json.loads() is used to extract the wanted text. + # Either way, it's a problem if a transform_source() can't + # handle the trailing text. + + # force an exception + kwargs['fatal'] = True + + # self._downloader._format_err(name, self._downloader.Styles.EMPHASIS) + for _ in range(2): + try: + # return self._parse_json(json_string, video_id, ignore_extra=True, **kwargs) + transform_source = kwargs.pop('transform_source', None) + if transform_source: + json_string = transform_source(json_string) + return self._parse_json(json_string, video_id, **compat_kwargs(kwargs)) + except ExtractorError as e: + end = int_or_none(self._search_regex(r'\(char\s+(\d+)', error_to_compat_str(e), 'end', default=None)) + if end is not None: + json_string = json_string[:end] + continue + msg = 'Unable to extract {0} - Failed to parse JSON'.format(name) + if fatal: + raise ExtractorError(msg, cause=e.cause, video_id=video_id) + elif not has_default: + self.report_warning( + '{0}: {1}'.format(msg, error_to_compat_str(e)), video_id=video_id) + return default + def _html_search_regex(self, pattern, string, name, default=NO_DEFAULT, fatal=True, flags=0, group=None): """ Like _search_regex, but strips HTML tags and unescapes entities. """ res = self._search_regex(pattern, string, name, default, fatal, flags, group) - if res: - return clean_html(res).strip() - else: - return res + if isinstance(res, tuple): + return tuple(map(clean_html, res)) + return clean_html(res) 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): + if self.get_param('usenetrc', False): try: + netrc_machine = netrc_machine or self._NETRC_MACHINE info = netrc.netrc().authenticators(netrc_machine) if info is not None: username = info[0] @@ -804,8 +1172,8 @@ class InfoExtractor(object): else: raise netrc.NetrcParseError( 'No authenticators for %s' % netrc_machine) - except (IOError, netrc.NetrcParseError) as err: - self._downloader.report_warning( + except (AttributeError, IOError, netrc.NetrcParseError) as err: + self.report_warning( 'parsing .netrc: %s' % error_to_compat_str(err)) return username, password @@ -842,18 +1210,18 @@ class InfoExtractor(object): """ if self._downloader is None: return None - downloader_params = self._downloader.params - if downloader_params.get('twofactor') is not None: - return downloader_params['twofactor'] + twofactor = self.get_param('twofactor') + if twofactor is not None: + return twofactor return compat_getpass('Type %s and press [Return]: ' % note) # Helper functions for extracting OpenGraph info @staticmethod def _og_regexes(prop): - content_re = r'content=(?:"([^"]+?)"|\'([^\']+?)\'|\s*([^\s"\'=<>`]+?))' - property_re = (r'(?:name|property)=(?:\'og:%(prop)s\'|"og:%(prop)s"|\s*og:%(prop)s\b)' + content_re = r'content=(?:"([^"]+?)"|\'([^\']+?)\'|\s*([^\s"\'=<>`]+?)(?=\s|/?>))' + property_re = (r'(?:name|property)=(?:\'og[:-]%(prop)s\'|"og[:-]%(prop)s"|\s*og[:-]%(prop)s\b)' % {'prop': re.escape(prop)}) template = r'<meta[^>]+?%s[^>]+?%s' return [ @@ -936,7 +1304,8 @@ class InfoExtractor(object): def _family_friendly_search(self, html): # See http://schema.org/VideoObject - family_friendly = self._html_search_meta('isFamilyFriendly', html) + family_friendly = self._html_search_meta( + 'isFamilyFriendly', html, default=None) if not family_friendly: return None @@ -954,17 +1323,33 @@ class InfoExtractor(object): 'twitter card player') def _search_json_ld(self, html, video_id, expected_type=None, **kwargs): - json_ld = self._search_regex( - r'(?s)<script[^>]+type=(["\'])application/ld\+json\1[^>]*>(?P<json_ld>.+?)</script>', - html, 'JSON-LD', group='json_ld', **kwargs) + json_ld_list = list(re.finditer(JSON_LD_RE, html)) default = kwargs.get('default', NO_DEFAULT) - if not json_ld: - return default if default is not NO_DEFAULT else {} # JSON-LD may be malformed and thus `fatal` should be respected. # At the same time `default` may be passed that assumes `fatal=False` # for _search_regex. Let's simulate the same behavior here as well. fatal = kwargs.get('fatal', True) if default == NO_DEFAULT else False - return self._json_ld(json_ld, video_id, fatal=fatal, expected_type=expected_type) + json_ld = [] + for mobj in json_ld_list: + json_ld_item = self._parse_json( + mobj.group('json_ld'), video_id, fatal=fatal) + if not json_ld_item: + continue + if isinstance(json_ld_item, dict): + json_ld.append(json_ld_item) + elif isinstance(json_ld_item, (list, tuple)): + json_ld.extend(json_ld_item) + if json_ld: + json_ld = self._json_ld(json_ld, video_id, fatal=fatal, expected_type=expected_type) + if json_ld: + return json_ld + if default is not NO_DEFAULT: + return default + elif fatal: + raise RegexNotFoundError('Unable to extract JSON-LD') + else: + self.report_warning('unable to extract JSON-LD %s' % bug_reports_message()) + return {} def _json_ld(self, json_ld, video_id, fatal=True, expected_type=None): if isinstance(json_ld, compat_str): @@ -976,45 +1361,168 @@ class InfoExtractor(object): return info if isinstance(json_ld, dict): json_ld = [json_ld] + + INTERACTION_TYPE_MAP = { + 'CommentAction': 'comment', + 'AgreeAction': 'like', + 'DisagreeAction': 'dislike', + 'LikeAction': 'like', + 'DislikeAction': 'dislike', + 'ListenAction': 'view', + 'WatchAction': 'view', + 'ViewAction': 'view', + } + + def extract_interaction_type(e): + interaction_type = e.get('interactionType') + if isinstance(interaction_type, dict): + interaction_type = interaction_type.get('@type') + return str_or_none(interaction_type) + + def extract_interaction_statistic(e): + interaction_statistic = e.get('interactionStatistic') + if isinstance(interaction_statistic, dict): + interaction_statistic = [interaction_statistic] + if not isinstance(interaction_statistic, list): + return + for is_e in interaction_statistic: + if not isinstance(is_e, dict): + continue + if is_e.get('@type') != 'InteractionCounter': + continue + interaction_type = extract_interaction_type(is_e) + if not interaction_type: + continue + # For interaction count some sites provide string instead of + # an integer (as per spec) with non digit characters (e.g. ",") + # so extracting count with more relaxed str_to_int + interaction_count = str_to_int(is_e.get('userInteractionCount')) + if interaction_count is None: + continue + count_kind = INTERACTION_TYPE_MAP.get(interaction_type.split('/')[-1]) + if not count_kind: + continue + count_key = '%s_count' % count_kind + if info.get(count_key) is not None: + continue + info[count_key] = interaction_count + + def extract_video_object(e): + assert e['@type'] == 'VideoObject' + author = e.get('author') + info.update({ + 'url': url_or_none(e.get('contentUrl')), + 'title': unescapeHTML(e.get('name')), + 'description': unescapeHTML(e.get('description')), + 'thumbnail': url_or_none(e.get('thumbnailUrl') or e.get('thumbnailURL')), + 'duration': parse_duration(e.get('duration')), + 'timestamp': unified_timestamp(e.get('uploadDate')), + # author can be an instance of 'Organization' or 'Person' types. + # both types can have 'name' property(inherited from 'Thing' type). [1] + # however some websites are using 'Text' type instead. + # 1. https://schema.org/VideoObject + 'uploader': author.get('name') if isinstance(author, dict) else author if isinstance(author, compat_str) else None, + 'filesize': float_or_none(e.get('contentSize')), + 'tbr': int_or_none(e.get('bitrate')), + 'width': int_or_none(e.get('width')), + 'height': int_or_none(e.get('height')), + 'view_count': int_or_none(e.get('interactionCount')), + }) + extract_interaction_statistic(e) + for e in json_ld: - if e.get('@context') == 'http://schema.org': + if '@context' in e: item_type = e.get('@type') if expected_type is not None and expected_type != item_type: - return info - if item_type == 'TVEpisode': + continue + if item_type in ('TVEpisode', 'Episode'): + episode_name = unescapeHTML(e.get('name')) info.update({ - 'episode': unescapeHTML(e.get('name')), + 'episode': episode_name, 'episode_number': int_or_none(e.get('episodeNumber')), 'description': unescapeHTML(e.get('description')), }) + if not info.get('title') and episode_name: + info['title'] = episode_name part_of_season = e.get('partOfSeason') - if isinstance(part_of_season, dict) and part_of_season.get('@type') == 'TVSeason': - info['season_number'] = int_or_none(part_of_season.get('seasonNumber')) + if isinstance(part_of_season, dict) and part_of_season.get('@type') in ('TVSeason', 'Season', 'CreativeWorkSeason'): + info.update({ + 'season': unescapeHTML(part_of_season.get('name')), + 'season_number': int_or_none(part_of_season.get('seasonNumber')), + }) part_of_series = e.get('partOfSeries') or e.get('partOfTVSeries') - if isinstance(part_of_series, dict) and part_of_series.get('@type') == 'TVSeries': + if isinstance(part_of_series, dict) and part_of_series.get('@type') in ('TVSeries', 'Series', 'CreativeWorkSeries'): info['series'] = unescapeHTML(part_of_series.get('name')) - elif item_type == 'Article': + elif item_type == 'Movie': + info.update({ + 'title': unescapeHTML(e.get('name')), + 'description': unescapeHTML(e.get('description')), + 'duration': parse_duration(e.get('duration')), + 'timestamp': unified_timestamp(e.get('dateCreated')), + }) + elif item_type in ('Article', 'NewsArticle'): info.update({ 'timestamp': parse_iso8601(e.get('datePublished')), 'title': unescapeHTML(e.get('headline')), 'description': unescapeHTML(e.get('articleBody')), }) elif item_type == 'VideoObject': - info.update({ - 'url': e.get('contentUrl'), - 'title': unescapeHTML(e.get('name')), - 'description': unescapeHTML(e.get('description')), - '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')), - 'tbr': int_or_none(e.get('bitrate')), - 'width': int_or_none(e.get('width')), - 'height': int_or_none(e.get('height')), - }) - break + extract_video_object(e) + if expected_type is None: + continue + else: + break + video = e.get('video') + if isinstance(video, dict) and video.get('@type') == 'VideoObject': + extract_video_object(video) + if expected_type is None: + continue + else: + break return dict((k, v) for k, v in info.items() if v is not None) + def _search_nextjs_data(self, webpage, video_id, **kw): + # ..., *, transform_source=None, fatal=True, default=NO_DEFAULT + + # TODO: remove this backward compat + default = kw.get('default', NO_DEFAULT) + if default == '{}': + kw['default'] = {} + kw = compat_kwargs(kw) + + return self._search_json( + r'''<script\s[^>]*?\bid\s*=\s*('|")__NEXT_DATA__\1[^>]*>''', + webpage, 'next.js data', video_id, end_pattern='</script>', + **kw) + + def _search_nuxt_data(self, webpage, video_id, *args, **kwargs): + """Parses Nuxt.js metadata. This works as long as the function __NUXT__ invokes is a pure function""" + + # self, webpage, video_id, context_name='__NUXT__', *, fatal=True, traverse=('data', 0) + context_name = args[0] if len(args) > 0 else kwargs.get('context_name', '__NUXT__') + fatal = kwargs.get('fatal', True) + traverse = kwargs.get('traverse', ('data', 0)) + + re_ctx = re.escape(context_name) + + FUNCTION_RE = (r'\(\s*function\s*\((?P<arg_keys>[\s\S]*?)\)\s*\{\s*' + r'return\s+(?P<js>\{[\s\S]*?})\s*;?\s*}\s*\((?P<arg_vals>[\s\S]*?)\)') + + js, arg_keys, arg_vals = self._search_regex( + (p.format(re_ctx, FUNCTION_RE) for p in + (r'<script>\s*window\s*\.\s*{0}\s*=\s*{1}\s*\)\s*;?\s*</script>', + r'{0}\s*\([\s\S]*?{1}')), + webpage, context_name, group=('js', 'arg_keys', 'arg_vals'), + default=NO_DEFAULT if fatal else (None, None, None)) + if js is None: + return {} + + args = dict(zip(arg_keys.split(','), map(json.dumps, self._parse_json( + '[{0}]'.format(arg_vals), video_id, transform_source=js_to_json, fatal=fatal) or ()))) + + ret = self._parse_json(js, video_id, transform_source=functools.partial(js_to_json, vars=args), fatal=fatal) + return traverse_obj(ret, traverse) or {} + @staticmethod def _hidden_inputs(html): html = re.sub(r'<!--(?:(?!<!--).)*-->', '', html) @@ -1071,7 +1579,7 @@ class InfoExtractor(object): if f.get('vcodec') == 'none': # audio only preference -= 50 - if self._downloader.params.get('prefer_free_formats'): + if self.get_param('prefer_free_formats'): ORDER = ['aac', 'mp3', 'm4a', 'webm', 'ogg', 'opus'] else: ORDER = ['webm', 'opus', 'ogg', 'mp3', 'aac', 'm4a'] @@ -1083,7 +1591,7 @@ class InfoExtractor(object): else: if f.get('acodec') == 'none': # video only preference -= 40 - if self._downloader.params.get('prefer_free_formats'): + if self.get_param('prefer_free_formats'): ORDER = ['flv', 'mp4', 'webm'] else: ORDER = ['webm', 'flv', 'mp4'] @@ -1140,17 +1648,16 @@ class InfoExtractor(object): 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): - self.to_screen( - '%s: %s URL is invalid, skipping' % (video_id, item)) - return False - raise + self.to_screen( + '%s: %s URL is invalid, skipping: %s' + % (video_id, item, error_to_compat_str(e.cause))) + return False def http_scheme(self): """ Either "http:" or "https:", depending on the user's preferences """ return ( 'http:' - if self._downloader.params.get('prefer_insecure', False) + if self.get_param('prefer_insecure', False) else 'https:') def _proto_relative_url(self, url, scheme=None): @@ -1172,14 +1679,14 @@ class InfoExtractor(object): def _extract_f4m_formats(self, manifest_url, video_id, preference=None, f4m_id=None, transform_source=lambda s: fix_xml_ampersands(s).strip(), - fatal=True, m3u8_id=None): + fatal=True, m3u8_id=None, data=None, headers={}, query={}): manifest = self._download_xml( manifest_url, video_id, 'Downloading f4m manifest', 'Unable to download f4m manifest', # Some manifests may be malformed, e.g. prosiebensat1 generated manifests - # (see https://github.com/rg3/youtube-dl/issues/6215#issuecomment-121704244) + # (see https://github.com/ytdl-org/youtube-dl/issues/6215#issuecomment-121704244) transform_source=transform_source, - fatal=fatal) + fatal=fatal, data=data, headers=headers, query=query) if manifest is False: return [] @@ -1191,6 +1698,9 @@ class InfoExtractor(object): def _parse_f4m_formats(self, manifest, manifest_url, video_id, preference=None, f4m_id=None, transform_source=lambda s: fix_xml_ampersands(s).strip(), fatal=True, m3u8_id=None): + if not isinstance(manifest, compat_etree_Element) and not fatal: + return [] + # currently youtube-dl cannot decode the playerVerificationChallenge as Akamai uses Adobe Alchemy akamai_pv = manifest.find('{http://ns.adobe.com/f4m/1.0}pv-2.0') if akamai_pv is not None and ';' in akamai_pv.text: @@ -1205,15 +1715,12 @@ class InfoExtractor(object): manifest_version = '2.0' media_nodes = manifest.findall('{http://ns.adobe.com/f4m/2.0}media') # Remove unsupported DRM protected media from final formats - # rendition (see https://github.com/rg3/youtube-dl/issues/8573). + # rendition (see https://github.com/ytdl-org/youtube-dl/issues/8573). media_nodes = remove_encrypted_media(media_nodes) if not media_nodes: return formats - base_url = xpath_text( - manifest, ['{http://ns.adobe.com/f4m/1.0}baseURL', '{http://ns.adobe.com/f4m/2.0}baseURL'], - 'base URL', default=None) - if base_url: - base_url = base_url.strip() + + manifest_base_url = get_base_url(manifest) bootstrap_info = xpath_element( manifest, ['{http://ns.adobe.com/f4m/1.0}bootstrapInfo', '{http://ns.adobe.com/f4m/2.0}bootstrapInfo'], @@ -1245,7 +1752,7 @@ class InfoExtractor(object): continue manifest_url = ( media_url if media_url.startswith('http://') or media_url.startswith('https://') - else ((base_url or '/'.join(manifest_url.split('/')[:-1])) + '/' + media_url)) + else ((manifest_base_url or '/'.join(manifest_url.split('/')[:-1])) + '/' + media_url)) # If media_url is itself a f4m manifest do the recursive extraction # since bitrates in parent manifest (this one) and media_url manifest # may differ leading to inability to resolve the format by requested @@ -1280,6 +1787,7 @@ class InfoExtractor(object): 'url': manifest_url, 'manifest_url': manifest_url, 'ext': 'flv' if bootstrap_info is not None else None, + 'protocol': 'f4m', 'tbr': tbr, 'width': width, 'height': height, @@ -1299,44 +1807,65 @@ class InfoExtractor(object): 'format_note': 'Quality selection URL', } + def _report_ignoring_subs(self, name): + self.report_warning(bug_reports_message( + 'Ignoring subtitle tracks found in the {0} manifest; ' + 'if any subtitle tracks are missing,'.format(name) + ), only_once=True) + def _extract_m3u8_formats(self, m3u8_url, video_id, ext=None, entry_protocol='m3u8', preference=None, m3u8_id=None, note=None, errnote=None, - fatal=True, live=False): - + fatal=True, live=False, data=None, headers={}, + query={}): res = self._download_webpage_handle( m3u8_url, video_id, note=note or 'Downloading m3u8 information', errnote=errnote or 'Failed to download m3u8 information', - fatal=fatal) + fatal=fatal, data=data, headers=headers, query=query) + if res is False: return [] + m3u8_doc, urlh = res m3u8_url = urlh.geturl() + return self._parse_m3u8_formats( + m3u8_doc, m3u8_url, ext=ext, entry_protocol=entry_protocol, + preference=preference, m3u8_id=m3u8_id, live=live) + + def _parse_m3u8_formats(self, m3u8_doc, m3u8_url, ext=None, + entry_protocol='m3u8', preference=None, + m3u8_id=None, live=False): if '#EXT-X-FAXS-CM:' in m3u8_doc: # Adobe Flash Access return [] - formats = [self._m3u8_meta_format(m3u8_url, ext, preference, m3u8_id)] + if re.search(r'#EXT-X-SESSION-KEY:.*?URI="skd://', m3u8_doc): # Apple FairPlay + return [] + + formats = [] 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 - # without qualities renditions. + # References: + # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-21 + # 2. https://github.com/ytdl-org/youtube-dl/issues/12211 + # 3. https://github.com/ytdl-org/youtube-dl/issues/18923 + + # We should try extracting formats only from master playlists [1, 4.3.4], + # i.e. playlists that describe available qualities. On the other hand + # media playlists [1, 4.3.3] should be returned as is since they contain + # just the media without qualities renditions. # Fortunately, master playlist can be easily distinguished from media - # playlist based on particular tags availability. As of [1, 2] master - # playlist tags MUST NOT appear in a media playist and vice versa. - # As of [3] #EXT-X-TARGETDURATION tag is REQUIRED for every media playlist - # and MUST NOT appear in master playlist thus we can clearly detect media - # playlist with this criterion. - # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.4 - # 2. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3 - # 3. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.1 + # playlist based on particular tags availability. As of [1, 4.3.3, 4.3.4] + # master playlist tags MUST NOT appear in a media playlist and vice versa. + # As of [1, 4.3.3.1] #EXT-X-TARGETDURATION tag is REQUIRED for every + # media playlist and MUST NOT appear in master playlist thus we can + # clearly detect media playlist with this criterion. + if '#EXT-X-TARGETDURATION' in m3u8_doc: # media playlist, return as is return [{ 'url': m3u8_url, @@ -1345,52 +1874,77 @@ class InfoExtractor(object): 'protocol': entry_protocol, 'preference': preference, }] - audio_in_video_stream = {} - last_info = {} - last_media = {} + + groups = {} + last_stream_inf = {} + + def extract_media(x_media_line): + media = parse_m3u8_attributes(x_media_line) + # As per [1, 4.3.4.1] TYPE, GROUP-ID and NAME are REQUIRED + media_type, group_id, name = media.get('TYPE'), media.get('GROUP-ID'), media.get('NAME') + if not (media_type and group_id and name): + return + groups.setdefault(group_id, []).append(media) + if media_type not in ('VIDEO', 'AUDIO'): + return + media_url = media.get('URI') + if media_url: + format_id = [] + for v in (m3u8_id, group_id, name): + if v: + format_id.append(v) + f = { + 'format_id': '-'.join(format_id), + 'url': format_url(media_url), + 'manifest_url': m3u8_url, + 'language': media.get('LANGUAGE'), + 'ext': ext, + 'protocol': entry_protocol, + 'preference': preference, + } + if media_type == 'AUDIO': + f['vcodec'] = 'none' + formats.append(f) + + def build_stream_name(): + # Despite specification does not mention NAME attribute for + # EXT-X-STREAM-INF tag it still sometimes may be present (see [1] + # or vidio test in TestInfoExtractor.test_parse_m3u8_formats) + # 1. http://www.vidio.com/watch/165683-dj_ambred-booyah-live-2015 + stream_name = last_stream_inf.get('NAME') + if stream_name: + return stream_name + # If there is no NAME in EXT-X-STREAM-INF it will be obtained + # from corresponding rendition group + stream_group_id = last_stream_inf.get('VIDEO') + if not stream_group_id: + return + stream_group = groups.get(stream_group_id) + if not stream_group: + return stream_group_id + rendition = stream_group[0] + return rendition.get('NAME') or stream_group_id + + # parse EXT-X-MEDIA tags before EXT-X-STREAM-INF in order to have the + # chance to detect video only formats when EXT-X-STREAM-INF tags + # precede EXT-X-MEDIA tags in HLS manifest such as [3]. + for line in m3u8_doc.splitlines(): + if line.startswith('#EXT-X-MEDIA:'): + extract_media(line) + 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:'): - 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 + last_stream_inf = parse_m3u8_attributes(line) elif line.startswith('#') or not line.strip(): continue else: - tbr = int_or_none(last_info.get('AVERAGE-BANDWIDTH') or last_info.get('BANDWIDTH'), scale=1000) + tbr = float_or_none( + last_stream_inf.get('AVERAGE-BANDWIDTH') + or last_stream_inf.get('BANDWIDTH'), scale=1000) format_id = [] if m3u8_id: format_id.append(m3u8_id) - # 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.get('NAME') + stream_name = build_stream_name() # Bandwidth of live streams may differ over time thus making # format_id unpredictable. So it's better to keep provided # format_id intact. @@ -1400,14 +1954,14 @@ class InfoExtractor(object): f = { 'format_id': '-'.join(format_id), 'url': manifest_url, - 'manifest_url': manifest_url, + 'manifest_url': m3u8_url, 'tbr': tbr, 'ext': ext, - 'fps': float_or_none(last_info.get('FRAME-RATE')), + 'fps': float_or_none(last_stream_inf.get('FRAME-RATE')), 'protocol': entry_protocol, 'preference': preference, } - resolution = last_info.get('RESOLUTION') + resolution = last_stream_inf.get('RESOLUTION') if resolution: mobj = re.search(r'(?P<width>\d+)[xX](?P<height>\d+)', resolution) if mobj: @@ -1423,13 +1977,39 @@ class InfoExtractor(object): 'vbr': vbr, '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' + codecs = parse_codecs(last_stream_inf.get('CODECS')) + f.update(codecs) + audio_group_id = last_stream_inf.get('AUDIO') + # As per [1, 4.3.4.1.1] any EXT-X-STREAM-INF tag which + # references a rendition group MUST have a CODECS attribute. + # However, this is not always respected, for example, [2] + # contains EXT-X-STREAM-INF tag which references AUDIO + # rendition group but does not have CODECS and despite + # referencing an audio group it represents a complete + # (with audio and video) format. So, for such cases we will + # ignore references to rendition groups and treat them + # as complete formats. + if audio_group_id and codecs and f.get('vcodec') != 'none': + audio_group = groups.get(audio_group_id) + if audio_group and audio_group[0].get('URI'): + # TODO: update acodec for audio only formats with + # the same GROUP-ID + f['acodec'] = 'none' formats.append(f) - last_info = {} - last_media = {} + + # for DailyMotion + progressive_uri = last_stream_inf.get('PROGRESSIVE-URI') + if progressive_uri: + http_f = f.copy() + del http_f['manifest_url'] + http_f.update({ + 'format_id': f['format_id'].replace('hls-', 'http-'), + 'protocol': 'http', + 'url': progressive_uri, + }) + formats.append(http_f) + + last_stream_inf = {} return formats @staticmethod @@ -1576,9 +2156,7 @@ class InfoExtractor(object): 'height': height, }) formats.extend(m3u8_formats) - continue - - if src_ext == 'f4m': + elif src_ext == 'f4m': f4m_url = src_url if not f4m_params: f4m_params = { @@ -1588,9 +2166,13 @@ class InfoExtractor(object): f4m_url += '&' if '?' in f4m_url else '?' f4m_url += compat_urllib_parse_urlencode(f4m_params) formats.extend(self._extract_f4m_formats(f4m_url, video_id, f4m_id='hds', fatal=False)) - continue - - if src_url.startswith('http') and self._is_valid_url(src, video_id): + elif src_ext == 'mpd': + formats.extend(self._extract_mpd_formats( + src_url, video_id, mpd_id='dash', fatal=False)) + elif re.search(r'\.ism/[Mm]anifest', src_url): + formats.extend(self._extract_ism_formats( + src_url, video_id, ism_id='mss', fatal=False)) + elif src_url.startswith('http') and self._is_valid_url(src, video_id): http_count += 1 formats.append({ 'url': src_url, @@ -1601,7 +2183,6 @@ class InfoExtractor(object): 'width': width, 'height': height, }) - continue return formats @@ -1621,22 +2202,24 @@ class InfoExtractor(object): }) return subtitles - def _extract_xspf_playlist(self, playlist_url, playlist_id, fatal=True): + def _extract_xspf_playlist(self, xspf_url, playlist_id, fatal=True): xspf = self._download_xml( - playlist_url, playlist_id, 'Downloading xpsf playlist', + xspf_url, playlist_id, 'Downloading xpsf playlist', 'Unable to download xspf manifest', fatal=fatal) if xspf is False: return [] - return self._parse_xspf(xspf, playlist_id) + return self._parse_xspf( + xspf, playlist_id, xspf_url=xspf_url, + xspf_base_url=base_url(xspf_url)) - def _parse_xspf(self, playlist, playlist_id): + def _parse_xspf(self, xspf_doc, playlist_id, xspf_url=None, xspf_base_url=None): NS_MAP = { 'xspf': 'http://xspf.org/ns/0/', 's1': 'http://static.streamone.nl/player/ns/0', } entries = [] - for track in playlist.findall(xpath_with_ns('./xspf:trackList/xspf:track', NS_MAP)): + for track in xspf_doc.findall(xpath_with_ns('./xspf:trackList/xspf:track', NS_MAP)): title = xpath_text( track, xpath_with_ns('./xspf:title', NS_MAP), 'title', default=playlist_id) description = xpath_text( @@ -1646,12 +2229,18 @@ class InfoExtractor(object): duration = float_or_none( xpath_text(track, xpath_with_ns('./xspf:duration', NS_MAP), 'duration'), 1000) - formats = [{ - 'url': location.text, - 'format_id': location.get(xpath_with_ns('s1:label', NS_MAP)), - 'width': int_or_none(location.get(xpath_with_ns('s1:width', NS_MAP))), - 'height': int_or_none(location.get(xpath_with_ns('s1:height', NS_MAP))), - } for location in track.findall(xpath_with_ns('./xspf:location', NS_MAP))] + formats = [] + for location in track.findall(xpath_with_ns('./xspf:location', NS_MAP)): + format_url = urljoin(xspf_base_url, location.text) + if not format_url: + continue + formats.append({ + 'url': format_url, + 'manifest_url': xspf_url, + 'format_id': location.get(xpath_with_ns('s1:label', NS_MAP)), + 'width': int_or_none(location.get(xpath_with_ns('s1:width', NS_MAP))), + 'height': int_or_none(location.get(xpath_with_ns('s1:height', NS_MAP))), + }) self._sort_formats(formats) entries.append({ @@ -1664,22 +2253,46 @@ class InfoExtractor(object): }) return entries - def _extract_mpd_formats(self, mpd_url, video_id, mpd_id=None, note=None, errnote=None, fatal=True, formats_dict={}): - res = self._download_webpage_handle( - mpd_url, video_id, - note=note or 'Downloading MPD manifest', - errnote=errnote or 'Failed to download MPD manifest', - fatal=fatal) - if res is False: - return [] - mpd, urlh = res - mpd_base_url = base_url(urlh.geturl()) + def _extract_mpd_formats(self, *args, **kwargs): + fmts, subs = self._extract_mpd_formats_and_subtitles(*args, **kwargs) + if subs: + self._report_ignoring_subs('DASH') + return fmts + + def _extract_mpd_formats_and_subtitles( + self, mpd_url, video_id, mpd_id=None, note=None, errnote=None, + fatal=True, data=None, headers=None, query=None): - return self._parse_mpd_formats( - compat_etree_fromstring(mpd.encode('utf-8')), mpd_id, mpd_base_url, - formats_dict=formats_dict, mpd_url=mpd_url) + # TODO: or not? param not yet implemented + if self.get_param('ignore_no_formats_error'): + fatal = False - def _parse_mpd_formats(self, mpd_doc, mpd_id=None, mpd_base_url='', formats_dict={}, mpd_url=None): + res = self._download_xml_handle( + mpd_url, video_id, + note='Downloading MPD manifest' if note is None else note, + errnote='Failed to download MPD manifest' if errnote is None else errnote, + fatal=fatal, data=data, headers=headers or {}, query=query or {}) + if res is False: + return [], {} + mpd_doc, urlh = res + if mpd_doc is None: + return [], {} + + # We could have been redirected to a new url when we retrieved our mpd file. + mpd_url = urlh.geturl() + mpd_base_url = base_url(mpd_url) + + return self._parse_mpd_formats_and_subtitles( + mpd_doc, mpd_id, mpd_base_url, mpd_url) + + def _parse_mpd_formats(self, *args, **kwargs): + fmts, subs = self._parse_mpd_formats_and_subtitles(*args, **kwargs) + if subs: + self._report_ignoring_subs('DASH') + return fmts + + def _parse_mpd_formats_and_subtitles( + self, mpd_doc, mpd_id=None, mpd_base_url='', mpd_url=None): """ Parse formats from MPD manifest. References: @@ -1687,8 +2300,10 @@ class InfoExtractor(object): http://standards.iso.org/ittf/PubliclyAvailableStandards/c065274_ISO_IEC_23009-1_2014.zip 2. https://en.wikipedia.org/wiki/Dynamic_Adaptive_Streaming_over_HTTP """ - if mpd_doc.get('type') == 'dynamic': - return [] + # TODO: param not yet implemented: default like previous yt-dl logic + if not self.get_param('dynamic_mpd', False): + if mpd_doc.get('type') == 'dynamic': + return [], {} namespace = self._search_regex(r'(?i)^{([^}]+)?}MPD$', mpd_doc.tag, 'namespace', default=None) @@ -1698,8 +2313,24 @@ class InfoExtractor(object): def is_drm_protected(element): return element.find(_add_ns('ContentProtection')) is not None + from ..utils import YoutubeDLHandler + fix_path = YoutubeDLHandler._fix_path + + def resolve_base_url(element, parent_base_url=None): + # TODO: use native XML traversal when ready + b_url = traverse_obj(element, ( + T(lambda e: e.find(_add_ns('BaseURL')).text))) + if parent_base_url and b_url: + if not parent_base_url[-1] in ('/', ':'): + parent_base_url += '/' + b_url = compat_urlparse.urljoin(parent_base_url, b_url) + if b_url: + b_url = fix_path(b_url) + return b_url or parent_base_url + def extract_multisegment_info(element, ms_parent_info): ms_info = ms_parent_info.copy() + base_url = ms_info['base_url'] = resolve_base_url(element, ms_info.get('base_url')) # As per [1, 5.3.9.2.2] SegmentList and SegmentTemplate share some # common attributes and elements. We will only extract relevant @@ -1728,20 +2359,32 @@ class InfoExtractor(object): ms_info['timescale'] = int(timescale) segment_duration = source.get('duration') if segment_duration: - ms_info['segment_duration'] = int(segment_duration) + ms_info['segment_duration'] = float(segment_duration) def extract_Initialization(source): initialization = source.find(_add_ns('Initialization')) if initialization is not None: - ms_info['initialization_url'] = initialization.attrib['sourceURL'] + ms_info['initialization_url'] = initialization.get('sourceURL') or base_url + initialization_url_range = initialization.get('range') + if initialization_url_range: + ms_info['initialization_url_range'] = initialization_url_range 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] + segment_urls = traverse_obj(segment_urls_e, ( + Ellipsis, T(lambda e: e.attrib), 'media')) + if segment_urls: + ms_info['segment_urls'] = segment_urls + segment_urls_range = traverse_obj(segment_urls_e, ( + Ellipsis, T(lambda e: e.attrib), 'mediaRange', + T(lambda r: re.findall(r'^\d+-\d+$', r)), 0)) + if segment_urls_range: + ms_info['segment_urls_range'] = segment_urls_range + if not segment_urls: + ms_info['segment_urls'] = [base_url for _ in segment_urls_range] else: segment_template = element.find(_add_ns('SegmentTemplate')) if segment_template is not None: @@ -1757,17 +2400,20 @@ class InfoExtractor(object): return ms_info mpd_duration = parse_duration(mpd_doc.get('mediaPresentationDuration')) - formats = [] + formats, subtitles = [], {} + stream_numbers = collections.defaultdict(int) + mpd_base_url = resolve_base_url(mpd_doc, mpd_base_url or mpd_url) for period in mpd_doc.findall(_add_ns('Period')): period_duration = parse_duration(period.get('duration')) or mpd_duration period_ms_info = extract_multisegment_info(period, { 'start_number': 1, 'timescale': 1, + 'base_url': mpd_base_url, }) for adaptation_set in period.findall(_add_ns('AdaptationSet')): if is_drm_protected(adaptation_set): continue - adaption_set_ms_info = extract_multisegment_info(adaptation_set, period_ms_info) + adaptation_set_ms_info = extract_multisegment_info(adaptation_set, period_ms_info) for representation in adaptation_set.findall(_add_ns('Representation')): if is_drm_protected(representation): continue @@ -1775,175 +2421,278 @@ class InfoExtractor(object): representation_attrib.update(representation.attrib) # According to [1, 5.3.7.2, Table 9, page 41], @mimeType is mandatory mime_type = representation_attrib['mimeType'] - content_type = mime_type.split('/')[0] - if content_type == 'text': - # TODO implement WebVTT downloading - pass - elif content_type in ('video', 'audio'): - base_url = '' - for element in (representation, adaptation_set, period, mpd_doc): - base_url_e = element.find(_add_ns('BaseURL')) - if base_url_e is not None: - base_url = base_url_e.text + base_url - if re.match(r'^https?://', base_url): - break - if mpd_base_url and not re.match(r'^https?://', base_url): - if not mpd_base_url.endswith('/') and not base_url.startswith('/'): - mpd_base_url += '/' - base_url = mpd_base_url + base_url - representation_id = representation_attrib.get('id') - 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')) + content_type = representation_attrib.get('contentType') or mime_type.split('/')[0] + codec_str = representation_attrib.get('codecs', '') + # Some kind of binary subtitle found in some youtube livestreams + if mime_type == 'application/x-rawcc': + codecs = {'scodec': codec_str} + else: + codecs = parse_codecs(codec_str) + if content_type not in ('video', 'audio', 'text'): + if mime_type == 'image/jpeg': + content_type = mime_type + elif codecs.get('vcodec', 'none') != 'none': + content_type = 'video' + elif codecs.get('acodec', 'none') != 'none': + content_type = 'audio' + elif codecs.get('scodec', 'none') != 'none': + content_type = 'text' + elif mimetype2ext(mime_type) in ('tt', 'dfxp', 'ttml', 'xml', 'json'): + content_type = 'text' + else: + self.report_warning('Unknown MIME type %s in DASH manifest' % mime_type) + continue + + representation_id = representation_attrib.get('id') + lang = representation_attrib.get('lang') + url_el = representation.find(_add_ns('BaseURL')) + filesize = int_or_none(url_el.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')) + format_id = join_nonempty(representation_id or content_type, mpd_id) + if content_type in ('video', 'audio'): 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(bandwidth, 1000), + 'tbr': float_or_none(bandwidth, 1000), 'asr': int_or_none(representation_attrib.get('audioSamplingRate')), 'fps': int_or_none(representation_attrib.get('frameRate')), 'language': lang if lang not in ('mul', 'und', 'zxx', 'mis') else None, 'format_note': 'DASH %s' % content_type, 'filesize': filesize, + 'container': mimetype2ext(mime_type) + '_dash', + } + f.update(codecs) + elif content_type == 'text': + f = { + 'ext': mimetype2ext(mime_type), + 'manifest_url': mpd_url, + 'filesize': filesize, + } + elif content_type == 'image/jpeg': + # See test case in VikiIE + # https://www.viki.com/videos/1175236v-choosing-spouse-by-lottery-episode-1 + f = { + 'format_id': format_id, + 'ext': 'mhtml', + 'manifest_url': mpd_url, + 'format_note': 'DASH storyboards (jpeg)', + 'acodec': 'none', + 'vcodec': 'none', + } + if is_drm_protected(adaptation_set) or is_drm_protected(representation): + f['has_drm'] = True + representation_ms_info = extract_multisegment_info(representation, adaptation_set_ms_info) + + def prepare_template(template_name, identifiers): + tmpl = representation_ms_info[template_name] + # First of, % characters outside $...$ templates + # must be escaped by doubling for proper processing + # by % operator string formatting used further (see + # https://github.com/ytdl-org/youtube-dl/issues/16867). + t = '' + in_template = False + for c in tmpl: + t += c + if c == '$': + in_template = not in_template + elif c == '%' and not in_template: + t += c + # Next, $...$ templates are translated to their + # %(...) counterparts to be used with % operator + 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('$$', '$') + return t + + # @initialization is a regular template like @media one + # so it should be handled just the same way (see + # https://github.com/ytdl-org/youtube-dl/issues/11605) + if 'initialization' in representation_ms_info: + initialization_template = prepare_template( + 'initialization', + # As per [1, 5.3.9.4.2, Table 15, page 54] $Number$ and + # $Time$ shall not be included for @initialization thus + # only $Bandwidth$ remains + ('Bandwidth', )) + representation_ms_info['initialization_url'] = initialization_template % { + 'Bandwidth': bandwidth, } - f.update(parse_codecs(representation_attrib.get('codecs'))) - representation_ms_info = extract_multisegment_info(representation, adaption_set_ms_info) - - def prepare_template(template_name, identifiers): - t = representation_ms_info[template_name] - t = t.replace('$RepresentationID$', representation_id) - t = re.sub(r'\$(%s)\$' % '|'.join(identifiers), r'%(\1)d', t) - t = re.sub(r'\$(%s)%%([^$]+)\$' % '|'.join(identifiers), r'%(\1)\2', t) - t.replace('$$', '$') - 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 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': 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: - # $Number*$ 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(): - 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'] + def location_key(location): + return 'url' if re.match(r'^https?://', location) else 'path' + + def calc_segment_duration(): + return float_or_none( + representation_ms_info['segment_duration'], + representation_ms_info['timescale']) if 'segment_duration' in representation_ms_info else None + + if 'segment_urls' not in representation_ms_info and 'media' in representation_ms_info: + + media_template = prepare_template('media', ('Number', 'Bandwidth', 'Time')) + media_location_key = location_key(media_template) + + # 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 and 's' not in representation_ms_info: + segment_duration = None + if 'total_number' not in representation_ms_info and 'segment_duration' in representation_ms_info: + segment_duration = float_or_none(representation_ms_info['segment_duration'], representation_ms_info['timescale']) + representation_ms_info['total_number'] = int(math.ceil( + float_or_none(period_duration, segment_duration, default=0))) + representation_ms_info['fragments'] = [{ + media_location_key: media_template % { + 'Number': segment_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: + # $Number*$ 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(): + segment_url = media_template % { + 'Time': segment_time, + 'Bandwidth': bandwidth, + 'Number': segment_number, + } + representation_ms_info['fragments'].append({ + media_location_key: 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 += segment_d add_segment_url() segment_number += 1 - for r in range(s.get('r', 0)): - segment_time += segment_d - add_segment_url() - segment_number += 1 - segment_time += segment_d - elif 'segment_urls' in representation_ms_info and 's' in representation_ms_info: + segment_time += segment_d + elif 'segment_urls' in representation_ms_info: + fragments = [] + if '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): + segment_uri = representation_ms_info['segment_urls'][segment_index] fragments.append({ - 'url': representation_ms_info['segment_urls'][segment_index], + location_key(segment_uri): segment_uri, '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({ - 'fragments': [], - 'protocol': 'http_dash_segments', - }) - if 'initialization_url' in representation_ms_info: - 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 - if fo['format_id'] == representation_id) - except StopIteration: - full_info = formats_dict.get(representation_id, {}).copy() - full_info.update(f) - formats.append(full_info) + elif 'segment_urls_range' in representation_ms_info: + # Segment URLs with mediaRange + # Example: https://kinescope.io/200615537/master.mpd + # https://github.com/ytdl-org/youtube-dl/issues/30235 + # or any mpd generated with Bento4 `mp4dash --no-split --use-segment-list` + segment_duration = calc_segment_duration() + for segment_url, segment_url_range in zip( + representation_ms_info['segment_urls'], representation_ms_info['segment_urls_range']): + fragments.append({ + location_key(segment_url): segment_url, + 'range': segment_url_range, + 'duration': segment_duration, + }) else: - existing_format.update(f) + # Segment URLs with no SegmentTimeline + # Example: https://www.seznam.cz/zpravy/clanek/cesko-zasahne-vitr-o-sile-vichrice-muze-byt-i-zivotu-nebezpecny-39091 + # https://github.com/ytdl-org/youtube-dl/pull/14844 + segment_duration = calc_segment_duration() + for segment_url in representation_ms_info['segment_urls']: + fragments.append({ + location_key(segment_url): segment_url, + 'duration': segment_duration, + }) + representation_ms_info['fragments'] = fragments + + # If there is a fragments key available then we correctly recognized fragmented media. + # Otherwise we will assume unfragmented media with direct access. Technically, such + # assumption is not necessarily correct since we may simply have no support for + # some forms of fragmented media renditions yet, but for now we'll use this fallback. + if 'fragments' in representation_ms_info: + base_url = representation_ms_info['base_url'] + f.update({ + # NB: mpd_url may be empty when MPD manifest is parsed from a string + 'url': mpd_url or base_url, + 'fragment_base_url': base_url, + 'fragments': [], + 'protocol': 'http_dash_segments', + }) + if 'initialization_url' in representation_ms_info and 'initialization_url_range' in representation_ms_info: + # Initialization URL with range (accompanied by Segment URLs with mediaRange above) + # https://github.com/ytdl-org/youtube-dl/issues/30235 + initialization_url = representation_ms_info['initialization_url'] + f['fragments'].append({ + location_key(initialization_url): initialization_url, + 'range': representation_ms_info['initialization_url_range'], + }) + elif 'initialization_url' in representation_ms_info: + initialization_url = representation_ms_info['initialization_url'] + if not f.get('url'): + f['url'] = initialization_url + f['fragments'].append({location_key(initialization_url): initialization_url}) + elif 'initialization_url_range' in representation_ms_info: + # no Initialization URL but range (accompanied by no Segment URLs but mediaRange above) + # https://github.com/ytdl-org/youtube-dl/issues/27575 + f['fragments'].append({ + location_key(base_url): base_url, + 'range': representation_ms_info['initialization_url_range'], + }) + f['fragments'].extend(representation_ms_info['fragments']) + if not period_duration: + period_duration = sum(traverse_obj(representation_ms_info, ( + 'fragments', Ellipsis, 'duration', T(float_or_none)))) else: - self.report_warning('Unknown MIME type %s in DASH manifest' % mime_type) - return formats + # Assuming direct URL to unfragmented media. + f['url'] = representation_ms_info['base_url'] + if content_type in ('video', 'audio', 'image/jpeg'): + f['manifest_stream_number'] = stream_numbers[f['url']] + stream_numbers[f['url']] += 1 + formats.append(f) + elif content_type == 'text': + subtitles.setdefault(lang or 'und', []).append(f) + return formats, subtitles - def _extract_ism_formats(self, ism_url, video_id, ism_id=None, note=None, errnote=None, fatal=True): - res = self._download_webpage_handle( + def _extract_ism_formats(self, ism_url, video_id, ism_id=None, note=None, errnote=None, fatal=True, data=None, headers={}, query={}): + res = self._download_xml_handle( ism_url, video_id, note=note or 'Downloading ISM manifest', errnote=errnote or 'Failed to download ISM manifest', - fatal=fatal) + fatal=fatal, data=data, headers=headers, query=query) if res is False: return [] - ism, urlh = res + ism_doc, urlh = res + if ism_doc is None: + return [] - return self._parse_ism_formats( - compat_etree_fromstring(ism.encode('utf-8')), urlh.geturl(), ism_id) + return self._parse_ism_formats(ism_doc, urlh.geturl(), ism_id) def _parse_ism_formats(self, ism_doc, ism_url, ism_id=None): + """ + Parse formats from ISM manifest. + References: + 1. [MS-SSTR]: Smooth Streaming Protocol, + https://msdn.microsoft.com/en-us/library/ff469518.aspx + """ if ism_doc.get('IsLive') == 'TRUE' or ism_doc.find('Protection') is not None: return [] @@ -1959,14 +2708,17 @@ class InfoExtractor(object): 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') + fourcc = track.get('FourCC', 'AACL' if track.get('AudioTag') == '255' else None) # 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')) + # [1] does not mention Width and Height attributes. However, + # they're often present while MaxWidth and MaxHeight are + # missing, so should be used as fallbacks + width = int_or_none(track.get('MaxWidth') or track.get('Width')) + height = int_or_none(track.get('MaxHeight') or track.get('Height')) sampling_rate = int_or_none(track.get('SamplingRate')) track_url_pattern = re.sub(r'{[Bb]itrate}', track.attrib['Bitrate'], url_pattern) @@ -2030,8 +2782,8 @@ class InfoExtractor(object): return formats def _parse_html5_media_entries(self, base_url, webpage, video_id, m3u8_id=None, m3u8_entry_protocol='m3u8', mpd_id=None, preference=None): - def absolute_url(video_url): - return compat_urlparse.urljoin(base_url, video_url) + def absolute_url(item_url): + return urljoin(base_url, item_url) def parse_content_type(content_type): if not content_type: @@ -2044,57 +2796,96 @@ class InfoExtractor(object): return f return {} - def _media_formats(src, cur_media_type): + def _media_formats(src, cur_media_type, type_info=None): + type_info = type_info or {} full_url = absolute_url(src) - ext = determine_ext(full_url) + ext = type_info.get('ext') or 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, - preference=preference) + preference=preference, fatal=False) elif ext == 'mpd': is_plain_url = False formats = self._extract_mpd_formats( - full_url, video_id, mpd_id=mpd_id) + full_url, video_id, mpd_id=mpd_id, fatal=False) else: is_plain_url = True formats = [{ 'url': full_url, 'vcodec': 'none' if cur_media_type == 'audio' else None, + 'ext': ext, }] return is_plain_url, formats entries = [] - media_tags = [(media_tag, media_type, '') - for media_tag, media_type - in re.findall(r'(?s)(<(video|audio)[^>]*/>)', webpage)] + # amp-video and amp-audio are very similar to their HTML5 counterparts + # so we wll include them right here (see + # https://www.ampproject.org/docs/reference/components/amp-video) + # For dl8-* tags see https://delight-vr.com/documentation/dl8-video/ + _MEDIA_TAG_NAME_RE = r'(?:(?:amp|dl8(?:-live)?)-)?(video(?:-js)?|audio)' + media_tags = [(media_tag, media_tag_name, media_type, '') + for media_tag, media_tag_name, media_type + in re.findall(r'(?s)(<(%s)[^>]*/>)' % _MEDIA_TAG_NAME_RE, 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: + # https://github.com/ytdl-org/youtube-dl/issues/11979, example URL: # http://www.porntrex.com/maps/videositemap.xml). - r'(?s)(<(?P<tag>video|audio)(?:\s+[^>]*)?>)(.*?)</(?P=tag)>', webpage)) - for media_tag, media_type, media_content in media_tags: + r'(?s)(<(?P<tag>%s)(?:\s+[^>]*)?>)(.*?)</(?P=tag)>' % _MEDIA_TAG_NAME_RE, webpage)) + for media_tag, _, media_type, media_content in media_tags: media_info = { 'formats': [], 'subtitles': {}, } media_attributes = extract_attributes(media_tag) - src = media_attributes.get('src') + src = strip_or_none(media_attributes.get('src')) if src: - _, formats = _media_formats(src, media_type) + f = parse_content_type(media_attributes.get('type')) + _, formats = _media_formats(src, media_type, f) media_info['formats'].extend(formats) - media_info['thumbnail'] = media_attributes.get('poster') + media_info['thumbnail'] = absolute_url(media_attributes.get('poster')) if media_content: for source_tag in re.findall(r'<source[^>]+>', media_content): - source_attributes = extract_attributes(source_tag) - src = source_attributes.get('src') + s_attr = extract_attributes(source_tag) + # data-video-src and data-src are non standard but seen + # several times in the wild + src = strip_or_none(dict_get(s_attr, ('src', 'data-video-src', 'data-src'))) if not src: continue - is_plain_url, formats = _media_formats(src, media_type) + f = parse_content_type(s_attr.get('type')) + is_plain_url, formats = _media_formats(src, media_type, f) if is_plain_url: - f = parse_content_type(source_attributes.get('type')) + # width, height, res, label and title attributes are + # all not standard but seen several times in the wild + labels = [ + s_attr.get(lbl) + for lbl in ('label', 'title') + if str_or_none(s_attr.get(lbl)) + ] + width = int_or_none(s_attr.get('width')) + height = (int_or_none(s_attr.get('height')) + or int_or_none(s_attr.get('res'))) + if not width or not height: + for lbl in labels: + resolution = parse_resolution(lbl) + if not resolution: + continue + width = width or resolution.get('width') + height = height or resolution.get('height') + for lbl in labels: + tbr = parse_bitrate(lbl) + if tbr: + break + else: + tbr = None + f.update({ + 'width': width, + 'height': height, + 'tbr': tbr, + 'format_id': s_attr.get('label') or s_attr.get('title'), + }) f.update(formats[0]) media_info['formats'].append(f) else: @@ -2103,21 +2894,31 @@ class InfoExtractor(object): track_attributes = extract_attributes(track_tag) kind = track_attributes.get('kind') if not kind or kind in ('subtitles', 'captions'): - src = track_attributes.get('src') + src = strip_or_none(track_attributes.get('src')) if not src: continue lang = track_attributes.get('srclang') or track_attributes.get('lang') or track_attributes.get('label') media_info['subtitles'].setdefault(lang, []).append({ 'url': absolute_url(src), }) + for f in media_info['formats']: + f.setdefault('http_headers', {})['Referer'] = base_url if media_info['formats'] or media_info['subtitles']: entries.append(media_info) return entries def _extract_akamai_formats(self, manifest_url, video_id, hosts={}): + signed = 'hdnea=' in manifest_url + if not signed: + # https://learn.akamai.com/en-us/webhelp/media-services-on-demand/stream-packaging-user-guide/GUID-BE6C0F73-1E06-483B-B0EA-57984B91B7F9.html + manifest_url = re.sub( + r'(?:b=[\d,-]+|(?:__a__|attributes)=off|__b__=\d+)&?', + '', manifest_url).strip('?') + 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) @@ -2128,36 +2929,71 @@ 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') 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_formats = self._extract_m3u8_formats( m3u8_url, video_id, 'mp4', 'm3u8_native', - m3u8_id='hls', fatal=False)) + m3u8_id='hls', fatal=False) + formats.extend(m3u8_formats) + + http_host = hosts.get('http') + if http_host and m3u8_formats and not signed: + REPL_REGEX = r'https?://[^/]+/i/([^,]+),([^/]+),([^/]+)\.csmil/.+' + qualities = re.match(REPL_REGEX, m3u8_url).group(2).split(',') + qualities_length = len(qualities) + if len(m3u8_formats) in (qualities_length, qualities_length + 1): + i = 0 + for f in m3u8_formats: + if f['vcodec'] != 'none': + for protocol in ('http', 'https'): + http_f = f.copy() + del http_f['manifest_url'] + http_url = re.sub( + REPL_REGEX, protocol + r'://%s/\g<1>%s\3' % (http_host, qualities[i]), f['url']) + http_f.update({ + 'format_id': http_f['format_id'].replace('hls-', protocol + '-'), + 'url': http_url, + 'protocol': protocol, + }) + formats.append(http_f) + i += 1 + return formats def _extract_wowza_formats(self, url, video_id, m3u8_entry_protocol='m3u8_native', skip_protocols=[]): + query = compat_urlparse.urlparse(url).query 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 + mobj = re.search( + r'(?:(?:http|rtmp|rtsp)(?P<s>s)?:)?(?P<url>//[^?]+)', url) + url_base = mobj.group('url') + http_base_url = '%s%s:%s' % ('http', mobj.group('s') or '', url_base) formats = [] + + def manifest_url(manifest): + m_url = '%s/%s' % (http_base_url, manifest) + if query: + m_url += '?%s' % query + return m_url + if 'm3u8' not in skip_protocols: formats.extend(self._extract_m3u8_formats( - http_base_url + '/playlist.m3u8', video_id, 'mp4', + manifest_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', + manifest_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', + manifest_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', + manifest_url('jwplayer.smil'), video_id, fatal=False) for rtmp_format in rtmp_formats: rtsp_format = rtmp_format.copy() @@ -2174,38 +3010,39 @@ class InfoExtractor(object): for protocol in ('rtmp', 'rtsp'): if protocol not in skip_protocols: formats.append({ - 'url': protocol + url_base, + 'url': '%s:%s' % (protocol, url_base), 'format_id': protocol, 'protocol': protocol, }) return formats def _find_jwplayer_data(self, webpage, video_id=None, transform_source=js_to_json): - mobj = re.search( - r'jwplayer\((?P<quote>[\'"])[^\'" ]+(?P=quote)\)\.setup\s*\((?P<options>[^)]+)\)', - webpage) - if mobj: - try: - jwplayer_data = self._parse_json(mobj.group('options'), - video_id=video_id, - transform_source=transform_source) - except ExtractorError: - pass - else: - if isinstance(jwplayer_data, dict): - return jwplayer_data + return self._search_json( + r'''(?<!-)\bjwplayer\s*\(\s*(?P<q>'|")(?!(?P=q)).+(?P=q)\s*\)(?:(?!</script>).)*?\.\s*(?:setup\s*\(|(?P<load>load)\s*\(\s*\[)''', + webpage, 'JWPlayer data', video_id, + # must be a {...} or sequence, ending + contains_pattern=r'\{[\s\S]*}(?(load)(?:\s*,\s*\{[\s\S]*})*)', end_pattern=r'(?(load)\]|\))', + transform_source=transform_source, default=None) def _extract_jwplayer_data(self, webpage, video_id, *args, **kwargs): - jwplayer_data = self._find_jwplayer_data( - webpage, video_id, transform_source=js_to_json) - return self._parse_jwplayer_data( - jwplayer_data, video_id, *args, **kwargs) + # allow passing `transform_source` through to _find_jwplayer_data() + transform_source = kwargs.pop('transform_source', None) + kwfind = compat_kwargs({'transform_source': transform_source}) if transform_source else {} + + jwplayer_data = self._find_jwplayer_data(webpage, video_id, **kwfind) + + 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): + flat_pl = try_get(jwplayer_data, lambda x: x.get('playlist') or True) + if flat_pl is None: + # not even a dict + return [] + # 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: + if flat_pl is True: jwplayer_data = {'playlist': [jwplayer_data]} entries = [] @@ -2226,31 +3063,45 @@ class InfoExtractor(object): formats = self._parse_jwplayer_formats( video_data['sources'], video_id=this_video_id, m3u8_id=m3u8_id, mpd_id=mpd_id, rtmp_params=rtmp_params, base_url=base_url) - 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) - }) + for track in traverse_obj(video_data, ( + 'tracks', lambda _, t: t.get('kind').lower() in ('captions', 'subtitles'))): + 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({ + entry = { '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')), + 'title': unescapeHTML(video_data['title'] if require_title else video_data.get('title')), + 'description': clean_html(video_data.get('description')), + 'thumbnail': urljoin(base_url, 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, - }) + 'alt_title': clean_html(video_data.get('subtitle')), # attributes used e.g. by Tele5 ... + 'genre': clean_html(video_data.get('genre')), + 'channel': clean_html(dict_get(video_data, ('category', 'channel'))), + 'season_number': int_or_none(video_data.get('season')), + 'episode_number': int_or_none(video_data.get('episode')), + 'release_year': int_or_none(video_data.get('releasedate')), + 'age_limit': int_or_none(video_data.get('age_restriction')), + } + # https://github.com/jwplayer/jwplayer/blob/master/src/js/utils/validator.js#L32 + if len(formats) == 1 and re.search(r'^(?:http|//).*(?:youtube\.com|youtu\.be)/.+', formats[0]['url']): + entry.update({ + '_type': 'url_transparent', + 'url': formats[0]['url'], + }) + else: + # avoid exception in case of only sttls + if formats: + self._sort_formats(formats) + entry['formats'] = formats + entries.append(entry) if len(entries) == 1: return entries[0] else: @@ -2258,18 +3109,23 @@ class InfoExtractor(object): def _parse_jwplayer_formats(self, jwplayer_sources_data, video_id=None, m3u8_id=None, mpd_id=None, rtmp_params=None, base_url=None): + urls = set() formats = [] for source in jwplayer_sources_data: - source_url = self._proto_relative_url(source['file']) - if base_url: - source_url = compat_urlparse.urljoin(base_url, source_url) + if not isinstance(source, dict): + continue + source_url = urljoin( + base_url, self._proto_relative_url(source.get('file'))) + if not source_url or source_url in urls: + continue + urls.add(source_url) source_type = source.get('type') or '' ext = mimetype2ext(source_type) or determine_ext(source_url) - if source_type == 'hls' or ext == 'm3u8': + if source_type == 'hls' or ext == 'm3u8' or 'format=m3u8-aapl' in source_url: formats.extend(self._extract_m3u8_formats( source_url, video_id, 'mp4', entry_protocol='m3u8_native', m3u8_id=m3u8_id, fatal=False)) - elif ext == 'mpd': + elif source_type == 'dash' or ext == 'mpd' or 'format=mpd-time-csf' in source_url: formats.extend(self._extract_mpd_formats( source_url, video_id, mpd_id=mpd_id, fatal=False)) elif ext == 'smil': @@ -2284,26 +3140,29 @@ class InfoExtractor(object): 'ext': ext, }) else: + format_id = str_or_none(source.get('label')) height = int_or_none(source.get('height')) - if height is None: + if height is None and format_id: # Often no height is provided but there is a label in # format like "1080p", "720p SD", or 1080. - height = int_or_none(self._search_regex( - r'^(\d{3,4})[pP]?(?:\b|$)', compat_str(source.get('label') or ''), - 'height', default=None)) + height = parse_resolution(format_id).get('height') a_format = { 'url': source_url, 'width': int_or_none(source.get('width')), 'height': height, - 'tbr': int_or_none(source.get('bitrate')), + 'tbr': int_or_none(source.get('bitrate'), scale=1000), + 'filesize': int_or_none(source.get('filesize')), 'ext': ext, } + if format_id: + a_format['format_id'] = format_id + 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) + r'((?:mp4|mp3|flv):)', source_url, maxsplit=1) if len(rtmp_url_parts) == 3: rtmp_url, prefix, play_path = rtmp_url_parts a_format.update({ @@ -2330,7 +3189,7 @@ class InfoExtractor(object): if fatal: raise ExtractorError(msg) else: - self._downloader.report_warning(msg) + self.report_warning(msg) return res def _float(self, v, name, fatal=False, **kwargs): @@ -2340,20 +3199,49 @@ class InfoExtractor(object): if fatal: raise ExtractorError(msg) else: - self._downloader.report_warning(msg) + self.report_warning(msg) return res - def _set_cookie(self, domain, name, value, expire_time=None): - cookie = compat_cookiejar.Cookie( - 0, name, value, None, None, domain, None, - None, '/', True, False, expire_time, '', None, None, None) - self._downloader.cookiejar.set_cookie(cookie) + def _set_cookie(self, domain, name, value, expire_time=None, port=None, + path='/', secure=False, discard=False, rest={}, **kwargs): + cookie = compat_cookiejar_Cookie( + 0, name, value, port, port is not None, domain, True, + domain.startswith('.'), path, True, secure, expire_time, + discard, None, None, rest) + self.cookiejar.set_cookie(cookie) def _get_cookies(self, url): - """ Return a compat_cookies.SimpleCookie with the cookies for the url """ + """ Return a compat_cookies_SimpleCookie with the cookies for the url """ req = sanitized_Request(url) - self._downloader.cookiejar.add_cookie_header(req) - return compat_cookies.SimpleCookie(req.get_header('Cookie')) + self.cookiejar.add_cookie_header(req) + return compat_cookies_SimpleCookie(req.get_header('Cookie')) + + def _apply_first_set_cookie_header(self, url_handle, cookie): + """ + Apply first Set-Cookie header instead of the last. Experimental. + + Some sites (e.g. [1-3]) may serve two cookies under the same name + in Set-Cookie header and expect the first (old) one to be set rather + than second (new). However, as of RFC6265 the newer one cookie + should be set into cookie store what actually happens. + We will workaround this issue by resetting the cookie to + the first one manually. + 1. https://new.vk.com/ + 2. https://github.com/ytdl-org/youtube-dl/issues/9841#issuecomment-227871201 + 3. https://learning.oreilly.com/ + """ + for header, cookies in url_handle.headers.items(): + if header.lower() != 'set-cookie': + continue + if sys.version_info[0] >= 3: + cookies = cookies.encode('iso-8859-1') + cookies = cookies.decode('utf-8') + cookie_value = re.search( + r'%s=(.+?);.*?\b[Dd]omain=(.+?)(?:[,;]|$)' % cookie, cookies) + if cookie_value: + value, domain = cookie_value.groups() + self._set_cookie(domain, cookie, value) + break def get_testcases(self, include_onlymatching=False): t = getattr(self, '_TEST', None) @@ -2385,8 +3273,8 @@ class InfoExtractor(object): return not any_restricted def extract_subtitles(self, *args, **kwargs): - if (self._downloader.params.get('writesubtitles', False) or - self._downloader.params.get('listsubtitles')): + if (self.get_param('writesubtitles', False) + or self.get_param('listsubtitles')): return self._get_subtitles(*args, **kwargs) return {} @@ -2403,16 +3291,24 @@ class InfoExtractor(object): return ret @classmethod - def _merge_subtitles(cls, subtitle_dict1, subtitle_dict2): - """ Merge two subtitle dictionaries, language by language. """ - ret = dict(subtitle_dict1) - for lang in subtitle_dict2: - ret[lang] = cls._merge_subtitle_items(subtitle_dict1.get(lang, []), subtitle_dict2[lang]) - return ret + def _merge_subtitles(cls, subtitle_dict1, *subtitle_dicts, **kwargs): + """ Merge subtitle dictionaries, language by language. """ + + # ..., * , target=None + target = kwargs.get('target') + if target is None: + target = dict(subtitle_dict1) + else: + subtitle_dicts = (subtitle_dict1,) + subtitle_dicts + + for subtitle_dict in subtitle_dicts: + for lang in subtitle_dict: + target[lang] = cls._merge_subtitle_items(target.get(lang, []), subtitle_dict[lang]) + return target def extract_automatic_captions(self, *args, **kwargs): - if (self._downloader.params.get('writeautomaticsub', False) or - self._downloader.params.get('listsubtitles')): + if (self.get_param('writeautomaticsub', False) + or self.get_param('listsubtitles')): return self._get_automatic_captions(*args, **kwargs) return {} @@ -2420,9 +3316,9 @@ class InfoExtractor(object): raise NotImplementedError('This method must be implemented by subclasses') def mark_watched(self, *args, **kwargs): - if (self._downloader.params.get('mark_watched', False) and - (self._get_login_info()[0] is not None or - self._downloader.params.get('cookiefile') is not None)): + if (self.get_param('mark_watched', False) + and (self._get_login_info()[0] is not None + or self.get_param('cookiefile') is not None)): self._mark_watched(*args, **kwargs) def _mark_watched(self, *args, **kwargs): @@ -2430,7 +3326,7 @@ class InfoExtractor(object): def geo_verification_headers(self): headers = {} - geo_verification_proxy = self._downloader.params.get('geo_verification_proxy') + geo_verification_proxy = self.get_param('geo_verification_proxy') if geo_verification_proxy: headers['Ytdl-request-proxy'] = geo_verification_proxy return headers @@ -2441,6 +3337,29 @@ class InfoExtractor(object): def _generic_title(self, url): return compat_urllib_parse_unquote(os.path.splitext(url_basename(url))[0]) + def _yes_playlist(self, playlist_id, video_id, *args, **kwargs): + # smuggled_data=None, *, playlist_label='playlist', video_label='video' + smuggled_data = args[0] if len(args) == 1 else kwargs.get('smuggled_data') + playlist_label = kwargs.get('playlist_label', 'playlist') + video_label = kwargs.get('video_label', 'video') + + if not playlist_id or not video_id: + return not video_id + + no_playlist = (smuggled_data or {}).get('force_noplaylist') + if no_playlist is not None: + return not no_playlist + + video_id = '' if video_id is True else ' ' + video_id + noplaylist = self.get_param('noplaylist') + self.to_screen( + 'Downloading just the {0}{1} because of --no-playlist'.format(video_label, video_id) + if noplaylist else + 'Downloading {0}{1} - add --no-playlist to download just the {2}{3}'.format( + playlist_label, '' if playlist_id is True else ' ' + playlist_id, + video_label, video_id)) + return not noplaylist + class SearchInfoExtractor(InfoExtractor): """ |