]> Raphaƫl G. Git Repositories - youtubedl/blob - youtube_dl/extractor/common.py
Prepare new version
[youtubedl] / youtube_dl / extractor / common.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 import base64
5 import datetime
6 import hashlib
7 import json
8 import netrc
9 import os
10 import random
11 import re
12 import socket
13 import sys
14 import time
15 import math
16
17 from ..compat import (
18 compat_cookiejar,
19 compat_cookies,
20 compat_etree_Element,
21 compat_etree_fromstring,
22 compat_getpass,
23 compat_integer_types,
24 compat_http_client,
25 compat_os_name,
26 compat_str,
27 compat_urllib_error,
28 compat_urllib_parse_unquote,
29 compat_urllib_parse_urlencode,
30 compat_urllib_request,
31 compat_urlparse,
32 compat_xml_parse_error,
33 )
34 from ..downloader.f4m import (
35 get_base_url,
36 remove_encrypted_media,
37 )
38 from ..utils import (
39 NO_DEFAULT,
40 age_restricted,
41 base_url,
42 bug_reports_message,
43 clean_html,
44 compiled_regex_type,
45 determine_ext,
46 determine_protocol,
47 dict_get,
48 error_to_compat_str,
49 ExtractorError,
50 extract_attributes,
51 fix_xml_ampersands,
52 float_or_none,
53 GeoRestrictedError,
54 GeoUtils,
55 int_or_none,
56 js_to_json,
57 JSON_LD_RE,
58 mimetype2ext,
59 orderedSet,
60 parse_bitrate,
61 parse_codecs,
62 parse_duration,
63 parse_iso8601,
64 parse_m3u8_attributes,
65 parse_resolution,
66 RegexNotFoundError,
67 sanitized_Request,
68 sanitize_filename,
69 str_or_none,
70 strip_or_none,
71 unescapeHTML,
72 unified_strdate,
73 unified_timestamp,
74 update_Request,
75 update_url_query,
76 urljoin,
77 url_basename,
78 url_or_none,
79 xpath_element,
80 xpath_text,
81 xpath_with_ns,
82 )
83
84
85 class InfoExtractor(object):
86 """Information Extractor class.
87
88 Information extractors are the classes that, given a URL, extract
89 information about the video (or videos) the URL refers to. This
90 information includes the real video URL, the video title, author and
91 others. The information is stored in a dictionary which is then
92 passed to the YoutubeDL. The YoutubeDL processes this
93 information possibly downloading the video to the file system, among
94 other possible outcomes.
95
96 The type field determines the type of the result.
97 By far the most common value (and the default if _type is missing) is
98 "video", which indicates a single video.
99
100 For a video, the dictionaries must include the following fields:
101
102 id: Video identifier.
103 title: Video title, unescaped.
104
105 Additionally, it must contain either a formats entry or a url one:
106
107 formats: A list of dictionaries for each format available, ordered
108 from worst to best quality.
109
110 Potential fields:
111 * url The mandatory URL representing the media:
112 for plain file media - HTTP URL of this file,
113 for RTMP - RTMP URL,
114 for HLS - URL of the M3U8 media playlist,
115 for HDS - URL of the F4M manifest,
116 for DASH
117 - HTTP URL to plain file media (in case of
118 unfragmented media)
119 - URL of the MPD manifest or base URL
120 representing the media if MPD manifest
121 is parsed from a string (in case of
122 fragmented media)
123 for MSS - URL of the ISM manifest.
124 * manifest_url
125 The URL of the manifest file in case of
126 fragmented media:
127 for HLS - URL of the M3U8 master playlist,
128 for HDS - URL of the F4M manifest,
129 for DASH - URL of the MPD manifest,
130 for MSS - URL of the ISM manifest.
131 * ext Will be calculated from URL if missing
132 * format A human-readable description of the format
133 ("mp4 container with h264/opus").
134 Calculated from the format_id, width, height.
135 and format_note fields if missing.
136 * format_id A short description of the format
137 ("mp4_h264_opus" or "19").
138 Technically optional, but strongly recommended.
139 * format_note Additional info about the format
140 ("3D" or "DASH video")
141 * width Width of the video, if known
142 * height Height of the video, if known
143 * resolution Textual description of width and height
144 * tbr Average bitrate of audio and video in KBit/s
145 * abr Average audio bitrate in KBit/s
146 * acodec Name of the audio codec in use
147 * asr Audio sampling rate in Hertz
148 * vbr Average video bitrate in KBit/s
149 * fps Frame rate
150 * vcodec Name of the video codec in use
151 * container Name of the container format
152 * filesize The number of bytes, if known in advance
153 * filesize_approx An estimate for the number of bytes
154 * player_url SWF Player URL (used for rtmpdump).
155 * protocol The protocol that will be used for the actual
156 download, lower-case.
157 "http", "https", "rtsp", "rtmp", "rtmpe",
158 "m3u8", "m3u8_native" or "http_dash_segments".
159 * fragment_base_url
160 Base URL for fragments. Each fragment's path
161 value (if present) will be relative to
162 this URL.
163 * fragments A list of fragments of a fragmented media.
164 Each fragment entry must contain either an url
165 or a path. If an url is present it should be
166 considered by a client. Otherwise both path and
167 fragment_base_url must be present. Here is
168 the list of all potential fields:
169 * "url" - fragment's URL
170 * "path" - fragment's path relative to
171 fragment_base_url
172 * "duration" (optional, int or float)
173 * "filesize" (optional, int)
174 * preference Order number of this format. If this field is
175 present and not None, the formats get sorted
176 by this field, regardless of all other values.
177 -1 for default (order by other properties),
178 -2 or smaller for less than default.
179 < -1000 to hide the format (if there is
180 another one which is strictly better)
181 * language Language code, e.g. "de" or "en-US".
182 * language_preference Is this in the language mentioned in
183 the URL?
184 10 if it's what the URL is about,
185 -1 for default (don't know),
186 -10 otherwise, other values reserved for now.
187 * quality Order number of the video quality of this
188 format, irrespective of the file format.
189 -1 for default (order by other properties),
190 -2 or smaller for less than default.
191 * source_preference Order number for this video source
192 (quality takes higher priority)
193 -1 for default (order by other properties),
194 -2 or smaller for less than default.
195 * http_headers A dictionary of additional HTTP headers
196 to add to the request.
197 * stretched_ratio If given and not 1, indicates that the
198 video's pixels are not square.
199 width : height ratio as float.
200 * no_resume The server does not support resuming the
201 (HTTP or RTMP) download. Boolean.
202 * downloader_options A dictionary of downloader options as
203 described in FileDownloader
204
205 url: Final video URL.
206 ext: Video filename extension.
207 format: The video format, defaults to ext (used for --get-format)
208 player_url: SWF Player URL (used for rtmpdump).
209
210 The following fields are optional:
211
212 alt_title: A secondary title of the video.
213 display_id An alternative identifier for the video, not necessarily
214 unique, but available before title. Typically, id is
215 something like "4234987", title "Dancing naked mole rats",
216 and display_id "dancing-naked-mole-rats"
217 thumbnails: A list of dictionaries, with the following entries:
218 * "id" (optional, string) - Thumbnail format ID
219 * "url"
220 * "preference" (optional, int) - quality of the image
221 * "width" (optional, int)
222 * "height" (optional, int)
223 * "resolution" (optional, string "{width}x{height}",
224 deprecated)
225 * "filesize" (optional, int)
226 thumbnail: Full URL to a video thumbnail image.
227 description: Full video description.
228 uploader: Full name of the video uploader.
229 license: License name the video is licensed under.
230 creator: The creator of the video.
231 release_date: The date (YYYYMMDD) when the video was released.
232 timestamp: UNIX timestamp of the moment the video became available.
233 upload_date: Video upload date (YYYYMMDD).
234 If not explicitly set, calculated from timestamp.
235 uploader_id: Nickname or id of the video uploader.
236 uploader_url: Full URL to a personal webpage of the video uploader.
237 channel: Full name of the channel the video is uploaded on.
238 Note that channel fields may or may not repeat uploader
239 fields. This depends on a particular extractor.
240 channel_id: Id of the channel.
241 channel_url: Full URL to a channel webpage.
242 location: Physical location where the video was filmed.
243 subtitles: The available subtitles as a dictionary in the format
244 {tag: subformats}. "tag" is usually a language code, and
245 "subformats" is a list sorted from lower to higher
246 preference, each element is a dictionary with the "ext"
247 entry and one of:
248 * "data": The subtitles file contents
249 * "url": A URL pointing to the subtitles file
250 "ext" will be calculated from URL if missing
251 automatic_captions: Like 'subtitles', used by the YoutubeIE for
252 automatically generated captions
253 duration: Length of the video in seconds, as an integer or float.
254 view_count: How many users have watched the video on the platform.
255 like_count: Number of positive ratings of the video
256 dislike_count: Number of negative ratings of the video
257 repost_count: Number of reposts of the video
258 average_rating: Average rating give by users, the scale used depends on the webpage
259 comment_count: Number of comments on the video
260 comments: A list of comments, each with one or more of the following
261 properties (all but one of text or html optional):
262 * "author" - human-readable name of the comment author
263 * "author_id" - user ID of the comment author
264 * "id" - Comment ID
265 * "html" - Comment as HTML
266 * "text" - Plain text of the comment
267 * "timestamp" - UNIX timestamp of comment
268 * "parent" - ID of the comment this one is replying to.
269 Set to "root" to indicate that this is a
270 comment to the original video.
271 age_limit: Age restriction for the video, as an integer (years)
272 webpage_url: The URL to the video webpage, if given to youtube-dl it
273 should allow to get the same result again. (It will be set
274 by YoutubeDL if it's missing)
275 categories: A list of categories that the video falls in, for example
276 ["Sports", "Berlin"]
277 tags: A list of tags assigned to the video, e.g. ["sweden", "pop music"]
278 is_live: True, False, or None (=unknown). Whether this video is a
279 live stream that goes on instead of a fixed-length video.
280 start_time: Time in seconds where the reproduction should start, as
281 specified in the URL.
282 end_time: Time in seconds where the reproduction should end, as
283 specified in the URL.
284 chapters: A list of dictionaries, with the following entries:
285 * "start_time" - The start time of the chapter in seconds
286 * "end_time" - The end time of the chapter in seconds
287 * "title" (optional, string)
288
289 The following fields should only be used when the video belongs to some logical
290 chapter or section:
291
292 chapter: Name or title of the chapter the video belongs to.
293 chapter_number: Number of the chapter the video belongs to, as an integer.
294 chapter_id: Id of the chapter the video belongs to, as a unicode string.
295
296 The following fields should only be used when the video is an episode of some
297 series, programme or podcast:
298
299 series: Title of the series or programme the video episode belongs to.
300 season: Title of the season the video episode belongs to.
301 season_number: Number of the season the video episode belongs to, as an integer.
302 season_id: Id of the season the video episode belongs to, as a unicode string.
303 episode: Title of the video episode. Unlike mandatory video title field,
304 this field should denote the exact title of the video episode
305 without any kind of decoration.
306 episode_number: Number of the video episode within a season, as an integer.
307 episode_id: Id of the video episode, as a unicode string.
308
309 The following fields should only be used when the media is a track or a part of
310 a music album:
311
312 track: Title of the track.
313 track_number: Number of the track within an album or a disc, as an integer.
314 track_id: Id of the track (useful in case of custom indexing, e.g. 6.iii),
315 as a unicode string.
316 artist: Artist(s) of the track.
317 genre: Genre(s) of the track.
318 album: Title of the album the track belongs to.
319 album_type: Type of the album (e.g. "Demo", "Full-length", "Split", "Compilation", etc).
320 album_artist: List of all artists appeared on the album (e.g.
321 "Ash Borer / Fell Voices" or "Various Artists", useful for splits
322 and compilations).
323 disc_number: Number of the disc or other physical medium the track belongs to,
324 as an integer.
325 release_year: Year (YYYY) when the album was released.
326
327 Unless mentioned otherwise, the fields should be Unicode strings.
328
329 Unless mentioned otherwise, None is equivalent to absence of information.
330
331
332 _type "playlist" indicates multiple videos.
333 There must be a key "entries", which is a list, an iterable, or a PagedList
334 object, each element of which is a valid dictionary by this specification.
335
336 Additionally, playlists can have "id", "title", "description", "uploader",
337 "uploader_id", "uploader_url" attributes with the same semantics as videos
338 (see above).
339
340
341 _type "multi_video" indicates that there are multiple videos that
342 form a single show, for examples multiple acts of an opera or TV episode.
343 It must have an entries key like a playlist and contain all the keys
344 required for a video at the same time.
345
346
347 _type "url" indicates that the video must be extracted from another
348 location, possibly by a different extractor. Its only required key is:
349 "url" - the next URL to extract.
350 The key "ie_key" can be set to the class name (minus the trailing "IE",
351 e.g. "Youtube") if the extractor class is known in advance.
352 Additionally, the dictionary may have any properties of the resolved entity
353 known in advance, for example "title" if the title of the referred video is
354 known ahead of time.
355
356
357 _type "url_transparent" entities have the same specification as "url", but
358 indicate that the given additional information is more precise than the one
359 associated with the resolved URL.
360 This is useful when a site employs a video service that hosts the video and
361 its technical metadata, but that video service does not embed a useful
362 title, description etc.
363
364
365 Subclasses of this one should re-define the _real_initialize() and
366 _real_extract() methods and define a _VALID_URL regexp.
367 Probably, they should also be added to the list of extractors.
368
369 _GEO_BYPASS attribute may be set to False in order to disable
370 geo restriction bypass mechanisms for a particular extractor.
371 Though it won't disable explicit geo restriction bypass based on
372 country code provided with geo_bypass_country.
373
374 _GEO_COUNTRIES attribute may contain a list of presumably geo unrestricted
375 countries for this extractor. One of these countries will be used by
376 geo restriction bypass mechanism right away in order to bypass
377 geo restriction, of course, if the mechanism is not disabled.
378
379 _GEO_IP_BLOCKS attribute may contain a list of presumably geo unrestricted
380 IP blocks in CIDR notation for this extractor. One of these IP blocks
381 will be used by geo restriction bypass mechanism similarly
382 to _GEO_COUNTRIES.
383
384 Finally, the _WORKING attribute should be set to False for broken IEs
385 in order to warn the users and skip the tests.
386 """
387
388 _ready = False
389 _downloader = None
390 _x_forwarded_for_ip = None
391 _GEO_BYPASS = True
392 _GEO_COUNTRIES = None
393 _GEO_IP_BLOCKS = None
394 _WORKING = True
395
396 def __init__(self, downloader=None):
397 """Constructor. Receives an optional downloader."""
398 self._ready = False
399 self._x_forwarded_for_ip = None
400 self.set_downloader(downloader)
401
402 @classmethod
403 def suitable(cls, url):
404 """Receives a URL and returns True if suitable for this IE."""
405
406 # This does not use has/getattr intentionally - we want to know whether
407 # we have cached the regexp for *this* class, whereas getattr would also
408 # match the superclass
409 if '_VALID_URL_RE' not in cls.__dict__:
410 cls._VALID_URL_RE = re.compile(cls._VALID_URL)
411 return cls._VALID_URL_RE.match(url) is not None
412
413 @classmethod
414 def _match_id(cls, url):
415 if '_VALID_URL_RE' not in cls.__dict__:
416 cls._VALID_URL_RE = re.compile(cls._VALID_URL)
417 m = cls._VALID_URL_RE.match(url)
418 assert m
419 return compat_str(m.group('id'))
420
421 @classmethod
422 def working(cls):
423 """Getter method for _WORKING."""
424 return cls._WORKING
425
426 def initialize(self):
427 """Initializes an instance (authentication, etc)."""
428 self._initialize_geo_bypass({
429 'countries': self._GEO_COUNTRIES,
430 'ip_blocks': self._GEO_IP_BLOCKS,
431 })
432 if not self._ready:
433 self._real_initialize()
434 self._ready = True
435
436 def _initialize_geo_bypass(self, geo_bypass_context):
437 """
438 Initialize geo restriction bypass mechanism.
439
440 This method is used to initialize geo bypass mechanism based on faking
441 X-Forwarded-For HTTP header. A random country from provided country list
442 is selected and a random IP belonging to this country is generated. This
443 IP will be passed as X-Forwarded-For HTTP header in all subsequent
444 HTTP requests.
445
446 This method will be used for initial geo bypass mechanism initialization
447 during the instance initialization with _GEO_COUNTRIES and
448 _GEO_IP_BLOCKS.
449
450 You may also manually call it from extractor's code if geo bypass
451 information is not available beforehand (e.g. obtained during
452 extraction) or due to some other reason. In this case you should pass
453 this information in geo bypass context passed as first argument. It may
454 contain following fields:
455
456 countries: List of geo unrestricted countries (similar
457 to _GEO_COUNTRIES)
458 ip_blocks: List of geo unrestricted IP blocks in CIDR notation
459 (similar to _GEO_IP_BLOCKS)
460
461 """
462 if not self._x_forwarded_for_ip:
463
464 # Geo bypass mechanism is explicitly disabled by user
465 if not self._downloader.params.get('geo_bypass', True):
466 return
467
468 if not geo_bypass_context:
469 geo_bypass_context = {}
470
471 # Backward compatibility: previously _initialize_geo_bypass
472 # expected a list of countries, some 3rd party code may still use
473 # it this way
474 if isinstance(geo_bypass_context, (list, tuple)):
475 geo_bypass_context = {
476 'countries': geo_bypass_context,
477 }
478
479 # The whole point of geo bypass mechanism is to fake IP
480 # as X-Forwarded-For HTTP header based on some IP block or
481 # country code.
482
483 # Path 1: bypassing based on IP block in CIDR notation
484
485 # Explicit IP block specified by user, use it right away
486 # regardless of whether extractor is geo bypassable or not
487 ip_block = self._downloader.params.get('geo_bypass_ip_block', None)
488
489 # Otherwise use random IP block from geo bypass context but only
490 # if extractor is known as geo bypassable
491 if not ip_block:
492 ip_blocks = geo_bypass_context.get('ip_blocks')
493 if self._GEO_BYPASS and ip_blocks:
494 ip_block = random.choice(ip_blocks)
495
496 if ip_block:
497 self._x_forwarded_for_ip = GeoUtils.random_ipv4(ip_block)
498 if self._downloader.params.get('verbose', False):
499 self._downloader.to_screen(
500 '[debug] Using fake IP %s as X-Forwarded-For.'
501 % self._x_forwarded_for_ip)
502 return
503
504 # Path 2: bypassing based on country code
505
506 # Explicit country code specified by user, use it right away
507 # regardless of whether extractor is geo bypassable or not
508 country = self._downloader.params.get('geo_bypass_country', None)
509
510 # Otherwise use random country code from geo bypass context but
511 # only if extractor is known as geo bypassable
512 if not country:
513 countries = geo_bypass_context.get('countries')
514 if self._GEO_BYPASS and countries:
515 country = random.choice(countries)
516
517 if country:
518 self._x_forwarded_for_ip = GeoUtils.random_ipv4(country)
519 if self._downloader.params.get('verbose', False):
520 self._downloader.to_screen(
521 '[debug] Using fake IP %s (%s) as X-Forwarded-For.'
522 % (self._x_forwarded_for_ip, country.upper()))
523
524 def extract(self, url):
525 """Extracts URL information and returns it in list of dicts."""
526 try:
527 for _ in range(2):
528 try:
529 self.initialize()
530 ie_result = self._real_extract(url)
531 if self._x_forwarded_for_ip:
532 ie_result['__x_forwarded_for_ip'] = self._x_forwarded_for_ip
533 return ie_result
534 except GeoRestrictedError as e:
535 if self.__maybe_fake_ip_and_retry(e.countries):
536 continue
537 raise
538 except ExtractorError:
539 raise
540 except compat_http_client.IncompleteRead as e:
541 raise ExtractorError('A network error has occurred.', cause=e, expected=True)
542 except (KeyError, StopIteration) as e:
543 raise ExtractorError('An extractor error has occurred.', cause=e)
544
545 def __maybe_fake_ip_and_retry(self, countries):
546 if (not self._downloader.params.get('geo_bypass_country', None)
547 and self._GEO_BYPASS
548 and self._downloader.params.get('geo_bypass', True)
549 and not self._x_forwarded_for_ip
550 and countries):
551 country_code = random.choice(countries)
552 self._x_forwarded_for_ip = GeoUtils.random_ipv4(country_code)
553 if self._x_forwarded_for_ip:
554 self.report_warning(
555 'Video is geo restricted. Retrying extraction with fake IP %s (%s) as X-Forwarded-For.'
556 % (self._x_forwarded_for_ip, country_code.upper()))
557 return True
558 return False
559
560 def set_downloader(self, downloader):
561 """Sets the downloader for this IE."""
562 self._downloader = downloader
563
564 def _real_initialize(self):
565 """Real initialization process. Redefine in subclasses."""
566 pass
567
568 def _real_extract(self, url):
569 """Real extraction process. Redefine in subclasses."""
570 pass
571
572 @classmethod
573 def ie_key(cls):
574 """A string for getting the InfoExtractor with get_info_extractor"""
575 return compat_str(cls.__name__[:-2])
576
577 @property
578 def IE_NAME(self):
579 return compat_str(type(self).__name__[:-2])
580
581 @staticmethod
582 def __can_accept_status_code(err, expected_status):
583 assert isinstance(err, compat_urllib_error.HTTPError)
584 if expected_status is None:
585 return False
586 if isinstance(expected_status, compat_integer_types):
587 return err.code == expected_status
588 elif isinstance(expected_status, (list, tuple)):
589 return err.code in expected_status
590 elif callable(expected_status):
591 return expected_status(err.code) is True
592 else:
593 assert False
594
595 def _request_webpage(self, url_or_request, video_id, note=None, errnote=None, fatal=True, data=None, headers={}, query={}, expected_status=None):
596 """
597 Return the response handle.
598
599 See _download_webpage docstring for arguments specification.
600 """
601 if note is None:
602 self.report_download_webpage(video_id)
603 elif note is not False:
604 if video_id is None:
605 self.to_screen('%s' % (note,))
606 else:
607 self.to_screen('%s: %s' % (video_id, note))
608
609 # Some sites check X-Forwarded-For HTTP header in order to figure out
610 # the origin of the client behind proxy. This allows bypassing geo
611 # restriction by faking this header's value to IP that belongs to some
612 # geo unrestricted country. We will do so once we encounter any
613 # geo restriction error.
614 if self._x_forwarded_for_ip:
615 if 'X-Forwarded-For' not in headers:
616 headers['X-Forwarded-For'] = self._x_forwarded_for_ip
617
618 if isinstance(url_or_request, compat_urllib_request.Request):
619 url_or_request = update_Request(
620 url_or_request, data=data, headers=headers, query=query)
621 else:
622 if query:
623 url_or_request = update_url_query(url_or_request, query)
624 if data is not None or headers:
625 url_or_request = sanitized_Request(url_or_request, data, headers)
626 try:
627 return self._downloader.urlopen(url_or_request)
628 except (compat_urllib_error.URLError, compat_http_client.HTTPException, socket.error) as err:
629 if isinstance(err, compat_urllib_error.HTTPError):
630 if self.__can_accept_status_code(err, expected_status):
631 # Retain reference to error to prevent file object from
632 # being closed before it can be read. Works around the
633 # effects of <https://bugs.python.org/issue15002>
634 # introduced in Python 3.4.1.
635 err.fp._error = err
636 return err.fp
637
638 if errnote is False:
639 return False
640 if errnote is None:
641 errnote = 'Unable to download webpage'
642
643 errmsg = '%s: %s' % (errnote, error_to_compat_str(err))
644 if fatal:
645 raise ExtractorError(errmsg, sys.exc_info()[2], cause=err)
646 else:
647 self._downloader.report_warning(errmsg)
648 return False
649
650 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):
651 """
652 Return a tuple (page content as string, URL handle).
653
654 See _download_webpage docstring for arguments specification.
655 """
656 # Strip hashes from the URL (#1038)
657 if isinstance(url_or_request, (compat_str, str)):
658 url_or_request = url_or_request.partition('#')[0]
659
660 urlh = self._request_webpage(url_or_request, video_id, note, errnote, fatal, data=data, headers=headers, query=query, expected_status=expected_status)
661 if urlh is False:
662 assert not fatal
663 return False
664 content = self._webpage_read_content(urlh, url_or_request, video_id, note, errnote, fatal, encoding=encoding)
665 return (content, urlh)
666
667 @staticmethod
668 def _guess_encoding_from_content(content_type, webpage_bytes):
669 m = re.match(r'[a-zA-Z0-9_.-]+/[a-zA-Z0-9_.-]+\s*;\s*charset=(.+)', content_type)
670 if m:
671 encoding = m.group(1)
672 else:
673 m = re.search(br'<meta[^>]+charset=[\'"]?([^\'")]+)[ /\'">]',
674 webpage_bytes[:1024])
675 if m:
676 encoding = m.group(1).decode('ascii')
677 elif webpage_bytes.startswith(b'\xff\xfe'):
678 encoding = 'utf-16'
679 else:
680 encoding = 'utf-8'
681
682 return encoding
683
684 def __check_blocked(self, content):
685 first_block = content[:512]
686 if ('<title>Access to this site is blocked</title>' in content
687 and 'Websense' in first_block):
688 msg = 'Access to this webpage has been blocked by Websense filtering software in your network.'
689 blocked_iframe = self._html_search_regex(
690 r'<iframe src="([^"]+)"', content,
691 'Websense information URL', default=None)
692 if blocked_iframe:
693 msg += ' Visit %s for more details' % blocked_iframe
694 raise ExtractorError(msg, expected=True)
695 if '<title>The URL you requested has been blocked</title>' in first_block:
696 msg = (
697 'Access to this webpage has been blocked by Indian censorship. '
698 'Use a VPN or proxy server (with --proxy) to route around it.')
699 block_msg = self._html_search_regex(
700 r'</h1><p>(.*?)</p>',
701 content, 'block message', default=None)
702 if block_msg:
703 msg += ' (Message: "%s")' % block_msg.replace('\n', ' ')
704 raise ExtractorError(msg, expected=True)
705 if ('<title>TTK :: Š”Š¾ŃŃ‚ŃƒŠæ Šŗ рŠµŃŃƒŃ€ŃŃƒ Š¾Š³Ń€Š°Š½ŠøчŠµŠ½</title>' in content
706 and 'blocklist.rkn.gov.ru' in content):
707 raise ExtractorError(
708 'Access to this webpage has been blocked by decision of the Russian government. '
709 'Visit http://blocklist.rkn.gov.ru/ for a block reason.',
710 expected=True)
711
712 def _webpage_read_content(self, urlh, url_or_request, video_id, note=None, errnote=None, fatal=True, prefix=None, encoding=None):
713 content_type = urlh.headers.get('Content-Type', '')
714 webpage_bytes = urlh.read()
715 if prefix is not None:
716 webpage_bytes = prefix + webpage_bytes
717 if not encoding:
718 encoding = self._guess_encoding_from_content(content_type, webpage_bytes)
719 if self._downloader.params.get('dump_intermediate_pages', False):
720 self.to_screen('Dumping request to ' + urlh.geturl())
721 dump = base64.b64encode(webpage_bytes).decode('ascii')
722 self._downloader.to_screen(dump)
723 if self._downloader.params.get('write_pages', False):
724 basen = '%s_%s' % (video_id, urlh.geturl())
725 if len(basen) > 240:
726 h = '___' + hashlib.md5(basen.encode('utf-8')).hexdigest()
727 basen = basen[:240 - len(h)] + h
728 raw_filename = basen + '.dump'
729 filename = sanitize_filename(raw_filename, restricted=True)
730 self.to_screen('Saving request to ' + filename)
731 # Working around MAX_PATH limitation on Windows (see
732 # http://msdn.microsoft.com/en-us/library/windows/desktop/aa365247(v=vs.85).aspx)
733 if compat_os_name == 'nt':
734 absfilepath = os.path.abspath(filename)
735 if len(absfilepath) > 259:
736 filename = '\\\\?\\' + absfilepath
737 with open(filename, 'wb') as outf:
738 outf.write(webpage_bytes)
739
740 try:
741 content = webpage_bytes.decode(encoding, 'replace')
742 except LookupError:
743 content = webpage_bytes.decode('utf-8', 'replace')
744
745 self.__check_blocked(content)
746
747 return content
748
749 def _download_webpage(
750 self, url_or_request, video_id, note=None, errnote=None,
751 fatal=True, tries=1, timeout=5, encoding=None, data=None,
752 headers={}, query={}, expected_status=None):
753 """
754 Return the data of the page as a string.
755
756 Arguments:
757 url_or_request -- plain text URL as a string or
758 a compat_urllib_request.Requestobject
759 video_id -- Video/playlist/item identifier (string)
760
761 Keyword arguments:
762 note -- note printed before downloading (string)
763 errnote -- note printed in case of an error (string)
764 fatal -- flag denoting whether error should be considered fatal,
765 i.e. whether it should cause ExtractionError to be raised,
766 otherwise a warning will be reported and extraction continued
767 tries -- number of tries
768 timeout -- sleep interval between tries
769 encoding -- encoding for a page content decoding, guessed automatically
770 when not explicitly specified
771 data -- POST data (bytes)
772 headers -- HTTP headers (dict)
773 query -- URL query (dict)
774 expected_status -- allows to accept failed HTTP requests (non 2xx
775 status code) by explicitly specifying a set of accepted status
776 codes. Can be any of the following entities:
777 - an integer type specifying an exact failed status code to
778 accept
779 - a list or a tuple of integer types specifying a list of
780 failed status codes to accept
781 - a callable accepting an actual failed status code and
782 returning True if it should be accepted
783 Note that this argument does not affect success status codes (2xx)
784 which are always accepted.
785 """
786
787 success = False
788 try_count = 0
789 while success is False:
790 try:
791 res = self._download_webpage_handle(
792 url_or_request, video_id, note, errnote, fatal,
793 encoding=encoding, data=data, headers=headers, query=query,
794 expected_status=expected_status)
795 success = True
796 except compat_http_client.IncompleteRead as e:
797 try_count += 1
798 if try_count >= tries:
799 raise e
800 self._sleep(timeout, video_id)
801 if res is False:
802 return res
803 else:
804 content, _ = res
805 return content
806
807 def _download_xml_handle(
808 self, url_or_request, video_id, note='Downloading XML',
809 errnote='Unable to download XML', transform_source=None,
810 fatal=True, encoding=None, data=None, headers={}, query={},
811 expected_status=None):
812 """
813 Return a tuple (xml as an compat_etree_Element, URL handle).
814
815 See _download_webpage docstring for arguments specification.
816 """
817 res = self._download_webpage_handle(
818 url_or_request, video_id, note, errnote, fatal=fatal,
819 encoding=encoding, data=data, headers=headers, query=query,
820 expected_status=expected_status)
821 if res is False:
822 return res
823 xml_string, urlh = res
824 return self._parse_xml(
825 xml_string, video_id, transform_source=transform_source,
826 fatal=fatal), urlh
827
828 def _download_xml(
829 self, url_or_request, video_id,
830 note='Downloading XML', errnote='Unable to download XML',
831 transform_source=None, fatal=True, encoding=None,
832 data=None, headers={}, query={}, expected_status=None):
833 """
834 Return the xml as an compat_etree_Element.
835
836 See _download_webpage docstring for arguments specification.
837 """
838 res = self._download_xml_handle(
839 url_or_request, video_id, note=note, errnote=errnote,
840 transform_source=transform_source, fatal=fatal, encoding=encoding,
841 data=data, headers=headers, query=query,
842 expected_status=expected_status)
843 return res if res is False else res[0]
844
845 def _parse_xml(self, xml_string, video_id, transform_source=None, fatal=True):
846 if transform_source:
847 xml_string = transform_source(xml_string)
848 try:
849 return compat_etree_fromstring(xml_string.encode('utf-8'))
850 except compat_xml_parse_error as ve:
851 errmsg = '%s: Failed to parse XML ' % video_id
852 if fatal:
853 raise ExtractorError(errmsg, cause=ve)
854 else:
855 self.report_warning(errmsg + str(ve))
856
857 def _download_json_handle(
858 self, url_or_request, video_id, note='Downloading JSON metadata',
859 errnote='Unable to download JSON metadata', transform_source=None,
860 fatal=True, encoding=None, data=None, headers={}, query={},
861 expected_status=None):
862 """
863 Return a tuple (JSON object, URL handle).
864
865 See _download_webpage docstring for arguments specification.
866 """
867 res = self._download_webpage_handle(
868 url_or_request, video_id, note, errnote, fatal=fatal,
869 encoding=encoding, data=data, headers=headers, query=query,
870 expected_status=expected_status)
871 if res is False:
872 return res
873 json_string, urlh = res
874 return self._parse_json(
875 json_string, video_id, transform_source=transform_source,
876 fatal=fatal), urlh
877
878 def _download_json(
879 self, url_or_request, video_id, note='Downloading JSON metadata',
880 errnote='Unable to download JSON metadata', transform_source=None,
881 fatal=True, encoding=None, data=None, headers={}, query={},
882 expected_status=None):
883 """
884 Return the JSON object as a dict.
885
886 See _download_webpage docstring for arguments specification.
887 """
888 res = self._download_json_handle(
889 url_or_request, video_id, note=note, errnote=errnote,
890 transform_source=transform_source, fatal=fatal, encoding=encoding,
891 data=data, headers=headers, query=query,
892 expected_status=expected_status)
893 return res if res is False else res[0]
894
895 def _parse_json(self, json_string, video_id, transform_source=None, fatal=True):
896 if transform_source:
897 json_string = transform_source(json_string)
898 try:
899 return json.loads(json_string)
900 except ValueError as ve:
901 errmsg = '%s: Failed to parse JSON ' % video_id
902 if fatal:
903 raise ExtractorError(errmsg, cause=ve)
904 else:
905 self.report_warning(errmsg + str(ve))
906
907 def report_warning(self, msg, video_id=None):
908 idstr = '' if video_id is None else '%s: ' % video_id
909 self._downloader.report_warning(
910 '[%s] %s%s' % (self.IE_NAME, idstr, msg))
911
912 def to_screen(self, msg):
913 """Print msg to screen, prefixing it with '[ie_name]'"""
914 self._downloader.to_screen('[%s] %s' % (self.IE_NAME, msg))
915
916 def report_extraction(self, id_or_name):
917 """Report information extraction."""
918 self.to_screen('%s: Extracting information' % id_or_name)
919
920 def report_download_webpage(self, video_id):
921 """Report webpage download."""
922 self.to_screen('%s: Downloading webpage' % video_id)
923
924 def report_age_confirmation(self):
925 """Report attempt to confirm age."""
926 self.to_screen('Confirming age')
927
928 def report_login(self):
929 """Report attempt to log in."""
930 self.to_screen('Logging in')
931
932 @staticmethod
933 def raise_login_required(msg='This video is only available for registered users'):
934 raise ExtractorError(
935 '%s. Use --username and --password or --netrc to provide account credentials.' % msg,
936 expected=True)
937
938 @staticmethod
939 def raise_geo_restricted(msg='This video is not available from your location due to geo restriction', countries=None):
940 raise GeoRestrictedError(msg, countries=countries)
941
942 # Methods for following #608
943 @staticmethod
944 def url_result(url, ie=None, video_id=None, video_title=None):
945 """Returns a URL that points to a page that should be processed"""
946 # TODO: ie should be the class used for getting the info
947 video_info = {'_type': 'url',
948 'url': url,
949 'ie_key': ie}
950 if video_id is not None:
951 video_info['id'] = video_id
952 if video_title is not None:
953 video_info['title'] = video_title
954 return video_info
955
956 def playlist_from_matches(self, matches, playlist_id=None, playlist_title=None, getter=None, ie=None):
957 urls = orderedSet(
958 self.url_result(self._proto_relative_url(getter(m) if getter else m), ie)
959 for m in matches)
960 return self.playlist_result(
961 urls, playlist_id=playlist_id, playlist_title=playlist_title)
962
963 @staticmethod
964 def playlist_result(entries, playlist_id=None, playlist_title=None, playlist_description=None):
965 """Returns a playlist"""
966 video_info = {'_type': 'playlist',
967 'entries': entries}
968 if playlist_id:
969 video_info['id'] = playlist_id
970 if playlist_title:
971 video_info['title'] = playlist_title
972 if playlist_description:
973 video_info['description'] = playlist_description
974 return video_info
975
976 def _search_regex(self, pattern, string, name, default=NO_DEFAULT, fatal=True, flags=0, group=None):
977 """
978 Perform a regex search on the given string, using a single or a list of
979 patterns returning the first matching group.
980 In case of failure return a default value or raise a WARNING or a
981 RegexNotFoundError, depending on fatal, specifying the field name.
982 """
983 if isinstance(pattern, (str, compat_str, compiled_regex_type)):
984 mobj = re.search(pattern, string, flags)
985 else:
986 for p in pattern:
987 mobj = re.search(p, string, flags)
988 if mobj:
989 break
990
991 if not self._downloader.params.get('no_color') and compat_os_name != 'nt' and sys.stderr.isatty():
992 _name = '\033[0;34m%s\033[0m' % name
993 else:
994 _name = name
995
996 if mobj:
997 if group is None:
998 # return the first matching group
999 return next(g for g in mobj.groups() if g is not None)
1000 else:
1001 return mobj.group(group)
1002 elif default is not NO_DEFAULT:
1003 return default
1004 elif fatal:
1005 raise RegexNotFoundError('Unable to extract %s' % _name)
1006 else:
1007 self._downloader.report_warning('unable to extract %s' % _name + bug_reports_message())
1008 return None
1009
1010 def _html_search_regex(self, pattern, string, name, default=NO_DEFAULT, fatal=True, flags=0, group=None):
1011 """
1012 Like _search_regex, but strips HTML tags and unescapes entities.
1013 """
1014 res = self._search_regex(pattern, string, name, default, fatal, flags, group)
1015 if res:
1016 return clean_html(res).strip()
1017 else:
1018 return res
1019
1020 def _get_netrc_login_info(self, netrc_machine=None):
1021 username = None
1022 password = None
1023 netrc_machine = netrc_machine or self._NETRC_MACHINE
1024
1025 if self._downloader.params.get('usenetrc', False):
1026 try:
1027 info = netrc.netrc().authenticators(netrc_machine)
1028 if info is not None:
1029 username = info[0]
1030 password = info[2]
1031 else:
1032 raise netrc.NetrcParseError(
1033 'No authenticators for %s' % netrc_machine)
1034 except (IOError, netrc.NetrcParseError) as err:
1035 self._downloader.report_warning(
1036 'parsing .netrc: %s' % error_to_compat_str(err))
1037
1038 return username, password
1039
1040 def _get_login_info(self, username_option='username', password_option='password', netrc_machine=None):
1041 """
1042 Get the login info as (username, password)
1043 First look for the manually specified credentials using username_option
1044 and password_option as keys in params dictionary. If no such credentials
1045 available look in the netrc file using the netrc_machine or _NETRC_MACHINE
1046 value.
1047 If there's no info available, return (None, None)
1048 """
1049 if self._downloader is None:
1050 return (None, None)
1051
1052 downloader_params = self._downloader.params
1053
1054 # Attempt to use provided username and password or .netrc data
1055 if downloader_params.get(username_option) is not None:
1056 username = downloader_params[username_option]
1057 password = downloader_params[password_option]
1058 else:
1059 username, password = self._get_netrc_login_info(netrc_machine)
1060
1061 return username, password
1062
1063 def _get_tfa_info(self, note='two-factor verification code'):
1064 """
1065 Get the two-factor authentication info
1066 TODO - asking the user will be required for sms/phone verify
1067 currently just uses the command line option
1068 If there's no info available, return None
1069 """
1070 if self._downloader is None:
1071 return None
1072 downloader_params = self._downloader.params
1073
1074 if downloader_params.get('twofactor') is not None:
1075 return downloader_params['twofactor']
1076
1077 return compat_getpass('Type %s and press [Return]: ' % note)
1078
1079 # Helper functions for extracting OpenGraph info
1080 @staticmethod
1081 def _og_regexes(prop):
1082 content_re = r'content=(?:"([^"]+?)"|\'([^\']+?)\'|\s*([^\s"\'=<>`]+?))'
1083 property_re = (r'(?:name|property)=(?:\'og[:-]%(prop)s\'|"og[:-]%(prop)s"|\s*og[:-]%(prop)s\b)'
1084 % {'prop': re.escape(prop)})
1085 template = r'<meta[^>]+?%s[^>]+?%s'
1086 return [
1087 template % (property_re, content_re),
1088 template % (content_re, property_re),
1089 ]
1090
1091 @staticmethod
1092 def _meta_regex(prop):
1093 return r'''(?isx)<meta
1094 (?=[^>]+(?:itemprop|name|property|id|http-equiv)=(["\']?)%s\1)
1095 [^>]+?content=(["\'])(?P<content>.*?)\2''' % re.escape(prop)
1096
1097 def _og_search_property(self, prop, html, name=None, **kargs):
1098 if not isinstance(prop, (list, tuple)):
1099 prop = [prop]
1100 if name is None:
1101 name = 'OpenGraph %s' % prop[0]
1102 og_regexes = []
1103 for p in prop:
1104 og_regexes.extend(self._og_regexes(p))
1105 escaped = self._search_regex(og_regexes, html, name, flags=re.DOTALL, **kargs)
1106 if escaped is None:
1107 return None
1108 return unescapeHTML(escaped)
1109
1110 def _og_search_thumbnail(self, html, **kargs):
1111 return self._og_search_property('image', html, 'thumbnail URL', fatal=False, **kargs)
1112
1113 def _og_search_description(self, html, **kargs):
1114 return self._og_search_property('description', html, fatal=False, **kargs)
1115
1116 def _og_search_title(self, html, **kargs):
1117 return self._og_search_property('title', html, **kargs)
1118
1119 def _og_search_video_url(self, html, name='video url', secure=True, **kargs):
1120 regexes = self._og_regexes('video') + self._og_regexes('video:url')
1121 if secure:
1122 regexes = self._og_regexes('video:secure_url') + regexes
1123 return self._html_search_regex(regexes, html, name, **kargs)
1124
1125 def _og_search_url(self, html, **kargs):
1126 return self._og_search_property('url', html, **kargs)
1127
1128 def _html_search_meta(self, name, html, display_name=None, fatal=False, **kwargs):
1129 if not isinstance(name, (list, tuple)):
1130 name = [name]
1131 if display_name is None:
1132 display_name = name[0]
1133 return self._html_search_regex(
1134 [self._meta_regex(n) for n in name],
1135 html, display_name, fatal=fatal, group='content', **kwargs)
1136
1137 def _dc_search_uploader(self, html):
1138 return self._html_search_meta('dc.creator', html, 'uploader')
1139
1140 def _rta_search(self, html):
1141 # See http://www.rtalabel.org/index.php?content=howtofaq#single
1142 if re.search(r'(?ix)<meta\s+name="rating"\s+'
1143 r' content="RTA-5042-1996-1400-1577-RTA"',
1144 html):
1145 return 18
1146 return 0
1147
1148 def _media_rating_search(self, html):
1149 # See http://www.tjg-designs.com/WP/metadata-code-examples-adding-metadata-to-your-web-pages/
1150 rating = self._html_search_meta('rating', html)
1151
1152 if not rating:
1153 return None
1154
1155 RATING_TABLE = {
1156 'safe for kids': 0,
1157 'general': 8,
1158 '14 years': 14,
1159 'mature': 17,
1160 'restricted': 19,
1161 }
1162 return RATING_TABLE.get(rating.lower())
1163
1164 def _family_friendly_search(self, html):
1165 # See http://schema.org/VideoObject
1166 family_friendly = self._html_search_meta(
1167 'isFamilyFriendly', html, default=None)
1168
1169 if not family_friendly:
1170 return None
1171
1172 RATING_TABLE = {
1173 '1': 0,
1174 'true': 0,
1175 '0': 18,
1176 'false': 18,
1177 }
1178 return RATING_TABLE.get(family_friendly.lower())
1179
1180 def _twitter_search_player(self, html):
1181 return self._html_search_meta('twitter:player', html,
1182 'twitter card player')
1183
1184 def _search_json_ld(self, html, video_id, expected_type=None, **kwargs):
1185 json_ld = self._search_regex(
1186 JSON_LD_RE, html, 'JSON-LD', group='json_ld', **kwargs)
1187 default = kwargs.get('default', NO_DEFAULT)
1188 if not json_ld:
1189 return default if default is not NO_DEFAULT else {}
1190 # JSON-LD may be malformed and thus `fatal` should be respected.
1191 # At the same time `default` may be passed that assumes `fatal=False`
1192 # for _search_regex. Let's simulate the same behavior here as well.
1193 fatal = kwargs.get('fatal', True) if default == NO_DEFAULT else False
1194 return self._json_ld(json_ld, video_id, fatal=fatal, expected_type=expected_type)
1195
1196 def _json_ld(self, json_ld, video_id, fatal=True, expected_type=None):
1197 if isinstance(json_ld, compat_str):
1198 json_ld = self._parse_json(json_ld, video_id, fatal=fatal)
1199 if not json_ld:
1200 return {}
1201 info = {}
1202 if not isinstance(json_ld, (list, tuple, dict)):
1203 return info
1204 if isinstance(json_ld, dict):
1205 json_ld = [json_ld]
1206
1207 INTERACTION_TYPE_MAP = {
1208 'CommentAction': 'comment',
1209 'AgreeAction': 'like',
1210 'DisagreeAction': 'dislike',
1211 'LikeAction': 'like',
1212 'DislikeAction': 'dislike',
1213 'ListenAction': 'view',
1214 'WatchAction': 'view',
1215 'ViewAction': 'view',
1216 }
1217
1218 def extract_interaction_statistic(e):
1219 interaction_statistic = e.get('interactionStatistic')
1220 if not isinstance(interaction_statistic, list):
1221 return
1222 for is_e in interaction_statistic:
1223 if not isinstance(is_e, dict):
1224 continue
1225 if is_e.get('@type') != 'InteractionCounter':
1226 continue
1227 interaction_type = is_e.get('interactionType')
1228 if not isinstance(interaction_type, compat_str):
1229 continue
1230 interaction_count = int_or_none(is_e.get('userInteractionCount'))
1231 if interaction_count is None:
1232 continue
1233 count_kind = INTERACTION_TYPE_MAP.get(interaction_type.split('/')[-1])
1234 if not count_kind:
1235 continue
1236 count_key = '%s_count' % count_kind
1237 if info.get(count_key) is not None:
1238 continue
1239 info[count_key] = interaction_count
1240
1241 def extract_video_object(e):
1242 assert e['@type'] == 'VideoObject'
1243 info.update({
1244 'url': url_or_none(e.get('contentUrl')),
1245 'title': unescapeHTML(e.get('name')),
1246 'description': unescapeHTML(e.get('description')),
1247 'thumbnail': url_or_none(e.get('thumbnailUrl') or e.get('thumbnailURL')),
1248 'duration': parse_duration(e.get('duration')),
1249 'timestamp': unified_timestamp(e.get('uploadDate')),
1250 'filesize': float_or_none(e.get('contentSize')),
1251 'tbr': int_or_none(e.get('bitrate')),
1252 'width': int_or_none(e.get('width')),
1253 'height': int_or_none(e.get('height')),
1254 'view_count': int_or_none(e.get('interactionCount')),
1255 })
1256 extract_interaction_statistic(e)
1257
1258 for e in json_ld:
1259 if isinstance(e.get('@context'), compat_str) and re.match(r'^https?://schema.org/?$', e.get('@context')):
1260 item_type = e.get('@type')
1261 if expected_type is not None and expected_type != item_type:
1262 return info
1263 if item_type in ('TVEpisode', 'Episode'):
1264 episode_name = unescapeHTML(e.get('name'))
1265 info.update({
1266 'episode': episode_name,
1267 'episode_number': int_or_none(e.get('episodeNumber')),
1268 'description': unescapeHTML(e.get('description')),
1269 })
1270 if not info.get('title') and episode_name:
1271 info['title'] = episode_name
1272 part_of_season = e.get('partOfSeason')
1273 if isinstance(part_of_season, dict) and part_of_season.get('@type') in ('TVSeason', 'Season', 'CreativeWorkSeason'):
1274 info.update({
1275 'season': unescapeHTML(part_of_season.get('name')),
1276 'season_number': int_or_none(part_of_season.get('seasonNumber')),
1277 })
1278 part_of_series = e.get('partOfSeries') or e.get('partOfTVSeries')
1279 if isinstance(part_of_series, dict) and part_of_series.get('@type') in ('TVSeries', 'Series', 'CreativeWorkSeries'):
1280 info['series'] = unescapeHTML(part_of_series.get('name'))
1281 elif item_type == 'Movie':
1282 info.update({
1283 'title': unescapeHTML(e.get('name')),
1284 'description': unescapeHTML(e.get('description')),
1285 'duration': parse_duration(e.get('duration')),
1286 'timestamp': unified_timestamp(e.get('dateCreated')),
1287 })
1288 elif item_type in ('Article', 'NewsArticle'):
1289 info.update({
1290 'timestamp': parse_iso8601(e.get('datePublished')),
1291 'title': unescapeHTML(e.get('headline')),
1292 'description': unescapeHTML(e.get('articleBody')),
1293 })
1294 elif item_type == 'VideoObject':
1295 extract_video_object(e)
1296 continue
1297 video = e.get('video')
1298 if isinstance(video, dict) and video.get('@type') == 'VideoObject':
1299 extract_video_object(video)
1300 break
1301 return dict((k, v) for k, v in info.items() if v is not None)
1302
1303 @staticmethod
1304 def _hidden_inputs(html):
1305 html = re.sub(r'<!--(?:(?!<!--).)*-->', '', html)
1306 hidden_inputs = {}
1307 for input in re.findall(r'(?i)(<input[^>]+>)', html):
1308 attrs = extract_attributes(input)
1309 if not input:
1310 continue
1311 if attrs.get('type') not in ('hidden', 'submit'):
1312 continue
1313 name = attrs.get('name') or attrs.get('id')
1314 value = attrs.get('value')
1315 if name and value is not None:
1316 hidden_inputs[name] = value
1317 return hidden_inputs
1318
1319 def _form_hidden_inputs(self, form_id, html):
1320 form = self._search_regex(
1321 r'(?is)<form[^>]+?id=(["\'])%s\1[^>]*>(?P<form>.+?)</form>' % form_id,
1322 html, '%s form' % form_id, group='form')
1323 return self._hidden_inputs(form)
1324
1325 def _sort_formats(self, formats, field_preference=None):
1326 if not formats:
1327 raise ExtractorError('No video formats found')
1328
1329 for f in formats:
1330 # Automatically determine tbr when missing based on abr and vbr (improves
1331 # formats sorting in some cases)
1332 if 'tbr' not in f and f.get('abr') is not None and f.get('vbr') is not None:
1333 f['tbr'] = f['abr'] + f['vbr']
1334
1335 def _formats_key(f):
1336 # TODO remove the following workaround
1337 from ..utils import determine_ext
1338 if not f.get('ext') and 'url' in f:
1339 f['ext'] = determine_ext(f['url'])
1340
1341 if isinstance(field_preference, (list, tuple)):
1342 return tuple(
1343 f.get(field)
1344 if f.get(field) is not None
1345 else ('' if field == 'format_id' else -1)
1346 for field in field_preference)
1347
1348 preference = f.get('preference')
1349 if preference is None:
1350 preference = 0
1351 if f.get('ext') in ['f4f', 'f4m']: # Not yet supported
1352 preference -= 0.5
1353
1354 protocol = f.get('protocol') or determine_protocol(f)
1355 proto_preference = 0 if protocol in ['http', 'https'] else (-0.5 if protocol == 'rtsp' else -0.1)
1356
1357 if f.get('vcodec') == 'none': # audio only
1358 preference -= 50
1359 if self._downloader.params.get('prefer_free_formats'):
1360 ORDER = ['aac', 'mp3', 'm4a', 'webm', 'ogg', 'opus']
1361 else:
1362 ORDER = ['webm', 'opus', 'ogg', 'mp3', 'aac', 'm4a']
1363 ext_preference = 0
1364 try:
1365 audio_ext_preference = ORDER.index(f['ext'])
1366 except ValueError:
1367 audio_ext_preference = -1
1368 else:
1369 if f.get('acodec') == 'none': # video only
1370 preference -= 40
1371 if self._downloader.params.get('prefer_free_formats'):
1372 ORDER = ['flv', 'mp4', 'webm']
1373 else:
1374 ORDER = ['webm', 'flv', 'mp4']
1375 try:
1376 ext_preference = ORDER.index(f['ext'])
1377 except ValueError:
1378 ext_preference = -1
1379 audio_ext_preference = 0
1380
1381 return (
1382 preference,
1383 f.get('language_preference') if f.get('language_preference') is not None else -1,
1384 f.get('quality') if f.get('quality') is not None else -1,
1385 f.get('tbr') if f.get('tbr') is not None else -1,
1386 f.get('filesize') if f.get('filesize') is not None else -1,
1387 f.get('vbr') if f.get('vbr') is not None else -1,
1388 f.get('height') if f.get('height') is not None else -1,
1389 f.get('width') if f.get('width') is not None else -1,
1390 proto_preference,
1391 ext_preference,
1392 f.get('abr') if f.get('abr') is not None else -1,
1393 audio_ext_preference,
1394 f.get('fps') if f.get('fps') is not None else -1,
1395 f.get('filesize_approx') if f.get('filesize_approx') is not None else -1,
1396 f.get('source_preference') if f.get('source_preference') is not None else -1,
1397 f.get('format_id') if f.get('format_id') is not None else '',
1398 )
1399 formats.sort(key=_formats_key)
1400
1401 def _check_formats(self, formats, video_id):
1402 if formats:
1403 formats[:] = filter(
1404 lambda f: self._is_valid_url(
1405 f['url'], video_id,
1406 item='%s video format' % f.get('format_id') if f.get('format_id') else 'video'),
1407 formats)
1408
1409 @staticmethod
1410 def _remove_duplicate_formats(formats):
1411 format_urls = set()
1412 unique_formats = []
1413 for f in formats:
1414 if f['url'] not in format_urls:
1415 format_urls.add(f['url'])
1416 unique_formats.append(f)
1417 formats[:] = unique_formats
1418
1419 def _is_valid_url(self, url, video_id, item='video', headers={}):
1420 url = self._proto_relative_url(url, scheme='http:')
1421 # For now assume non HTTP(S) URLs always valid
1422 if not (url.startswith('http://') or url.startswith('https://')):
1423 return True
1424 try:
1425 self._request_webpage(url, video_id, 'Checking %s URL' % item, headers=headers)
1426 return True
1427 except ExtractorError as e:
1428 if isinstance(e.cause, compat_urllib_error.URLError):
1429 self.to_screen(
1430 '%s: %s URL is invalid, skipping' % (video_id, item))
1431 return False
1432 raise
1433
1434 def http_scheme(self):
1435 """ Either "http:" or "https:", depending on the user's preferences """
1436 return (
1437 'http:'
1438 if self._downloader.params.get('prefer_insecure', False)
1439 else 'https:')
1440
1441 def _proto_relative_url(self, url, scheme=None):
1442 if url is None:
1443 return url
1444 if url.startswith('//'):
1445 if scheme is None:
1446 scheme = self.http_scheme()
1447 return scheme + url
1448 else:
1449 return url
1450
1451 def _sleep(self, timeout, video_id, msg_template=None):
1452 if msg_template is None:
1453 msg_template = '%(video_id)s: Waiting for %(timeout)s seconds'
1454 msg = msg_template % {'video_id': video_id, 'timeout': timeout}
1455 self.to_screen(msg)
1456 time.sleep(timeout)
1457
1458 def _extract_f4m_formats(self, manifest_url, video_id, preference=None, f4m_id=None,
1459 transform_source=lambda s: fix_xml_ampersands(s).strip(),
1460 fatal=True, m3u8_id=None):
1461 manifest = self._download_xml(
1462 manifest_url, video_id, 'Downloading f4m manifest',
1463 'Unable to download f4m manifest',
1464 # Some manifests may be malformed, e.g. prosiebensat1 generated manifests
1465 # (see https://github.com/ytdl-org/youtube-dl/issues/6215#issuecomment-121704244)
1466 transform_source=transform_source,
1467 fatal=fatal)
1468
1469 if manifest is False:
1470 return []
1471
1472 return self._parse_f4m_formats(
1473 manifest, manifest_url, video_id, preference=preference, f4m_id=f4m_id,
1474 transform_source=transform_source, fatal=fatal, m3u8_id=m3u8_id)
1475
1476 def _parse_f4m_formats(self, manifest, manifest_url, video_id, preference=None, f4m_id=None,
1477 transform_source=lambda s: fix_xml_ampersands(s).strip(),
1478 fatal=True, m3u8_id=None):
1479 if not isinstance(manifest, compat_etree_Element) and not fatal:
1480 return []
1481
1482 # currently youtube-dl cannot decode the playerVerificationChallenge as Akamai uses Adobe Alchemy
1483 akamai_pv = manifest.find('{http://ns.adobe.com/f4m/1.0}pv-2.0')
1484 if akamai_pv is not None and ';' in akamai_pv.text:
1485 playerVerificationChallenge = akamai_pv.text.split(';')[0]
1486 if playerVerificationChallenge.strip() != '':
1487 return []
1488
1489 formats = []
1490 manifest_version = '1.0'
1491 media_nodes = manifest.findall('{http://ns.adobe.com/f4m/1.0}media')
1492 if not media_nodes:
1493 manifest_version = '2.0'
1494 media_nodes = manifest.findall('{http://ns.adobe.com/f4m/2.0}media')
1495 # Remove unsupported DRM protected media from final formats
1496 # rendition (see https://github.com/ytdl-org/youtube-dl/issues/8573).
1497 media_nodes = remove_encrypted_media(media_nodes)
1498 if not media_nodes:
1499 return formats
1500
1501 manifest_base_url = get_base_url(manifest)
1502
1503 bootstrap_info = xpath_element(
1504 manifest, ['{http://ns.adobe.com/f4m/1.0}bootstrapInfo', '{http://ns.adobe.com/f4m/2.0}bootstrapInfo'],
1505 'bootstrap info', default=None)
1506
1507 vcodec = None
1508 mime_type = xpath_text(
1509 manifest, ['{http://ns.adobe.com/f4m/1.0}mimeType', '{http://ns.adobe.com/f4m/2.0}mimeType'],
1510 'base URL', default=None)
1511 if mime_type and mime_type.startswith('audio/'):
1512 vcodec = 'none'
1513
1514 for i, media_el in enumerate(media_nodes):
1515 tbr = int_or_none(media_el.attrib.get('bitrate'))
1516 width = int_or_none(media_el.attrib.get('width'))
1517 height = int_or_none(media_el.attrib.get('height'))
1518 format_id = '-'.join(filter(None, [f4m_id, compat_str(i if tbr is None else tbr)]))
1519 # If <bootstrapInfo> is present, the specified f4m is a
1520 # stream-level manifest, and only set-level manifests may refer to
1521 # external resources. See section 11.4 and section 4 of F4M spec
1522 if bootstrap_info is None:
1523 media_url = None
1524 # @href is introduced in 2.0, see section 11.6 of F4M spec
1525 if manifest_version == '2.0':
1526 media_url = media_el.attrib.get('href')
1527 if media_url is None:
1528 media_url = media_el.attrib.get('url')
1529 if not media_url:
1530 continue
1531 manifest_url = (
1532 media_url if media_url.startswith('http://') or media_url.startswith('https://')
1533 else ((manifest_base_url or '/'.join(manifest_url.split('/')[:-1])) + '/' + media_url))
1534 # If media_url is itself a f4m manifest do the recursive extraction
1535 # since bitrates in parent manifest (this one) and media_url manifest
1536 # may differ leading to inability to resolve the format by requested
1537 # bitrate in f4m downloader
1538 ext = determine_ext(manifest_url)
1539 if ext == 'f4m':
1540 f4m_formats = self._extract_f4m_formats(
1541 manifest_url, video_id, preference=preference, f4m_id=f4m_id,
1542 transform_source=transform_source, fatal=fatal)
1543 # Sometimes stream-level manifest contains single media entry that
1544 # does not contain any quality metadata (e.g. http://matchtv.ru/#live-player).
1545 # At the same time parent's media entry in set-level manifest may
1546 # contain it. We will copy it from parent in such cases.
1547 if len(f4m_formats) == 1:
1548 f = f4m_formats[0]
1549 f.update({
1550 'tbr': f.get('tbr') or tbr,
1551 'width': f.get('width') or width,
1552 'height': f.get('height') or height,
1553 'format_id': f.get('format_id') if not tbr else format_id,
1554 'vcodec': vcodec,
1555 })
1556 formats.extend(f4m_formats)
1557 continue
1558 elif ext == 'm3u8':
1559 formats.extend(self._extract_m3u8_formats(
1560 manifest_url, video_id, 'mp4', preference=preference,
1561 m3u8_id=m3u8_id, fatal=fatal))
1562 continue
1563 formats.append({
1564 'format_id': format_id,
1565 'url': manifest_url,
1566 'manifest_url': manifest_url,
1567 'ext': 'flv' if bootstrap_info is not None else None,
1568 'protocol': 'f4m',
1569 'tbr': tbr,
1570 'width': width,
1571 'height': height,
1572 'vcodec': vcodec,
1573 'preference': preference,
1574 })
1575 return formats
1576
1577 def _m3u8_meta_format(self, m3u8_url, ext=None, preference=None, m3u8_id=None):
1578 return {
1579 'format_id': '-'.join(filter(None, [m3u8_id, 'meta'])),
1580 'url': m3u8_url,
1581 'ext': ext,
1582 'protocol': 'm3u8',
1583 'preference': preference - 100 if preference else -100,
1584 'resolution': 'multiple',
1585 'format_note': 'Quality selection URL',
1586 }
1587
1588 def _extract_m3u8_formats(self, m3u8_url, video_id, ext=None,
1589 entry_protocol='m3u8', preference=None,
1590 m3u8_id=None, note=None, errnote=None,
1591 fatal=True, live=False):
1592 res = self._download_webpage_handle(
1593 m3u8_url, video_id,
1594 note=note or 'Downloading m3u8 information',
1595 errnote=errnote or 'Failed to download m3u8 information',
1596 fatal=fatal)
1597
1598 if res is False:
1599 return []
1600
1601 m3u8_doc, urlh = res
1602 m3u8_url = urlh.geturl()
1603
1604 return self._parse_m3u8_formats(
1605 m3u8_doc, m3u8_url, ext=ext, entry_protocol=entry_protocol,
1606 preference=preference, m3u8_id=m3u8_id, live=live)
1607
1608 def _parse_m3u8_formats(self, m3u8_doc, m3u8_url, ext=None,
1609 entry_protocol='m3u8', preference=None,
1610 m3u8_id=None, live=False):
1611 if '#EXT-X-FAXS-CM:' in m3u8_doc: # Adobe Flash Access
1612 return []
1613
1614 if re.search(r'#EXT-X-SESSION-KEY:.*?URI="skd://', m3u8_doc): # Apple FairPlay
1615 return []
1616
1617 formats = []
1618
1619 format_url = lambda u: (
1620 u
1621 if re.match(r'^https?://', u)
1622 else compat_urlparse.urljoin(m3u8_url, u))
1623
1624 # References:
1625 # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-21
1626 # 2. https://github.com/ytdl-org/youtube-dl/issues/12211
1627 # 3. https://github.com/ytdl-org/youtube-dl/issues/18923
1628
1629 # We should try extracting formats only from master playlists [1, 4.3.4],
1630 # i.e. playlists that describe available qualities. On the other hand
1631 # media playlists [1, 4.3.3] should be returned as is since they contain
1632 # just the media without qualities renditions.
1633 # Fortunately, master playlist can be easily distinguished from media
1634 # playlist based on particular tags availability. As of [1, 4.3.3, 4.3.4]
1635 # master playlist tags MUST NOT appear in a media playist and vice versa.
1636 # As of [1, 4.3.3.1] #EXT-X-TARGETDURATION tag is REQUIRED for every
1637 # media playlist and MUST NOT appear in master playlist thus we can
1638 # clearly detect media playlist with this criterion.
1639
1640 if '#EXT-X-TARGETDURATION' in m3u8_doc: # media playlist, return as is
1641 return [{
1642 'url': m3u8_url,
1643 'format_id': m3u8_id,
1644 'ext': ext,
1645 'protocol': entry_protocol,
1646 'preference': preference,
1647 }]
1648
1649 groups = {}
1650 last_stream_inf = {}
1651
1652 def extract_media(x_media_line):
1653 media = parse_m3u8_attributes(x_media_line)
1654 # As per [1, 4.3.4.1] TYPE, GROUP-ID and NAME are REQUIRED
1655 media_type, group_id, name = media.get('TYPE'), media.get('GROUP-ID'), media.get('NAME')
1656 if not (media_type and group_id and name):
1657 return
1658 groups.setdefault(group_id, []).append(media)
1659 if media_type not in ('VIDEO', 'AUDIO'):
1660 return
1661 media_url = media.get('URI')
1662 if media_url:
1663 format_id = []
1664 for v in (m3u8_id, group_id, name):
1665 if v:
1666 format_id.append(v)
1667 f = {
1668 'format_id': '-'.join(format_id),
1669 'url': format_url(media_url),
1670 'manifest_url': m3u8_url,
1671 'language': media.get('LANGUAGE'),
1672 'ext': ext,
1673 'protocol': entry_protocol,
1674 'preference': preference,
1675 }
1676 if media_type == 'AUDIO':
1677 f['vcodec'] = 'none'
1678 formats.append(f)
1679
1680 def build_stream_name():
1681 # Despite specification does not mention NAME attribute for
1682 # EXT-X-STREAM-INF tag it still sometimes may be present (see [1]
1683 # or vidio test in TestInfoExtractor.test_parse_m3u8_formats)
1684 # 1. http://www.vidio.com/watch/165683-dj_ambred-booyah-live-2015
1685 stream_name = last_stream_inf.get('NAME')
1686 if stream_name:
1687 return stream_name
1688 # If there is no NAME in EXT-X-STREAM-INF it will be obtained
1689 # from corresponding rendition group
1690 stream_group_id = last_stream_inf.get('VIDEO')
1691 if not stream_group_id:
1692 return
1693 stream_group = groups.get(stream_group_id)
1694 if not stream_group:
1695 return stream_group_id
1696 rendition = stream_group[0]
1697 return rendition.get('NAME') or stream_group_id
1698
1699 # parse EXT-X-MEDIA tags before EXT-X-STREAM-INF in order to have the
1700 # chance to detect video only formats when EXT-X-STREAM-INF tags
1701 # precede EXT-X-MEDIA tags in HLS manifest such as [3].
1702 for line in m3u8_doc.splitlines():
1703 if line.startswith('#EXT-X-MEDIA:'):
1704 extract_media(line)
1705
1706 for line in m3u8_doc.splitlines():
1707 if line.startswith('#EXT-X-STREAM-INF:'):
1708 last_stream_inf = parse_m3u8_attributes(line)
1709 elif line.startswith('#') or not line.strip():
1710 continue
1711 else:
1712 tbr = float_or_none(
1713 last_stream_inf.get('AVERAGE-BANDWIDTH')
1714 or last_stream_inf.get('BANDWIDTH'), scale=1000)
1715 format_id = []
1716 if m3u8_id:
1717 format_id.append(m3u8_id)
1718 stream_name = build_stream_name()
1719 # Bandwidth of live streams may differ over time thus making
1720 # format_id unpredictable. So it's better to keep provided
1721 # format_id intact.
1722 if not live:
1723 format_id.append(stream_name if stream_name else '%d' % (tbr if tbr else len(formats)))
1724 manifest_url = format_url(line.strip())
1725 f = {
1726 'format_id': '-'.join(format_id),
1727 'url': manifest_url,
1728 'manifest_url': m3u8_url,
1729 'tbr': tbr,
1730 'ext': ext,
1731 'fps': float_or_none(last_stream_inf.get('FRAME-RATE')),
1732 'protocol': entry_protocol,
1733 'preference': preference,
1734 }
1735 resolution = last_stream_inf.get('RESOLUTION')
1736 if resolution:
1737 mobj = re.search(r'(?P<width>\d+)[xX](?P<height>\d+)', resolution)
1738 if mobj:
1739 f['width'] = int(mobj.group('width'))
1740 f['height'] = int(mobj.group('height'))
1741 # Unified Streaming Platform
1742 mobj = re.search(
1743 r'audio.*?(?:%3D|=)(\d+)(?:-video.*?(?:%3D|=)(\d+))?', f['url'])
1744 if mobj:
1745 abr, vbr = mobj.groups()
1746 abr, vbr = float_or_none(abr, 1000), float_or_none(vbr, 1000)
1747 f.update({
1748 'vbr': vbr,
1749 'abr': abr,
1750 })
1751 codecs = parse_codecs(last_stream_inf.get('CODECS'))
1752 f.update(codecs)
1753 audio_group_id = last_stream_inf.get('AUDIO')
1754 # As per [1, 4.3.4.1.1] any EXT-X-STREAM-INF tag which
1755 # references a rendition group MUST have a CODECS attribute.
1756 # However, this is not always respected, for example, [2]
1757 # contains EXT-X-STREAM-INF tag which references AUDIO
1758 # rendition group but does not have CODECS and despite
1759 # referencing an audio group it represents a complete
1760 # (with audio and video) format. So, for such cases we will
1761 # ignore references to rendition groups and treat them
1762 # as complete formats.
1763 if audio_group_id and codecs and f.get('vcodec') != 'none':
1764 audio_group = groups.get(audio_group_id)
1765 if audio_group and audio_group[0].get('URI'):
1766 # TODO: update acodec for audio only formats with
1767 # the same GROUP-ID
1768 f['acodec'] = 'none'
1769 formats.append(f)
1770 last_stream_inf = {}
1771 return formats
1772
1773 @staticmethod
1774 def _xpath_ns(path, namespace=None):
1775 if not namespace:
1776 return path
1777 out = []
1778 for c in path.split('/'):
1779 if not c or c == '.':
1780 out.append(c)
1781 else:
1782 out.append('{%s}%s' % (namespace, c))
1783 return '/'.join(out)
1784
1785 def _extract_smil_formats(self, smil_url, video_id, fatal=True, f4m_params=None, transform_source=None):
1786 smil = self._download_smil(smil_url, video_id, fatal=fatal, transform_source=transform_source)
1787
1788 if smil is False:
1789 assert not fatal
1790 return []
1791
1792 namespace = self._parse_smil_namespace(smil)
1793
1794 return self._parse_smil_formats(
1795 smil, smil_url, video_id, namespace=namespace, f4m_params=f4m_params)
1796
1797 def _extract_smil_info(self, smil_url, video_id, fatal=True, f4m_params=None):
1798 smil = self._download_smil(smil_url, video_id, fatal=fatal)
1799 if smil is False:
1800 return {}
1801 return self._parse_smil(smil, smil_url, video_id, f4m_params=f4m_params)
1802
1803 def _download_smil(self, smil_url, video_id, fatal=True, transform_source=None):
1804 return self._download_xml(
1805 smil_url, video_id, 'Downloading SMIL file',
1806 'Unable to download SMIL file', fatal=fatal, transform_source=transform_source)
1807
1808 def _parse_smil(self, smil, smil_url, video_id, f4m_params=None):
1809 namespace = self._parse_smil_namespace(smil)
1810
1811 formats = self._parse_smil_formats(
1812 smil, smil_url, video_id, namespace=namespace, f4m_params=f4m_params)
1813 subtitles = self._parse_smil_subtitles(smil, namespace=namespace)
1814
1815 video_id = os.path.splitext(url_basename(smil_url))[0]
1816 title = None
1817 description = None
1818 upload_date = None
1819 for meta in smil.findall(self._xpath_ns('./head/meta', namespace)):
1820 name = meta.attrib.get('name')
1821 content = meta.attrib.get('content')
1822 if not name or not content:
1823 continue
1824 if not title and name == 'title':
1825 title = content
1826 elif not description and name in ('description', 'abstract'):
1827 description = content
1828 elif not upload_date and name == 'date':
1829 upload_date = unified_strdate(content)
1830
1831 thumbnails = [{
1832 'id': image.get('type'),
1833 'url': image.get('src'),
1834 'width': int_or_none(image.get('width')),
1835 'height': int_or_none(image.get('height')),
1836 } for image in smil.findall(self._xpath_ns('.//image', namespace)) if image.get('src')]
1837
1838 return {
1839 'id': video_id,
1840 'title': title or video_id,
1841 'description': description,
1842 'upload_date': upload_date,
1843 'thumbnails': thumbnails,
1844 'formats': formats,
1845 'subtitles': subtitles,
1846 }
1847
1848 def _parse_smil_namespace(self, smil):
1849 return self._search_regex(
1850 r'(?i)^{([^}]+)?}smil$', smil.tag, 'namespace', default=None)
1851
1852 def _parse_smil_formats(self, smil, smil_url, video_id, namespace=None, f4m_params=None, transform_rtmp_url=None):
1853 base = smil_url
1854 for meta in smil.findall(self._xpath_ns('./head/meta', namespace)):
1855 b = meta.get('base') or meta.get('httpBase')
1856 if b:
1857 base = b
1858 break
1859
1860 formats = []
1861 rtmp_count = 0
1862 http_count = 0
1863 m3u8_count = 0
1864
1865 srcs = []
1866 media = smil.findall(self._xpath_ns('.//video', namespace)) + smil.findall(self._xpath_ns('.//audio', namespace))
1867 for medium in media:
1868 src = medium.get('src')
1869 if not src or src in srcs:
1870 continue
1871 srcs.append(src)
1872
1873 bitrate = float_or_none(medium.get('system-bitrate') or medium.get('systemBitrate'), 1000)
1874 filesize = int_or_none(medium.get('size') or medium.get('fileSize'))
1875 width = int_or_none(medium.get('width'))
1876 height = int_or_none(medium.get('height'))
1877 proto = medium.get('proto')
1878 ext = medium.get('ext')
1879 src_ext = determine_ext(src)
1880 streamer = medium.get('streamer') or base
1881
1882 if proto == 'rtmp' or streamer.startswith('rtmp'):
1883 rtmp_count += 1
1884 formats.append({
1885 'url': streamer,
1886 'play_path': src,
1887 'ext': 'flv',
1888 'format_id': 'rtmp-%d' % (rtmp_count if bitrate is None else bitrate),
1889 'tbr': bitrate,
1890 'filesize': filesize,
1891 'width': width,
1892 'height': height,
1893 })
1894 if transform_rtmp_url:
1895 streamer, src = transform_rtmp_url(streamer, src)
1896 formats[-1].update({
1897 'url': streamer,
1898 'play_path': src,
1899 })
1900 continue
1901
1902 src_url = src if src.startswith('http') else compat_urlparse.urljoin(base, src)
1903 src_url = src_url.strip()
1904
1905 if proto == 'm3u8' or src_ext == 'm3u8':
1906 m3u8_formats = self._extract_m3u8_formats(
1907 src_url, video_id, ext or 'mp4', m3u8_id='hls', fatal=False)
1908 if len(m3u8_formats) == 1:
1909 m3u8_count += 1
1910 m3u8_formats[0].update({
1911 'format_id': 'hls-%d' % (m3u8_count if bitrate is None else bitrate),
1912 'tbr': bitrate,
1913 'width': width,
1914 'height': height,
1915 })
1916 formats.extend(m3u8_formats)
1917 elif src_ext == 'f4m':
1918 f4m_url = src_url
1919 if not f4m_params:
1920 f4m_params = {
1921 'hdcore': '3.2.0',
1922 'plugin': 'flowplayer-3.2.0.1',
1923 }
1924 f4m_url += '&' if '?' in f4m_url else '?'
1925 f4m_url += compat_urllib_parse_urlencode(f4m_params)
1926 formats.extend(self._extract_f4m_formats(f4m_url, video_id, f4m_id='hds', fatal=False))
1927 elif src_ext == 'mpd':
1928 formats.extend(self._extract_mpd_formats(
1929 src_url, video_id, mpd_id='dash', fatal=False))
1930 elif re.search(r'\.ism/[Mm]anifest', src_url):
1931 formats.extend(self._extract_ism_formats(
1932 src_url, video_id, ism_id='mss', fatal=False))
1933 elif src_url.startswith('http') and self._is_valid_url(src, video_id):
1934 http_count += 1
1935 formats.append({
1936 'url': src_url,
1937 'ext': ext or src_ext or 'flv',
1938 'format_id': 'http-%d' % (bitrate or http_count),
1939 'tbr': bitrate,
1940 'filesize': filesize,
1941 'width': width,
1942 'height': height,
1943 })
1944
1945 return formats
1946
1947 def _parse_smil_subtitles(self, smil, namespace=None, subtitles_lang='en'):
1948 urls = []
1949 subtitles = {}
1950 for num, textstream in enumerate(smil.findall(self._xpath_ns('.//textstream', namespace))):
1951 src = textstream.get('src')
1952 if not src or src in urls:
1953 continue
1954 urls.append(src)
1955 ext = textstream.get('ext') or mimetype2ext(textstream.get('type')) or determine_ext(src)
1956 lang = textstream.get('systemLanguage') or textstream.get('systemLanguageName') or textstream.get('lang') or subtitles_lang
1957 subtitles.setdefault(lang, []).append({
1958 'url': src,
1959 'ext': ext,
1960 })
1961 return subtitles
1962
1963 def _extract_xspf_playlist(self, xspf_url, playlist_id, fatal=True):
1964 xspf = self._download_xml(
1965 xspf_url, playlist_id, 'Downloading xpsf playlist',
1966 'Unable to download xspf manifest', fatal=fatal)
1967 if xspf is False:
1968 return []
1969 return self._parse_xspf(
1970 xspf, playlist_id, xspf_url=xspf_url,
1971 xspf_base_url=base_url(xspf_url))
1972
1973 def _parse_xspf(self, xspf_doc, playlist_id, xspf_url=None, xspf_base_url=None):
1974 NS_MAP = {
1975 'xspf': 'http://xspf.org/ns/0/',
1976 's1': 'http://static.streamone.nl/player/ns/0',
1977 }
1978
1979 entries = []
1980 for track in xspf_doc.findall(xpath_with_ns('./xspf:trackList/xspf:track', NS_MAP)):
1981 title = xpath_text(
1982 track, xpath_with_ns('./xspf:title', NS_MAP), 'title', default=playlist_id)
1983 description = xpath_text(
1984 track, xpath_with_ns('./xspf:annotation', NS_MAP), 'description')
1985 thumbnail = xpath_text(
1986 track, xpath_with_ns('./xspf:image', NS_MAP), 'thumbnail')
1987 duration = float_or_none(
1988 xpath_text(track, xpath_with_ns('./xspf:duration', NS_MAP), 'duration'), 1000)
1989
1990 formats = []
1991 for location in track.findall(xpath_with_ns('./xspf:location', NS_MAP)):
1992 format_url = urljoin(xspf_base_url, location.text)
1993 if not format_url:
1994 continue
1995 formats.append({
1996 'url': format_url,
1997 'manifest_url': xspf_url,
1998 'format_id': location.get(xpath_with_ns('s1:label', NS_MAP)),
1999 'width': int_or_none(location.get(xpath_with_ns('s1:width', NS_MAP))),
2000 'height': int_or_none(location.get(xpath_with_ns('s1:height', NS_MAP))),
2001 })
2002 self._sort_formats(formats)
2003
2004 entries.append({
2005 'id': playlist_id,
2006 'title': title,
2007 'description': description,
2008 'thumbnail': thumbnail,
2009 'duration': duration,
2010 'formats': formats,
2011 })
2012 return entries
2013
2014 def _extract_mpd_formats(self, mpd_url, video_id, mpd_id=None, note=None, errnote=None, fatal=True, formats_dict={}):
2015 res = self._download_xml_handle(
2016 mpd_url, video_id,
2017 note=note or 'Downloading MPD manifest',
2018 errnote=errnote or 'Failed to download MPD manifest',
2019 fatal=fatal)
2020 if res is False:
2021 return []
2022 mpd_doc, urlh = res
2023 if mpd_doc is None:
2024 return []
2025 mpd_base_url = base_url(urlh.geturl())
2026
2027 return self._parse_mpd_formats(
2028 mpd_doc, mpd_id=mpd_id, mpd_base_url=mpd_base_url,
2029 formats_dict=formats_dict, mpd_url=mpd_url)
2030
2031 def _parse_mpd_formats(self, mpd_doc, mpd_id=None, mpd_base_url='', formats_dict={}, mpd_url=None):
2032 """
2033 Parse formats from MPD manifest.
2034 References:
2035 1. MPEG-DASH Standard, ISO/IEC 23009-1:2014(E),
2036 http://standards.iso.org/ittf/PubliclyAvailableStandards/c065274_ISO_IEC_23009-1_2014.zip
2037 2. https://en.wikipedia.org/wiki/Dynamic_Adaptive_Streaming_over_HTTP
2038 """
2039 if mpd_doc.get('type') == 'dynamic':
2040 return []
2041
2042 namespace = self._search_regex(r'(?i)^{([^}]+)?}MPD$', mpd_doc.tag, 'namespace', default=None)
2043
2044 def _add_ns(path):
2045 return self._xpath_ns(path, namespace)
2046
2047 def is_drm_protected(element):
2048 return element.find(_add_ns('ContentProtection')) is not None
2049
2050 def extract_multisegment_info(element, ms_parent_info):
2051 ms_info = ms_parent_info.copy()
2052
2053 # As per [1, 5.3.9.2.2] SegmentList and SegmentTemplate share some
2054 # common attributes and elements. We will only extract relevant
2055 # for us.
2056 def extract_common(source):
2057 segment_timeline = source.find(_add_ns('SegmentTimeline'))
2058 if segment_timeline is not None:
2059 s_e = segment_timeline.findall(_add_ns('S'))
2060 if s_e:
2061 ms_info['total_number'] = 0
2062 ms_info['s'] = []
2063 for s in s_e:
2064 r = int(s.get('r', 0))
2065 ms_info['total_number'] += 1 + r
2066 ms_info['s'].append({
2067 't': int(s.get('t', 0)),
2068 # @d is mandatory (see [1, 5.3.9.6.2, Table 17, page 60])
2069 'd': int(s.attrib['d']),
2070 'r': r,
2071 })
2072 start_number = source.get('startNumber')
2073 if start_number:
2074 ms_info['start_number'] = int(start_number)
2075 timescale = source.get('timescale')
2076 if timescale:
2077 ms_info['timescale'] = int(timescale)
2078 segment_duration = source.get('duration')
2079 if segment_duration:
2080 ms_info['segment_duration'] = float(segment_duration)
2081
2082 def extract_Initialization(source):
2083 initialization = source.find(_add_ns('Initialization'))
2084 if initialization is not None:
2085 ms_info['initialization_url'] = initialization.attrib['sourceURL']
2086
2087 segment_list = element.find(_add_ns('SegmentList'))
2088 if segment_list is not None:
2089 extract_common(segment_list)
2090 extract_Initialization(segment_list)
2091 segment_urls_e = segment_list.findall(_add_ns('SegmentURL'))
2092 if segment_urls_e:
2093 ms_info['segment_urls'] = [segment.attrib['media'] for segment in segment_urls_e]
2094 else:
2095 segment_template = element.find(_add_ns('SegmentTemplate'))
2096 if segment_template is not None:
2097 extract_common(segment_template)
2098 media = segment_template.get('media')
2099 if media:
2100 ms_info['media'] = media
2101 initialization = segment_template.get('initialization')
2102 if initialization:
2103 ms_info['initialization'] = initialization
2104 else:
2105 extract_Initialization(segment_template)
2106 return ms_info
2107
2108 mpd_duration = parse_duration(mpd_doc.get('mediaPresentationDuration'))
2109 formats = []
2110 for period in mpd_doc.findall(_add_ns('Period')):
2111 period_duration = parse_duration(period.get('duration')) or mpd_duration
2112 period_ms_info = extract_multisegment_info(period, {
2113 'start_number': 1,
2114 'timescale': 1,
2115 })
2116 for adaptation_set in period.findall(_add_ns('AdaptationSet')):
2117 if is_drm_protected(adaptation_set):
2118 continue
2119 adaption_set_ms_info = extract_multisegment_info(adaptation_set, period_ms_info)
2120 for representation in adaptation_set.findall(_add_ns('Representation')):
2121 if is_drm_protected(representation):
2122 continue
2123 representation_attrib = adaptation_set.attrib.copy()
2124 representation_attrib.update(representation.attrib)
2125 # According to [1, 5.3.7.2, Table 9, page 41], @mimeType is mandatory
2126 mime_type = representation_attrib['mimeType']
2127 content_type = mime_type.split('/')[0]
2128 if content_type == 'text':
2129 # TODO implement WebVTT downloading
2130 pass
2131 elif content_type in ('video', 'audio'):
2132 base_url = ''
2133 for element in (representation, adaptation_set, period, mpd_doc):
2134 base_url_e = element.find(_add_ns('BaseURL'))
2135 if base_url_e is not None:
2136 base_url = base_url_e.text + base_url
2137 if re.match(r'^https?://', base_url):
2138 break
2139 if mpd_base_url and not re.match(r'^https?://', base_url):
2140 if not mpd_base_url.endswith('/') and not base_url.startswith('/'):
2141 mpd_base_url += '/'
2142 base_url = mpd_base_url + base_url
2143 representation_id = representation_attrib.get('id')
2144 lang = representation_attrib.get('lang')
2145 url_el = representation.find(_add_ns('BaseURL'))
2146 filesize = int_or_none(url_el.attrib.get('{http://youtube.com/yt/2012/10/10}contentLength') if url_el is not None else None)
2147 bandwidth = int_or_none(representation_attrib.get('bandwidth'))
2148 f = {
2149 'format_id': '%s-%s' % (mpd_id, representation_id) if mpd_id else representation_id,
2150 'manifest_url': mpd_url,
2151 'ext': mimetype2ext(mime_type),
2152 'width': int_or_none(representation_attrib.get('width')),
2153 'height': int_or_none(representation_attrib.get('height')),
2154 'tbr': float_or_none(bandwidth, 1000),
2155 'asr': int_or_none(representation_attrib.get('audioSamplingRate')),
2156 'fps': int_or_none(representation_attrib.get('frameRate')),
2157 'language': lang if lang not in ('mul', 'und', 'zxx', 'mis') else None,
2158 'format_note': 'DASH %s' % content_type,
2159 'filesize': filesize,
2160 'container': mimetype2ext(mime_type) + '_dash',
2161 }
2162 f.update(parse_codecs(representation_attrib.get('codecs')))
2163 representation_ms_info = extract_multisegment_info(representation, adaption_set_ms_info)
2164
2165 def prepare_template(template_name, identifiers):
2166 tmpl = representation_ms_info[template_name]
2167 # First of, % characters outside $...$ templates
2168 # must be escaped by doubling for proper processing
2169 # by % operator string formatting used further (see
2170 # https://github.com/ytdl-org/youtube-dl/issues/16867).
2171 t = ''
2172 in_template = False
2173 for c in tmpl:
2174 t += c
2175 if c == '$':
2176 in_template = not in_template
2177 elif c == '%' and not in_template:
2178 t += c
2179 # Next, $...$ templates are translated to their
2180 # %(...) counterparts to be used with % operator
2181 t = t.replace('$RepresentationID$', representation_id)
2182 t = re.sub(r'\$(%s)\$' % '|'.join(identifiers), r'%(\1)d', t)
2183 t = re.sub(r'\$(%s)%%([^$]+)\$' % '|'.join(identifiers), r'%(\1)\2', t)
2184 t.replace('$$', '$')
2185 return t
2186
2187 # @initialization is a regular template like @media one
2188 # so it should be handled just the same way (see
2189 # https://github.com/ytdl-org/youtube-dl/issues/11605)
2190 if 'initialization' in representation_ms_info:
2191 initialization_template = prepare_template(
2192 'initialization',
2193 # As per [1, 5.3.9.4.2, Table 15, page 54] $Number$ and
2194 # $Time$ shall not be included for @initialization thus
2195 # only $Bandwidth$ remains
2196 ('Bandwidth', ))
2197 representation_ms_info['initialization_url'] = initialization_template % {
2198 'Bandwidth': bandwidth,
2199 }
2200
2201 def location_key(location):
2202 return 'url' if re.match(r'^https?://', location) else 'path'
2203
2204 if 'segment_urls' not in representation_ms_info and 'media' in representation_ms_info:
2205
2206 media_template = prepare_template('media', ('Number', 'Bandwidth', 'Time'))
2207 media_location_key = location_key(media_template)
2208
2209 # As per [1, 5.3.9.4.4, Table 16, page 55] $Number$ and $Time$
2210 # can't be used at the same time
2211 if '%(Number' in media_template and 's' not in representation_ms_info:
2212 segment_duration = None
2213 if 'total_number' not in representation_ms_info and 'segment_duration' in representation_ms_info:
2214 segment_duration = float_or_none(representation_ms_info['segment_duration'], representation_ms_info['timescale'])
2215 representation_ms_info['total_number'] = int(math.ceil(float(period_duration) / segment_duration))
2216 representation_ms_info['fragments'] = [{
2217 media_location_key: media_template % {
2218 'Number': segment_number,
2219 'Bandwidth': bandwidth,
2220 },
2221 'duration': segment_duration,
2222 } for segment_number in range(
2223 representation_ms_info['start_number'],
2224 representation_ms_info['total_number'] + representation_ms_info['start_number'])]
2225 else:
2226 # $Number*$ or $Time$ in media template with S list available
2227 # Example $Number*$: http://www.svtplay.se/klipp/9023742/stopptid-om-bjorn-borg
2228 # Example $Time$: https://play.arkena.com/embed/avp/v2/player/media/b41dda37-d8e7-4d3f-b1b5-9a9db578bdfe/1/129411
2229 representation_ms_info['fragments'] = []
2230 segment_time = 0
2231 segment_d = None
2232 segment_number = representation_ms_info['start_number']
2233
2234 def add_segment_url():
2235 segment_url = media_template % {
2236 'Time': segment_time,
2237 'Bandwidth': bandwidth,
2238 'Number': segment_number,
2239 }
2240 representation_ms_info['fragments'].append({
2241 media_location_key: segment_url,
2242 'duration': float_or_none(segment_d, representation_ms_info['timescale']),
2243 })
2244
2245 for num, s in enumerate(representation_ms_info['s']):
2246 segment_time = s.get('t') or segment_time
2247 segment_d = s['d']
2248 add_segment_url()
2249 segment_number += 1
2250 for r in range(s.get('r', 0)):
2251 segment_time += segment_d
2252 add_segment_url()
2253 segment_number += 1
2254 segment_time += segment_d
2255 elif 'segment_urls' in representation_ms_info and 's' in representation_ms_info:
2256 # No media template
2257 # Example: https://www.youtube.com/watch?v=iXZV5uAYMJI
2258 # or any YouTube dashsegments video
2259 fragments = []
2260 segment_index = 0
2261 timescale = representation_ms_info['timescale']
2262 for s in representation_ms_info['s']:
2263 duration = float_or_none(s['d'], timescale)
2264 for r in range(s.get('r', 0) + 1):
2265 segment_uri = representation_ms_info['segment_urls'][segment_index]
2266 fragments.append({
2267 location_key(segment_uri): segment_uri,
2268 'duration': duration,
2269 })
2270 segment_index += 1
2271 representation_ms_info['fragments'] = fragments
2272 elif 'segment_urls' in representation_ms_info:
2273 # Segment URLs with no SegmentTimeline
2274 # Example: https://www.seznam.cz/zpravy/clanek/cesko-zasahne-vitr-o-sile-vichrice-muze-byt-i-zivotu-nebezpecny-39091
2275 # https://github.com/ytdl-org/youtube-dl/pull/14844
2276 fragments = []
2277 segment_duration = float_or_none(
2278 representation_ms_info['segment_duration'],
2279 representation_ms_info['timescale']) if 'segment_duration' in representation_ms_info else None
2280 for segment_url in representation_ms_info['segment_urls']:
2281 fragment = {
2282 location_key(segment_url): segment_url,
2283 }
2284 if segment_duration:
2285 fragment['duration'] = segment_duration
2286 fragments.append(fragment)
2287 representation_ms_info['fragments'] = fragments
2288 # If there is a fragments key available then we correctly recognized fragmented media.
2289 # Otherwise we will assume unfragmented media with direct access. Technically, such
2290 # assumption is not necessarily correct since we may simply have no support for
2291 # some forms of fragmented media renditions yet, but for now we'll use this fallback.
2292 if 'fragments' in representation_ms_info:
2293 f.update({
2294 # NB: mpd_url may be empty when MPD manifest is parsed from a string
2295 'url': mpd_url or base_url,
2296 'fragment_base_url': base_url,
2297 'fragments': [],
2298 'protocol': 'http_dash_segments',
2299 })
2300 if 'initialization_url' in representation_ms_info:
2301 initialization_url = representation_ms_info['initialization_url']
2302 if not f.get('url'):
2303 f['url'] = initialization_url
2304 f['fragments'].append({location_key(initialization_url): initialization_url})
2305 f['fragments'].extend(representation_ms_info['fragments'])
2306 else:
2307 # Assuming direct URL to unfragmented media.
2308 f['url'] = base_url
2309
2310 # According to [1, 5.3.5.2, Table 7, page 35] @id of Representation
2311 # is not necessarily unique within a Period thus formats with
2312 # the same `format_id` are quite possible. There are numerous examples
2313 # of such manifests (see https://github.com/ytdl-org/youtube-dl/issues/15111,
2314 # https://github.com/ytdl-org/youtube-dl/issues/13919)
2315 full_info = formats_dict.get(representation_id, {}).copy()
2316 full_info.update(f)
2317 formats.append(full_info)
2318 else:
2319 self.report_warning('Unknown MIME type %s in DASH manifest' % mime_type)
2320 return formats
2321
2322 def _extract_ism_formats(self, ism_url, video_id, ism_id=None, note=None, errnote=None, fatal=True):
2323 res = self._download_xml_handle(
2324 ism_url, video_id,
2325 note=note or 'Downloading ISM manifest',
2326 errnote=errnote or 'Failed to download ISM manifest',
2327 fatal=fatal)
2328 if res is False:
2329 return []
2330 ism_doc, urlh = res
2331
2332 return self._parse_ism_formats(ism_doc, urlh.geturl(), ism_id)
2333
2334 def _parse_ism_formats(self, ism_doc, ism_url, ism_id=None):
2335 """
2336 Parse formats from ISM manifest.
2337 References:
2338 1. [MS-SSTR]: Smooth Streaming Protocol,
2339 https://msdn.microsoft.com/en-us/library/ff469518.aspx
2340 """
2341 if ism_doc.get('IsLive') == 'TRUE' or ism_doc.find('Protection') is not None:
2342 return []
2343
2344 duration = int(ism_doc.attrib['Duration'])
2345 timescale = int_or_none(ism_doc.get('TimeScale')) or 10000000
2346
2347 formats = []
2348 for stream in ism_doc.findall('StreamIndex'):
2349 stream_type = stream.get('Type')
2350 if stream_type not in ('video', 'audio'):
2351 continue
2352 url_pattern = stream.attrib['Url']
2353 stream_timescale = int_or_none(stream.get('TimeScale')) or timescale
2354 stream_name = stream.get('Name')
2355 for track in stream.findall('QualityLevel'):
2356 fourcc = track.get('FourCC', 'AACL' if track.get('AudioTag') == '255' else None)
2357 # TODO: add support for WVC1 and WMAP
2358 if fourcc not in ('H264', 'AVC1', 'AACL'):
2359 self.report_warning('%s is not a supported codec' % fourcc)
2360 continue
2361 tbr = int(track.attrib['Bitrate']) // 1000
2362 # [1] does not mention Width and Height attributes. However,
2363 # they're often present while MaxWidth and MaxHeight are
2364 # missing, so should be used as fallbacks
2365 width = int_or_none(track.get('MaxWidth') or track.get('Width'))
2366 height = int_or_none(track.get('MaxHeight') or track.get('Height'))
2367 sampling_rate = int_or_none(track.get('SamplingRate'))
2368
2369 track_url_pattern = re.sub(r'{[Bb]itrate}', track.attrib['Bitrate'], url_pattern)
2370 track_url_pattern = compat_urlparse.urljoin(ism_url, track_url_pattern)
2371
2372 fragments = []
2373 fragment_ctx = {
2374 'time': 0,
2375 }
2376 stream_fragments = stream.findall('c')
2377 for stream_fragment_index, stream_fragment in enumerate(stream_fragments):
2378 fragment_ctx['time'] = int_or_none(stream_fragment.get('t')) or fragment_ctx['time']
2379 fragment_repeat = int_or_none(stream_fragment.get('r')) or 1
2380 fragment_ctx['duration'] = int_or_none(stream_fragment.get('d'))
2381 if not fragment_ctx['duration']:
2382 try:
2383 next_fragment_time = int(stream_fragment[stream_fragment_index + 1].attrib['t'])
2384 except IndexError:
2385 next_fragment_time = duration
2386 fragment_ctx['duration'] = (next_fragment_time - fragment_ctx['time']) / fragment_repeat
2387 for _ in range(fragment_repeat):
2388 fragments.append({
2389 'url': re.sub(r'{start[ _]time}', compat_str(fragment_ctx['time']), track_url_pattern),
2390 'duration': fragment_ctx['duration'] / stream_timescale,
2391 })
2392 fragment_ctx['time'] += fragment_ctx['duration']
2393
2394 format_id = []
2395 if ism_id:
2396 format_id.append(ism_id)
2397 if stream_name:
2398 format_id.append(stream_name)
2399 format_id.append(compat_str(tbr))
2400
2401 formats.append({
2402 'format_id': '-'.join(format_id),
2403 'url': ism_url,
2404 'manifest_url': ism_url,
2405 'ext': 'ismv' if stream_type == 'video' else 'isma',
2406 'width': width,
2407 'height': height,
2408 'tbr': tbr,
2409 'asr': sampling_rate,
2410 'vcodec': 'none' if stream_type == 'audio' else fourcc,
2411 'acodec': 'none' if stream_type == 'video' else fourcc,
2412 'protocol': 'ism',
2413 'fragments': fragments,
2414 '_download_params': {
2415 'duration': duration,
2416 'timescale': stream_timescale,
2417 'width': width or 0,
2418 'height': height or 0,
2419 'fourcc': fourcc,
2420 'codec_private_data': track.get('CodecPrivateData'),
2421 'sampling_rate': sampling_rate,
2422 'channels': int_or_none(track.get('Channels', 2)),
2423 'bits_per_sample': int_or_none(track.get('BitsPerSample', 16)),
2424 'nal_unit_length_field': int_or_none(track.get('NALUnitLengthField', 4)),
2425 },
2426 })
2427 return formats
2428
2429 def _parse_html5_media_entries(self, base_url, webpage, video_id, m3u8_id=None, m3u8_entry_protocol='m3u8', mpd_id=None, preference=None):
2430 def absolute_url(item_url):
2431 return urljoin(base_url, item_url)
2432
2433 def parse_content_type(content_type):
2434 if not content_type:
2435 return {}
2436 ctr = re.search(r'(?P<mimetype>[^/]+/[^;]+)(?:;\s*codecs="?(?P<codecs>[^"]+))?', content_type)
2437 if ctr:
2438 mimetype, codecs = ctr.groups()
2439 f = parse_codecs(codecs)
2440 f['ext'] = mimetype2ext(mimetype)
2441 return f
2442 return {}
2443
2444 def _media_formats(src, cur_media_type, type_info={}):
2445 full_url = absolute_url(src)
2446 ext = type_info.get('ext') or determine_ext(full_url)
2447 if ext == 'm3u8':
2448 is_plain_url = False
2449 formats = self._extract_m3u8_formats(
2450 full_url, video_id, ext='mp4',
2451 entry_protocol=m3u8_entry_protocol, m3u8_id=m3u8_id,
2452 preference=preference, fatal=False)
2453 elif ext == 'mpd':
2454 is_plain_url = False
2455 formats = self._extract_mpd_formats(
2456 full_url, video_id, mpd_id=mpd_id, fatal=False)
2457 else:
2458 is_plain_url = True
2459 formats = [{
2460 'url': full_url,
2461 'vcodec': 'none' if cur_media_type == 'audio' else None,
2462 }]
2463 return is_plain_url, formats
2464
2465 entries = []
2466 # amp-video and amp-audio are very similar to their HTML5 counterparts
2467 # so we wll include them right here (see
2468 # https://www.ampproject.org/docs/reference/components/amp-video)
2469 media_tags = [(media_tag, media_type, '')
2470 for media_tag, media_type
2471 in re.findall(r'(?s)(<(?:amp-)?(video|audio)[^>]*/>)', webpage)]
2472 media_tags.extend(re.findall(
2473 # We only allow video|audio followed by a whitespace or '>'.
2474 # Allowing more characters may end up in significant slow down (see
2475 # https://github.com/ytdl-org/youtube-dl/issues/11979, example URL:
2476 # http://www.porntrex.com/maps/videositemap.xml).
2477 r'(?s)(<(?P<tag>(?:amp-)?(?:video|audio))(?:\s+[^>]*)?>)(.*?)</(?P=tag)>', webpage))
2478 for media_tag, media_type, media_content in media_tags:
2479 media_info = {
2480 'formats': [],
2481 'subtitles': {},
2482 }
2483 media_attributes = extract_attributes(media_tag)
2484 src = strip_or_none(media_attributes.get('src'))
2485 if src:
2486 _, formats = _media_formats(src, media_type)
2487 media_info['formats'].extend(formats)
2488 media_info['thumbnail'] = absolute_url(media_attributes.get('poster'))
2489 if media_content:
2490 for source_tag in re.findall(r'<source[^>]+>', media_content):
2491 s_attr = extract_attributes(source_tag)
2492 # data-video-src and data-src are non standard but seen
2493 # several times in the wild
2494 src = strip_or_none(dict_get(s_attr, ('src', 'data-video-src', 'data-src')))
2495 if not src:
2496 continue
2497 f = parse_content_type(s_attr.get('type'))
2498 is_plain_url, formats = _media_formats(src, media_type, f)
2499 if is_plain_url:
2500 # width, height, res, label and title attributes are
2501 # all not standard but seen several times in the wild
2502 labels = [
2503 s_attr.get(lbl)
2504 for lbl in ('label', 'title')
2505 if str_or_none(s_attr.get(lbl))
2506 ]
2507 width = int_or_none(s_attr.get('width'))
2508 height = (int_or_none(s_attr.get('height'))
2509 or int_or_none(s_attr.get('res')))
2510 if not width or not height:
2511 for lbl in labels:
2512 resolution = parse_resolution(lbl)
2513 if not resolution:
2514 continue
2515 width = width or resolution.get('width')
2516 height = height or resolution.get('height')
2517 for lbl in labels:
2518 tbr = parse_bitrate(lbl)
2519 if tbr:
2520 break
2521 else:
2522 tbr = None
2523 f.update({
2524 'width': width,
2525 'height': height,
2526 'tbr': tbr,
2527 'format_id': s_attr.get('label') or s_attr.get('title'),
2528 })
2529 f.update(formats[0])
2530 media_info['formats'].append(f)
2531 else:
2532 media_info['formats'].extend(formats)
2533 for track_tag in re.findall(r'<track[^>]+>', media_content):
2534 track_attributes = extract_attributes(track_tag)
2535 kind = track_attributes.get('kind')
2536 if not kind or kind in ('subtitles', 'captions'):
2537 src = strip_or_none(track_attributes.get('src'))
2538 if not src:
2539 continue
2540 lang = track_attributes.get('srclang') or track_attributes.get('lang') or track_attributes.get('label')
2541 media_info['subtitles'].setdefault(lang, []).append({
2542 'url': absolute_url(src),
2543 })
2544 for f in media_info['formats']:
2545 f.setdefault('http_headers', {})['Referer'] = base_url
2546 if media_info['formats'] or media_info['subtitles']:
2547 entries.append(media_info)
2548 return entries
2549
2550 def _extract_akamai_formats(self, manifest_url, video_id, hosts={}):
2551 formats = []
2552 hdcore_sign = 'hdcore=3.7.0'
2553 f4m_url = re.sub(r'(https?://[^/]+)/i/', r'\1/z/', manifest_url).replace('/master.m3u8', '/manifest.f4m')
2554 hds_host = hosts.get('hds')
2555 if hds_host:
2556 f4m_url = re.sub(r'(https?://)[^/]+', r'\1' + hds_host, f4m_url)
2557 if 'hdcore=' not in f4m_url:
2558 f4m_url += ('&' if '?' in f4m_url else '?') + hdcore_sign
2559 f4m_formats = self._extract_f4m_formats(
2560 f4m_url, video_id, f4m_id='hds', fatal=False)
2561 for entry in f4m_formats:
2562 entry.update({'extra_param_to_segment_url': hdcore_sign})
2563 formats.extend(f4m_formats)
2564 m3u8_url = re.sub(r'(https?://[^/]+)/z/', r'\1/i/', manifest_url).replace('/manifest.f4m', '/master.m3u8')
2565 hls_host = hosts.get('hls')
2566 if hls_host:
2567 m3u8_url = re.sub(r'(https?://)[^/]+', r'\1' + hls_host, m3u8_url)
2568 formats.extend(self._extract_m3u8_formats(
2569 m3u8_url, video_id, 'mp4', 'm3u8_native',
2570 m3u8_id='hls', fatal=False))
2571 return formats
2572
2573 def _extract_wowza_formats(self, url, video_id, m3u8_entry_protocol='m3u8_native', skip_protocols=[]):
2574 query = compat_urlparse.urlparse(url).query
2575 url = re.sub(r'/(?:manifest|playlist|jwplayer)\.(?:m3u8|f4m|mpd|smil)', '', url)
2576 mobj = re.search(
2577 r'(?:(?:http|rtmp|rtsp)(?P<s>s)?:)?(?P<url>//[^?]+)', url)
2578 url_base = mobj.group('url')
2579 http_base_url = '%s%s:%s' % ('http', mobj.group('s') or '', url_base)
2580 formats = []
2581
2582 def manifest_url(manifest):
2583 m_url = '%s/%s' % (http_base_url, manifest)
2584 if query:
2585 m_url += '?%s' % query
2586 return m_url
2587
2588 if 'm3u8' not in skip_protocols:
2589 formats.extend(self._extract_m3u8_formats(
2590 manifest_url('playlist.m3u8'), video_id, 'mp4',
2591 m3u8_entry_protocol, m3u8_id='hls', fatal=False))
2592 if 'f4m' not in skip_protocols:
2593 formats.extend(self._extract_f4m_formats(
2594 manifest_url('manifest.f4m'),
2595 video_id, f4m_id='hds', fatal=False))
2596 if 'dash' not in skip_protocols:
2597 formats.extend(self._extract_mpd_formats(
2598 manifest_url('manifest.mpd'),
2599 video_id, mpd_id='dash', fatal=False))
2600 if re.search(r'(?:/smil:|\.smil)', url_base):
2601 if 'smil' not in skip_protocols:
2602 rtmp_formats = self._extract_smil_formats(
2603 manifest_url('jwplayer.smil'),
2604 video_id, fatal=False)
2605 for rtmp_format in rtmp_formats:
2606 rtsp_format = rtmp_format.copy()
2607 rtsp_format['url'] = '%s/%s' % (rtmp_format['url'], rtmp_format['play_path'])
2608 del rtsp_format['play_path']
2609 del rtsp_format['ext']
2610 rtsp_format.update({
2611 'url': rtsp_format['url'].replace('rtmp://', 'rtsp://'),
2612 'format_id': rtmp_format['format_id'].replace('rtmp', 'rtsp'),
2613 'protocol': 'rtsp',
2614 })
2615 formats.extend([rtmp_format, rtsp_format])
2616 else:
2617 for protocol in ('rtmp', 'rtsp'):
2618 if protocol not in skip_protocols:
2619 formats.append({
2620 'url': '%s:%s' % (protocol, url_base),
2621 'format_id': protocol,
2622 'protocol': protocol,
2623 })
2624 return formats
2625
2626 def _find_jwplayer_data(self, webpage, video_id=None, transform_source=js_to_json):
2627 mobj = re.search(
2628 r'(?s)jwplayer\((?P<quote>[\'"])[^\'" ]+(?P=quote)\)(?!</script>).*?\.setup\s*\((?P<options>[^)]+)\)',
2629 webpage)
2630 if mobj:
2631 try:
2632 jwplayer_data = self._parse_json(mobj.group('options'),
2633 video_id=video_id,
2634 transform_source=transform_source)
2635 except ExtractorError:
2636 pass
2637 else:
2638 if isinstance(jwplayer_data, dict):
2639 return jwplayer_data
2640
2641 def _extract_jwplayer_data(self, webpage, video_id, *args, **kwargs):
2642 jwplayer_data = self._find_jwplayer_data(
2643 webpage, video_id, transform_source=js_to_json)
2644 return self._parse_jwplayer_data(
2645 jwplayer_data, video_id, *args, **kwargs)
2646
2647 def _parse_jwplayer_data(self, jwplayer_data, video_id=None, require_title=True,
2648 m3u8_id=None, mpd_id=None, rtmp_params=None, base_url=None):
2649 # JWPlayer backward compatibility: flattened playlists
2650 # https://github.com/jwplayer/jwplayer/blob/v7.4.3/src/js/api/config.js#L81-L96
2651 if 'playlist' not in jwplayer_data:
2652 jwplayer_data = {'playlist': [jwplayer_data]}
2653
2654 entries = []
2655
2656 # JWPlayer backward compatibility: single playlist item
2657 # https://github.com/jwplayer/jwplayer/blob/v7.7.0/src/js/playlist/playlist.js#L10
2658 if not isinstance(jwplayer_data['playlist'], list):
2659 jwplayer_data['playlist'] = [jwplayer_data['playlist']]
2660
2661 for video_data in jwplayer_data['playlist']:
2662 # JWPlayer backward compatibility: flattened sources
2663 # https://github.com/jwplayer/jwplayer/blob/v7.4.3/src/js/playlist/item.js#L29-L35
2664 if 'sources' not in video_data:
2665 video_data['sources'] = [video_data]
2666
2667 this_video_id = video_id or video_data['mediaid']
2668
2669 formats = self._parse_jwplayer_formats(
2670 video_data['sources'], video_id=this_video_id, m3u8_id=m3u8_id,
2671 mpd_id=mpd_id, rtmp_params=rtmp_params, base_url=base_url)
2672
2673 subtitles = {}
2674 tracks = video_data.get('tracks')
2675 if tracks and isinstance(tracks, list):
2676 for track in tracks:
2677 if not isinstance(track, dict):
2678 continue
2679 track_kind = track.get('kind')
2680 if not track_kind or not isinstance(track_kind, compat_str):
2681 continue
2682 if track_kind.lower() not in ('captions', 'subtitles'):
2683 continue
2684 track_url = urljoin(base_url, track.get('file'))
2685 if not track_url:
2686 continue
2687 subtitles.setdefault(track.get('label') or 'en', []).append({
2688 'url': self._proto_relative_url(track_url)
2689 })
2690
2691 entry = {
2692 'id': this_video_id,
2693 'title': unescapeHTML(video_data['title'] if require_title else video_data.get('title')),
2694 'description': video_data.get('description'),
2695 'thumbnail': urljoin(base_url, self._proto_relative_url(video_data.get('image'))),
2696 'timestamp': int_or_none(video_data.get('pubdate')),
2697 'duration': float_or_none(jwplayer_data.get('duration') or video_data.get('duration')),
2698 'subtitles': subtitles,
2699 }
2700 # https://github.com/jwplayer/jwplayer/blob/master/src/js/utils/validator.js#L32
2701 if len(formats) == 1 and re.search(r'^(?:http|//).*(?:youtube\.com|youtu\.be)/.+', formats[0]['url']):
2702 entry.update({
2703 '_type': 'url_transparent',
2704 'url': formats[0]['url'],
2705 })
2706 else:
2707 self._sort_formats(formats)
2708 entry['formats'] = formats
2709 entries.append(entry)
2710 if len(entries) == 1:
2711 return entries[0]
2712 else:
2713 return self.playlist_result(entries)
2714
2715 def _parse_jwplayer_formats(self, jwplayer_sources_data, video_id=None,
2716 m3u8_id=None, mpd_id=None, rtmp_params=None, base_url=None):
2717 urls = []
2718 formats = []
2719 for source in jwplayer_sources_data:
2720 if not isinstance(source, dict):
2721 continue
2722 source_url = urljoin(
2723 base_url, self._proto_relative_url(source.get('file')))
2724 if not source_url or source_url in urls:
2725 continue
2726 urls.append(source_url)
2727 source_type = source.get('type') or ''
2728 ext = mimetype2ext(source_type) or determine_ext(source_url)
2729 if source_type == 'hls' or ext == 'm3u8':
2730 formats.extend(self._extract_m3u8_formats(
2731 source_url, video_id, 'mp4', entry_protocol='m3u8_native',
2732 m3u8_id=m3u8_id, fatal=False))
2733 elif source_type == 'dash' or ext == 'mpd':
2734 formats.extend(self._extract_mpd_formats(
2735 source_url, video_id, mpd_id=mpd_id, fatal=False))
2736 elif ext == 'smil':
2737 formats.extend(self._extract_smil_formats(
2738 source_url, video_id, fatal=False))
2739 # https://github.com/jwplayer/jwplayer/blob/master/src/js/providers/default.js#L67
2740 elif source_type.startswith('audio') or ext in (
2741 'oga', 'aac', 'mp3', 'mpeg', 'vorbis'):
2742 formats.append({
2743 'url': source_url,
2744 'vcodec': 'none',
2745 'ext': ext,
2746 })
2747 else:
2748 height = int_or_none(source.get('height'))
2749 if height is None:
2750 # Often no height is provided but there is a label in
2751 # format like "1080p", "720p SD", or 1080.
2752 height = int_or_none(self._search_regex(
2753 r'^(\d{3,4})[pP]?(?:\b|$)', compat_str(source.get('label') or ''),
2754 'height', default=None))
2755 a_format = {
2756 'url': source_url,
2757 'width': int_or_none(source.get('width')),
2758 'height': height,
2759 'tbr': int_or_none(source.get('bitrate')),
2760 'ext': ext,
2761 }
2762 if source_url.startswith('rtmp'):
2763 a_format['ext'] = 'flv'
2764 # See com/longtailvideo/jwplayer/media/RTMPMediaProvider.as
2765 # of jwplayer.flash.swf
2766 rtmp_url_parts = re.split(
2767 r'((?:mp4|mp3|flv):)', source_url, 1)
2768 if len(rtmp_url_parts) == 3:
2769 rtmp_url, prefix, play_path = rtmp_url_parts
2770 a_format.update({
2771 'url': rtmp_url,
2772 'play_path': prefix + play_path,
2773 })
2774 if rtmp_params:
2775 a_format.update(rtmp_params)
2776 formats.append(a_format)
2777 return formats
2778
2779 def _live_title(self, name):
2780 """ Generate the title for a live video """
2781 now = datetime.datetime.now()
2782 now_str = now.strftime('%Y-%m-%d %H:%M')
2783 return name + ' ' + now_str
2784
2785 def _int(self, v, name, fatal=False, **kwargs):
2786 res = int_or_none(v, **kwargs)
2787 if 'get_attr' in kwargs:
2788 print(getattr(v, kwargs['get_attr']))
2789 if res is None:
2790 msg = 'Failed to extract %s: Could not parse value %r' % (name, v)
2791 if fatal:
2792 raise ExtractorError(msg)
2793 else:
2794 self._downloader.report_warning(msg)
2795 return res
2796
2797 def _float(self, v, name, fatal=False, **kwargs):
2798 res = float_or_none(v, **kwargs)
2799 if res is None:
2800 msg = 'Failed to extract %s: Could not parse value %r' % (name, v)
2801 if fatal:
2802 raise ExtractorError(msg)
2803 else:
2804 self._downloader.report_warning(msg)
2805 return res
2806
2807 def _set_cookie(self, domain, name, value, expire_time=None, port=None,
2808 path='/', secure=False, discard=False, rest={}, **kwargs):
2809 cookie = compat_cookiejar.Cookie(
2810 0, name, value, port, port is not None, domain, True,
2811 domain.startswith('.'), path, True, secure, expire_time,
2812 discard, None, None, rest)
2813 self._downloader.cookiejar.set_cookie(cookie)
2814
2815 def _get_cookies(self, url):
2816 """ Return a compat_cookies.SimpleCookie with the cookies for the url """
2817 req = sanitized_Request(url)
2818 self._downloader.cookiejar.add_cookie_header(req)
2819 return compat_cookies.SimpleCookie(req.get_header('Cookie'))
2820
2821 def _apply_first_set_cookie_header(self, url_handle, cookie):
2822 """
2823 Apply first Set-Cookie header instead of the last. Experimental.
2824
2825 Some sites (e.g. [1-3]) may serve two cookies under the same name
2826 in Set-Cookie header and expect the first (old) one to be set rather
2827 than second (new). However, as of RFC6265 the newer one cookie
2828 should be set into cookie store what actually happens.
2829 We will workaround this issue by resetting the cookie to
2830 the first one manually.
2831 1. https://new.vk.com/
2832 2. https://github.com/ytdl-org/youtube-dl/issues/9841#issuecomment-227871201
2833 3. https://learning.oreilly.com/
2834 """
2835 for header, cookies in url_handle.headers.items():
2836 if header.lower() != 'set-cookie':
2837 continue
2838 if sys.version_info[0] >= 3:
2839 cookies = cookies.encode('iso-8859-1')
2840 cookies = cookies.decode('utf-8')
2841 cookie_value = re.search(
2842 r'%s=(.+?);.*?\b[Dd]omain=(.+?)(?:[,;]|$)' % cookie, cookies)
2843 if cookie_value:
2844 value, domain = cookie_value.groups()
2845 self._set_cookie(domain, cookie, value)
2846 break
2847
2848 def get_testcases(self, include_onlymatching=False):
2849 t = getattr(self, '_TEST', None)
2850 if t:
2851 assert not hasattr(self, '_TESTS'), \
2852 '%s has _TEST and _TESTS' % type(self).__name__
2853 tests = [t]
2854 else:
2855 tests = getattr(self, '_TESTS', [])
2856 for t in tests:
2857 if not include_onlymatching and t.get('only_matching', False):
2858 continue
2859 t['name'] = type(self).__name__[:-len('IE')]
2860 yield t
2861
2862 def is_suitable(self, age_limit):
2863 """ Test whether the extractor is generally suitable for the given
2864 age limit (i.e. pornographic sites are not, all others usually are) """
2865
2866 any_restricted = False
2867 for tc in self.get_testcases(include_onlymatching=False):
2868 if tc.get('playlist', []):
2869 tc = tc['playlist'][0]
2870 is_restricted = age_restricted(
2871 tc.get('info_dict', {}).get('age_limit'), age_limit)
2872 if not is_restricted:
2873 return True
2874 any_restricted = any_restricted or is_restricted
2875 return not any_restricted
2876
2877 def extract_subtitles(self, *args, **kwargs):
2878 if (self._downloader.params.get('writesubtitles', False)
2879 or self._downloader.params.get('listsubtitles')):
2880 return self._get_subtitles(*args, **kwargs)
2881 return {}
2882
2883 def _get_subtitles(self, *args, **kwargs):
2884 raise NotImplementedError('This method must be implemented by subclasses')
2885
2886 @staticmethod
2887 def _merge_subtitle_items(subtitle_list1, subtitle_list2):
2888 """ Merge subtitle items for one language. Items with duplicated URLs
2889 will be dropped. """
2890 list1_urls = set([item['url'] for item in subtitle_list1])
2891 ret = list(subtitle_list1)
2892 ret.extend([item for item in subtitle_list2 if item['url'] not in list1_urls])
2893 return ret
2894
2895 @classmethod
2896 def _merge_subtitles(cls, subtitle_dict1, subtitle_dict2):
2897 """ Merge two subtitle dictionaries, language by language. """
2898 ret = dict(subtitle_dict1)
2899 for lang in subtitle_dict2:
2900 ret[lang] = cls._merge_subtitle_items(subtitle_dict1.get(lang, []), subtitle_dict2[lang])
2901 return ret
2902
2903 def extract_automatic_captions(self, *args, **kwargs):
2904 if (self._downloader.params.get('writeautomaticsub', False)
2905 or self._downloader.params.get('listsubtitles')):
2906 return self._get_automatic_captions(*args, **kwargs)
2907 return {}
2908
2909 def _get_automatic_captions(self, *args, **kwargs):
2910 raise NotImplementedError('This method must be implemented by subclasses')
2911
2912 def mark_watched(self, *args, **kwargs):
2913 if (self._downloader.params.get('mark_watched', False)
2914 and (self._get_login_info()[0] is not None
2915 or self._downloader.params.get('cookiefile') is not None)):
2916 self._mark_watched(*args, **kwargs)
2917
2918 def _mark_watched(self, *args, **kwargs):
2919 raise NotImplementedError('This method must be implemented by subclasses')
2920
2921 def geo_verification_headers(self):
2922 headers = {}
2923 geo_verification_proxy = self._downloader.params.get('geo_verification_proxy')
2924 if geo_verification_proxy:
2925 headers['Ytdl-request-proxy'] = geo_verification_proxy
2926 return headers
2927
2928 def _generic_id(self, url):
2929 return compat_urllib_parse_unquote(os.path.splitext(url.rstrip('/').split('/')[-1])[0])
2930
2931 def _generic_title(self, url):
2932 return compat_urllib_parse_unquote(os.path.splitext(url_basename(url))[0])
2933
2934
2935 class SearchInfoExtractor(InfoExtractor):
2936 """
2937 Base class for paged search queries extractors.
2938 They accept URLs in the format _SEARCH_KEY(|all|[0-9]):{query}
2939 Instances should define _SEARCH_KEY and _MAX_RESULTS.
2940 """
2941
2942 @classmethod
2943 def _make_valid_url(cls):
2944 return r'%s(?P<prefix>|[1-9][0-9]*|all):(?P<query>[\s\S]+)' % cls._SEARCH_KEY
2945
2946 @classmethod
2947 def suitable(cls, url):
2948 return re.match(cls._make_valid_url(), url) is not None
2949
2950 def _real_extract(self, query):
2951 mobj = re.match(self._make_valid_url(), query)
2952 if mobj is None:
2953 raise ExtractorError('Invalid search query "%s"' % query)
2954
2955 prefix = mobj.group('prefix')
2956 query = mobj.group('query')
2957 if prefix == '':
2958 return self._get_n_results(query, 1)
2959 elif prefix == 'all':
2960 return self._get_n_results(query, self._MAX_RESULTS)
2961 else:
2962 n = int(prefix)
2963 if n <= 0:
2964 raise ExtractorError('invalid download number %s for query "%s"' % (n, query))
2965 elif n > self._MAX_RESULTS:
2966 self._downloader.report_warning('%s returns max %i results (you requested %i)' % (self._SEARCH_KEY, self._MAX_RESULTS, n))
2967 n = self._MAX_RESULTS
2968 return self._get_n_results(query, n)
2969
2970 def _get_n_results(self, query, n):
2971 """Get a specified number of results for a query"""
2972 raise NotImplementedError('This method must be implemented by subclasses')
2973
2974 @property
2975 def SEARCH_KEY(self):
2976 return self._SEARCH_KEY