]> Raphaƫl G. Git Repositories - youtubedl/blob - youtube_dl/extractor/common.py
e2d9f52b018c25abc5a58a93785473ff88d90b74
[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_fromstring,
21 compat_getpass,
22 compat_http_client,
23 compat_os_name,
24 compat_str,
25 compat_urllib_error,
26 compat_urllib_parse_unquote,
27 compat_urllib_parse_urlencode,
28 compat_urllib_request,
29 compat_urlparse,
30 compat_xml_parse_error,
31 )
32 from ..downloader.f4m import (
33 get_base_url,
34 remove_encrypted_media,
35 )
36 from ..utils import (
37 NO_DEFAULT,
38 age_restricted,
39 base_url,
40 bug_reports_message,
41 clean_html,
42 compiled_regex_type,
43 determine_ext,
44 determine_protocol,
45 error_to_compat_str,
46 ExtractorError,
47 extract_attributes,
48 fix_xml_ampersands,
49 float_or_none,
50 GeoRestrictedError,
51 GeoUtils,
52 int_or_none,
53 js_to_json,
54 mimetype2ext,
55 orderedSet,
56 parse_codecs,
57 parse_duration,
58 parse_iso8601,
59 parse_m3u8_attributes,
60 RegexNotFoundError,
61 sanitized_Request,
62 sanitize_filename,
63 unescapeHTML,
64 unified_strdate,
65 unified_timestamp,
66 update_Request,
67 update_url_query,
68 urljoin,
69 url_basename,
70 xpath_element,
71 xpath_text,
72 xpath_with_ns,
73 )
74
75
76 class InfoExtractor(object):
77 """Information Extractor class.
78
79 Information extractors are the classes that, given a URL, extract
80 information about the video (or videos) the URL refers to. This
81 information includes the real video URL, the video title, author and
82 others. The information is stored in a dictionary which is then
83 passed to the YoutubeDL. The YoutubeDL processes this
84 information possibly downloading the video to the file system, among
85 other possible outcomes.
86
87 The type field determines the type of the result.
88 By far the most common value (and the default if _type is missing) is
89 "video", which indicates a single video.
90
91 For a video, the dictionaries must include the following fields:
92
93 id: Video identifier.
94 title: Video title, unescaped.
95
96 Additionally, it must contain either a formats entry or a url one:
97
98 formats: A list of dictionaries for each format available, ordered
99 from worst to best quality.
100
101 Potential fields:
102 * url Mandatory. The URL of the video file
103 * manifest_url
104 The URL of the manifest file in case of
105 fragmented media (DASH, hls, hds)
106 * ext Will be calculated from URL if missing
107 * format A human-readable description of the format
108 ("mp4 container with h264/opus").
109 Calculated from the format_id, width, height.
110 and format_note fields if missing.
111 * format_id A short description of the format
112 ("mp4_h264_opus" or "19").
113 Technically optional, but strongly recommended.
114 * format_note Additional info about the format
115 ("3D" or "DASH video")
116 * width Width of the video, if known
117 * height Height of the video, if known
118 * resolution Textual description of width and height
119 * tbr Average bitrate of audio and video in KBit/s
120 * abr Average audio bitrate in KBit/s
121 * acodec Name of the audio codec in use
122 * asr Audio sampling rate in Hertz
123 * vbr Average video bitrate in KBit/s
124 * fps Frame rate
125 * vcodec Name of the video codec in use
126 * container Name of the container format
127 * filesize The number of bytes, if known in advance
128 * filesize_approx An estimate for the number of bytes
129 * player_url SWF Player URL (used for rtmpdump).
130 * protocol The protocol that will be used for the actual
131 download, lower-case.
132 "http", "https", "rtsp", "rtmp", "rtmpe",
133 "m3u8", "m3u8_native" or "http_dash_segments".
134 * fragment_base_url
135 Base URL for fragments. Each fragment's path
136 value (if present) will be relative to
137 this URL.
138 * fragments A list of fragments of a fragmented media.
139 Each fragment entry must contain either an url
140 or a path. If an url is present it should be
141 considered by a client. Otherwise both path and
142 fragment_base_url must be present. Here is
143 the list of all potential fields:
144 * "url" - fragment's URL
145 * "path" - fragment's path relative to
146 fragment_base_url
147 * "duration" (optional, int or float)
148 * "filesize" (optional, int)
149 * preference Order number of this format. If this field is
150 present and not None, the formats get sorted
151 by this field, regardless of all other values.
152 -1 for default (order by other properties),
153 -2 or smaller for less than default.
154 < -1000 to hide the format (if there is
155 another one which is strictly better)
156 * language Language code, e.g. "de" or "en-US".
157 * language_preference Is this in the language mentioned in
158 the URL?
159 10 if it's what the URL is about,
160 -1 for default (don't know),
161 -10 otherwise, other values reserved for now.
162 * quality Order number of the video quality of this
163 format, irrespective of the file format.
164 -1 for default (order by other properties),
165 -2 or smaller for less than default.
166 * source_preference Order number for this video source
167 (quality takes higher priority)
168 -1 for default (order by other properties),
169 -2 or smaller for less than default.
170 * http_headers A dictionary of additional HTTP headers
171 to add to the request.
172 * stretched_ratio If given and not 1, indicates that the
173 video's pixels are not square.
174 width : height ratio as float.
175 * no_resume The server does not support resuming the
176 (HTTP or RTMP) download. Boolean.
177
178 url: Final video URL.
179 ext: Video filename extension.
180 format: The video format, defaults to ext (used for --get-format)
181 player_url: SWF Player URL (used for rtmpdump).
182
183 The following fields are optional:
184
185 alt_title: A secondary title of the video.
186 display_id An alternative identifier for the video, not necessarily
187 unique, but available before title. Typically, id is
188 something like "4234987", title "Dancing naked mole rats",
189 and display_id "dancing-naked-mole-rats"
190 thumbnails: A list of dictionaries, with the following entries:
191 * "id" (optional, string) - Thumbnail format ID
192 * "url"
193 * "preference" (optional, int) - quality of the image
194 * "width" (optional, int)
195 * "height" (optional, int)
196 * "resolution" (optional, string "{width}x{height"},
197 deprecated)
198 * "filesize" (optional, int)
199 thumbnail: Full URL to a video thumbnail image.
200 description: Full video description.
201 uploader: Full name of the video uploader.
202 license: License name the video is licensed under.
203 creator: The creator of the video.
204 release_date: The date (YYYYMMDD) when the video was released.
205 timestamp: UNIX timestamp of the moment the video became available.
206 upload_date: Video upload date (YYYYMMDD).
207 If not explicitly set, calculated from timestamp.
208 uploader_id: Nickname or id of the video uploader.
209 uploader_url: Full URL to a personal webpage of the video uploader.
210 location: Physical location where the video was filmed.
211 subtitles: The available subtitles as a dictionary in the format
212 {tag: subformats}. "tag" is usually a language code, and
213 "subformats" is a list sorted from lower to higher
214 preference, each element is a dictionary with the "ext"
215 entry and one of:
216 * "data": The subtitles file contents
217 * "url": A URL pointing to the subtitles file
218 "ext" will be calculated from URL if missing
219 automatic_captions: Like 'subtitles', used by the YoutubeIE for
220 automatically generated captions
221 duration: Length of the video in seconds, as an integer or float.
222 view_count: How many users have watched the video on the platform.
223 like_count: Number of positive ratings of the video
224 dislike_count: Number of negative ratings of the video
225 repost_count: Number of reposts of the video
226 average_rating: Average rating give by users, the scale used depends on the webpage
227 comment_count: Number of comments on the video
228 comments: A list of comments, each with one or more of the following
229 properties (all but one of text or html optional):
230 * "author" - human-readable name of the comment author
231 * "author_id" - user ID of the comment author
232 * "id" - Comment ID
233 * "html" - Comment as HTML
234 * "text" - Plain text of the comment
235 * "timestamp" - UNIX timestamp of comment
236 * "parent" - ID of the comment this one is replying to.
237 Set to "root" to indicate that this is a
238 comment to the original video.
239 age_limit: Age restriction for the video, as an integer (years)
240 webpage_url: The URL to the video webpage, if given to youtube-dl it
241 should allow to get the same result again. (It will be set
242 by YoutubeDL if it's missing)
243 categories: A list of categories that the video falls in, for example
244 ["Sports", "Berlin"]
245 tags: A list of tags assigned to the video, e.g. ["sweden", "pop music"]
246 is_live: True, False, or None (=unknown). Whether this video is a
247 live stream that goes on instead of a fixed-length video.
248 start_time: Time in seconds where the reproduction should start, as
249 specified in the URL.
250 end_time: Time in seconds where the reproduction should end, as
251 specified in the URL.
252 chapters: A list of dictionaries, with the following entries:
253 * "start_time" - The start time of the chapter in seconds
254 * "end_time" - The end time of the chapter in seconds
255 * "title" (optional, string)
256
257 The following fields should only be used when the video belongs to some logical
258 chapter or section:
259
260 chapter: Name or title of the chapter the video belongs to.
261 chapter_number: Number of the chapter the video belongs to, as an integer.
262 chapter_id: Id of the chapter the video belongs to, as a unicode string.
263
264 The following fields should only be used when the video is an episode of some
265 series, programme or podcast:
266
267 series: Title of the series or programme the video episode belongs to.
268 season: Title of the season the video episode belongs to.
269 season_number: Number of the season the video episode belongs to, as an integer.
270 season_id: Id of the season the video episode belongs to, as a unicode string.
271 episode: Title of the video episode. Unlike mandatory video title field,
272 this field should denote the exact title of the video episode
273 without any kind of decoration.
274 episode_number: Number of the video episode within a season, as an integer.
275 episode_id: Id of the video episode, as a unicode string.
276
277 The following fields should only be used when the media is a track or a part of
278 a music album:
279
280 track: Title of the track.
281 track_number: Number of the track within an album or a disc, as an integer.
282 track_id: Id of the track (useful in case of custom indexing, e.g. 6.iii),
283 as a unicode string.
284 artist: Artist(s) of the track.
285 genre: Genre(s) of the track.
286 album: Title of the album the track belongs to.
287 album_type: Type of the album (e.g. "Demo", "Full-length", "Split", "Compilation", etc).
288 album_artist: List of all artists appeared on the album (e.g.
289 "Ash Borer / Fell Voices" or "Various Artists", useful for splits
290 and compilations).
291 disc_number: Number of the disc or other physical medium the track belongs to,
292 as an integer.
293 release_year: Year (YYYY) when the album was released.
294
295 Unless mentioned otherwise, the fields should be Unicode strings.
296
297 Unless mentioned otherwise, None is equivalent to absence of information.
298
299
300 _type "playlist" indicates multiple videos.
301 There must be a key "entries", which is a list, an iterable, or a PagedList
302 object, each element of which is a valid dictionary by this specification.
303
304 Additionally, playlists can have "title", "description" and "id" attributes
305 with the same semantics as videos (see above).
306
307
308 _type "multi_video" indicates that there are multiple videos that
309 form a single show, for examples multiple acts of an opera or TV episode.
310 It must have an entries key like a playlist and contain all the keys
311 required for a video at the same time.
312
313
314 _type "url" indicates that the video must be extracted from another
315 location, possibly by a different extractor. Its only required key is:
316 "url" - the next URL to extract.
317 The key "ie_key" can be set to the class name (minus the trailing "IE",
318 e.g. "Youtube") if the extractor class is known in advance.
319 Additionally, the dictionary may have any properties of the resolved entity
320 known in advance, for example "title" if the title of the referred video is
321 known ahead of time.
322
323
324 _type "url_transparent" entities have the same specification as "url", but
325 indicate that the given additional information is more precise than the one
326 associated with the resolved URL.
327 This is useful when a site employs a video service that hosts the video and
328 its technical metadata, but that video service does not embed a useful
329 title, description etc.
330
331
332 Subclasses of this one should re-define the _real_initialize() and
333 _real_extract() methods and define a _VALID_URL regexp.
334 Probably, they should also be added to the list of extractors.
335
336 _GEO_BYPASS attribute may be set to False in order to disable
337 geo restriction bypass mechanisms for a particular extractor.
338 Though it won't disable explicit geo restriction bypass based on
339 country code provided with geo_bypass_country. (experimental)
340
341 _GEO_COUNTRIES attribute may contain a list of presumably geo unrestricted
342 countries for this extractor. One of these countries will be used by
343 geo restriction bypass mechanism right away in order to bypass
344 geo restriction, of course, if the mechanism is not disabled. (experimental)
345
346 NB: both these geo attributes are experimental and may change in future
347 or be completely removed.
348
349 Finally, the _WORKING attribute should be set to False for broken IEs
350 in order to warn the users and skip the tests.
351 """
352
353 _ready = False
354 _downloader = None
355 _x_forwarded_for_ip = None
356 _GEO_BYPASS = True
357 _GEO_COUNTRIES = None
358 _WORKING = True
359
360 def __init__(self, downloader=None):
361 """Constructor. Receives an optional downloader."""
362 self._ready = False
363 self._x_forwarded_for_ip = None
364 self.set_downloader(downloader)
365
366 @classmethod
367 def suitable(cls, url):
368 """Receives a URL and returns True if suitable for this IE."""
369
370 # This does not use has/getattr intentionally - we want to know whether
371 # we have cached the regexp for *this* class, whereas getattr would also
372 # match the superclass
373 if '_VALID_URL_RE' not in cls.__dict__:
374 cls._VALID_URL_RE = re.compile(cls._VALID_URL)
375 return cls._VALID_URL_RE.match(url) is not None
376
377 @classmethod
378 def _match_id(cls, url):
379 if '_VALID_URL_RE' not in cls.__dict__:
380 cls._VALID_URL_RE = re.compile(cls._VALID_URL)
381 m = cls._VALID_URL_RE.match(url)
382 assert m
383 return compat_str(m.group('id'))
384
385 @classmethod
386 def working(cls):
387 """Getter method for _WORKING."""
388 return cls._WORKING
389
390 def initialize(self):
391 """Initializes an instance (authentication, etc)."""
392 self._initialize_geo_bypass(self._GEO_COUNTRIES)
393 if not self._ready:
394 self._real_initialize()
395 self._ready = True
396
397 def _initialize_geo_bypass(self, countries):
398 """
399 Initialize geo restriction bypass mechanism.
400
401 This method is used to initialize geo bypass mechanism based on faking
402 X-Forwarded-For HTTP header. A random country from provided country list
403 is selected and a random IP belonging to this country is generated. This
404 IP will be passed as X-Forwarded-For HTTP header in all subsequent
405 HTTP requests.
406
407 This method will be used for initial geo bypass mechanism initialization
408 during the instance initialization with _GEO_COUNTRIES.
409
410 You may also manually call it from extractor's code if geo countries
411 information is not available beforehand (e.g. obtained during
412 extraction) or due to some another reason.
413 """
414 if not self._x_forwarded_for_ip:
415 country_code = self._downloader.params.get('geo_bypass_country', None)
416 # If there is no explicit country for geo bypass specified and
417 # the extractor is known to be geo restricted let's fake IP
418 # as X-Forwarded-For right away.
419 if (not country_code and
420 self._GEO_BYPASS and
421 self._downloader.params.get('geo_bypass', True) and
422 countries):
423 country_code = random.choice(countries)
424 if country_code:
425 self._x_forwarded_for_ip = GeoUtils.random_ipv4(country_code)
426 if self._downloader.params.get('verbose', False):
427 self._downloader.to_screen(
428 '[debug] Using fake IP %s (%s) as X-Forwarded-For.'
429 % (self._x_forwarded_for_ip, country_code.upper()))
430
431 def extract(self, url):
432 """Extracts URL information and returns it in list of dicts."""
433 try:
434 for _ in range(2):
435 try:
436 self.initialize()
437 ie_result = self._real_extract(url)
438 if self._x_forwarded_for_ip:
439 ie_result['__x_forwarded_for_ip'] = self._x_forwarded_for_ip
440 return ie_result
441 except GeoRestrictedError as e:
442 if self.__maybe_fake_ip_and_retry(e.countries):
443 continue
444 raise
445 except ExtractorError:
446 raise
447 except compat_http_client.IncompleteRead as e:
448 raise ExtractorError('A network error has occurred.', cause=e, expected=True)
449 except (KeyError, StopIteration) as e:
450 raise ExtractorError('An extractor error has occurred.', cause=e)
451
452 def __maybe_fake_ip_and_retry(self, countries):
453 if (not self._downloader.params.get('geo_bypass_country', None) and
454 self._GEO_BYPASS and
455 self._downloader.params.get('geo_bypass', True) and
456 not self._x_forwarded_for_ip and
457 countries):
458 country_code = random.choice(countries)
459 self._x_forwarded_for_ip = GeoUtils.random_ipv4(country_code)
460 if self._x_forwarded_for_ip:
461 self.report_warning(
462 'Video is geo restricted. Retrying extraction with fake IP %s (%s) as X-Forwarded-For.'
463 % (self._x_forwarded_for_ip, country_code.upper()))
464 return True
465 return False
466
467 def set_downloader(self, downloader):
468 """Sets the downloader for this IE."""
469 self._downloader = downloader
470
471 def _real_initialize(self):
472 """Real initialization process. Redefine in subclasses."""
473 pass
474
475 def _real_extract(self, url):
476 """Real extraction process. Redefine in subclasses."""
477 pass
478
479 @classmethod
480 def ie_key(cls):
481 """A string for getting the InfoExtractor with get_info_extractor"""
482 return compat_str(cls.__name__[:-2])
483
484 @property
485 def IE_NAME(self):
486 return compat_str(type(self).__name__[:-2])
487
488 def _request_webpage(self, url_or_request, video_id, note=None, errnote=None, fatal=True, data=None, headers={}, query={}):
489 """ Returns the response handle """
490 if note is None:
491 self.report_download_webpage(video_id)
492 elif note is not False:
493 if video_id is None:
494 self.to_screen('%s' % (note,))
495 else:
496 self.to_screen('%s: %s' % (video_id, note))
497 if isinstance(url_or_request, compat_urllib_request.Request):
498 url_or_request = update_Request(
499 url_or_request, data=data, headers=headers, query=query)
500 else:
501 if query:
502 url_or_request = update_url_query(url_or_request, query)
503 if data is not None or headers:
504 url_or_request = sanitized_Request(url_or_request, data, headers)
505 try:
506 return self._downloader.urlopen(url_or_request)
507 except (compat_urllib_error.URLError, compat_http_client.HTTPException, socket.error) as err:
508 if errnote is False:
509 return False
510 if errnote is None:
511 errnote = 'Unable to download webpage'
512
513 errmsg = '%s: %s' % (errnote, error_to_compat_str(err))
514 if fatal:
515 raise ExtractorError(errmsg, sys.exc_info()[2], cause=err)
516 else:
517 self._downloader.report_warning(errmsg)
518 return False
519
520 def _download_webpage_handle(self, url_or_request, video_id, note=None, errnote=None, fatal=True, encoding=None, data=None, headers={}, query={}):
521 """ Returns a tuple (page content as string, URL handle) """
522 # Strip hashes from the URL (#1038)
523 if isinstance(url_or_request, (compat_str, str)):
524 url_or_request = url_or_request.partition('#')[0]
525
526 # Some sites check X-Forwarded-For HTTP header in order to figure out
527 # the origin of the client behind proxy. This allows bypassing geo
528 # restriction by faking this header's value to IP that belongs to some
529 # geo unrestricted country. We will do so once we encounter any
530 # geo restriction error.
531 if self._x_forwarded_for_ip:
532 if 'X-Forwarded-For' not in headers:
533 headers['X-Forwarded-For'] = self._x_forwarded_for_ip
534
535 urlh = self._request_webpage(url_or_request, video_id, note, errnote, fatal, data=data, headers=headers, query=query)
536 if urlh is False:
537 assert not fatal
538 return False
539 content = self._webpage_read_content(urlh, url_or_request, video_id, note, errnote, fatal, encoding=encoding)
540 return (content, urlh)
541
542 @staticmethod
543 def _guess_encoding_from_content(content_type, webpage_bytes):
544 m = re.match(r'[a-zA-Z0-9_.-]+/[a-zA-Z0-9_.-]+\s*;\s*charset=(.+)', content_type)
545 if m:
546 encoding = m.group(1)
547 else:
548 m = re.search(br'<meta[^>]+charset=[\'"]?([^\'")]+)[ /\'">]',
549 webpage_bytes[:1024])
550 if m:
551 encoding = m.group(1).decode('ascii')
552 elif webpage_bytes.startswith(b'\xff\xfe'):
553 encoding = 'utf-16'
554 else:
555 encoding = 'utf-8'
556
557 return encoding
558
559 def __check_blocked(self, content):
560 first_block = content[:512]
561 if ('<title>Access to this site is blocked</title>' in content and
562 'Websense' in first_block):
563 msg = 'Access to this webpage has been blocked by Websense filtering software in your network.'
564 blocked_iframe = self._html_search_regex(
565 r'<iframe src="([^"]+)"', content,
566 'Websense information URL', default=None)
567 if blocked_iframe:
568 msg += ' Visit %s for more details' % blocked_iframe
569 raise ExtractorError(msg, expected=True)
570 if '<title>The URL you requested has been blocked</title>' in first_block:
571 msg = (
572 'Access to this webpage has been blocked by Indian censorship. '
573 'Use a VPN or proxy server (with --proxy) to route around it.')
574 block_msg = self._html_search_regex(
575 r'</h1><p>(.*?)</p>',
576 content, 'block message', default=None)
577 if block_msg:
578 msg += ' (Message: "%s")' % block_msg.replace('\n', ' ')
579 raise ExtractorError(msg, expected=True)
580 if ('<title>TTK :: Š”Š¾ŃŃ‚ŃƒŠæ Šŗ рŠµŃŃƒŃ€ŃŃƒ Š¾Š³Ń€Š°Š½ŠøчŠµŠ½</title>' in content and
581 'blocklist.rkn.gov.ru' in content):
582 raise ExtractorError(
583 'Access to this webpage has been blocked by decision of the Russian government. '
584 'Visit http://blocklist.rkn.gov.ru/ for a block reason.',
585 expected=True)
586
587 def _webpage_read_content(self, urlh, url_or_request, video_id, note=None, errnote=None, fatal=True, prefix=None, encoding=None):
588 content_type = urlh.headers.get('Content-Type', '')
589 webpage_bytes = urlh.read()
590 if prefix is not None:
591 webpage_bytes = prefix + webpage_bytes
592 if not encoding:
593 encoding = self._guess_encoding_from_content(content_type, webpage_bytes)
594 if self._downloader.params.get('dump_intermediate_pages', False):
595 try:
596 url = url_or_request.get_full_url()
597 except AttributeError:
598 url = url_or_request
599 self.to_screen('Dumping request to ' + url)
600 dump = base64.b64encode(webpage_bytes).decode('ascii')
601 self._downloader.to_screen(dump)
602 if self._downloader.params.get('write_pages', False):
603 try:
604 url = url_or_request.get_full_url()
605 except AttributeError:
606 url = url_or_request
607 basen = '%s_%s' % (video_id, url)
608 if len(basen) > 240:
609 h = '___' + hashlib.md5(basen.encode('utf-8')).hexdigest()
610 basen = basen[:240 - len(h)] + h
611 raw_filename = basen + '.dump'
612 filename = sanitize_filename(raw_filename, restricted=True)
613 self.to_screen('Saving request to ' + filename)
614 # Working around MAX_PATH limitation on Windows (see
615 # http://msdn.microsoft.com/en-us/library/windows/desktop/aa365247(v=vs.85).aspx)
616 if compat_os_name == 'nt':
617 absfilepath = os.path.abspath(filename)
618 if len(absfilepath) > 259:
619 filename = '\\\\?\\' + absfilepath
620 with open(filename, 'wb') as outf:
621 outf.write(webpage_bytes)
622
623 try:
624 content = webpage_bytes.decode(encoding, 'replace')
625 except LookupError:
626 content = webpage_bytes.decode('utf-8', 'replace')
627
628 self.__check_blocked(content)
629
630 return content
631
632 def _download_webpage(self, url_or_request, video_id, note=None, errnote=None, fatal=True, tries=1, timeout=5, encoding=None, data=None, headers={}, query={}):
633 """ Returns the data of the page as a string """
634 success = False
635 try_count = 0
636 while success is False:
637 try:
638 res = self._download_webpage_handle(url_or_request, video_id, note, errnote, fatal, encoding=encoding, data=data, headers=headers, query=query)
639 success = True
640 except compat_http_client.IncompleteRead as e:
641 try_count += 1
642 if try_count >= tries:
643 raise e
644 self._sleep(timeout, video_id)
645 if res is False:
646 return res
647 else:
648 content, _ = res
649 return content
650
651 def _download_xml(self, url_or_request, video_id,
652 note='Downloading XML', errnote='Unable to download XML',
653 transform_source=None, fatal=True, encoding=None,
654 data=None, headers={}, query={}):
655 """Return the xml as an xml.etree.ElementTree.Element"""
656 xml_string = self._download_webpage(
657 url_or_request, video_id, note, errnote, fatal=fatal,
658 encoding=encoding, data=data, headers=headers, query=query)
659 if xml_string is False:
660 return xml_string
661 return self._parse_xml(
662 xml_string, video_id, transform_source=transform_source,
663 fatal=fatal)
664
665 def _parse_xml(self, xml_string, video_id, transform_source=None, fatal=True):
666 if transform_source:
667 xml_string = transform_source(xml_string)
668 try:
669 return compat_etree_fromstring(xml_string.encode('utf-8'))
670 except compat_xml_parse_error as ve:
671 errmsg = '%s: Failed to parse XML ' % video_id
672 if fatal:
673 raise ExtractorError(errmsg, cause=ve)
674 else:
675 self.report_warning(errmsg + str(ve))
676
677 def _download_json(self, url_or_request, video_id,
678 note='Downloading JSON metadata',
679 errnote='Unable to download JSON metadata',
680 transform_source=None,
681 fatal=True, encoding=None, data=None, headers={}, query={}):
682 json_string = self._download_webpage(
683 url_or_request, video_id, note, errnote, fatal=fatal,
684 encoding=encoding, data=data, headers=headers, query=query)
685 if (not fatal) and json_string is False:
686 return None
687 return self._parse_json(
688 json_string, video_id, transform_source=transform_source, fatal=fatal)
689
690 def _parse_json(self, json_string, video_id, transform_source=None, fatal=True):
691 if transform_source:
692 json_string = transform_source(json_string)
693 try:
694 return json.loads(json_string)
695 except ValueError as ve:
696 errmsg = '%s: Failed to parse JSON ' % video_id
697 if fatal:
698 raise ExtractorError(errmsg, cause=ve)
699 else:
700 self.report_warning(errmsg + str(ve))
701
702 def report_warning(self, msg, video_id=None):
703 idstr = '' if video_id is None else '%s: ' % video_id
704 self._downloader.report_warning(
705 '[%s] %s%s' % (self.IE_NAME, idstr, msg))
706
707 def to_screen(self, msg):
708 """Print msg to screen, prefixing it with '[ie_name]'"""
709 self._downloader.to_screen('[%s] %s' % (self.IE_NAME, msg))
710
711 def report_extraction(self, id_or_name):
712 """Report information extraction."""
713 self.to_screen('%s: Extracting information' % id_or_name)
714
715 def report_download_webpage(self, video_id):
716 """Report webpage download."""
717 self.to_screen('%s: Downloading webpage' % video_id)
718
719 def report_age_confirmation(self):
720 """Report attempt to confirm age."""
721 self.to_screen('Confirming age')
722
723 def report_login(self):
724 """Report attempt to log in."""
725 self.to_screen('Logging in')
726
727 @staticmethod
728 def raise_login_required(msg='This video is only available for registered users'):
729 raise ExtractorError(
730 '%s. Use --username and --password or --netrc to provide account credentials.' % msg,
731 expected=True)
732
733 @staticmethod
734 def raise_geo_restricted(msg='This video is not available from your location due to geo restriction', countries=None):
735 raise GeoRestrictedError(msg, countries=countries)
736
737 # Methods for following #608
738 @staticmethod
739 def url_result(url, ie=None, video_id=None, video_title=None):
740 """Returns a URL that points to a page that should be processed"""
741 # TODO: ie should be the class used for getting the info
742 video_info = {'_type': 'url',
743 'url': url,
744 'ie_key': ie}
745 if video_id is not None:
746 video_info['id'] = video_id
747 if video_title is not None:
748 video_info['title'] = video_title
749 return video_info
750
751 def playlist_from_matches(self, matches, playlist_id=None, playlist_title=None, getter=None, ie=None):
752 urls = orderedSet(
753 self.url_result(self._proto_relative_url(getter(m) if getter else m), ie)
754 for m in matches)
755 return self.playlist_result(
756 urls, playlist_id=playlist_id, playlist_title=playlist_title)
757
758 @staticmethod
759 def playlist_result(entries, playlist_id=None, playlist_title=None, playlist_description=None):
760 """Returns a playlist"""
761 video_info = {'_type': 'playlist',
762 'entries': entries}
763 if playlist_id:
764 video_info['id'] = playlist_id
765 if playlist_title:
766 video_info['title'] = playlist_title
767 if playlist_description:
768 video_info['description'] = playlist_description
769 return video_info
770
771 def _search_regex(self, pattern, string, name, default=NO_DEFAULT, fatal=True, flags=0, group=None):
772 """
773 Perform a regex search on the given string, using a single or a list of
774 patterns returning the first matching group.
775 In case of failure return a default value or raise a WARNING or a
776 RegexNotFoundError, depending on fatal, specifying the field name.
777 """
778 if isinstance(pattern, (str, compat_str, compiled_regex_type)):
779 mobj = re.search(pattern, string, flags)
780 else:
781 for p in pattern:
782 mobj = re.search(p, string, flags)
783 if mobj:
784 break
785
786 if not self._downloader.params.get('no_color') and compat_os_name != 'nt' and sys.stderr.isatty():
787 _name = '\033[0;34m%s\033[0m' % name
788 else:
789 _name = name
790
791 if mobj:
792 if group is None:
793 # return the first matching group
794 return next(g for g in mobj.groups() if g is not None)
795 else:
796 return mobj.group(group)
797 elif default is not NO_DEFAULT:
798 return default
799 elif fatal:
800 raise RegexNotFoundError('Unable to extract %s' % _name)
801 else:
802 self._downloader.report_warning('unable to extract %s' % _name + bug_reports_message())
803 return None
804
805 def _html_search_regex(self, pattern, string, name, default=NO_DEFAULT, fatal=True, flags=0, group=None):
806 """
807 Like _search_regex, but strips HTML tags and unescapes entities.
808 """
809 res = self._search_regex(pattern, string, name, default, fatal, flags, group)
810 if res:
811 return clean_html(res).strip()
812 else:
813 return res
814
815 def _get_netrc_login_info(self, netrc_machine=None):
816 username = None
817 password = None
818 netrc_machine = netrc_machine or self._NETRC_MACHINE
819
820 if self._downloader.params.get('usenetrc', False):
821 try:
822 info = netrc.netrc().authenticators(netrc_machine)
823 if info is not None:
824 username = info[0]
825 password = info[2]
826 else:
827 raise netrc.NetrcParseError(
828 'No authenticators for %s' % netrc_machine)
829 except (IOError, netrc.NetrcParseError) as err:
830 self._downloader.report_warning(
831 'parsing .netrc: %s' % error_to_compat_str(err))
832
833 return username, password
834
835 def _get_login_info(self, username_option='username', password_option='password', netrc_machine=None):
836 """
837 Get the login info as (username, password)
838 First look for the manually specified credentials using username_option
839 and password_option as keys in params dictionary. If no such credentials
840 available look in the netrc file using the netrc_machine or _NETRC_MACHINE
841 value.
842 If there's no info available, return (None, None)
843 """
844 if self._downloader is None:
845 return (None, None)
846
847 downloader_params = self._downloader.params
848
849 # Attempt to use provided username and password or .netrc data
850 if downloader_params.get(username_option) is not None:
851 username = downloader_params[username_option]
852 password = downloader_params[password_option]
853 else:
854 username, password = self._get_netrc_login_info(netrc_machine)
855
856 return username, password
857
858 def _get_tfa_info(self, note='two-factor verification code'):
859 """
860 Get the two-factor authentication info
861 TODO - asking the user will be required for sms/phone verify
862 currently just uses the command line option
863 If there's no info available, return None
864 """
865 if self._downloader is None:
866 return None
867 downloader_params = self._downloader.params
868
869 if downloader_params.get('twofactor') is not None:
870 return downloader_params['twofactor']
871
872 return compat_getpass('Type %s and press [Return]: ' % note)
873
874 # Helper functions for extracting OpenGraph info
875 @staticmethod
876 def _og_regexes(prop):
877 content_re = r'content=(?:"([^"]+?)"|\'([^\']+?)\'|\s*([^\s"\'=<>`]+?))'
878 property_re = (r'(?:name|property)=(?:\'og:%(prop)s\'|"og:%(prop)s"|\s*og:%(prop)s\b)'
879 % {'prop': re.escape(prop)})
880 template = r'<meta[^>]+?%s[^>]+?%s'
881 return [
882 template % (property_re, content_re),
883 template % (content_re, property_re),
884 ]
885
886 @staticmethod
887 def _meta_regex(prop):
888 return r'''(?isx)<meta
889 (?=[^>]+(?:itemprop|name|property|id|http-equiv)=(["\']?)%s\1)
890 [^>]+?content=(["\'])(?P<content>.*?)\2''' % re.escape(prop)
891
892 def _og_search_property(self, prop, html, name=None, **kargs):
893 if not isinstance(prop, (list, tuple)):
894 prop = [prop]
895 if name is None:
896 name = 'OpenGraph %s' % prop[0]
897 og_regexes = []
898 for p in prop:
899 og_regexes.extend(self._og_regexes(p))
900 escaped = self._search_regex(og_regexes, html, name, flags=re.DOTALL, **kargs)
901 if escaped is None:
902 return None
903 return unescapeHTML(escaped)
904
905 def _og_search_thumbnail(self, html, **kargs):
906 return self._og_search_property('image', html, 'thumbnail URL', fatal=False, **kargs)
907
908 def _og_search_description(self, html, **kargs):
909 return self._og_search_property('description', html, fatal=False, **kargs)
910
911 def _og_search_title(self, html, **kargs):
912 return self._og_search_property('title', html, **kargs)
913
914 def _og_search_video_url(self, html, name='video url', secure=True, **kargs):
915 regexes = self._og_regexes('video') + self._og_regexes('video:url')
916 if secure:
917 regexes = self._og_regexes('video:secure_url') + regexes
918 return self._html_search_regex(regexes, html, name, **kargs)
919
920 def _og_search_url(self, html, **kargs):
921 return self._og_search_property('url', html, **kargs)
922
923 def _html_search_meta(self, name, html, display_name=None, fatal=False, **kwargs):
924 if not isinstance(name, (list, tuple)):
925 name = [name]
926 if display_name is None:
927 display_name = name[0]
928 return self._html_search_regex(
929 [self._meta_regex(n) for n in name],
930 html, display_name, fatal=fatal, group='content', **kwargs)
931
932 def _dc_search_uploader(self, html):
933 return self._html_search_meta('dc.creator', html, 'uploader')
934
935 def _rta_search(self, html):
936 # See http://www.rtalabel.org/index.php?content=howtofaq#single
937 if re.search(r'(?ix)<meta\s+name="rating"\s+'
938 r' content="RTA-5042-1996-1400-1577-RTA"',
939 html):
940 return 18
941 return 0
942
943 def _media_rating_search(self, html):
944 # See http://www.tjg-designs.com/WP/metadata-code-examples-adding-metadata-to-your-web-pages/
945 rating = self._html_search_meta('rating', html)
946
947 if not rating:
948 return None
949
950 RATING_TABLE = {
951 'safe for kids': 0,
952 'general': 8,
953 '14 years': 14,
954 'mature': 17,
955 'restricted': 19,
956 }
957 return RATING_TABLE.get(rating.lower())
958
959 def _family_friendly_search(self, html):
960 # See http://schema.org/VideoObject
961 family_friendly = self._html_search_meta(
962 'isFamilyFriendly', html, default=None)
963
964 if not family_friendly:
965 return None
966
967 RATING_TABLE = {
968 '1': 0,
969 'true': 0,
970 '0': 18,
971 'false': 18,
972 }
973 return RATING_TABLE.get(family_friendly.lower())
974
975 def _twitter_search_player(self, html):
976 return self._html_search_meta('twitter:player', html,
977 'twitter card player')
978
979 def _search_json_ld(self, html, video_id, expected_type=None, **kwargs):
980 json_ld = self._search_regex(
981 r'(?s)<script[^>]+type=(["\'])application/ld\+json\1[^>]*>(?P<json_ld>.+?)</script>',
982 html, 'JSON-LD', group='json_ld', **kwargs)
983 default = kwargs.get('default', NO_DEFAULT)
984 if not json_ld:
985 return default if default is not NO_DEFAULT else {}
986 # JSON-LD may be malformed and thus `fatal` should be respected.
987 # At the same time `default` may be passed that assumes `fatal=False`
988 # for _search_regex. Let's simulate the same behavior here as well.
989 fatal = kwargs.get('fatal', True) if default == NO_DEFAULT else False
990 return self._json_ld(json_ld, video_id, fatal=fatal, expected_type=expected_type)
991
992 def _json_ld(self, json_ld, video_id, fatal=True, expected_type=None):
993 if isinstance(json_ld, compat_str):
994 json_ld = self._parse_json(json_ld, video_id, fatal=fatal)
995 if not json_ld:
996 return {}
997 info = {}
998 if not isinstance(json_ld, (list, tuple, dict)):
999 return info
1000 if isinstance(json_ld, dict):
1001 json_ld = [json_ld]
1002
1003 def extract_video_object(e):
1004 assert e['@type'] == 'VideoObject'
1005 info.update({
1006 'url': e.get('contentUrl'),
1007 'title': unescapeHTML(e.get('name')),
1008 'description': unescapeHTML(e.get('description')),
1009 'thumbnail': e.get('thumbnailUrl') or e.get('thumbnailURL'),
1010 'duration': parse_duration(e.get('duration')),
1011 'timestamp': unified_timestamp(e.get('uploadDate')),
1012 'filesize': float_or_none(e.get('contentSize')),
1013 'tbr': int_or_none(e.get('bitrate')),
1014 'width': int_or_none(e.get('width')),
1015 'height': int_or_none(e.get('height')),
1016 'view_count': int_or_none(e.get('interactionCount')),
1017 })
1018
1019 for e in json_ld:
1020 if e.get('@context') == 'http://schema.org':
1021 item_type = e.get('@type')
1022 if expected_type is not None and expected_type != item_type:
1023 return info
1024 if item_type in ('TVEpisode', 'Episode'):
1025 info.update({
1026 'episode': unescapeHTML(e.get('name')),
1027 'episode_number': int_or_none(e.get('episodeNumber')),
1028 'description': unescapeHTML(e.get('description')),
1029 })
1030 part_of_season = e.get('partOfSeason')
1031 if isinstance(part_of_season, dict) and part_of_season.get('@type') in ('TVSeason', 'Season', 'CreativeWorkSeason'):
1032 info['season_number'] = int_or_none(part_of_season.get('seasonNumber'))
1033 part_of_series = e.get('partOfSeries') or e.get('partOfTVSeries')
1034 if isinstance(part_of_series, dict) and part_of_series.get('@type') in ('TVSeries', 'Series', 'CreativeWorkSeries'):
1035 info['series'] = unescapeHTML(part_of_series.get('name'))
1036 elif item_type == 'Article':
1037 info.update({
1038 'timestamp': parse_iso8601(e.get('datePublished')),
1039 'title': unescapeHTML(e.get('headline')),
1040 'description': unescapeHTML(e.get('articleBody')),
1041 })
1042 elif item_type == 'VideoObject':
1043 extract_video_object(e)
1044 continue
1045 video = e.get('video')
1046 if isinstance(video, dict) and video.get('@type') == 'VideoObject':
1047 extract_video_object(video)
1048 break
1049 return dict((k, v) for k, v in info.items() if v is not None)
1050
1051 @staticmethod
1052 def _hidden_inputs(html):
1053 html = re.sub(r'<!--(?:(?!<!--).)*-->', '', html)
1054 hidden_inputs = {}
1055 for input in re.findall(r'(?i)(<input[^>]+>)', html):
1056 attrs = extract_attributes(input)
1057 if not input:
1058 continue
1059 if attrs.get('type') not in ('hidden', 'submit'):
1060 continue
1061 name = attrs.get('name') or attrs.get('id')
1062 value = attrs.get('value')
1063 if name and value is not None:
1064 hidden_inputs[name] = value
1065 return hidden_inputs
1066
1067 def _form_hidden_inputs(self, form_id, html):
1068 form = self._search_regex(
1069 r'(?is)<form[^>]+?id=(["\'])%s\1[^>]*>(?P<form>.+?)</form>' % form_id,
1070 html, '%s form' % form_id, group='form')
1071 return self._hidden_inputs(form)
1072
1073 def _sort_formats(self, formats, field_preference=None):
1074 if not formats:
1075 raise ExtractorError('No video formats found')
1076
1077 for f in formats:
1078 # Automatically determine tbr when missing based on abr and vbr (improves
1079 # formats sorting in some cases)
1080 if 'tbr' not in f and f.get('abr') is not None and f.get('vbr') is not None:
1081 f['tbr'] = f['abr'] + f['vbr']
1082
1083 def _formats_key(f):
1084 # TODO remove the following workaround
1085 from ..utils import determine_ext
1086 if not f.get('ext') and 'url' in f:
1087 f['ext'] = determine_ext(f['url'])
1088
1089 if isinstance(field_preference, (list, tuple)):
1090 return tuple(
1091 f.get(field)
1092 if f.get(field) is not None
1093 else ('' if field == 'format_id' else -1)
1094 for field in field_preference)
1095
1096 preference = f.get('preference')
1097 if preference is None:
1098 preference = 0
1099 if f.get('ext') in ['f4f', 'f4m']: # Not yet supported
1100 preference -= 0.5
1101
1102 protocol = f.get('protocol') or determine_protocol(f)
1103 proto_preference = 0 if protocol in ['http', 'https'] else (-0.5 if protocol == 'rtsp' else -0.1)
1104
1105 if f.get('vcodec') == 'none': # audio only
1106 preference -= 50
1107 if self._downloader.params.get('prefer_free_formats'):
1108 ORDER = ['aac', 'mp3', 'm4a', 'webm', 'ogg', 'opus']
1109 else:
1110 ORDER = ['webm', 'opus', 'ogg', 'mp3', 'aac', 'm4a']
1111 ext_preference = 0
1112 try:
1113 audio_ext_preference = ORDER.index(f['ext'])
1114 except ValueError:
1115 audio_ext_preference = -1
1116 else:
1117 if f.get('acodec') == 'none': # video only
1118 preference -= 40
1119 if self._downloader.params.get('prefer_free_formats'):
1120 ORDER = ['flv', 'mp4', 'webm']
1121 else:
1122 ORDER = ['webm', 'flv', 'mp4']
1123 try:
1124 ext_preference = ORDER.index(f['ext'])
1125 except ValueError:
1126 ext_preference = -1
1127 audio_ext_preference = 0
1128
1129 return (
1130 preference,
1131 f.get('language_preference') if f.get('language_preference') is not None else -1,
1132 f.get('quality') if f.get('quality') is not None else -1,
1133 f.get('tbr') if f.get('tbr') is not None else -1,
1134 f.get('filesize') if f.get('filesize') is not None else -1,
1135 f.get('vbr') if f.get('vbr') is not None else -1,
1136 f.get('height') if f.get('height') is not None else -1,
1137 f.get('width') if f.get('width') is not None else -1,
1138 proto_preference,
1139 ext_preference,
1140 f.get('abr') if f.get('abr') is not None else -1,
1141 audio_ext_preference,
1142 f.get('fps') if f.get('fps') is not None else -1,
1143 f.get('filesize_approx') if f.get('filesize_approx') is not None else -1,
1144 f.get('source_preference') if f.get('source_preference') is not None else -1,
1145 f.get('format_id') if f.get('format_id') is not None else '',
1146 )
1147 formats.sort(key=_formats_key)
1148
1149 def _check_formats(self, formats, video_id):
1150 if formats:
1151 formats[:] = filter(
1152 lambda f: self._is_valid_url(
1153 f['url'], video_id,
1154 item='%s video format' % f.get('format_id') if f.get('format_id') else 'video'),
1155 formats)
1156
1157 @staticmethod
1158 def _remove_duplicate_formats(formats):
1159 format_urls = set()
1160 unique_formats = []
1161 for f in formats:
1162 if f['url'] not in format_urls:
1163 format_urls.add(f['url'])
1164 unique_formats.append(f)
1165 formats[:] = unique_formats
1166
1167 def _is_valid_url(self, url, video_id, item='video', headers={}):
1168 url = self._proto_relative_url(url, scheme='http:')
1169 # For now assume non HTTP(S) URLs always valid
1170 if not (url.startswith('http://') or url.startswith('https://')):
1171 return True
1172 try:
1173 self._request_webpage(url, video_id, 'Checking %s URL' % item, headers=headers)
1174 return True
1175 except ExtractorError as e:
1176 if isinstance(e.cause, compat_urllib_error.URLError):
1177 self.to_screen(
1178 '%s: %s URL is invalid, skipping' % (video_id, item))
1179 return False
1180 raise
1181
1182 def http_scheme(self):
1183 """ Either "http:" or "https:", depending on the user's preferences """
1184 return (
1185 'http:'
1186 if self._downloader.params.get('prefer_insecure', False)
1187 else 'https:')
1188
1189 def _proto_relative_url(self, url, scheme=None):
1190 if url is None:
1191 return url
1192 if url.startswith('//'):
1193 if scheme is None:
1194 scheme = self.http_scheme()
1195 return scheme + url
1196 else:
1197 return url
1198
1199 def _sleep(self, timeout, video_id, msg_template=None):
1200 if msg_template is None:
1201 msg_template = '%(video_id)s: Waiting for %(timeout)s seconds'
1202 msg = msg_template % {'video_id': video_id, 'timeout': timeout}
1203 self.to_screen(msg)
1204 time.sleep(timeout)
1205
1206 def _extract_f4m_formats(self, manifest_url, video_id, preference=None, f4m_id=None,
1207 transform_source=lambda s: fix_xml_ampersands(s).strip(),
1208 fatal=True, m3u8_id=None):
1209 manifest = self._download_xml(
1210 manifest_url, video_id, 'Downloading f4m manifest',
1211 'Unable to download f4m manifest',
1212 # Some manifests may be malformed, e.g. prosiebensat1 generated manifests
1213 # (see https://github.com/rg3/youtube-dl/issues/6215#issuecomment-121704244)
1214 transform_source=transform_source,
1215 fatal=fatal)
1216
1217 if manifest is False:
1218 return []
1219
1220 return self._parse_f4m_formats(
1221 manifest, manifest_url, video_id, preference=preference, f4m_id=f4m_id,
1222 transform_source=transform_source, fatal=fatal, m3u8_id=m3u8_id)
1223
1224 def _parse_f4m_formats(self, manifest, manifest_url, video_id, preference=None, f4m_id=None,
1225 transform_source=lambda s: fix_xml_ampersands(s).strip(),
1226 fatal=True, m3u8_id=None):
1227 # currently youtube-dl cannot decode the playerVerificationChallenge as Akamai uses Adobe Alchemy
1228 akamai_pv = manifest.find('{http://ns.adobe.com/f4m/1.0}pv-2.0')
1229 if akamai_pv is not None and ';' in akamai_pv.text:
1230 playerVerificationChallenge = akamai_pv.text.split(';')[0]
1231 if playerVerificationChallenge.strip() != '':
1232 return []
1233
1234 formats = []
1235 manifest_version = '1.0'
1236 media_nodes = manifest.findall('{http://ns.adobe.com/f4m/1.0}media')
1237 if not media_nodes:
1238 manifest_version = '2.0'
1239 media_nodes = manifest.findall('{http://ns.adobe.com/f4m/2.0}media')
1240 # Remove unsupported DRM protected media from final formats
1241 # rendition (see https://github.com/rg3/youtube-dl/issues/8573).
1242 media_nodes = remove_encrypted_media(media_nodes)
1243 if not media_nodes:
1244 return formats
1245
1246 manifest_base_url = get_base_url(manifest)
1247
1248 bootstrap_info = xpath_element(
1249 manifest, ['{http://ns.adobe.com/f4m/1.0}bootstrapInfo', '{http://ns.adobe.com/f4m/2.0}bootstrapInfo'],
1250 'bootstrap info', default=None)
1251
1252 vcodec = None
1253 mime_type = xpath_text(
1254 manifest, ['{http://ns.adobe.com/f4m/1.0}mimeType', '{http://ns.adobe.com/f4m/2.0}mimeType'],
1255 'base URL', default=None)
1256 if mime_type and mime_type.startswith('audio/'):
1257 vcodec = 'none'
1258
1259 for i, media_el in enumerate(media_nodes):
1260 tbr = int_or_none(media_el.attrib.get('bitrate'))
1261 width = int_or_none(media_el.attrib.get('width'))
1262 height = int_or_none(media_el.attrib.get('height'))
1263 format_id = '-'.join(filter(None, [f4m_id, compat_str(i if tbr is None else tbr)]))
1264 # If <bootstrapInfo> is present, the specified f4m is a
1265 # stream-level manifest, and only set-level manifests may refer to
1266 # external resources. See section 11.4 and section 4 of F4M spec
1267 if bootstrap_info is None:
1268 media_url = None
1269 # @href is introduced in 2.0, see section 11.6 of F4M spec
1270 if manifest_version == '2.0':
1271 media_url = media_el.attrib.get('href')
1272 if media_url is None:
1273 media_url = media_el.attrib.get('url')
1274 if not media_url:
1275 continue
1276 manifest_url = (
1277 media_url if media_url.startswith('http://') or media_url.startswith('https://')
1278 else ((manifest_base_url or '/'.join(manifest_url.split('/')[:-1])) + '/' + media_url))
1279 # If media_url is itself a f4m manifest do the recursive extraction
1280 # since bitrates in parent manifest (this one) and media_url manifest
1281 # may differ leading to inability to resolve the format by requested
1282 # bitrate in f4m downloader
1283 ext = determine_ext(manifest_url)
1284 if ext == 'f4m':
1285 f4m_formats = self._extract_f4m_formats(
1286 manifest_url, video_id, preference=preference, f4m_id=f4m_id,
1287 transform_source=transform_source, fatal=fatal)
1288 # Sometimes stream-level manifest contains single media entry that
1289 # does not contain any quality metadata (e.g. http://matchtv.ru/#live-player).
1290 # At the same time parent's media entry in set-level manifest may
1291 # contain it. We will copy it from parent in such cases.
1292 if len(f4m_formats) == 1:
1293 f = f4m_formats[0]
1294 f.update({
1295 'tbr': f.get('tbr') or tbr,
1296 'width': f.get('width') or width,
1297 'height': f.get('height') or height,
1298 'format_id': f.get('format_id') if not tbr else format_id,
1299 'vcodec': vcodec,
1300 })
1301 formats.extend(f4m_formats)
1302 continue
1303 elif ext == 'm3u8':
1304 formats.extend(self._extract_m3u8_formats(
1305 manifest_url, video_id, 'mp4', preference=preference,
1306 m3u8_id=m3u8_id, fatal=fatal))
1307 continue
1308 formats.append({
1309 'format_id': format_id,
1310 'url': manifest_url,
1311 'manifest_url': manifest_url,
1312 'ext': 'flv' if bootstrap_info is not None else None,
1313 'protocol': 'f4m',
1314 'tbr': tbr,
1315 'width': width,
1316 'height': height,
1317 'vcodec': vcodec,
1318 'preference': preference,
1319 })
1320 return formats
1321
1322 def _m3u8_meta_format(self, m3u8_url, ext=None, preference=None, m3u8_id=None):
1323 return {
1324 'format_id': '-'.join(filter(None, [m3u8_id, 'meta'])),
1325 'url': m3u8_url,
1326 'ext': ext,
1327 'protocol': 'm3u8',
1328 'preference': preference - 100 if preference else -100,
1329 'resolution': 'multiple',
1330 'format_note': 'Quality selection URL',
1331 }
1332
1333 def _extract_m3u8_formats(self, m3u8_url, video_id, ext=None,
1334 entry_protocol='m3u8', preference=None,
1335 m3u8_id=None, note=None, errnote=None,
1336 fatal=True, live=False):
1337 res = self._download_webpage_handle(
1338 m3u8_url, video_id,
1339 note=note or 'Downloading m3u8 information',
1340 errnote=errnote or 'Failed to download m3u8 information',
1341 fatal=fatal)
1342
1343 if res is False:
1344 return []
1345
1346 m3u8_doc, urlh = res
1347 m3u8_url = urlh.geturl()
1348
1349 return self._parse_m3u8_formats(
1350 m3u8_doc, m3u8_url, ext=ext, entry_protocol=entry_protocol,
1351 preference=preference, m3u8_id=m3u8_id, live=live)
1352
1353 def _parse_m3u8_formats(self, m3u8_doc, m3u8_url, ext=None,
1354 entry_protocol='m3u8', preference=None,
1355 m3u8_id=None, live=False):
1356 if '#EXT-X-FAXS-CM:' in m3u8_doc: # Adobe Flash Access
1357 return []
1358
1359 formats = []
1360
1361 format_url = lambda u: (
1362 u
1363 if re.match(r'^https?://', u)
1364 else compat_urlparse.urljoin(m3u8_url, u))
1365
1366 # References:
1367 # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-21
1368 # 2. https://github.com/rg3/youtube-dl/issues/12211
1369
1370 # We should try extracting formats only from master playlists [1, 4.3.4],
1371 # i.e. playlists that describe available qualities. On the other hand
1372 # media playlists [1, 4.3.3] should be returned as is since they contain
1373 # just the media without qualities renditions.
1374 # Fortunately, master playlist can be easily distinguished from media
1375 # playlist based on particular tags availability. As of [1, 4.3.3, 4.3.4]
1376 # master playlist tags MUST NOT appear in a media playist and vice versa.
1377 # As of [1, 4.3.3.1] #EXT-X-TARGETDURATION tag is REQUIRED for every
1378 # media playlist and MUST NOT appear in master playlist thus we can
1379 # clearly detect media playlist with this criterion.
1380
1381 if '#EXT-X-TARGETDURATION' in m3u8_doc: # media playlist, return as is
1382 return [{
1383 'url': m3u8_url,
1384 'format_id': m3u8_id,
1385 'ext': ext,
1386 'protocol': entry_protocol,
1387 'preference': preference,
1388 }]
1389
1390 groups = {}
1391 last_stream_inf = {}
1392
1393 def extract_media(x_media_line):
1394 media = parse_m3u8_attributes(x_media_line)
1395 # As per [1, 4.3.4.1] TYPE, GROUP-ID and NAME are REQUIRED
1396 media_type, group_id, name = media.get('TYPE'), media.get('GROUP-ID'), media.get('NAME')
1397 if not (media_type and group_id and name):
1398 return
1399 groups.setdefault(group_id, []).append(media)
1400 if media_type not in ('VIDEO', 'AUDIO'):
1401 return
1402 media_url = media.get('URI')
1403 if media_url:
1404 format_id = []
1405 for v in (m3u8_id, group_id, name):
1406 if v:
1407 format_id.append(v)
1408 f = {
1409 'format_id': '-'.join(format_id),
1410 'url': format_url(media_url),
1411 'manifest_url': m3u8_url,
1412 'language': media.get('LANGUAGE'),
1413 'ext': ext,
1414 'protocol': entry_protocol,
1415 'preference': preference,
1416 }
1417 if media_type == 'AUDIO':
1418 f['vcodec'] = 'none'
1419 formats.append(f)
1420
1421 def build_stream_name():
1422 # Despite specification does not mention NAME attribute for
1423 # EXT-X-STREAM-INF tag it still sometimes may be present (see [1]
1424 # or vidio test in TestInfoExtractor.test_parse_m3u8_formats)
1425 # 1. http://www.vidio.com/watch/165683-dj_ambred-booyah-live-2015
1426 stream_name = last_stream_inf.get('NAME')
1427 if stream_name:
1428 return stream_name
1429 # If there is no NAME in EXT-X-STREAM-INF it will be obtained
1430 # from corresponding rendition group
1431 stream_group_id = last_stream_inf.get('VIDEO')
1432 if not stream_group_id:
1433 return
1434 stream_group = groups.get(stream_group_id)
1435 if not stream_group:
1436 return stream_group_id
1437 rendition = stream_group[0]
1438 return rendition.get('NAME') or stream_group_id
1439
1440 for line in m3u8_doc.splitlines():
1441 if line.startswith('#EXT-X-STREAM-INF:'):
1442 last_stream_inf = parse_m3u8_attributes(line)
1443 elif line.startswith('#EXT-X-MEDIA:'):
1444 extract_media(line)
1445 elif line.startswith('#') or not line.strip():
1446 continue
1447 else:
1448 tbr = float_or_none(
1449 last_stream_inf.get('AVERAGE-BANDWIDTH') or
1450 last_stream_inf.get('BANDWIDTH'), scale=1000)
1451 format_id = []
1452 if m3u8_id:
1453 format_id.append(m3u8_id)
1454 stream_name = build_stream_name()
1455 # Bandwidth of live streams may differ over time thus making
1456 # format_id unpredictable. So it's better to keep provided
1457 # format_id intact.
1458 if not live:
1459 format_id.append(stream_name if stream_name else '%d' % (tbr if tbr else len(formats)))
1460 manifest_url = format_url(line.strip())
1461 f = {
1462 'format_id': '-'.join(format_id),
1463 'url': manifest_url,
1464 'manifest_url': m3u8_url,
1465 'tbr': tbr,
1466 'ext': ext,
1467 'fps': float_or_none(last_stream_inf.get('FRAME-RATE')),
1468 'protocol': entry_protocol,
1469 'preference': preference,
1470 }
1471 resolution = last_stream_inf.get('RESOLUTION')
1472 if resolution:
1473 mobj = re.search(r'(?P<width>\d+)[xX](?P<height>\d+)', resolution)
1474 if mobj:
1475 f['width'] = int(mobj.group('width'))
1476 f['height'] = int(mobj.group('height'))
1477 # Unified Streaming Platform
1478 mobj = re.search(
1479 r'audio.*?(?:%3D|=)(\d+)(?:-video.*?(?:%3D|=)(\d+))?', f['url'])
1480 if mobj:
1481 abr, vbr = mobj.groups()
1482 abr, vbr = float_or_none(abr, 1000), float_or_none(vbr, 1000)
1483 f.update({
1484 'vbr': vbr,
1485 'abr': abr,
1486 })
1487 codecs = parse_codecs(last_stream_inf.get('CODECS'))
1488 f.update(codecs)
1489 audio_group_id = last_stream_inf.get('AUDIO')
1490 # As per [1, 4.3.4.1.1] any EXT-X-STREAM-INF tag which
1491 # references a rendition group MUST have a CODECS attribute.
1492 # However, this is not always respected, for example, [2]
1493 # contains EXT-X-STREAM-INF tag which references AUDIO
1494 # rendition group but does not have CODECS and despite
1495 # referencing audio group an audio group, it represents
1496 # a complete (with audio and video) format. So, for such cases
1497 # we will ignore references to rendition groups and treat them
1498 # as complete formats.
1499 if audio_group_id and codecs and f.get('vcodec') != 'none':
1500 audio_group = groups.get(audio_group_id)
1501 if audio_group and audio_group[0].get('URI'):
1502 # TODO: update acodec for audio only formats with
1503 # the same GROUP-ID
1504 f['acodec'] = 'none'
1505 formats.append(f)
1506 last_stream_inf = {}
1507 return formats
1508
1509 @staticmethod
1510 def _xpath_ns(path, namespace=None):
1511 if not namespace:
1512 return path
1513 out = []
1514 for c in path.split('/'):
1515 if not c or c == '.':
1516 out.append(c)
1517 else:
1518 out.append('{%s}%s' % (namespace, c))
1519 return '/'.join(out)
1520
1521 def _extract_smil_formats(self, smil_url, video_id, fatal=True, f4m_params=None, transform_source=None):
1522 smil = self._download_smil(smil_url, video_id, fatal=fatal, transform_source=transform_source)
1523
1524 if smil is False:
1525 assert not fatal
1526 return []
1527
1528 namespace = self._parse_smil_namespace(smil)
1529
1530 return self._parse_smil_formats(
1531 smil, smil_url, video_id, namespace=namespace, f4m_params=f4m_params)
1532
1533 def _extract_smil_info(self, smil_url, video_id, fatal=True, f4m_params=None):
1534 smil = self._download_smil(smil_url, video_id, fatal=fatal)
1535 if smil is False:
1536 return {}
1537 return self._parse_smil(smil, smil_url, video_id, f4m_params=f4m_params)
1538
1539 def _download_smil(self, smil_url, video_id, fatal=True, transform_source=None):
1540 return self._download_xml(
1541 smil_url, video_id, 'Downloading SMIL file',
1542 'Unable to download SMIL file', fatal=fatal, transform_source=transform_source)
1543
1544 def _parse_smil(self, smil, smil_url, video_id, f4m_params=None):
1545 namespace = self._parse_smil_namespace(smil)
1546
1547 formats = self._parse_smil_formats(
1548 smil, smil_url, video_id, namespace=namespace, f4m_params=f4m_params)
1549 subtitles = self._parse_smil_subtitles(smil, namespace=namespace)
1550
1551 video_id = os.path.splitext(url_basename(smil_url))[0]
1552 title = None
1553 description = None
1554 upload_date = None
1555 for meta in smil.findall(self._xpath_ns('./head/meta', namespace)):
1556 name = meta.attrib.get('name')
1557 content = meta.attrib.get('content')
1558 if not name or not content:
1559 continue
1560 if not title and name == 'title':
1561 title = content
1562 elif not description and name in ('description', 'abstract'):
1563 description = content
1564 elif not upload_date and name == 'date':
1565 upload_date = unified_strdate(content)
1566
1567 thumbnails = [{
1568 'id': image.get('type'),
1569 'url': image.get('src'),
1570 'width': int_or_none(image.get('width')),
1571 'height': int_or_none(image.get('height')),
1572 } for image in smil.findall(self._xpath_ns('.//image', namespace)) if image.get('src')]
1573
1574 return {
1575 'id': video_id,
1576 'title': title or video_id,
1577 'description': description,
1578 'upload_date': upload_date,
1579 'thumbnails': thumbnails,
1580 'formats': formats,
1581 'subtitles': subtitles,
1582 }
1583
1584 def _parse_smil_namespace(self, smil):
1585 return self._search_regex(
1586 r'(?i)^{([^}]+)?}smil$', smil.tag, 'namespace', default=None)
1587
1588 def _parse_smil_formats(self, smil, smil_url, video_id, namespace=None, f4m_params=None, transform_rtmp_url=None):
1589 base = smil_url
1590 for meta in smil.findall(self._xpath_ns('./head/meta', namespace)):
1591 b = meta.get('base') or meta.get('httpBase')
1592 if b:
1593 base = b
1594 break
1595
1596 formats = []
1597 rtmp_count = 0
1598 http_count = 0
1599 m3u8_count = 0
1600
1601 srcs = []
1602 media = smil.findall(self._xpath_ns('.//video', namespace)) + smil.findall(self._xpath_ns('.//audio', namespace))
1603 for medium in media:
1604 src = medium.get('src')
1605 if not src or src in srcs:
1606 continue
1607 srcs.append(src)
1608
1609 bitrate = float_or_none(medium.get('system-bitrate') or medium.get('systemBitrate'), 1000)
1610 filesize = int_or_none(medium.get('size') or medium.get('fileSize'))
1611 width = int_or_none(medium.get('width'))
1612 height = int_or_none(medium.get('height'))
1613 proto = medium.get('proto')
1614 ext = medium.get('ext')
1615 src_ext = determine_ext(src)
1616 streamer = medium.get('streamer') or base
1617
1618 if proto == 'rtmp' or streamer.startswith('rtmp'):
1619 rtmp_count += 1
1620 formats.append({
1621 'url': streamer,
1622 'play_path': src,
1623 'ext': 'flv',
1624 'format_id': 'rtmp-%d' % (rtmp_count if bitrate is None else bitrate),
1625 'tbr': bitrate,
1626 'filesize': filesize,
1627 'width': width,
1628 'height': height,
1629 })
1630 if transform_rtmp_url:
1631 streamer, src = transform_rtmp_url(streamer, src)
1632 formats[-1].update({
1633 'url': streamer,
1634 'play_path': src,
1635 })
1636 continue
1637
1638 src_url = src if src.startswith('http') else compat_urlparse.urljoin(base, src)
1639 src_url = src_url.strip()
1640
1641 if proto == 'm3u8' or src_ext == 'm3u8':
1642 m3u8_formats = self._extract_m3u8_formats(
1643 src_url, video_id, ext or 'mp4', m3u8_id='hls', fatal=False)
1644 if len(m3u8_formats) == 1:
1645 m3u8_count += 1
1646 m3u8_formats[0].update({
1647 'format_id': 'hls-%d' % (m3u8_count if bitrate is None else bitrate),
1648 'tbr': bitrate,
1649 'width': width,
1650 'height': height,
1651 })
1652 formats.extend(m3u8_formats)
1653 continue
1654
1655 if src_ext == 'f4m':
1656 f4m_url = src_url
1657 if not f4m_params:
1658 f4m_params = {
1659 'hdcore': '3.2.0',
1660 'plugin': 'flowplayer-3.2.0.1',
1661 }
1662 f4m_url += '&' if '?' in f4m_url else '?'
1663 f4m_url += compat_urllib_parse_urlencode(f4m_params)
1664 formats.extend(self._extract_f4m_formats(f4m_url, video_id, f4m_id='hds', fatal=False))
1665 continue
1666
1667 if src_url.startswith('http') and self._is_valid_url(src, video_id):
1668 http_count += 1
1669 formats.append({
1670 'url': src_url,
1671 'ext': ext or src_ext or 'flv',
1672 'format_id': 'http-%d' % (bitrate or http_count),
1673 'tbr': bitrate,
1674 'filesize': filesize,
1675 'width': width,
1676 'height': height,
1677 })
1678 continue
1679
1680 return formats
1681
1682 def _parse_smil_subtitles(self, smil, namespace=None, subtitles_lang='en'):
1683 urls = []
1684 subtitles = {}
1685 for num, textstream in enumerate(smil.findall(self._xpath_ns('.//textstream', namespace))):
1686 src = textstream.get('src')
1687 if not src or src in urls:
1688 continue
1689 urls.append(src)
1690 ext = textstream.get('ext') or mimetype2ext(textstream.get('type')) or determine_ext(src)
1691 lang = textstream.get('systemLanguage') or textstream.get('systemLanguageName') or textstream.get('lang') or subtitles_lang
1692 subtitles.setdefault(lang, []).append({
1693 'url': src,
1694 'ext': ext,
1695 })
1696 return subtitles
1697
1698 def _extract_xspf_playlist(self, playlist_url, playlist_id, fatal=True):
1699 xspf = self._download_xml(
1700 playlist_url, playlist_id, 'Downloading xpsf playlist',
1701 'Unable to download xspf manifest', fatal=fatal)
1702 if xspf is False:
1703 return []
1704 return self._parse_xspf(xspf, playlist_id)
1705
1706 def _parse_xspf(self, playlist, playlist_id):
1707 NS_MAP = {
1708 'xspf': 'http://xspf.org/ns/0/',
1709 's1': 'http://static.streamone.nl/player/ns/0',
1710 }
1711
1712 entries = []
1713 for track in playlist.findall(xpath_with_ns('./xspf:trackList/xspf:track', NS_MAP)):
1714 title = xpath_text(
1715 track, xpath_with_ns('./xspf:title', NS_MAP), 'title', default=playlist_id)
1716 description = xpath_text(
1717 track, xpath_with_ns('./xspf:annotation', NS_MAP), 'description')
1718 thumbnail = xpath_text(
1719 track, xpath_with_ns('./xspf:image', NS_MAP), 'thumbnail')
1720 duration = float_or_none(
1721 xpath_text(track, xpath_with_ns('./xspf:duration', NS_MAP), 'duration'), 1000)
1722
1723 formats = [{
1724 'url': location.text,
1725 'format_id': location.get(xpath_with_ns('s1:label', NS_MAP)),
1726 'width': int_or_none(location.get(xpath_with_ns('s1:width', NS_MAP))),
1727 'height': int_or_none(location.get(xpath_with_ns('s1:height', NS_MAP))),
1728 } for location in track.findall(xpath_with_ns('./xspf:location', NS_MAP))]
1729 self._sort_formats(formats)
1730
1731 entries.append({
1732 'id': playlist_id,
1733 'title': title,
1734 'description': description,
1735 'thumbnail': thumbnail,
1736 'duration': duration,
1737 'formats': formats,
1738 })
1739 return entries
1740
1741 def _extract_mpd_formats(self, mpd_url, video_id, mpd_id=None, note=None, errnote=None, fatal=True, formats_dict={}):
1742 res = self._download_webpage_handle(
1743 mpd_url, video_id,
1744 note=note or 'Downloading MPD manifest',
1745 errnote=errnote or 'Failed to download MPD manifest',
1746 fatal=fatal)
1747 if res is False:
1748 return []
1749 mpd, urlh = res
1750 mpd_base_url = base_url(urlh.geturl())
1751
1752 return self._parse_mpd_formats(
1753 compat_etree_fromstring(mpd.encode('utf-8')), mpd_id, mpd_base_url,
1754 formats_dict=formats_dict, mpd_url=mpd_url)
1755
1756 def _parse_mpd_formats(self, mpd_doc, mpd_id=None, mpd_base_url='', formats_dict={}, mpd_url=None):
1757 """
1758 Parse formats from MPD manifest.
1759 References:
1760 1. MPEG-DASH Standard, ISO/IEC 23009-1:2014(E),
1761 http://standards.iso.org/ittf/PubliclyAvailableStandards/c065274_ISO_IEC_23009-1_2014.zip
1762 2. https://en.wikipedia.org/wiki/Dynamic_Adaptive_Streaming_over_HTTP
1763 """
1764 if mpd_doc.get('type') == 'dynamic':
1765 return []
1766
1767 namespace = self._search_regex(r'(?i)^{([^}]+)?}MPD$', mpd_doc.tag, 'namespace', default=None)
1768
1769 def _add_ns(path):
1770 return self._xpath_ns(path, namespace)
1771
1772 def is_drm_protected(element):
1773 return element.find(_add_ns('ContentProtection')) is not None
1774
1775 def extract_multisegment_info(element, ms_parent_info):
1776 ms_info = ms_parent_info.copy()
1777
1778 # As per [1, 5.3.9.2.2] SegmentList and SegmentTemplate share some
1779 # common attributes and elements. We will only extract relevant
1780 # for us.
1781 def extract_common(source):
1782 segment_timeline = source.find(_add_ns('SegmentTimeline'))
1783 if segment_timeline is not None:
1784 s_e = segment_timeline.findall(_add_ns('S'))
1785 if s_e:
1786 ms_info['total_number'] = 0
1787 ms_info['s'] = []
1788 for s in s_e:
1789 r = int(s.get('r', 0))
1790 ms_info['total_number'] += 1 + r
1791 ms_info['s'].append({
1792 't': int(s.get('t', 0)),
1793 # @d is mandatory (see [1, 5.3.9.6.2, Table 17, page 60])
1794 'd': int(s.attrib['d']),
1795 'r': r,
1796 })
1797 start_number = source.get('startNumber')
1798 if start_number:
1799 ms_info['start_number'] = int(start_number)
1800 timescale = source.get('timescale')
1801 if timescale:
1802 ms_info['timescale'] = int(timescale)
1803 segment_duration = source.get('duration')
1804 if segment_duration:
1805 ms_info['segment_duration'] = float(segment_duration)
1806
1807 def extract_Initialization(source):
1808 initialization = source.find(_add_ns('Initialization'))
1809 if initialization is not None:
1810 ms_info['initialization_url'] = initialization.attrib['sourceURL']
1811
1812 segment_list = element.find(_add_ns('SegmentList'))
1813 if segment_list is not None:
1814 extract_common(segment_list)
1815 extract_Initialization(segment_list)
1816 segment_urls_e = segment_list.findall(_add_ns('SegmentURL'))
1817 if segment_urls_e:
1818 ms_info['segment_urls'] = [segment.attrib['media'] for segment in segment_urls_e]
1819 else:
1820 segment_template = element.find(_add_ns('SegmentTemplate'))
1821 if segment_template is not None:
1822 extract_common(segment_template)
1823 media = segment_template.get('media')
1824 if media:
1825 ms_info['media'] = media
1826 initialization = segment_template.get('initialization')
1827 if initialization:
1828 ms_info['initialization'] = initialization
1829 else:
1830 extract_Initialization(segment_template)
1831 return ms_info
1832
1833 mpd_duration = parse_duration(mpd_doc.get('mediaPresentationDuration'))
1834 formats = []
1835 for period in mpd_doc.findall(_add_ns('Period')):
1836 period_duration = parse_duration(period.get('duration')) or mpd_duration
1837 period_ms_info = extract_multisegment_info(period, {
1838 'start_number': 1,
1839 'timescale': 1,
1840 })
1841 for adaptation_set in period.findall(_add_ns('AdaptationSet')):
1842 if is_drm_protected(adaptation_set):
1843 continue
1844 adaption_set_ms_info = extract_multisegment_info(adaptation_set, period_ms_info)
1845 for representation in adaptation_set.findall(_add_ns('Representation')):
1846 if is_drm_protected(representation):
1847 continue
1848 representation_attrib = adaptation_set.attrib.copy()
1849 representation_attrib.update(representation.attrib)
1850 # According to [1, 5.3.7.2, Table 9, page 41], @mimeType is mandatory
1851 mime_type = representation_attrib['mimeType']
1852 content_type = mime_type.split('/')[0]
1853 if content_type == 'text':
1854 # TODO implement WebVTT downloading
1855 pass
1856 elif content_type in ('video', 'audio'):
1857 base_url = ''
1858 for element in (representation, adaptation_set, period, mpd_doc):
1859 base_url_e = element.find(_add_ns('BaseURL'))
1860 if base_url_e is not None:
1861 base_url = base_url_e.text + base_url
1862 if re.match(r'^https?://', base_url):
1863 break
1864 if mpd_base_url and not re.match(r'^https?://', base_url):
1865 if not mpd_base_url.endswith('/') and not base_url.startswith('/'):
1866 mpd_base_url += '/'
1867 base_url = mpd_base_url + base_url
1868 representation_id = representation_attrib.get('id')
1869 lang = representation_attrib.get('lang')
1870 url_el = representation.find(_add_ns('BaseURL'))
1871 filesize = int_or_none(url_el.attrib.get('{http://youtube.com/yt/2012/10/10}contentLength') if url_el is not None else None)
1872 bandwidth = int_or_none(representation_attrib.get('bandwidth'))
1873 f = {
1874 'format_id': '%s-%s' % (mpd_id, representation_id) if mpd_id else representation_id,
1875 'url': base_url,
1876 'manifest_url': mpd_url,
1877 'ext': mimetype2ext(mime_type),
1878 'width': int_or_none(representation_attrib.get('width')),
1879 'height': int_or_none(representation_attrib.get('height')),
1880 'tbr': float_or_none(bandwidth, 1000),
1881 'asr': int_or_none(representation_attrib.get('audioSamplingRate')),
1882 'fps': int_or_none(representation_attrib.get('frameRate')),
1883 'language': lang if lang not in ('mul', 'und', 'zxx', 'mis') else None,
1884 'format_note': 'DASH %s' % content_type,
1885 'filesize': filesize,
1886 }
1887 f.update(parse_codecs(representation_attrib.get('codecs')))
1888 representation_ms_info = extract_multisegment_info(representation, adaption_set_ms_info)
1889
1890 def prepare_template(template_name, identifiers):
1891 t = representation_ms_info[template_name]
1892 t = t.replace('$RepresentationID$', representation_id)
1893 t = re.sub(r'\$(%s)\$' % '|'.join(identifiers), r'%(\1)d', t)
1894 t = re.sub(r'\$(%s)%%([^$]+)\$' % '|'.join(identifiers), r'%(\1)\2', t)
1895 t.replace('$$', '$')
1896 return t
1897
1898 # @initialization is a regular template like @media one
1899 # so it should be handled just the same way (see
1900 # https://github.com/rg3/youtube-dl/issues/11605)
1901 if 'initialization' in representation_ms_info:
1902 initialization_template = prepare_template(
1903 'initialization',
1904 # As per [1, 5.3.9.4.2, Table 15, page 54] $Number$ and
1905 # $Time$ shall not be included for @initialization thus
1906 # only $Bandwidth$ remains
1907 ('Bandwidth', ))
1908 representation_ms_info['initialization_url'] = initialization_template % {
1909 'Bandwidth': bandwidth,
1910 }
1911
1912 def location_key(location):
1913 return 'url' if re.match(r'^https?://', location) else 'path'
1914
1915 if 'segment_urls' not in representation_ms_info and 'media' in representation_ms_info:
1916
1917 media_template = prepare_template('media', ('Number', 'Bandwidth', 'Time'))
1918 media_location_key = location_key(media_template)
1919
1920 # As per [1, 5.3.9.4.4, Table 16, page 55] $Number$ and $Time$
1921 # can't be used at the same time
1922 if '%(Number' in media_template and 's' not in representation_ms_info:
1923 segment_duration = None
1924 if 'total_number' not in representation_ms_info and 'segment_duration' in representation_ms_info:
1925 segment_duration = float_or_none(representation_ms_info['segment_duration'], representation_ms_info['timescale'])
1926 representation_ms_info['total_number'] = int(math.ceil(float(period_duration) / segment_duration))
1927 representation_ms_info['fragments'] = [{
1928 media_location_key: media_template % {
1929 'Number': segment_number,
1930 'Bandwidth': bandwidth,
1931 },
1932 'duration': segment_duration,
1933 } for segment_number in range(
1934 representation_ms_info['start_number'],
1935 representation_ms_info['total_number'] + representation_ms_info['start_number'])]
1936 else:
1937 # $Number*$ or $Time$ in media template with S list available
1938 # Example $Number*$: http://www.svtplay.se/klipp/9023742/stopptid-om-bjorn-borg
1939 # Example $Time$: https://play.arkena.com/embed/avp/v2/player/media/b41dda37-d8e7-4d3f-b1b5-9a9db578bdfe/1/129411
1940 representation_ms_info['fragments'] = []
1941 segment_time = 0
1942 segment_d = None
1943 segment_number = representation_ms_info['start_number']
1944
1945 def add_segment_url():
1946 segment_url = media_template % {
1947 'Time': segment_time,
1948 'Bandwidth': bandwidth,
1949 'Number': segment_number,
1950 }
1951 representation_ms_info['fragments'].append({
1952 media_location_key: segment_url,
1953 'duration': float_or_none(segment_d, representation_ms_info['timescale']),
1954 })
1955
1956 for num, s in enumerate(representation_ms_info['s']):
1957 segment_time = s.get('t') or segment_time
1958 segment_d = s['d']
1959 add_segment_url()
1960 segment_number += 1
1961 for r in range(s.get('r', 0)):
1962 segment_time += segment_d
1963 add_segment_url()
1964 segment_number += 1
1965 segment_time += segment_d
1966 elif 'segment_urls' in representation_ms_info and 's' in representation_ms_info:
1967 # No media template
1968 # Example: https://www.youtube.com/watch?v=iXZV5uAYMJI
1969 # or any YouTube dashsegments video
1970 fragments = []
1971 segment_index = 0
1972 timescale = representation_ms_info['timescale']
1973 for s in representation_ms_info['s']:
1974 duration = float_or_none(s['d'], timescale)
1975 for r in range(s.get('r', 0) + 1):
1976 segment_uri = representation_ms_info['segment_urls'][segment_index]
1977 fragments.append({
1978 location_key(segment_uri): segment_uri,
1979 'duration': duration,
1980 })
1981 segment_index += 1
1982 representation_ms_info['fragments'] = fragments
1983 # NB: MPD manifest may contain direct URLs to unfragmented media.
1984 # No fragments key is present in this case.
1985 if 'fragments' in representation_ms_info:
1986 f.update({
1987 'fragment_base_url': base_url,
1988 'fragments': [],
1989 'protocol': 'http_dash_segments',
1990 })
1991 if 'initialization_url' in representation_ms_info:
1992 initialization_url = representation_ms_info['initialization_url']
1993 if not f.get('url'):
1994 f['url'] = initialization_url
1995 f['fragments'].append({location_key(initialization_url): initialization_url})
1996 f['fragments'].extend(representation_ms_info['fragments'])
1997 try:
1998 existing_format = next(
1999 fo for fo in formats
2000 if fo['format_id'] == representation_id)
2001 except StopIteration:
2002 full_info = formats_dict.get(representation_id, {}).copy()
2003 full_info.update(f)
2004 formats.append(full_info)
2005 else:
2006 existing_format.update(f)
2007 else:
2008 self.report_warning('Unknown MIME type %s in DASH manifest' % mime_type)
2009 return formats
2010
2011 def _extract_ism_formats(self, ism_url, video_id, ism_id=None, note=None, errnote=None, fatal=True):
2012 res = self._download_webpage_handle(
2013 ism_url, video_id,
2014 note=note or 'Downloading ISM manifest',
2015 errnote=errnote or 'Failed to download ISM manifest',
2016 fatal=fatal)
2017 if res is False:
2018 return []
2019 ism, urlh = res
2020
2021 return self._parse_ism_formats(
2022 compat_etree_fromstring(ism.encode('utf-8')), urlh.geturl(), ism_id)
2023
2024 def _parse_ism_formats(self, ism_doc, ism_url, ism_id=None):
2025 """
2026 Parse formats from ISM manifest.
2027 References:
2028 1. [MS-SSTR]: Smooth Streaming Protocol,
2029 https://msdn.microsoft.com/en-us/library/ff469518.aspx
2030 """
2031 if ism_doc.get('IsLive') == 'TRUE' or ism_doc.find('Protection') is not None:
2032 return []
2033
2034 duration = int(ism_doc.attrib['Duration'])
2035 timescale = int_or_none(ism_doc.get('TimeScale')) or 10000000
2036
2037 formats = []
2038 for stream in ism_doc.findall('StreamIndex'):
2039 stream_type = stream.get('Type')
2040 if stream_type not in ('video', 'audio'):
2041 continue
2042 url_pattern = stream.attrib['Url']
2043 stream_timescale = int_or_none(stream.get('TimeScale')) or timescale
2044 stream_name = stream.get('Name')
2045 for track in stream.findall('QualityLevel'):
2046 fourcc = track.get('FourCC')
2047 # TODO: add support for WVC1 and WMAP
2048 if fourcc not in ('H264', 'AVC1', 'AACL'):
2049 self.report_warning('%s is not a supported codec' % fourcc)
2050 continue
2051 tbr = int(track.attrib['Bitrate']) // 1000
2052 # [1] does not mention Width and Height attributes. However,
2053 # they're often present while MaxWidth and MaxHeight are
2054 # missing, so should be used as fallbacks
2055 width = int_or_none(track.get('MaxWidth') or track.get('Width'))
2056 height = int_or_none(track.get('MaxHeight') or track.get('Height'))
2057 sampling_rate = int_or_none(track.get('SamplingRate'))
2058
2059 track_url_pattern = re.sub(r'{[Bb]itrate}', track.attrib['Bitrate'], url_pattern)
2060 track_url_pattern = compat_urlparse.urljoin(ism_url, track_url_pattern)
2061
2062 fragments = []
2063 fragment_ctx = {
2064 'time': 0,
2065 }
2066 stream_fragments = stream.findall('c')
2067 for stream_fragment_index, stream_fragment in enumerate(stream_fragments):
2068 fragment_ctx['time'] = int_or_none(stream_fragment.get('t')) or fragment_ctx['time']
2069 fragment_repeat = int_or_none(stream_fragment.get('r')) or 1
2070 fragment_ctx['duration'] = int_or_none(stream_fragment.get('d'))
2071 if not fragment_ctx['duration']:
2072 try:
2073 next_fragment_time = int(stream_fragment[stream_fragment_index + 1].attrib['t'])
2074 except IndexError:
2075 next_fragment_time = duration
2076 fragment_ctx['duration'] = (next_fragment_time - fragment_ctx['time']) / fragment_repeat
2077 for _ in range(fragment_repeat):
2078 fragments.append({
2079 'url': re.sub(r'{start[ _]time}', compat_str(fragment_ctx['time']), track_url_pattern),
2080 'duration': fragment_ctx['duration'] / stream_timescale,
2081 })
2082 fragment_ctx['time'] += fragment_ctx['duration']
2083
2084 format_id = []
2085 if ism_id:
2086 format_id.append(ism_id)
2087 if stream_name:
2088 format_id.append(stream_name)
2089 format_id.append(compat_str(tbr))
2090
2091 formats.append({
2092 'format_id': '-'.join(format_id),
2093 'url': ism_url,
2094 'manifest_url': ism_url,
2095 'ext': 'ismv' if stream_type == 'video' else 'isma',
2096 'width': width,
2097 'height': height,
2098 'tbr': tbr,
2099 'asr': sampling_rate,
2100 'vcodec': 'none' if stream_type == 'audio' else fourcc,
2101 'acodec': 'none' if stream_type == 'video' else fourcc,
2102 'protocol': 'ism',
2103 'fragments': fragments,
2104 '_download_params': {
2105 'duration': duration,
2106 'timescale': stream_timescale,
2107 'width': width or 0,
2108 'height': height or 0,
2109 'fourcc': fourcc,
2110 'codec_private_data': track.get('CodecPrivateData'),
2111 'sampling_rate': sampling_rate,
2112 'channels': int_or_none(track.get('Channels', 2)),
2113 'bits_per_sample': int_or_none(track.get('BitsPerSample', 16)),
2114 'nal_unit_length_field': int_or_none(track.get('NALUnitLengthField', 4)),
2115 },
2116 })
2117 return formats
2118
2119 def _parse_html5_media_entries(self, base_url, webpage, video_id, m3u8_id=None, m3u8_entry_protocol='m3u8', mpd_id=None, preference=None):
2120 def absolute_url(video_url):
2121 return compat_urlparse.urljoin(base_url, video_url)
2122
2123 def parse_content_type(content_type):
2124 if not content_type:
2125 return {}
2126 ctr = re.search(r'(?P<mimetype>[^/]+/[^;]+)(?:;\s*codecs="?(?P<codecs>[^"]+))?', content_type)
2127 if ctr:
2128 mimetype, codecs = ctr.groups()
2129 f = parse_codecs(codecs)
2130 f['ext'] = mimetype2ext(mimetype)
2131 return f
2132 return {}
2133
2134 def _media_formats(src, cur_media_type, type_info={}):
2135 full_url = absolute_url(src)
2136 ext = type_info.get('ext') or determine_ext(full_url)
2137 if ext == 'm3u8':
2138 is_plain_url = False
2139 formats = self._extract_m3u8_formats(
2140 full_url, video_id, ext='mp4',
2141 entry_protocol=m3u8_entry_protocol, m3u8_id=m3u8_id,
2142 preference=preference, fatal=False)
2143 elif ext == 'mpd':
2144 is_plain_url = False
2145 formats = self._extract_mpd_formats(
2146 full_url, video_id, mpd_id=mpd_id, fatal=False)
2147 else:
2148 is_plain_url = True
2149 formats = [{
2150 'url': full_url,
2151 'vcodec': 'none' if cur_media_type == 'audio' else None,
2152 }]
2153 return is_plain_url, formats
2154
2155 entries = []
2156 # amp-video and amp-audio are very similar to their HTML5 counterparts
2157 # so we wll include them right here (see
2158 # https://www.ampproject.org/docs/reference/components/amp-video)
2159 media_tags = [(media_tag, media_type, '')
2160 for media_tag, media_type
2161 in re.findall(r'(?s)(<(?:amp-)?(video|audio)[^>]*/>)', webpage)]
2162 media_tags.extend(re.findall(
2163 # We only allow video|audio followed by a whitespace or '>'.
2164 # Allowing more characters may end up in significant slow down (see
2165 # https://github.com/rg3/youtube-dl/issues/11979, example URL:
2166 # http://www.porntrex.com/maps/videositemap.xml).
2167 r'(?s)(<(?P<tag>(?:amp-)?(?:video|audio))(?:\s+[^>]*)?>)(.*?)</(?P=tag)>', webpage))
2168 for media_tag, media_type, media_content in media_tags:
2169 media_info = {
2170 'formats': [],
2171 'subtitles': {},
2172 }
2173 media_attributes = extract_attributes(media_tag)
2174 src = media_attributes.get('src')
2175 if src:
2176 _, formats = _media_formats(src, media_type)
2177 media_info['formats'].extend(formats)
2178 media_info['thumbnail'] = media_attributes.get('poster')
2179 if media_content:
2180 for source_tag in re.findall(r'<source[^>]+>', media_content):
2181 source_attributes = extract_attributes(source_tag)
2182 src = source_attributes.get('src')
2183 if not src:
2184 continue
2185 f = parse_content_type(source_attributes.get('type'))
2186 is_plain_url, formats = _media_formats(src, media_type, f)
2187 if is_plain_url:
2188 # res attribute is not standard but seen several times
2189 # in the wild
2190 f.update({
2191 'height': int_or_none(source_attributes.get('res')),
2192 'format_id': source_attributes.get('label'),
2193 })
2194 f.update(formats[0])
2195 media_info['formats'].append(f)
2196 else:
2197 media_info['formats'].extend(formats)
2198 for track_tag in re.findall(r'<track[^>]+>', media_content):
2199 track_attributes = extract_attributes(track_tag)
2200 kind = track_attributes.get('kind')
2201 if not kind or kind in ('subtitles', 'captions'):
2202 src = track_attributes.get('src')
2203 if not src:
2204 continue
2205 lang = track_attributes.get('srclang') or track_attributes.get('lang') or track_attributes.get('label')
2206 media_info['subtitles'].setdefault(lang, []).append({
2207 'url': absolute_url(src),
2208 })
2209 if media_info['formats'] or media_info['subtitles']:
2210 entries.append(media_info)
2211 return entries
2212
2213 def _extract_akamai_formats(self, manifest_url, video_id, hosts={}):
2214 formats = []
2215 hdcore_sign = 'hdcore=3.7.0'
2216 f4m_url = re.sub(r'(https?://[^/]+)/i/', r'\1/z/', manifest_url).replace('/master.m3u8', '/manifest.f4m')
2217 hds_host = hosts.get('hds')
2218 if hds_host:
2219 f4m_url = re.sub(r'(https?://)[^/]+', r'\1' + hds_host, f4m_url)
2220 if 'hdcore=' not in f4m_url:
2221 f4m_url += ('&' if '?' in f4m_url else '?') + hdcore_sign
2222 f4m_formats = self._extract_f4m_formats(
2223 f4m_url, video_id, f4m_id='hds', fatal=False)
2224 for entry in f4m_formats:
2225 entry.update({'extra_param_to_segment_url': hdcore_sign})
2226 formats.extend(f4m_formats)
2227 m3u8_url = re.sub(r'(https?://[^/]+)/z/', r'\1/i/', manifest_url).replace('/manifest.f4m', '/master.m3u8')
2228 hls_host = hosts.get('hls')
2229 if hls_host:
2230 m3u8_url = re.sub(r'(https?://)[^/]+', r'\1' + hls_host, m3u8_url)
2231 formats.extend(self._extract_m3u8_formats(
2232 m3u8_url, video_id, 'mp4', 'm3u8_native',
2233 m3u8_id='hls', fatal=False))
2234 return formats
2235
2236 def _extract_wowza_formats(self, url, video_id, m3u8_entry_protocol='m3u8_native', skip_protocols=[]):
2237 query = compat_urlparse.urlparse(url).query
2238 url = re.sub(r'/(?:manifest|playlist|jwplayer)\.(?:m3u8|f4m|mpd|smil)', '', url)
2239 url_base = self._search_regex(
2240 r'(?:(?:https?|rtmp|rtsp):)?(//[^?]+)', url, 'format url')
2241 http_base_url = '%s:%s' % ('http', url_base)
2242 formats = []
2243
2244 def manifest_url(manifest):
2245 m_url = '%s/%s' % (http_base_url, manifest)
2246 if query:
2247 m_url += '?%s' % query
2248 return m_url
2249
2250 if 'm3u8' not in skip_protocols:
2251 formats.extend(self._extract_m3u8_formats(
2252 manifest_url('playlist.m3u8'), video_id, 'mp4',
2253 m3u8_entry_protocol, m3u8_id='hls', fatal=False))
2254 if 'f4m' not in skip_protocols:
2255 formats.extend(self._extract_f4m_formats(
2256 manifest_url('manifest.f4m'),
2257 video_id, f4m_id='hds', fatal=False))
2258 if 'dash' not in skip_protocols:
2259 formats.extend(self._extract_mpd_formats(
2260 manifest_url('manifest.mpd'),
2261 video_id, mpd_id='dash', fatal=False))
2262 if re.search(r'(?:/smil:|\.smil)', url_base):
2263 if 'smil' not in skip_protocols:
2264 rtmp_formats = self._extract_smil_formats(
2265 manifest_url('jwplayer.smil'),
2266 video_id, fatal=False)
2267 for rtmp_format in rtmp_formats:
2268 rtsp_format = rtmp_format.copy()
2269 rtsp_format['url'] = '%s/%s' % (rtmp_format['url'], rtmp_format['play_path'])
2270 del rtsp_format['play_path']
2271 del rtsp_format['ext']
2272 rtsp_format.update({
2273 'url': rtsp_format['url'].replace('rtmp://', 'rtsp://'),
2274 'format_id': rtmp_format['format_id'].replace('rtmp', 'rtsp'),
2275 'protocol': 'rtsp',
2276 })
2277 formats.extend([rtmp_format, rtsp_format])
2278 else:
2279 for protocol in ('rtmp', 'rtsp'):
2280 if protocol not in skip_protocols:
2281 formats.append({
2282 'url': '%s:%s' % (protocol, url_base),
2283 'format_id': protocol,
2284 'protocol': protocol,
2285 })
2286 return formats
2287
2288 def _find_jwplayer_data(self, webpage, video_id=None, transform_source=js_to_json):
2289 mobj = re.search(
2290 r'(?s)jwplayer\((?P<quote>[\'"])[^\'" ]+(?P=quote)\)(?!</script>).*?\.setup\s*\((?P<options>[^)]+)\)',
2291 webpage)
2292 if mobj:
2293 try:
2294 jwplayer_data = self._parse_json(mobj.group('options'),
2295 video_id=video_id,
2296 transform_source=transform_source)
2297 except ExtractorError:
2298 pass
2299 else:
2300 if isinstance(jwplayer_data, dict):
2301 return jwplayer_data
2302
2303 def _extract_jwplayer_data(self, webpage, video_id, *args, **kwargs):
2304 jwplayer_data = self._find_jwplayer_data(
2305 webpage, video_id, transform_source=js_to_json)
2306 return self._parse_jwplayer_data(
2307 jwplayer_data, video_id, *args, **kwargs)
2308
2309 def _parse_jwplayer_data(self, jwplayer_data, video_id=None, require_title=True,
2310 m3u8_id=None, mpd_id=None, rtmp_params=None, base_url=None):
2311 # JWPlayer backward compatibility: flattened playlists
2312 # https://github.com/jwplayer/jwplayer/blob/v7.4.3/src/js/api/config.js#L81-L96
2313 if 'playlist' not in jwplayer_data:
2314 jwplayer_data = {'playlist': [jwplayer_data]}
2315
2316 entries = []
2317
2318 # JWPlayer backward compatibility: single playlist item
2319 # https://github.com/jwplayer/jwplayer/blob/v7.7.0/src/js/playlist/playlist.js#L10
2320 if not isinstance(jwplayer_data['playlist'], list):
2321 jwplayer_data['playlist'] = [jwplayer_data['playlist']]
2322
2323 for video_data in jwplayer_data['playlist']:
2324 # JWPlayer backward compatibility: flattened sources
2325 # https://github.com/jwplayer/jwplayer/blob/v7.4.3/src/js/playlist/item.js#L29-L35
2326 if 'sources' not in video_data:
2327 video_data['sources'] = [video_data]
2328
2329 this_video_id = video_id or video_data['mediaid']
2330
2331 formats = self._parse_jwplayer_formats(
2332 video_data['sources'], video_id=this_video_id, m3u8_id=m3u8_id,
2333 mpd_id=mpd_id, rtmp_params=rtmp_params, base_url=base_url)
2334
2335 subtitles = {}
2336 tracks = video_data.get('tracks')
2337 if tracks and isinstance(tracks, list):
2338 for track in tracks:
2339 if not isinstance(track, dict):
2340 continue
2341 if track.get('kind') != 'captions':
2342 continue
2343 track_url = urljoin(base_url, track.get('file'))
2344 if not track_url:
2345 continue
2346 subtitles.setdefault(track.get('label') or 'en', []).append({
2347 'url': self._proto_relative_url(track_url)
2348 })
2349
2350 entry = {
2351 'id': this_video_id,
2352 'title': unescapeHTML(video_data['title'] if require_title else video_data.get('title')),
2353 'description': video_data.get('description'),
2354 'thumbnail': self._proto_relative_url(video_data.get('image')),
2355 'timestamp': int_or_none(video_data.get('pubdate')),
2356 'duration': float_or_none(jwplayer_data.get('duration') or video_data.get('duration')),
2357 'subtitles': subtitles,
2358 }
2359 # https://github.com/jwplayer/jwplayer/blob/master/src/js/utils/validator.js#L32
2360 if len(formats) == 1 and re.search(r'^(?:http|//).*(?:youtube\.com|youtu\.be)/.+', formats[0]['url']):
2361 entry.update({
2362 '_type': 'url_transparent',
2363 'url': formats[0]['url'],
2364 })
2365 else:
2366 self._sort_formats(formats)
2367 entry['formats'] = formats
2368 entries.append(entry)
2369 if len(entries) == 1:
2370 return entries[0]
2371 else:
2372 return self.playlist_result(entries)
2373
2374 def _parse_jwplayer_formats(self, jwplayer_sources_data, video_id=None,
2375 m3u8_id=None, mpd_id=None, rtmp_params=None, base_url=None):
2376 urls = []
2377 formats = []
2378 for source in jwplayer_sources_data:
2379 if not isinstance(source, dict):
2380 continue
2381 source_url = self._proto_relative_url(source.get('file'))
2382 if not source_url:
2383 continue
2384 if base_url:
2385 source_url = compat_urlparse.urljoin(base_url, source_url)
2386 if source_url in urls:
2387 continue
2388 urls.append(source_url)
2389 source_type = source.get('type') or ''
2390 ext = mimetype2ext(source_type) or determine_ext(source_url)
2391 if source_type == 'hls' or ext == 'm3u8':
2392 formats.extend(self._extract_m3u8_formats(
2393 source_url, video_id, 'mp4', entry_protocol='m3u8_native',
2394 m3u8_id=m3u8_id, fatal=False))
2395 elif ext == 'mpd':
2396 formats.extend(self._extract_mpd_formats(
2397 source_url, video_id, mpd_id=mpd_id, fatal=False))
2398 elif ext == 'smil':
2399 formats.extend(self._extract_smil_formats(
2400 source_url, video_id, fatal=False))
2401 # https://github.com/jwplayer/jwplayer/blob/master/src/js/providers/default.js#L67
2402 elif source_type.startswith('audio') or ext in (
2403 'oga', 'aac', 'mp3', 'mpeg', 'vorbis'):
2404 formats.append({
2405 'url': source_url,
2406 'vcodec': 'none',
2407 'ext': ext,
2408 })
2409 else:
2410 height = int_or_none(source.get('height'))
2411 if height is None:
2412 # Often no height is provided but there is a label in
2413 # format like "1080p", "720p SD", or 1080.
2414 height = int_or_none(self._search_regex(
2415 r'^(\d{3,4})[pP]?(?:\b|$)', compat_str(source.get('label') or ''),
2416 'height', default=None))
2417 a_format = {
2418 'url': source_url,
2419 'width': int_or_none(source.get('width')),
2420 'height': height,
2421 'tbr': int_or_none(source.get('bitrate')),
2422 'ext': ext,
2423 }
2424 if source_url.startswith('rtmp'):
2425 a_format['ext'] = 'flv'
2426 # See com/longtailvideo/jwplayer/media/RTMPMediaProvider.as
2427 # of jwplayer.flash.swf
2428 rtmp_url_parts = re.split(
2429 r'((?:mp4|mp3|flv):)', source_url, 1)
2430 if len(rtmp_url_parts) == 3:
2431 rtmp_url, prefix, play_path = rtmp_url_parts
2432 a_format.update({
2433 'url': rtmp_url,
2434 'play_path': prefix + play_path,
2435 })
2436 if rtmp_params:
2437 a_format.update(rtmp_params)
2438 formats.append(a_format)
2439 return formats
2440
2441 def _live_title(self, name):
2442 """ Generate the title for a live video """
2443 now = datetime.datetime.now()
2444 now_str = now.strftime('%Y-%m-%d %H:%M')
2445 return name + ' ' + now_str
2446
2447 def _int(self, v, name, fatal=False, **kwargs):
2448 res = int_or_none(v, **kwargs)
2449 if 'get_attr' in kwargs:
2450 print(getattr(v, kwargs['get_attr']))
2451 if res is None:
2452 msg = 'Failed to extract %s: Could not parse value %r' % (name, v)
2453 if fatal:
2454 raise ExtractorError(msg)
2455 else:
2456 self._downloader.report_warning(msg)
2457 return res
2458
2459 def _float(self, v, name, fatal=False, **kwargs):
2460 res = float_or_none(v, **kwargs)
2461 if res is None:
2462 msg = 'Failed to extract %s: Could not parse value %r' % (name, v)
2463 if fatal:
2464 raise ExtractorError(msg)
2465 else:
2466 self._downloader.report_warning(msg)
2467 return res
2468
2469 def _set_cookie(self, domain, name, value, expire_time=None, port=None,
2470 path='/', secure=False, discard=False, rest={}, **kwargs):
2471 cookie = compat_cookiejar.Cookie(
2472 0, name, value, port, port is not None, domain, True,
2473 domain.startswith('.'), path, True, secure, expire_time,
2474 discard, None, None, rest)
2475 self._downloader.cookiejar.set_cookie(cookie)
2476
2477 def _get_cookies(self, url):
2478 """ Return a compat_cookies.SimpleCookie with the cookies for the url """
2479 req = sanitized_Request(url)
2480 self._downloader.cookiejar.add_cookie_header(req)
2481 return compat_cookies.SimpleCookie(req.get_header('Cookie'))
2482
2483 def get_testcases(self, include_onlymatching=False):
2484 t = getattr(self, '_TEST', None)
2485 if t:
2486 assert not hasattr(self, '_TESTS'), \
2487 '%s has _TEST and _TESTS' % type(self).__name__
2488 tests = [t]
2489 else:
2490 tests = getattr(self, '_TESTS', [])
2491 for t in tests:
2492 if not include_onlymatching and t.get('only_matching', False):
2493 continue
2494 t['name'] = type(self).__name__[:-len('IE')]
2495 yield t
2496
2497 def is_suitable(self, age_limit):
2498 """ Test whether the extractor is generally suitable for the given
2499 age limit (i.e. pornographic sites are not, all others usually are) """
2500
2501 any_restricted = False
2502 for tc in self.get_testcases(include_onlymatching=False):
2503 if tc.get('playlist', []):
2504 tc = tc['playlist'][0]
2505 is_restricted = age_restricted(
2506 tc.get('info_dict', {}).get('age_limit'), age_limit)
2507 if not is_restricted:
2508 return True
2509 any_restricted = any_restricted or is_restricted
2510 return not any_restricted
2511
2512 def extract_subtitles(self, *args, **kwargs):
2513 if (self._downloader.params.get('writesubtitles', False) or
2514 self._downloader.params.get('listsubtitles')):
2515 return self._get_subtitles(*args, **kwargs)
2516 return {}
2517
2518 def _get_subtitles(self, *args, **kwargs):
2519 raise NotImplementedError('This method must be implemented by subclasses')
2520
2521 @staticmethod
2522 def _merge_subtitle_items(subtitle_list1, subtitle_list2):
2523 """ Merge subtitle items for one language. Items with duplicated URLs
2524 will be dropped. """
2525 list1_urls = set([item['url'] for item in subtitle_list1])
2526 ret = list(subtitle_list1)
2527 ret.extend([item for item in subtitle_list2 if item['url'] not in list1_urls])
2528 return ret
2529
2530 @classmethod
2531 def _merge_subtitles(cls, subtitle_dict1, subtitle_dict2):
2532 """ Merge two subtitle dictionaries, language by language. """
2533 ret = dict(subtitle_dict1)
2534 for lang in subtitle_dict2:
2535 ret[lang] = cls._merge_subtitle_items(subtitle_dict1.get(lang, []), subtitle_dict2[lang])
2536 return ret
2537
2538 def extract_automatic_captions(self, *args, **kwargs):
2539 if (self._downloader.params.get('writeautomaticsub', False) or
2540 self._downloader.params.get('listsubtitles')):
2541 return self._get_automatic_captions(*args, **kwargs)
2542 return {}
2543
2544 def _get_automatic_captions(self, *args, **kwargs):
2545 raise NotImplementedError('This method must be implemented by subclasses')
2546
2547 def mark_watched(self, *args, **kwargs):
2548 if (self._downloader.params.get('mark_watched', False) and
2549 (self._get_login_info()[0] is not None or
2550 self._downloader.params.get('cookiefile') is not None)):
2551 self._mark_watched(*args, **kwargs)
2552
2553 def _mark_watched(self, *args, **kwargs):
2554 raise NotImplementedError('This method must be implemented by subclasses')
2555
2556 def geo_verification_headers(self):
2557 headers = {}
2558 geo_verification_proxy = self._downloader.params.get('geo_verification_proxy')
2559 if geo_verification_proxy:
2560 headers['Ytdl-request-proxy'] = geo_verification_proxy
2561 return headers
2562
2563 def _generic_id(self, url):
2564 return compat_urllib_parse_unquote(os.path.splitext(url.rstrip('/').split('/')[-1])[0])
2565
2566 def _generic_title(self, url):
2567 return compat_urllib_parse_unquote(os.path.splitext(url_basename(url))[0])
2568
2569
2570 class SearchInfoExtractor(InfoExtractor):
2571 """
2572 Base class for paged search queries extractors.
2573 They accept URLs in the format _SEARCH_KEY(|all|[0-9]):{query}
2574 Instances should define _SEARCH_KEY and _MAX_RESULTS.
2575 """
2576
2577 @classmethod
2578 def _make_valid_url(cls):
2579 return r'%s(?P<prefix>|[1-9][0-9]*|all):(?P<query>[\s\S]+)' % cls._SEARCH_KEY
2580
2581 @classmethod
2582 def suitable(cls, url):
2583 return re.match(cls._make_valid_url(), url) is not None
2584
2585 def _real_extract(self, query):
2586 mobj = re.match(self._make_valid_url(), query)
2587 if mobj is None:
2588 raise ExtractorError('Invalid search query "%s"' % query)
2589
2590 prefix = mobj.group('prefix')
2591 query = mobj.group('query')
2592 if prefix == '':
2593 return self._get_n_results(query, 1)
2594 elif prefix == 'all':
2595 return self._get_n_results(query, self._MAX_RESULTS)
2596 else:
2597 n = int(prefix)
2598 if n <= 0:
2599 raise ExtractorError('invalid download number %s for query "%s"' % (n, query))
2600 elif n > self._MAX_RESULTS:
2601 self._downloader.report_warning('%s returns max %i results (you requested %i)' % (self._SEARCH_KEY, self._MAX_RESULTS, n))
2602 n = self._MAX_RESULTS
2603 return self._get_n_results(query, n)
2604
2605 def _get_n_results(self, query, n):
2606 """Get a specified number of results for a query"""
2607 raise NotImplementedError('This method must be implemented by subclasses')
2608
2609 @property
2610 def SEARCH_KEY(self):
2611 return self._SEARCH_KEY