- # encoding: utf-8
- from __future__ import unicode_literals
-
- import json
- import re
- import itertools
-
- from .common import InfoExtractor
- from .subtitles import SubtitlesInfoExtractor
- from ..utils import (
- compat_HTTPError,
- compat_urllib_parse,
- compat_urllib_request,
- compat_urlparse,
- ExtractorError,
- InAdvancePagedList,
- int_or_none,
- RegexNotFoundError,
- std_headers,
- unsmuggle_url,
- urlencode_postdata,
- )
-
-
- class VimeoBaseInfoExtractor(InfoExtractor):
- _NETRC_MACHINE = 'vimeo'
- _LOGIN_REQUIRED = False
-
- def _login(self):
- (username, password) = self._get_login_info()
- if username is None:
- if self._LOGIN_REQUIRED:
- raise ExtractorError('No login info available, needed for using %s.' % self.IE_NAME, expected=True)
- return
- self.report_login()
- login_url = 'https://vimeo.com/log_in'
- webpage = self._download_webpage(login_url, None, False)
- token = self._search_regex(r'xsrft: \'(.*?)\'', webpage, 'login token')
- data = urlencode_postdata({
- 'email': username,
- 'password': password,
- 'action': 'login',
- 'service': 'vimeo',
- 'token': token,
- })
- login_request = compat_urllib_request.Request(login_url, data)
- login_request.add_header('Content-Type', 'application/x-www-form-urlencoded')
- login_request.add_header('Cookie', 'xsrft=%s' % token)
- self._download_webpage(login_request, None, False, 'Wrong login info')
-
-
- class VimeoIE(VimeoBaseInfoExtractor, SubtitlesInfoExtractor):
- """Information extractor for vimeo.com."""
-
- # _VALID_URL matches Vimeo URLs
- _VALID_URL = r'''(?x)
- https?://
- (?:(?:www|(?P<player>player))\.)?
- vimeo(?P<pro>pro)?\.com/
- (?!channels/[^/?#]+/?(?:$|[?#])|album/)
- (?:.*?/)?
- (?:(?:play_redirect_hls|moogaloop\.swf)\?clip_id=)?
- (?:videos?/)?
- (?P<id>[0-9]+)
- /?(?:[?&].*)?(?:[#].*)?$'''
- IE_NAME = 'vimeo'
- _TESTS = [
- {
- 'url': 'http://vimeo.com/56015672#at=0',
- 'md5': '8879b6cc097e987f02484baf890129e5',
- 'info_dict': {
- 'id': '56015672',
- 'ext': 'mp4',
- "upload_date": "20121220",
- "description": "This is a test case for youtube-dl.\nFor more information, see github.com/rg3/youtube-dl\nTest chars: \u2605 \" ' \u5e78 / \\ \u00e4 \u21ad \U0001d550",
- "uploader_id": "user7108434",
- "uploader": "Filippo Valsorda",
- "title": "youtube-dl test video - \u2605 \" ' \u5e78 / \\ \u00e4 \u21ad \U0001d550",
- "duration": 10,
- },
- },
- {
- 'url': 'http://vimeopro.com/openstreetmapus/state-of-the-map-us-2013/video/68093876',
- 'md5': '3b5ca6aa22b60dfeeadf50b72e44ed82',
- 'note': 'Vimeo Pro video (#1197)',
- 'info_dict': {
- 'id': '68093876',
- 'ext': 'mp4',
- 'uploader_id': 'openstreetmapus',
- 'uploader': 'OpenStreetMap US',
- 'title': 'Andy Allan - Putting the Carto into OpenStreetMap Cartography',
- 'description': 'md5:380943ec71b89736ff4bf27183233d09',
- 'duration': 1595,
- },
- },
- {
- 'url': 'http://player.vimeo.com/video/54469442',
- 'md5': '619b811a4417aa4abe78dc653becf511',
- 'note': 'Videos that embed the url in the player page',
- 'info_dict': {
- 'id': '54469442',
- 'ext': 'mp4',
- 'title': 'Kathy Sierra: Building the minimum Badass User, Business of Software 2012',
- 'uploader': 'The BLN & Business of Software',
- 'uploader_id': 'theblnbusinessofsoftware',
- 'duration': 3610,
- 'description': None,
- },
- },
- {
- 'url': 'http://vimeo.com/68375962',
- 'md5': 'aaf896bdb7ddd6476df50007a0ac0ae7',
- 'note': 'Video protected with password',
- 'info_dict': {
- 'id': '68375962',
- 'ext': 'mp4',
- 'title': 'youtube-dl password protected test video',
- 'upload_date': '20130614',
- 'uploader_id': 'user18948128',
- 'uploader': 'Jaime Marquínez Ferrándiz',
- 'duration': 10,
- 'description': 'This is "youtube-dl password protected test video" by Jaime Marquínez Ferrándiz on Vimeo, the home for high quality videos and the people who love them.',
- },
- 'params': {
- 'videopassword': 'youtube-dl',
- },
- },
- {
- 'url': 'http://vimeo.com/channels/keypeele/75629013',
- 'md5': '2f86a05afe9d7abc0b9126d229bbe15d',
- 'note': 'Video is freely available via original URL '
- 'and protected with password when accessed via http://vimeo.com/75629013',
- 'info_dict': {
- 'id': '75629013',
- 'ext': 'mp4',
- 'title': 'Key & Peele: Terrorist Interrogation',
- 'description': 'md5:8678b246399b070816b12313e8b4eb5c',
- 'uploader_id': 'atencio',
- 'uploader': 'Peter Atencio',
- 'duration': 187,
- },
- },
- {
- 'url': 'http://vimeo.com/76979871',
- 'md5': '3363dd6ffebe3784d56f4132317fd446',
- 'note': 'Video with subtitles',
- 'info_dict': {
- 'id': '76979871',
- 'ext': 'mp4',
- 'title': 'The New Vimeo Player (You Know, For Videos)',
- 'description': 'md5:2ec900bf97c3f389378a96aee11260ea',
- 'upload_date': '20131015',
- 'uploader_id': 'staff',
- 'uploader': 'Vimeo Staff',
- 'duration': 62,
- }
- },
- {
- # from https://www.ouya.tv/game/Pier-Solar-and-the-Great-Architects/
- 'url': 'https://player.vimeo.com/video/98044508',
- 'note': 'The js code contains assignments to the same variable as the config',
- 'info_dict': {
- 'id': '98044508',
- 'ext': 'mp4',
- 'title': 'Pier Solar OUYA Official Trailer',
- 'uploader': 'Tulio Gonçalves',
- 'uploader_id': 'user28849593',
- },
- },
- ]
-
- def _verify_video_password(self, url, video_id, webpage):
- password = self._downloader.params.get('videopassword', None)
- if password is None:
- raise ExtractorError('This video is protected by a password, use the --video-password option')
- token = self._search_regex(r'xsrft: \'(.*?)\'', webpage, 'login token')
- data = compat_urllib_parse.urlencode({
- 'password': password,
- 'token': token,
- })
- # I didn't manage to use the password with https
- if url.startswith('https'):
- pass_url = url.replace('https', 'http')
- else:
- pass_url = url
- password_request = compat_urllib_request.Request(pass_url + '/password', data)
- password_request.add_header('Content-Type', 'application/x-www-form-urlencoded')
- password_request.add_header('Cookie', 'xsrft=%s' % token)
- self._download_webpage(password_request, video_id,
- 'Verifying the password',
- 'Wrong password')
-
- def _verify_player_video_password(self, url, video_id):
- password = self._downloader.params.get('videopassword', None)
- if password is None:
- raise ExtractorError('This video is protected by a password, use the --video-password option')
- data = compat_urllib_parse.urlencode({'password': password})
- pass_url = url + '/check-password'
- password_request = compat_urllib_request.Request(pass_url, data)
- password_request.add_header('Content-Type', 'application/x-www-form-urlencoded')
- return self._download_json(
- password_request, video_id,
- 'Verifying the password',
- 'Wrong password')
-
- def _real_initialize(self):
- self._login()
-
- def _real_extract(self, url):
- url, data = unsmuggle_url(url)
- headers = std_headers
- if data is not None:
- headers = headers.copy()
- headers.update(data)
- if 'Referer' not in headers:
- headers['Referer'] = url
-
- # Extract ID from URL
- mobj = re.match(self._VALID_URL, url)
- video_id = mobj.group('id')
- orig_url = url
- if mobj.group('pro') or mobj.group('player'):
- url = 'http://player.vimeo.com/video/' + video_id
-
- # Retrieve video webpage to extract further information
- request = compat_urllib_request.Request(url, None, headers)
- try:
- webpage = self._download_webpage(request, video_id)
- except ExtractorError as ee:
- if isinstance(ee.cause, compat_HTTPError) and ee.cause.code == 403:
- errmsg = ee.cause.read()
- if b'Because of its privacy settings, this video cannot be played here' in errmsg:
- raise ExtractorError(
- 'Cannot download embed-only video without embedding '
- 'URL. Please call youtube-dl with the URL of the page '
- 'that embeds this video.',
- expected=True)
- raise
-
- # Now we begin extracting as much information as we can from what we
- # retrieved. First we extract the information common to all extractors,
- # and latter we extract those that are Vimeo specific.
- self.report_extraction(video_id)
-
- # Extract the config JSON
- try:
- try:
- config_url = self._html_search_regex(
- r' data-config-url="(.+?)"', webpage, 'config URL')
- config_json = self._download_webpage(config_url, video_id)
- config = json.loads(config_json)
- except RegexNotFoundError:
- # For pro videos or player.vimeo.com urls
- # We try to find out to which variable is assigned the config dic
- m_variable_name = re.search('(\w)\.video\.id', webpage)
- if m_variable_name is not None:
- config_re = r'%s=({[^}].+?});' % re.escape(m_variable_name.group(1))
- else:
- config_re = [r' = {config:({.+?}),assets:', r'(?:[abc])=({.+?});']
- config = self._search_regex(config_re, webpage, 'info section',
- flags=re.DOTALL)
- config = json.loads(config)
- except Exception as e:
- if re.search('The creator of this video has not given you permission to embed it on this domain.', webpage):
- raise ExtractorError('The author has restricted the access to this video, try with the "--referer" option')
-
- if re.search('<form[^>]+?id="pw_form"', webpage) is not None:
- self._verify_video_password(url, video_id, webpage)
- return self._real_extract(url)
- else:
- raise ExtractorError('Unable to extract info section',
- cause=e)
- else:
- if config.get('view') == 4:
- config = self._verify_player_video_password(url, video_id)
-
- # Extract title
- video_title = config["video"]["title"]
-
- # Extract uploader and uploader_id
- video_uploader = config["video"]["owner"]["name"]
- video_uploader_id = config["video"]["owner"]["url"].split('/')[-1] if config["video"]["owner"]["url"] else None
-
- # Extract video thumbnail
- video_thumbnail = config["video"].get("thumbnail")
- if video_thumbnail is None:
- video_thumbs = config["video"].get("thumbs")
- if video_thumbs and isinstance(video_thumbs, dict):
- _, video_thumbnail = sorted((int(width if width.isdigit() else 0), t_url) for (width, t_url) in video_thumbs.items())[-1]
-
- # Extract video description
-
- video_description = self._html_search_regex(
- r'(?s)<div\s+class="[^"]*description[^"]*"[^>]*>(.*?)</div>',
- webpage, 'description', default=None)
- if not video_description:
- video_description = self._html_search_meta(
- 'description', webpage, default=None)
- if not video_description and mobj.group('pro'):
- orig_webpage = self._download_webpage(
- orig_url, video_id,
- note='Downloading webpage for description',
- fatal=False)
- if orig_webpage:
- video_description = self._html_search_meta(
- 'description', orig_webpage, default=None)
- if not video_description and not mobj.group('player'):
- self._downloader.report_warning('Cannot find video description')
-
- # Extract video duration
- video_duration = int_or_none(config["video"].get("duration"))
-
- # Extract upload date
- video_upload_date = None
- mobj = re.search(r'<meta itemprop="dateCreated" content="(\d{4})-(\d{2})-(\d{2})T', webpage)
- if mobj is not None:
- video_upload_date = mobj.group(1) + mobj.group(2) + mobj.group(3)
-
- try:
- view_count = int(self._search_regex(r'UserPlays:(\d+)', webpage, 'view count'))
- like_count = int(self._search_regex(r'UserLikes:(\d+)', webpage, 'like count'))
- comment_count = int(self._search_regex(r'UserComments:(\d+)', webpage, 'comment count'))
- except RegexNotFoundError:
- # This info is only available in vimeo.com/{id} urls
- view_count = None
- like_count = None
- comment_count = None
-
- # Vimeo specific: extract request signature and timestamp
- sig = config['request']['signature']
- timestamp = config['request']['timestamp']
-
- # Vimeo specific: extract video codec and quality information
- # First consider quality, then codecs, then take everything
- codecs = [('vp6', 'flv'), ('vp8', 'flv'), ('h264', 'mp4')]
- files = {'hd': [], 'sd': [], 'other': []}
- config_files = config["video"].get("files") or config["request"].get("files")
- for codec_name, codec_extension in codecs:
- for quality in config_files.get(codec_name, []):
- format_id = '-'.join((codec_name, quality)).lower()
- key = quality if quality in files else 'other'
- video_url = None
- if isinstance(config_files[codec_name], dict):
- file_info = config_files[codec_name][quality]
- video_url = file_info.get('url')
- else:
- file_info = {}
- if video_url is None:
- video_url = "http://player.vimeo.com/play_redirect?clip_id=%s&sig=%s&time=%s&quality=%s&codecs=%s&type=moogaloop_local&embed_location=" \
- % (video_id, sig, timestamp, quality, codec_name.upper())
-
- files[key].append({
- 'ext': codec_extension,
- 'url': video_url,
- 'format_id': format_id,
- 'width': file_info.get('width'),
- 'height': file_info.get('height'),
- })
- formats = []
- for key in ('other', 'sd', 'hd'):
- formats += files[key]
- if len(formats) == 0:
- raise ExtractorError('No known codec found')
-
- subtitles = {}
- text_tracks = config['request'].get('text_tracks')
- if text_tracks:
- for tt in text_tracks:
- subtitles[tt['lang']] = 'http://vimeo.com' + tt['url']
-
- video_subtitles = self.extract_subtitles(video_id, subtitles)
- if self._downloader.params.get('listsubtitles', False):
- self._list_available_subtitles(video_id, subtitles)
- return
-
- return {
- 'id': video_id,
- 'uploader': video_uploader,
- 'uploader_id': video_uploader_id,
- 'upload_date': video_upload_date,
- 'title': video_title,
- 'thumbnail': video_thumbnail,
- 'description': video_description,
- 'duration': video_duration,
- 'formats': formats,
- 'webpage_url': url,
- 'view_count': view_count,
- 'like_count': like_count,
- 'comment_count': comment_count,
- 'subtitles': video_subtitles,
- }
-
-
- class VimeoChannelIE(InfoExtractor):
- IE_NAME = 'vimeo:channel'
- _VALID_URL = r'https?://vimeo\.com/channels/(?P<id>[^/?#]+)/?(?:$|[?#])'
- _MORE_PAGES_INDICATOR = r'<a.+?rel="next"'
- _TITLE_RE = r'<link rel="alternate"[^>]+?title="(.*?)"'
- _TESTS = [{
- 'url': 'http://vimeo.com/channels/tributes',
- 'info_dict': {
- 'title': 'Vimeo Tributes',
- },
- 'playlist_mincount': 25,
- }]
-
- def _page_url(self, base_url, pagenum):
- return '%s/videos/page:%d/' % (base_url, pagenum)
-
- def _extract_list_title(self, webpage):
- return self._html_search_regex(self._TITLE_RE, webpage, 'list title')
-
- def _extract_videos(self, list_id, base_url):
- video_ids = []
- for pagenum in itertools.count(1):
- webpage = self._download_webpage(
- self._page_url(base_url, pagenum), list_id,
- 'Downloading page %s' % pagenum)
- video_ids.extend(re.findall(r'id="clip_(\d+?)"', webpage))
- if re.search(self._MORE_PAGES_INDICATOR, webpage, re.DOTALL) is None:
- break
-
- entries = [self.url_result('http://vimeo.com/%s' % video_id, 'Vimeo')
- for video_id in video_ids]
- return {'_type': 'playlist',
- 'id': list_id,
- 'title': self._extract_list_title(webpage),
- 'entries': entries,
- }
-
- def _real_extract(self, url):
- mobj = re.match(self._VALID_URL, url)
- channel_id = mobj.group('id')
- return self._extract_videos(channel_id, 'http://vimeo.com/channels/%s' % channel_id)
-
-
- class VimeoUserIE(VimeoChannelIE):
- IE_NAME = 'vimeo:user'
- _VALID_URL = r'https?://vimeo\.com/(?![0-9]+(?:$|[?#/]))(?P<name>[^/]+)(?:/videos|[#?]|$)'
- _TITLE_RE = r'<a[^>]+?class="user">([^<>]+?)</a>'
- _TESTS = [{
- 'url': 'http://vimeo.com/nkistudio/videos',
- 'info_dict': {
- 'title': 'Nki',
- },
- 'playlist_mincount': 66,
- }]
-
- def _real_extract(self, url):
- mobj = re.match(self._VALID_URL, url)
- name = mobj.group('name')
- return self._extract_videos(name, 'http://vimeo.com/%s' % name)
-
-
- class VimeoAlbumIE(VimeoChannelIE):
- IE_NAME = 'vimeo:album'
- _VALID_URL = r'https?://vimeo\.com/album/(?P<id>\d+)'
- _TITLE_RE = r'<header id="page_header">\n\s*<h1>(.*?)</h1>'
- _TESTS = [{
- 'url': 'http://vimeo.com/album/2632481',
- 'info_dict': {
- 'title': 'Staff Favorites: November 2013',
- },
- 'playlist_mincount': 13,
- }]
-
- def _page_url(self, base_url, pagenum):
- return '%s/page:%d/' % (base_url, pagenum)
-
- def _real_extract(self, url):
- mobj = re.match(self._VALID_URL, url)
- album_id = mobj.group('id')
- return self._extract_videos(album_id, 'http://vimeo.com/album/%s' % album_id)
-
-
- class VimeoGroupsIE(VimeoAlbumIE):
- IE_NAME = 'vimeo:group'
- _VALID_URL = r'(?:https?://)?vimeo\.com/groups/(?P<name>[^/]+)'
- _TESTS = [{
- 'url': 'http://vimeo.com/groups/rolexawards',
- 'info_dict': {
- 'title': 'Rolex Awards for Enterprise',
- },
- 'playlist_mincount': 73,
- }]
-
- def _extract_list_title(self, webpage):
- return self._og_search_title(webpage)
-
- def _real_extract(self, url):
- mobj = re.match(self._VALID_URL, url)
- name = mobj.group('name')
- return self._extract_videos(name, 'http://vimeo.com/groups/%s' % name)
-
-
- class VimeoReviewIE(InfoExtractor):
- IE_NAME = 'vimeo:review'
- IE_DESC = 'Review pages on vimeo'
- _VALID_URL = r'https?://vimeo\.com/[^/]+/review/(?P<id>[^/]+)'
- _TESTS = [{
- 'url': 'https://vimeo.com/user21297594/review/75524534/3c257a1b5d',
- 'file': '75524534.mp4',
- 'md5': 'c507a72f780cacc12b2248bb4006d253',
- 'info_dict': {
- 'title': "DICK HARDWICK 'Comedian'",
- 'uploader': 'Richard Hardwick',
- }
- }, {
- 'note': 'video player needs Referer',
- 'url': 'http://vimeo.com/user22258446/review/91613211/13f927e053',
- 'md5': '6295fdab8f4bf6a002d058b2c6dce276',
- 'info_dict': {
- 'id': '91613211',
- 'ext': 'mp4',
- 'title': 're:(?i)^Death by dogma versus assembling agile . Sander Hoogendoorn',
- 'uploader': 'DevWeek Events',
- 'duration': 2773,
- 'thumbnail': 're:^https?://.*\.jpg$',
- }
- }]
-
- def _real_extract(self, url):
- mobj = re.match(self._VALID_URL, url)
- video_id = mobj.group('id')
- player_url = 'https://player.vimeo.com/player/' + video_id
- return self.url_result(player_url, 'Vimeo', video_id)
-
-
- class VimeoWatchLaterIE(VimeoBaseInfoExtractor, VimeoChannelIE):
- IE_NAME = 'vimeo:watchlater'
- IE_DESC = 'Vimeo watch later list, "vimeowatchlater" keyword (requires authentication)'
- _VALID_URL = r'https?://vimeo\.com/home/watchlater|:vimeowatchlater'
- _LOGIN_REQUIRED = True
- _TITLE_RE = r'href="/home/watchlater".*?>(.*?)<'
- _TESTS = [{
- 'url': 'http://vimeo.com/home/watchlater',
- 'only_matching': True,
- }]
-
- def _real_initialize(self):
- self._login()
-
- def _page_url(self, base_url, pagenum):
- url = '%s/page:%d/' % (base_url, pagenum)
- request = compat_urllib_request.Request(url)
- # Set the header to get a partial html page with the ids,
- # the normal page doesn't contain them.
- request.add_header('X-Requested-With', 'XMLHttpRequest')
- return request
-
- def _real_extract(self, url):
- return self._extract_videos('watchlater', 'https://vimeo.com/home/watchlater')
-
-
- class VimeoLikesIE(InfoExtractor):
- _VALID_URL = r'https?://(?:www\.)?vimeo\.com/user(?P<id>[0-9]+)/likes/?(?:$|[?#]|sort:)'
- IE_NAME = 'vimeo:likes'
- IE_DESC = 'Vimeo user likes'
- _TEST = {
- 'url': 'https://vimeo.com/user755559/likes/',
- 'playlist_mincount': 293,
- "info_dict": {
- "description": "See all the videos urza likes",
- "title": 'Videos urza likes',
- },
- }
-
- def _real_extract(self, url):
- user_id = self._match_id(url)
- webpage = self._download_webpage(url, user_id)
- page_count = self._int(
- self._search_regex(
- r'''(?x)<li><a\s+href="[^"]+"\s+data-page="([0-9]+)">
- .*?</a></li>\s*<li\s+class="pagination_next">
- ''', webpage, 'page count'),
- 'page count', fatal=True)
- PAGE_SIZE = 12
- title = self._html_search_regex(
- r'(?s)<h1>(.+?)</h1>', webpage, 'title', fatal=False)
- description = self._html_search_meta('description', webpage)
-
- def _get_page(idx):
- page_url = '%s//vimeo.com/user%s/likes/page:%d/sort:date' % (
- self.http_scheme(), user_id, idx + 1)
- webpage = self._download_webpage(
- page_url, user_id,
- note='Downloading page %d/%d' % (idx + 1, page_count))
- video_list = self._search_regex(
- r'(?s)<ol class="js-browse_list[^"]+"[^>]*>(.*?)</ol>',
- webpage, 'video content')
- paths = re.findall(
- r'<li[^>]*>\s*<a\s+href="([^"]+)"', video_list)
- for path in paths:
- yield {
- '_type': 'url',
- 'url': compat_urlparse.urljoin(page_url, path),
- }
-
- pl = InAdvancePagedList(_get_page, page_count, PAGE_SIZE)
-
- return {
- '_type': 'playlist',
- 'id': 'user%s_likes' % user_id,
- 'title': title,
- 'description': description,
- 'entries': pl,
- }
|