mirror of
https://github.com/ytdl-org/youtube-dl
synced 2024-11-27 15:16:57 +01:00
7f641d2c7a
- use OnDemandPagedList for list extractors - reduce show extraction requests - extract original video format and subtitles - add support for adobe tv embeds
289 lines
10 KiB
Python
289 lines
10 KiB
Python
from __future__ import unicode_literals
|
|
|
|
import functools
|
|
import re
|
|
|
|
from .common import InfoExtractor
|
|
from ..compat import compat_str
|
|
from ..utils import (
|
|
float_or_none,
|
|
int_or_none,
|
|
ISO639Utils,
|
|
OnDemandPagedList,
|
|
parse_duration,
|
|
str_or_none,
|
|
str_to_int,
|
|
unified_strdate,
|
|
)
|
|
|
|
|
|
class AdobeTVBaseIE(InfoExtractor):
|
|
def _call_api(self, path, video_id, query, note=None):
|
|
return self._download_json(
|
|
'http://tv.adobe.com/api/v4/' + path,
|
|
video_id, note, query=query)['data']
|
|
|
|
def _parse_subtitles(self, video_data, url_key):
|
|
subtitles = {}
|
|
for translation in video_data.get('translations', []):
|
|
vtt_path = translation.get(url_key)
|
|
if not vtt_path:
|
|
continue
|
|
lang = translation.get('language_w3c') or ISO639Utils.long2short(translation['language_medium'])
|
|
subtitles.setdefault(lang, []).append({
|
|
'ext': 'vtt',
|
|
'url': vtt_path,
|
|
})
|
|
return subtitles
|
|
|
|
def _parse_video_data(self, video_data):
|
|
video_id = compat_str(video_data['id'])
|
|
title = video_data['title']
|
|
|
|
s3_extracted = False
|
|
formats = []
|
|
for source in video_data.get('videos', []):
|
|
source_url = source.get('url')
|
|
if not source_url:
|
|
continue
|
|
f = {
|
|
'format_id': source.get('quality_level'),
|
|
'fps': int_or_none(source.get('frame_rate')),
|
|
'height': int_or_none(source.get('height')),
|
|
'tbr': int_or_none(source.get('video_data_rate')),
|
|
'width': int_or_none(source.get('width')),
|
|
'url': source_url,
|
|
}
|
|
original_filename = source.get('original_filename')
|
|
if original_filename:
|
|
if not (f.get('height') and f.get('width')):
|
|
mobj = re.search(r'_(\d+)x(\d+)', original_filename)
|
|
if mobj:
|
|
f.update({
|
|
'height': int(mobj.group(2)),
|
|
'width': int(mobj.group(1)),
|
|
})
|
|
if original_filename.startswith('s3://') and not s3_extracted:
|
|
formats.append({
|
|
'format_id': 'original',
|
|
'preference': 1,
|
|
'url': original_filename.replace('s3://', 'https://s3.amazonaws.com/'),
|
|
})
|
|
s3_extracted = True
|
|
formats.append(f)
|
|
self._sort_formats(formats)
|
|
|
|
return {
|
|
'id': video_id,
|
|
'title': title,
|
|
'description': video_data.get('description'),
|
|
'thumbnail': video_data.get('thumbnail'),
|
|
'upload_date': unified_strdate(video_data.get('start_date')),
|
|
'duration': parse_duration(video_data.get('duration')),
|
|
'view_count': str_to_int(video_data.get('playcount')),
|
|
'formats': formats,
|
|
'subtitles': self._parse_subtitles(video_data, 'vtt'),
|
|
}
|
|
|
|
|
|
class AdobeTVEmbedIE(AdobeTVBaseIE):
|
|
IE_NAME = 'adobetv:embed'
|
|
_VALID_URL = r'https?://tv\.adobe\.com/embed/\d+/(?P<id>\d+)'
|
|
_TEST = {
|
|
'url': 'https://tv.adobe.com/embed/22/4153',
|
|
'md5': 'c8c0461bf04d54574fc2b4d07ac6783a',
|
|
'info_dict': {
|
|
'id': '4153',
|
|
'ext': 'flv',
|
|
'title': 'Creating Graphics Optimized for BlackBerry',
|
|
'description': 'md5:eac6e8dced38bdaae51cd94447927459',
|
|
'thumbnail': r're:https?://.*\.jpg$',
|
|
'upload_date': '20091109',
|
|
'duration': 377,
|
|
'view_count': int,
|
|
},
|
|
}
|
|
|
|
def _real_extract(self, url):
|
|
video_id = self._match_id(url)
|
|
|
|
video_data = self._call_api(
|
|
'episode/' + video_id, video_id, {'disclosure': 'standard'})[0]
|
|
return self._parse_video_data(video_data)
|
|
|
|
|
|
class AdobeTVIE(AdobeTVBaseIE):
|
|
IE_NAME = 'adobetv'
|
|
_VALID_URL = r'https?://tv\.adobe\.com/(?:(?P<language>fr|de|es|jp)/)?watch/(?P<show_urlname>[^/]+)/(?P<id>[^/]+)'
|
|
|
|
_TEST = {
|
|
'url': 'http://tv.adobe.com/watch/the-complete-picture-with-julieanne-kost/quick-tip-how-to-draw-a-circle-around-an-object-in-photoshop/',
|
|
'md5': '9bc5727bcdd55251f35ad311ca74fa1e',
|
|
'info_dict': {
|
|
'id': '10981',
|
|
'ext': 'mp4',
|
|
'title': 'Quick Tip - How to Draw a Circle Around an Object in Photoshop',
|
|
'description': 'md5:99ec318dc909d7ba2a1f2b038f7d2311',
|
|
'thumbnail': r're:https?://.*\.jpg$',
|
|
'upload_date': '20110914',
|
|
'duration': 60,
|
|
'view_count': int,
|
|
},
|
|
}
|
|
|
|
def _real_extract(self, url):
|
|
language, show_urlname, urlname = re.match(self._VALID_URL, url).groups()
|
|
if not language:
|
|
language = 'en'
|
|
|
|
video_data = self._call_api(
|
|
'episode/get', urlname, {
|
|
'disclosure': 'standard',
|
|
'language': language,
|
|
'show_urlname': show_urlname,
|
|
'urlname': urlname,
|
|
})[0]
|
|
return self._parse_video_data(video_data)
|
|
|
|
|
|
class AdobeTVPlaylistBaseIE(AdobeTVBaseIE):
|
|
_PAGE_SIZE = 25
|
|
|
|
def _fetch_page(self, display_id, query, page):
|
|
page += 1
|
|
query['page'] = page
|
|
for element_data in self._call_api(
|
|
self._RESOURCE, display_id, query, 'Download Page %d' % page):
|
|
yield self._process_data(element_data)
|
|
|
|
def _extract_playlist_entries(self, display_id, query):
|
|
return OnDemandPagedList(functools.partial(
|
|
self._fetch_page, display_id, query), self._PAGE_SIZE)
|
|
|
|
|
|
class AdobeTVShowIE(AdobeTVPlaylistBaseIE):
|
|
IE_NAME = 'adobetv:show'
|
|
_VALID_URL = r'https?://tv\.adobe\.com/(?:(?P<language>fr|de|es|jp)/)?show/(?P<id>[^/]+)'
|
|
|
|
_TEST = {
|
|
'url': 'http://tv.adobe.com/show/the-complete-picture-with-julieanne-kost',
|
|
'info_dict': {
|
|
'id': '36',
|
|
'title': 'The Complete Picture with Julieanne Kost',
|
|
'description': 'md5:fa50867102dcd1aa0ddf2ab039311b27',
|
|
},
|
|
'playlist_mincount': 136,
|
|
}
|
|
_RESOURCE = 'episode'
|
|
_process_data = AdobeTVBaseIE._parse_video_data
|
|
|
|
def _real_extract(self, url):
|
|
language, show_urlname = re.match(self._VALID_URL, url).groups()
|
|
if not language:
|
|
language = 'en'
|
|
query = {
|
|
'disclosure': 'standard',
|
|
'language': language,
|
|
'show_urlname': show_urlname,
|
|
}
|
|
|
|
show_data = self._call_api(
|
|
'show/get', show_urlname, query)[0]
|
|
|
|
return self.playlist_result(
|
|
self._extract_playlist_entries(show_urlname, query),
|
|
str_or_none(show_data.get('id')),
|
|
show_data.get('show_name'),
|
|
show_data.get('show_description'))
|
|
|
|
|
|
class AdobeTVChannelIE(AdobeTVPlaylistBaseIE):
|
|
IE_NAME = 'adobetv:channel'
|
|
_VALID_URL = r'https?://tv\.adobe\.com/(?:(?P<language>fr|de|es|jp)/)?channel/(?P<id>[^/]+)(?:/(?P<category_urlname>[^/]+))?'
|
|
|
|
_TEST = {
|
|
'url': 'http://tv.adobe.com/channel/development',
|
|
'info_dict': {
|
|
'id': 'development',
|
|
},
|
|
'playlist_mincount': 96,
|
|
}
|
|
_RESOURCE = 'show'
|
|
|
|
def _process_data(self, show_data):
|
|
return self.url_result(
|
|
show_data['url'], 'AdobeTVShow', str_or_none(show_data.get('id')))
|
|
|
|
def _real_extract(self, url):
|
|
language, channel_urlname, category_urlname = re.match(self._VALID_URL, url).groups()
|
|
if not language:
|
|
language = 'en'
|
|
query = {
|
|
'channel_urlname': channel_urlname,
|
|
'language': language,
|
|
}
|
|
if category_urlname:
|
|
query['category_urlname'] = category_urlname
|
|
|
|
return self.playlist_result(
|
|
self._extract_playlist_entries(channel_urlname, query),
|
|
channel_urlname)
|
|
|
|
|
|
class AdobeTVVideoIE(AdobeTVBaseIE):
|
|
IE_NAME = 'adobetv:video'
|
|
_VALID_URL = r'https?://video\.tv\.adobe\.com/v/(?P<id>\d+)'
|
|
|
|
_TEST = {
|
|
# From https://helpx.adobe.com/acrobat/how-to/new-experience-acrobat-dc.html?set=acrobat--get-started--essential-beginners
|
|
'url': 'https://video.tv.adobe.com/v/2456/',
|
|
'md5': '43662b577c018ad707a63766462b1e87',
|
|
'info_dict': {
|
|
'id': '2456',
|
|
'ext': 'mp4',
|
|
'title': 'New experience with Acrobat DC',
|
|
'description': 'New experience with Acrobat DC',
|
|
'duration': 248.667,
|
|
},
|
|
}
|
|
|
|
def _real_extract(self, url):
|
|
video_id = self._match_id(url)
|
|
webpage = self._download_webpage(url, video_id)
|
|
|
|
video_data = self._parse_json(self._search_regex(
|
|
r'var\s+bridge\s*=\s*([^;]+);', webpage, 'bridged data'), video_id)
|
|
title = video_data['title']
|
|
|
|
formats = []
|
|
sources = video_data.get('sources') or []
|
|
for source in sources:
|
|
source_src = source.get('src')
|
|
if not source_src:
|
|
continue
|
|
formats.append({
|
|
'filesize': int_or_none(source.get('kilobytes') or None, invscale=1000),
|
|
'format_id': '-'.join(filter(None, [source.get('format'), source.get('label')])),
|
|
'height': int_or_none(source.get('height') or None),
|
|
'tbr': int_or_none(source.get('bitrate') or None),
|
|
'width': int_or_none(source.get('width') or None),
|
|
'url': source_src,
|
|
})
|
|
self._sort_formats(formats)
|
|
|
|
# For both metadata and downloaded files the duration varies among
|
|
# formats. I just pick the max one
|
|
duration = max(filter(None, [
|
|
float_or_none(source.get('duration'), scale=1000)
|
|
for source in sources]))
|
|
|
|
return {
|
|
'id': video_id,
|
|
'formats': formats,
|
|
'title': title,
|
|
'description': video_data.get('description'),
|
|
'thumbnail': video_data.get('video', {}).get('poster'),
|
|
'duration': duration,
|
|
'subtitles': self._parse_subtitles(video_data, 'vttPath'),
|
|
}
|