-# coding: utf-8
-from __future__ import unicode_literals
-
import hashlib
import itertools
-import re
+import urllib.parse
from .common import InfoExtractor, SearchInfoExtractor
-from ..compat import (
- compat_str,
- compat_urllib_parse,
-)
+from .youtube import YoutubeIE
from ..utils import (
- clean_html,
ExtractorError,
+ clean_html,
int_or_none,
mimetype2ext,
parse_iso8601,
- smuggle_url,
+ traverse_obj,
try_get,
url_or_none,
)
-from .brightcove import BrightcoveNewIE
-from .youtube import YoutubeIE
-
class YahooIE(InfoExtractor):
IE_DESC = 'Yahoo screen and movies'
_VALID_URL = r'(?P<url>https?://(?:(?P<country>[a-zA-Z]{2}(?:-[a-zA-Z]{2})?|malaysia)\.)?(?:[\da-zA-Z_-]+\.)?yahoo\.com/(?:[^/]+/)*(?P<id>[^?&#]*-[0-9]+(?:-[a-z]+)?)\.html)'
+ _EMBED_REGEX = [r'<iframe[^>]+?src=(["\'])(?P<url>https?://(?:screen|movies)\.yahoo\.com/.+?\.html\?format=embed)\1']
+
_TESTS = [{
'url': 'http://screen.yahoo.com/julian-smith-travis-legg-watch-214727115.html',
'info_dict': {
if not formats and msg == 'geo restricted':
self.raise_geo_restricted(metadata_available=True)
- self._sort_formats(formats)
-
thumbnails = []
for thumb in video.get('thumbnails', []):
thumb_url = thumb.get('url')
return {
'id': video_id,
- 'title': self._live_title(title) if is_live else title,
+ 'title': title,
'formats': formats,
'thumbnails': thumbnails,
'description': clean_html(video.get('description')),
if items.get('markup'):
entries.extend(
- self.url_result(yt_url) for yt_url in YoutubeIE._extract_urls(items['markup']))
+ self.url_result(yt_url) for yt_url in YoutubeIE._extract_embed_urls(url, items['markup']))
return self.playlist_result(
entries, item.get('uuid'),
IE_NAME = 'screen.yahoo:search'
_SEARCH_KEY = 'yvsearch'
- def _get_n_results(self, query, n):
- """Get a specified number of results for a query"""
- entries = []
+ def _search_results(self, query):
for pagenum in itertools.count(0):
- result_url = 'http://video.search.yahoo.com/search/?p=%s&fr=screen&o=js&gs=0&b=%d' % (compat_urllib_parse.quote_plus(query), pagenum * 30)
+ result_url = 'http://video.search.yahoo.com/search/?p=%s&fr=screen&o=js&gs=0&b=%d' % (urllib.parse.quote_plus(query), pagenum * 30)
info = self._download_json(result_url, query,
note='Downloading results page ' + str(pagenum + 1))
- m = info['m']
- results = info['results']
-
- for (i, r) in enumerate(results):
- if (pagenum * 30) + i >= n:
- break
- mobj = re.search(r'(?P<url>screen\.yahoo\.com/.*?-\d*?\.html)"', r)
- e = self.url_result('http://' + mobj.group('url'), 'Yahoo')
- entries.append(e)
- if (pagenum * 30 + i >= n) or (m['last'] >= (m['total'] - 1)):
+ yield from (self.url_result(result['rurl']) for result in info['results'])
+ if info['m']['last'] >= info['m']['total'] - 1:
break
- return {
- '_type': 'playlist',
- 'id': query,
- 'entries': entries,
- }
-
-
-class YahooGyaOPlayerIE(InfoExtractor):
- IE_NAME = 'yahoo:gyao:player'
- _VALID_URL = r'https?://(?:gyao\.yahoo\.co\.jp/(?:player|episode(?:/[^/]+)?)|streaming\.yahoo\.co\.jp/c/y)/(?P<id>\d+/v\d+/v\d+|[\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12})'
- _TESTS = [{
- 'url': 'https://gyao.yahoo.co.jp/player/00998/v00818/v0000000000000008564/',
- 'info_dict': {
- 'id': '5993125228001',
- 'ext': 'mp4',
- 'title': 'フューリー 【字幕版】',
- 'description': 'md5:21e691c798a15330eda4db17a8fe45a5',
- 'uploader_id': '4235717419001',
- 'upload_date': '20190124',
- 'timestamp': 1548294365,
- },
- 'params': {
- # m3u8 download
- 'skip_download': True,
- },
- }, {
- 'url': 'https://streaming.yahoo.co.jp/c/y/01034/v00133/v0000000000000000706/',
- 'only_matching': True,
- }, {
- 'url': 'https://gyao.yahoo.co.jp/episode/%E3%81%8D%E3%81%AE%E3%81%86%E4%BD%95%E9%A3%9F%E3%81%B9%E3%81%9F%EF%BC%9F%20%E7%AC%AC2%E8%A9%B1%202019%2F4%2F12%E6%94%BE%E9%80%81%E5%88%86/5cb02352-b725-409e-9f8d-88f947a9f682',
- 'only_matching': True,
- }, {
- 'url': 'https://gyao.yahoo.co.jp/episode/5fa1226c-ef8d-4e93-af7a-fd92f4e30597',
- 'only_matching': True,
- }]
- _GEO_BYPASS = False
-
- def _real_extract(self, url):
- video_id = self._match_id(url).replace('/', ':')
- headers = self.geo_verification_headers()
- headers['Accept'] = 'application/json'
- resp = self._download_json(
- 'https://gyao.yahoo.co.jp/apis/playback/graphql', video_id, query={
- 'appId': 'dj00aiZpPUNJeDh2cU1RazU3UCZzPWNvbnN1bWVyc2VjcmV0Jng9NTk-',
- 'query': '''{
- content(parameter: {contentId: "%s", logicaAgent: PC_WEB}) {
- video {
- delivery {
- id
- }
- title
- }
- }
-}''' % video_id,
- }, headers=headers)
- content = resp['data']['content']
- if not content:
- msg = resp['errors'][0]['message']
- if msg == 'not in japan':
- self.raise_geo_restricted(countries=['JP'])
- raise ExtractorError(msg)
- video = content['video']
- return {
- '_type': 'url_transparent',
- 'id': video_id,
- 'title': video['title'],
- 'url': smuggle_url(
- 'http://players.brightcove.net/4235717419001/SyG5P0gjb_default/index.html?videoId=' + video['delivery']['id'],
- {'geo_countries': ['JP']}),
- 'ie_key': BrightcoveNewIE.ie_key(),
- }
-
-
-class YahooGyaOIE(InfoExtractor):
- IE_NAME = 'yahoo:gyao'
- _VALID_URL = r'https?://(?:gyao\.yahoo\.co\.jp/(?:p|title(?:/[^/]+)?)|streaming\.yahoo\.co\.jp/p/y)/(?P<id>\d+/v\d+|[\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12})'
- _TESTS = [{
- 'url': 'https://gyao.yahoo.co.jp/p/00449/v03102/',
- 'info_dict': {
- 'id': '00449:v03102',
- },
- 'playlist_count': 2,
- }, {
- 'url': 'https://streaming.yahoo.co.jp/p/y/01034/v00133/',
- 'only_matching': True,
- }, {
- 'url': 'https://gyao.yahoo.co.jp/title/%E3%81%97%E3%82%83%E3%81%B9%E3%81%8F%E3%82%8A007/5b025a49-b2e5-4dc7-945c-09c6634afacf',
- 'only_matching': True,
- }, {
- 'url': 'https://gyao.yahoo.co.jp/title/5b025a49-b2e5-4dc7-945c-09c6634afacf',
- 'only_matching': True,
- }]
-
- def _real_extract(self, url):
- program_id = self._match_id(url).replace('/', ':')
- videos = self._download_json(
- 'https://gyao.yahoo.co.jp/api/programs/%s/videos' % program_id, program_id)['videos']
- entries = []
- for video in videos:
- video_id = video.get('id')
- if not video_id:
- continue
- entries.append(self.url_result(
- 'https://gyao.yahoo.co.jp/player/%s/' % video_id.replace(':', '/'),
- YahooGyaOPlayerIE.ie_key(), video_id))
- return self.playlist_result(entries, program_id)
-
class YahooJapanNewsIE(InfoExtractor):
IE_NAME = 'yahoo:japannews'
IE_DESC = 'Yahoo! Japan News'
- _VALID_URL = r'https?://(?P<host>(?:news|headlines)\.yahoo\.co\.jp)[^\d]*(?P<id>\d[\d-]*\d)?'
+ _VALID_URL = r'https?://news\.yahoo\.co\.jp/(?:articles|feature)/(?P<id>[a-zA-Z0-9]+)'
_GEO_COUNTRIES = ['JP']
_TESTS = [{
- 'url': 'https://headlines.yahoo.co.jp/videonews/ann?a=20190716-00000071-ann-int',
+ 'url': 'https://news.yahoo.co.jp/articles/a70fe3a064f1cfec937e2252c7fc6c1ba3201c0e',
'info_dict': {
- 'id': '1736242',
+ 'id': 'a70fe3a064f1cfec937e2252c7fc6c1ba3201c0e',
'ext': 'mp4',
- 'title': 'ã\83 ã\83³å¤§çµ±é \98ã\81\8c対æ\97¥æ\89¹å\88¤ã\82\92å¼·å\8c\96â\80\9cç\8f¾é\87\91å\8c\96â\80\9då\8a¹æ\9e\9cã\81¯ï¼\9fï¼\88ã\83\86ã\83¬ã\83\93æ\9c\9dæ\97¥ç³»ï¼\88ANNï¼\89ï¼\89 - Yahoo!ã\83\8bã\83¥ã\83¼ã\82¹',
- 'description': '韓国の元徴用工らを巡る裁判の原告が弁護士が差し押さえた三菱重工業の資産を売却して - Yahoo!ニュース(テレビ朝日系(ANN))',
- 'thumbnail': r're:^https?://.*\.[a-zA-Z\d]{3,4}$',
+ 'title': 'ã\80\90ç\8b¬è\87ªã\80\91å®\89å\80\8då\85\83ç·\8fç\90\86ã\80\8cå\9b½è\91¬ã\80\8dä¸æ¢æ±\82ã\82\81â\80\9cè\84\85è¿«ã\83¡ã\83¼ã\83«â\80\9dâ\80¦ã\80\8cå\90ã\81©ã\82\82èª\98æ\8b\90ã\80\8dâ\80\9cé\80\81ä¿¡è\80\85â\80\9dã\82\92追跡',
+ 'description': 'md5:1c06974575f930f692d8696fbcfdc546',
+ 'thumbnail': r're:https://.+',
},
'params': {
'skip_download': True,
},
- }, {
- # geo restricted
- 'url': 'https://headlines.yahoo.co.jp/hl?a=20190721-00000001-oxv-l04',
- 'only_matching': True,
- }, {
- 'url': 'https://headlines.yahoo.co.jp/videonews/',
- 'only_matching': True,
- }, {
- 'url': 'https://news.yahoo.co.jp',
- 'only_matching': True,
- }, {
- 'url': 'https://news.yahoo.co.jp/byline/hashimotojunji/20190628-00131977/',
- 'only_matching': True,
}, {
'url': 'https://news.yahoo.co.jp/feature/1356',
'only_matching': True
def _extract_formats(self, json_data, content_id):
formats = []
- video_data = try_get(
- json_data,
- lambda x: x['ResultSet']['Result'][0]['VideoUrlSet']['VideoUrl'],
- list)
- for vid in video_data or []:
+ for vid in traverse_obj(json_data, ('ResultSet', 'Result', ..., 'VideoUrlSet', 'VideoUrl', ...)) or []:
delivery = vid.get('delivery')
url = url_or_none(vid.get('Url'))
if not delivery or not url:
else:
formats.append({
'url': url,
- 'format_id': 'http-%s' % compat_str(vid.get('bitrate', '')),
+ 'format_id': f'http-{vid.get("bitrate")}',
'height': int_or_none(vid.get('height')),
'width': int_or_none(vid.get('width')),
'tbr': int_or_none(vid.get('bitrate')),
})
self._remove_duplicate_formats(formats)
- self._sort_formats(formats)
return formats
def _real_extract(self, url):
- mobj = self._match_valid_url(url)
- host = mobj.group('host')
- display_id = mobj.group('id') or host
-
- webpage = self._download_webpage(url, display_id)
-
- title = self._html_search_meta(
- ['og:title', 'twitter:title'], webpage, 'title', default=None
- ) or self._html_search_regex('<title>([^<]+)</title>', webpage, 'title')
-
- if display_id == host:
- # Headline page (w/ multiple BC playlists) ('news.yahoo.co.jp', 'headlines.yahoo.co.jp/videonews/', ...)
- stream_plists = re.findall(r'plist=(\d+)', webpage) or re.findall(r'plist["\']:\s*["\']([^"\']+)', webpage)
- entries = [
- self.url_result(
- smuggle_url(
- 'http://players.brightcove.net/5690807595001/HyZNerRl7_default/index.html?playlistId=%s' % plist_id,
- {'geo_countries': ['JP']}),
- ie='BrightcoveNew', video_id=plist_id)
- for plist_id in stream_plists]
- return self.playlist_result(entries, playlist_title=title)
-
- # Article page
- description = self._html_search_meta(
- ['og:description', 'description', 'twitter:description'],
- webpage, 'description', default=None)
- thumbnail = self._og_search_thumbnail(
- webpage, default=None) or self._html_search_meta(
- 'twitter:image', webpage, 'thumbnail', default=None)
- space_id = self._search_regex([
- r'<script[^>]+class=["\']yvpub-player["\'][^>]+spaceid=([^&"\']+)',
- r'YAHOO\.JP\.srch\.\w+link\.onLoad[^;]+spaceID["\' ]*:["\' ]+([^"\']+)',
- r'<!--\s+SpaceID=(\d+)'
- ], webpage, 'spaceid')
-
- content_id = self._search_regex(
- r'<script[^>]+class=["\']yvpub-player["\'][^>]+contentid=(?P<contentid>[^&"\']+)',
- webpage, 'contentid', group='contentid')
-
+ video_id = self._match_id(url)
+ webpage = self._download_webpage(url, video_id)
+ preloaded_state = self._search_json(r'__PRELOADED_STATE__\s*=', webpage, 'preloaded state', video_id)
+
+ content_id = traverse_obj(
+ preloaded_state, ('articleDetail', 'paragraphs', ..., 'objectItems', ..., 'video', 'vid'),
+ get_all=False, expected_type=int)
+ if content_id is None:
+ raise ExtractorError('This article does not contain a video', expected=True)
+
+ HOST = 'news.yahoo.co.jp'
+ space_id = traverse_obj(preloaded_state, ('pageData', 'spaceId'), expected_type=str)
json_data = self._download_json(
- 'https://feapi-yvpub.yahooapis.jp/v1/content/%s' % content_id,
- content_id,
- query={
+ f'https://feapi-yvpub.yahooapis.jp/v1/content/{content_id}',
+ video_id, query={
'appid': 'dj0zaiZpPVZMTVFJR0FwZWpiMyZzPWNvbnN1bWVyc2VjcmV0Jng9YjU-',
'output': 'json',
- 'space_id': space_id,
- 'domain': host,
- 'ak': hashlib.md5('_'.join((space_id, host)).encode()).hexdigest(),
+ 'domain': HOST,
+ 'ak': hashlib.md5('_'.join((space_id, HOST)).encode()).hexdigest() if space_id else '',
'device_type': '1100',
})
- formats = self._extract_formats(json_data, content_id)
+
+ title = (
+ traverse_obj(preloaded_state,
+ ('articleDetail', 'headline'), ('pageData', 'pageParam', 'title'),
+ expected_type=str)
+ or self._html_search_meta(('og:title', 'twitter:title'), webpage, 'title', default=None)
+ or self._html_extract_title(webpage))
+ description = (
+ traverse_obj(preloaded_state, ('pageData', 'description'), expected_type=str)
+ or self._html_search_meta(
+ ('og:description', 'description', 'twitter:description'),
+ webpage, 'description', default=None))
+ thumbnail = (
+ traverse_obj(preloaded_state, ('pageData', 'ogpImage'), expected_type=str)
+ or self._og_search_thumbnail(webpage, default=None)
+ or self._html_search_meta('twitter:image', webpage, 'thumbnail', default=None))
return {
- 'id': content_id,
+ 'id': video_id,
'title': title,
'description': description,
'thumbnail': thumbnail,
- 'formats': formats,
+ 'formats': self._extract_formats(json_data, video_id),
}