[byutv] Fix extraction (closes #14966, closes #14967)

This commit is contained in:
Sergey M․ 2017-12-13 23:14:30 +07:00
parent 15960255fe
commit c8be7d5f74
No known key found for this signature in database
GPG key ID: 2C393E0F18A9236D
2 changed files with 19 additions and 57 deletions

View file

@ -3,20 +3,19 @@
import re import re
from .common import InfoExtractor from .common import InfoExtractor
from ..utils import ExtractorError
class BYUtvIE(InfoExtractor): class BYUtvIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?byutv\.org/watch/(?!event/)(?P<id>[0-9a-f-]+)(?:/(?P<display_id>[^/?#&]+))?' _VALID_URL = r'https?://(?:www\.)?byutv\.org/(?:watch|player)/(?!event/)(?P<id>[0-9a-f-]+)(?:/(?P<display_id>[^/?#&]+))?'
_TESTS = [{ _TESTS = [{
'url': 'http://www.byutv.org/watch/6587b9a3-89d2-42a6-a7f7-fd2f81840a7d/studio-c-season-5-episode-5', 'url': 'http://www.byutv.org/watch/6587b9a3-89d2-42a6-a7f7-fd2f81840a7d/studio-c-season-5-episode-5',
'info_dict': { 'info_dict': {
'id': '6587b9a3-89d2-42a6-a7f7-fd2f81840a7d', 'id': 'ZvanRocTpW-G5_yZFeltTAMv6jxOU9KH',
'display_id': 'studio-c-season-5-episode-5', 'display_id': 'studio-c-season-5-episode-5',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Season 5 Episode 5', 'title': 'Season 5 Episode 5',
'description': 'md5:e07269172baff037f8e8bf9956bc9747', 'description': 'md5:1d31dc18ef4f075b28f6a65937d22c65',
'thumbnail': r're:^https?://.*\.jpg$', 'thumbnail': r're:^https?://.*',
'duration': 1486.486, 'duration': 1486.486,
}, },
'params': { 'params': {
@ -26,6 +25,9 @@ class BYUtvIE(InfoExtractor):
}, { }, {
'url': 'http://www.byutv.org/watch/6587b9a3-89d2-42a6-a7f7-fd2f81840a7d', 'url': 'http://www.byutv.org/watch/6587b9a3-89d2-42a6-a7f7-fd2f81840a7d',
'only_matching': True, 'only_matching': True,
}, {
'url': 'https://www.byutv.org/player/27741493-dc83-40b0-8420-e7ae38a2ae98/byu-football-toledo-vs-byu-93016?listid=4fe0fee5-0d3c-4a29-b725-e4948627f472&listindex=0&q=toledo',
'only_matching': True,
}] }]
def _real_extract(self, url): def _real_extract(self, url):
@ -33,16 +35,16 @@ def _real_extract(self, url):
video_id = mobj.group('id') video_id = mobj.group('id')
display_id = mobj.group('display_id') or video_id display_id = mobj.group('display_id') or video_id
webpage = self._download_webpage(url, display_id) ep = self._download_json(
episode_code = self._search_regex( 'https://api.byutv.org/api3/catalog/getvideosforcontent', video_id,
r'(?s)episode:(.*?\}),\s*\n', webpage, 'episode information') query={
'contentid': video_id,
ep = self._parse_json( 'channel': 'byutv_global',
episode_code, display_id, transform_source=lambda s: 'x-byutv-context': 'web$Global',
re.sub(r'(\n\s+)([a-zA-Z]+):\s+\'(.*?)\'', r'\1"\2": "\3"', s)) }, headers={
'x-byutv-context': 'web$Global',
if ep['providerType'] != 'Ooyala': 'x-byutv-platformkey': 'xsaaw9c7y5',
raise ExtractorError('Unsupported provider %s' % ep['provider']) })['ooyalaVOD']
return { return {
'_type': 'url_transparent', '_type': 'url_transparent',
@ -50,44 +52,7 @@ def _real_extract(self, url):
'url': 'ooyala:%s' % ep['providerId'], 'url': 'ooyala:%s' % ep['providerId'],
'id': video_id, 'id': video_id,
'display_id': display_id, 'display_id': display_id,
'title': ep['title'], 'title': ep.get('title'),
'description': ep.get('description'), 'description': ep.get('description'),
'thumbnail': ep.get('imageThumbnail'), 'thumbnail': ep.get('imageThumbnail'),
} }
class BYUtvEventIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?byutv\.org/watch/event/(?P<id>[0-9a-f-]+)'
_TEST = {
'url': 'http://www.byutv.org/watch/event/29941b9b-8bf6-48d2-aebf-7a87add9e34b',
'info_dict': {
'id': '29941b9b-8bf6-48d2-aebf-7a87add9e34b',
'ext': 'mp4',
'title': 'Toledo vs. BYU (9/30/16)',
},
'params': {
'skip_download': True,
},
'add_ie': ['Ooyala'],
}
def _real_extract(self, url):
video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
ooyala_id = self._search_regex(
r'providerId\s*:\s*(["\'])(?P<id>(?:(?!\1).)+)\1',
webpage, 'ooyala id', group='id')
title = self._search_regex(
r'class=["\']description["\'][^>]*>\s*<h1>([^<]+)</h1>', webpage,
'title').strip()
return {
'_type': 'url_transparent',
'ie_key': 'Ooyala',
'url': 'ooyala:%s' % ooyala_id,
'id': video_id,
'title': title,
}

View file

@ -138,10 +138,7 @@
BrightcoveNewIE, BrightcoveNewIE,
) )
from .buzzfeed import BuzzFeedIE from .buzzfeed import BuzzFeedIE
from .byutv import ( from .byutv import BYUtvIE
BYUtvIE,
BYUtvEventIE,
)
from .c56 import C56IE from .c56 import C56IE
from .camdemy import ( from .camdemy import (
CamdemyIE, CamdemyIE,