2 changed files with 110 additions and 0 deletions
Unified View
Diff Options
@ -0,0 +1,109 @@ |
|||||
|
# coding: utf-8 |
||||
|
from __future__ import unicode_literals |
||||
|
|
||||
|
import json |
||||
|
import re |
||||
|
|
||||
|
from .common import InfoExtractor |
||||
|
from ..compat import compat_HTTPError |
||||
|
from ..utils import ( |
||||
|
clean_html, |
||||
|
ExtractorError, |
||||
|
int_or_none, |
||||
|
PUTRequest, |
||||
|
) |
||||
|
|
||||
|
|
||||
|
class PlayPlusTVIE(InfoExtractor): |
||||
|
_VALID_URL = r'https?://(?:www\.)?playplus\.tv/VOD/(?P<project_id>[0-9]+)/(?P<id>[0-9a-f]{32})' |
||||
|
_TEST = { |
||||
|
'url': 'https://www.playplus.tv/VOD/7572/db8d274a5163424e967f35a30ddafb8e', |
||||
|
'md5': 'd078cb89d7ab6b9df37ce23c647aef72', |
||||
|
'info_dict': { |
||||
|
'id': 'db8d274a5163424e967f35a30ddafb8e', |
||||
|
'ext': 'mp4', |
||||
|
'title': 'Capítulo 179 - Final', |
||||
|
'description': 'md5:01085d62d8033a1e34121d3c3cabc838', |
||||
|
'timestamp': 1529992740, |
||||
|
'upload_date': '20180626', |
||||
|
}, |
||||
|
'skip': 'Requires account credential', |
||||
|
} |
||||
|
_NETRC_MACHINE = 'playplustv' |
||||
|
_GEO_COUNTRIES = ['BR'] |
||||
|
_token = None |
||||
|
_profile_id = None |
||||
|
|
||||
|
def _call_api(self, resource, video_id=None, query=None): |
||||
|
return self._download_json('https://api.playplus.tv/api/media/v2/get' + resource, video_id, headers={ |
||||
|
'Authorization': 'Bearer ' + self._token, |
||||
|
}, query=query) |
||||
|
|
||||
|
def _real_initialize(self): |
||||
|
email, password = self._get_login_info() |
||||
|
if email is None: |
||||
|
self.raise_login_required() |
||||
|
|
||||
|
req = PUTRequest( |
||||
|
'https://api.playplus.tv/api/web/login', json.dumps({ |
||||
|
'email': email, |
||||
|
'password': password, |
||||
|
}).encode(), { |
||||
|
'Content-Type': 'application/json; charset=utf-8', |
||||
|
}) |
||||
|
|
||||
|
try: |
||||
|
self._token = self._download_json(req, None)['token'] |
||||
|
except ExtractorError as e: |
||||
|
if isinstance(e.cause, compat_HTTPError) and e.cause.code == 401: |
||||
|
raise ExtractorError(self._parse_json( |
||||
|
e.cause.read(), None)['errorMessage'], expected=True) |
||||
|
raise |
||||
|
|
||||
|
self._profile = self._call_api('Profiles')['list'][0]['_id'] |
||||
|
|
||||
|
def _real_extract(self, url): |
||||
|
project_id, media_id = re.match(self._VALID_URL, url).groups() |
||||
|
media = self._call_api( |
||||
|
'Media', media_id, { |
||||
|
'profileId': self._profile, |
||||
|
'projectId': project_id, |
||||
|
'mediaId': media_id, |
||||
|
})['obj'] |
||||
|
title = media['title'] |
||||
|
|
||||
|
formats = [] |
||||
|
for f in media.get('files', []): |
||||
|
f_url = f.get('url') |
||||
|
if not f_url: |
||||
|
continue |
||||
|
file_info = f.get('fileInfo') or {} |
||||
|
formats.append({ |
||||
|
'url': f_url, |
||||
|
'width': int_or_none(file_info.get('width')), |
||||
|
'height': int_or_none(file_info.get('height')), |
||||
|
}) |
||||
|
self._sort_formats(formats) |
||||
|
|
||||
|
thumbnails = [] |
||||
|
for thumb in media.get('thumbs', []): |
||||
|
thumb_url = thumb.get('url') |
||||
|
if not thumb_url: |
||||
|
continue |
||||
|
thumbnails.append({ |
||||
|
'url': thumb_url, |
||||
|
'width': int_or_none(thumb.get('width')), |
||||
|
'height': int_or_none(thumb.get('height')), |
||||
|
}) |
||||
|
|
||||
|
return { |
||||
|
'id': media_id, |
||||
|
'title': title, |
||||
|
'formats': formats, |
||||
|
'thumbnails': thumbnails, |
||||
|
'description': clean_html(media.get('description')) or media.get('shortDescription'), |
||||
|
'timestamp': int_or_none(media.get('publishDate'), 1000), |
||||
|
'view_count': int_or_none(media.get('numberOfViews')), |
||||
|
'comment_count': int_or_none(media.get('numberOfComments')), |
||||
|
'tags': media.get('tags'), |
||||
|
} |
Write
Preview
Loading…
Cancel
Save