yt-dlp/yt_dlp/extractor/bleacherreport.py

from .amp import AMPIE
from .common import InfoExtractor
from ..utils import (
    ExtractorError,
    int_or_none,
    parse_iso8601,
    str_or_none,
)


class BleacherReportIE(InfoExtractor):
    _WORKING = False
    _VALID_URL = r'https?://(?:www\.)?bleacherreport\.com/articles/(?P<id>\d+)'
    _TESTS = [{
        'url': 'http://bleacherreport.com/articles/2496438-fsu-stat-projections-is-jalen-ramsey-best-defensive-player-in-college-football',
        'md5': 'a3ffc3dc73afdbc2010f02d98f990f20',
        'info_dict': {
            'id': '2496438',
            'ext': 'mp4',
            'title': 'FSU Stat Projections: Is Jalen Ramsey Best Defensive Player in College Football?',
            'uploader_id': '3992341',
            'description': 'CFB, ACC, Florida State',
            'timestamp': 1434380212,
            'upload_date': '20150615',
            'uploader': 'Team Stream Now ',
        },
        'skip': 'Video removed',
    }, {
        'url': 'http://bleacherreport.com/articles/2586817-aussie-golfers-get-fright-of-their-lives-after-being-chased-by-angry-kangaroo',
        'md5': '6a5cd403418c7b01719248ca97fb0692',
        'info_dict': {
            'id': '2586817',
            'ext': 'webm',
            'title': 'Aussie Golfers Get Fright of Their Lives After Being Chased by Angry Kangaroo',
            'timestamp': 1446839961,
            'uploader': 'Sean Fay',
            'description': 'md5:b1601e2314c4d8eec23b6eafe086a757',
            'uploader_id': '6466954',
            'upload_date': '20151011',
        },
        'add_ie': ['Youtube'],
    }]

    def _real_extract(self, url):
        article_id = self._match_id(url)

        article_data = self._download_json(f'http://api.bleacherreport.com/api/v1/articles/{article_id}', article_id)['article']

        thumbnails = []
        primary_photo = article_data.get('primaryPhoto')
        if primary_photo:
            thumbnails = [{
                'url': primary_photo['url'],
                'width': primary_photo.get('width'),
                'height': primary_photo.get('height'),
            }]

        info = {
            '_type': 'url_transparent',
            'id': article_id,
            'title': article_data['title'],
            'uploader': article_data.get('author', {}).get('name'),
            'uploader_id': str_or_none(article_data.get('authorId')),
            'timestamp': parse_iso8601(article_data.get('createdAt')),
            'thumbnails': thumbnails,
            'comment_count': int_or_none(article_data.get('commentsCount')),
            'view_count': int_or_none(article_data.get('hitCount')),
        }

        video = article_data.get('video')
        if video:
            video_type = video['type']
            if video_type in ('cms.bleacherreport.com', 'vid.bleacherreport.com'):
                info['url'] = 'http://bleacherreport.com/video_embed?id={}'.format(video['id'])
            elif video_type == 'youtube.com':
                info['url'] = video['id']
            elif video_type == 'vine.co':
                info['url'] = 'https://vine.co/v/{}'.format(video['id'])
            else:
                info['url'] = video_type + video['id']
            return info
        else:
            raise ExtractorError('no video in the article', expected=True)


class BleacherReportCMSIE(AMPIE):
    _WORKING = False
    _VALID_URL = r'https?://(?:www\.)?bleacherreport\.com/video_embed\?id=(?P<id>[0-9a-f-]{36}|\d{5})'
    _TESTS = [{
        'url': 'http://bleacherreport.com/video_embed?id=8fd44c2f-3dc5-4821-9118-2c825a98c0e1&library=video-cms',
        'md5': '670b2d73f48549da032861130488c681',
        'info_dict': {
            'id': '8fd44c2f-3dc5-4821-9118-2c825a98c0e1',
            'ext': 'mp4',
            'title': 'Cena vs. Rollins Would Expose the Heavyweight Division',
            'description': 'md5:984afb4ade2f9c0db35f3267ed88b36e',
            'upload_date': '20150723',
            'timestamp': 1437679032,

        },
        'expected_warnings': [
            'Unable to download f4m manifest',
        ],
    }]

    def _real_extract(self, url):
        video_id = self._match_id(url)
        info = self._extract_feed_info(f'http://vid.bleacherreport.com/videos/{video_id}.akamai')
        info['id'] = video_id
        return info