cbslocal.py (4117B)
1 # coding: utf-8 2 from __future__ import unicode_literals 3 4 from .anvato import AnvatoIE 5 from .sendtonews import SendtoNewsIE 6 from ..compat import compat_urlparse 7 from ..utils import ( 8 parse_iso8601, 9 unified_timestamp, 10 ) 11 12 13 class CBSLocalIE(AnvatoIE): 14 _VALID_URL_BASE = r'https?://[a-z]+\.cbslocal\.com/' 15 _VALID_URL = _VALID_URL_BASE + r'video/(?P<id>\d+)' 16 17 _TESTS = [{ 18 'url': 'http://newyork.cbslocal.com/video/3580809-a-very-blue-anniversary/', 19 'info_dict': { 20 'id': '3580809', 21 'ext': 'mp4', 22 'title': 'A Very Blue Anniversary', 23 'description': 'CBS2’s Cindy Hsu has more.', 24 'thumbnail': 're:^https?://.*', 25 'timestamp': int, 26 'upload_date': r're:^\d{8}$', 27 'uploader': 'CBS', 28 'subtitles': { 29 'en': 'mincount:5', 30 }, 31 'categories': [ 32 'Stations\\Spoken Word\\WCBSTV', 33 'Syndication\\AOL', 34 'Syndication\\MSN', 35 'Syndication\\NDN', 36 'Syndication\\Yahoo', 37 'Content\\News', 38 'Content\\News\\Local News', 39 ], 40 'tags': ['CBS 2 News Weekends', 'Cindy Hsu', 'Blue Man Group'], 41 }, 42 'params': { 43 'skip_download': True, 44 }, 45 }] 46 47 def _real_extract(self, url): 48 mcp_id = self._match_id(url) 49 return self.url_result( 50 'anvato:anvato_cbslocal_app_web_prod_547f3e49241ef0e5d30c79b2efbca5d92c698f67:' + mcp_id, 'Anvato', mcp_id) 51 52 53 class CBSLocalArticleIE(AnvatoIE): 54 _VALID_URL = CBSLocalIE._VALID_URL_BASE + r'\d+/\d+/\d+/(?P<id>[0-9a-z-]+)' 55 56 _TESTS = [{ 57 # Anvato backend 58 'url': 'http://losangeles.cbslocal.com/2016/05/16/safety-advocates-say-fatal-car-seat-failures-are-public-health-crisis', 59 'md5': 'f0ee3081e3843f575fccef901199b212', 60 'info_dict': { 61 'id': '3401037', 62 'ext': 'mp4', 63 'title': 'Safety Advocates Say Fatal Car Seat Failures Are \'Public Health Crisis\'', 64 'description': 'Collapsing seats have been the focus of scrutiny for decades, though experts say remarkably little has been done to address the issue. Randy Paige reports.', 65 'thumbnail': 're:^https?://.*', 66 'timestamp': 1463440500, 67 'upload_date': '20160516', 68 'uploader': 'CBS', 69 'subtitles': { 70 'en': 'mincount:5', 71 }, 72 'categories': [ 73 'Stations\\Spoken Word\\KCBSTV', 74 'Syndication\\MSN', 75 'Syndication\\NDN', 76 'Syndication\\AOL', 77 'Syndication\\Yahoo', 78 'Syndication\\Tribune', 79 'Syndication\\Curb.tv', 80 'Content\\News' 81 ], 82 'tags': ['CBS 2 News Evening'], 83 }, 84 }, { 85 # SendtoNews embed 86 'url': 'http://cleveland.cbslocal.com/2016/05/16/indians-score-season-high-15-runs-in-blowout-win-over-reds-rapid-reaction/', 87 'info_dict': { 88 'id': 'GxfCe0Zo7D-175909-5588', 89 }, 90 'playlist_count': 9, 91 'params': { 92 # m3u8 download 93 'skip_download': True, 94 }, 95 }] 96 97 def _real_extract(self, url): 98 display_id = self._match_id(url) 99 webpage = self._download_webpage(url, display_id) 100 101 sendtonews_url = SendtoNewsIE._extract_url(webpage) 102 if sendtonews_url: 103 return self.url_result( 104 compat_urlparse.urljoin(url, sendtonews_url), 105 ie=SendtoNewsIE.ie_key()) 106 107 info_dict = self._extract_anvato_videos(webpage, display_id) 108 109 timestamp = unified_timestamp(self._html_search_regex( 110 r'class="(?:entry|post)-date"[^>]*>([^<]+)', webpage, 111 'released date', default=None)) or parse_iso8601( 112 self._html_search_meta('uploadDate', webpage)) 113 114 info_dict.update({ 115 'display_id': display_id, 116 'timestamp': timestamp, 117 }) 118 119 return info_dict