yt_dlp/extractor/nzherald.py

   1 from .brightcove import BrightcoveNewIE
   2 from .common import InfoExtractor
   3
   4 from ..compat import compat_str
   5 from ..utils import (
   6     ExtractorError,
   7     traverse_obj
   8 )
   9
  10
  11 class NZHeraldIE(InfoExtractor):
  12     IE_NAME = 'nzherald'
  13     _VALID_URL = r'https?://(?:www\.)?nzherald\.co\.nz/[\w\/-]+\/(?P<id>[A-Z0-9]+)'
  14     _TESTS = [
  15         {
  16             'url': 'https://www.nzherald.co.nz/nz/weather-heavy-rain-gales-across-nz-most-days-this-week/PTG7QWY4E2225YHZ5NAIRBTYTQ/',
  17             'info_dict': {
  18                 'id': '6271084466001',
  19                 'ext': 'mp4',
  20                 'title': 'MetService severe weather warning: September 6th - 7th',
  21                 'timestamp': 1630891576,
  22                 'upload_date': '20210906',
  23                 'uploader_id': '1308227299001',
  24                 'description': 'md5:db6ca335a22e2cdf37ab9d2bcda52902'
  25             }
  26
  27         }, {
  28             # Webpage has brightcove embed player url
  29             'url': 'https://www.nzherald.co.nz/travel/pencarrow-coastal-trail/HDVTPJEPP46HJ2UEMK4EGD2DFI/',
  30             'info_dict': {
  31                 'id': '6261791733001',
  32                 'ext': 'mp4',
  33                 'title': 'Pencarrow Coastal Trail',
  34                 'timestamp': 1625102897,
  35                 'upload_date': '20210701',
  36                 'uploader_id': '1308227299001',
  37                 'description': 'md5:d361aaa0c6498f7ac1bc4fc0a0aec1e4'
  38             }
  39
  40         }, {
  41             # two video embeds of the same video
  42             'url': 'https://www.nzherald.co.nz/nz/truck-driver-captured-cutting-off-motorist-on-state-highway-1-in-canterbury/FIHNJB7PLLPHWQPK4S7ZBDUC4I/',
  43             'info_dict': {
  44                 'id': '6251114530001',
  45                 'ext': 'mp4',
  46                 'title': 'Truck travelling north from Rakaia runs car off road',
  47                 'timestamp': 1619730509,
  48                 'upload_date': '20210429',
  49                 'uploader_id': '1308227299001',
  50                 'description': 'md5:4cae7dfb7613ac4c73b9e73a75c6b5d7'
  51             }
  52         }, {
  53             'url': 'https://www.nzherald.co.nz/kahu/kaupapa-companies-my-taiao-supporting-maori-in-study-and-business/PQBO2J25WCG77VGRX7W7BVYEAI/',
  54             'only_matching': True
  55         }, {
  56             'url': 'https://nzherald.co.nz/the-country/video/focus-nzs-first-mass-covid-19-vaccination-event/N5I7IL3BRFLZSD33TLDLYJDGK4/',
  57             'only_matching': True
  58         }, {
  59             'url': 'https://www.nzherald.co.nz/the-vision-is-clear/news/tvic-damian-roper-planting-trees-an-addiction/AN2AAEPNRK5VLISDWQAJZB6ATQ',
  60             'only_matching': True
  61         }
  62     ]
  63
  64     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/1308227299001/S1BXZn8t_default/index.html?videoId=%s'
  65
  66     def _extract_bc_embed_url(self, webpage):
  67         """The initial webpage may include the brightcove player embed url"""
  68         bc_url = BrightcoveNewIE._extract_url(self, webpage)
  69         return bc_url or self._search_regex(
  70             r'(?:embedUrl)\"\s*:\s*\"(?P<embed_url>%s)' % BrightcoveNewIE._VALID_URL,
  71             webpage, 'embed url', default=None, group='embed_url')
  72
  73     def _real_extract(self, url):
  74         article_id = self._match_id(url)
  75         webpage = self._download_webpage(url, article_id)
  76         bc_url = self._extract_bc_embed_url(webpage)
  77
  78         if not bc_url:
  79             fusion_metadata = self._parse_json(
  80                 self._search_regex(r'Fusion\.globalContent\s*=\s*({.+?})\s*;', webpage, 'fusion metadata'), article_id)
  81
  82             video_metadata = fusion_metadata.get('video')
  83             bc_video_id = traverse_obj(
  84                 video_metadata or fusion_metadata,  # fusion metadata is the video metadata for video-only pages
  85                 'brightcoveId', ('content_elements', ..., 'referent', 'id'),
  86                 get_all=False, expected_type=compat_str)
  87
  88             if not bc_video_id:
  89                 if isinstance(video_metadata, dict) and len(video_metadata) == 0:
  90                     raise ExtractorError('This article does not have a video.', expected=True)
  91                 else:
  92                     raise ExtractorError('Failed to extract brightcove video id')
  93             bc_url = self.BRIGHTCOVE_URL_TEMPLATE % bc_video_id
  94
  95         return self.url_result(bc_url, 'BrightcoveNew')