yt_dlp/extractor/sverigesradio.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     determine_ext,
   4     int_or_none,
   5     str_or_none,
   6 )
   7
   8
   9 class SverigesRadioBaseIE(InfoExtractor):
  10     _BASE_URL = 'https://sverigesradio.se/sida/playerajax/'
  11     _QUALITIES = ['low', 'medium', 'high']
  12     _EXT_TO_CODEC_MAP = {
  13         'mp3': 'mp3',
  14         'm4a': 'aac',
  15     }
  16     _CODING_FORMAT_TO_ABR_MAP = {
  17         5: 128,
  18         11: 192,
  19         12: 32,
  20         13: 96,
  21     }
  22
  23     def _real_extract(self, url):
  24         audio_id = self._match_id(url)
  25         query = {
  26             'id': audio_id,
  27             'type': self._AUDIO_TYPE,
  28         }
  29
  30         item = self._download_json(
  31             self._BASE_URL + 'audiometadata', audio_id,
  32             'Downloading audio JSON metadata', query=query)['items'][0]
  33         title = item['subtitle']
  34
  35         query['format'] = 'iis'
  36         urls = []
  37         formats = []
  38         for quality in self._QUALITIES:
  39             query['quality'] = quality
  40             audio_url_data = self._download_json(
  41                 self._BASE_URL + 'getaudiourl', audio_id,
  42                 'Downloading %s format JSON metadata' % quality,
  43                 fatal=False, query=query) or {}
  44             audio_url = audio_url_data.get('audioUrl')
  45             if not audio_url or audio_url in urls:
  46                 continue
  47             urls.append(audio_url)
  48             ext = determine_ext(audio_url)
  49             coding_format = audio_url_data.get('codingFormat')
  50             abr = int_or_none(self._search_regex(
  51                 r'_a(\d+)\.m4a', audio_url, 'audio bitrate',
  52                 default=None)) or self._CODING_FORMAT_TO_ABR_MAP.get(coding_format)
  53             formats.append({
  54                 'abr': abr,
  55                 'acodec': self._EXT_TO_CODEC_MAP.get(ext),
  56                 'ext': ext,
  57                 'format_id': str_or_none(coding_format),
  58                 'vcodec': 'none',
  59                 'url': audio_url,
  60             })
  61
  62         return {
  63             'id': audio_id,
  64             'title': title,
  65             'formats': formats,
  66             'series': item.get('title'),
  67             'duration': int_or_none(item.get('duration')),
  68             'thumbnail': item.get('displayimageurl'),
  69             'description': item.get('description'),
  70         }
  71
  72
  73 class SverigesRadioPublicationIE(SverigesRadioBaseIE):
  74     IE_NAME = 'sverigesradio:publication'
  75     _VALID_URL = r'https?://(?:www\.)?sverigesradio\.se/sida/(?:artikel|gruppsida)\.aspx\?.*?\bartikel=(?P<id>[0-9]+)'
  76     _TESTS = [{
  77         'url': 'https://sverigesradio.se/sida/artikel.aspx?programid=83&artikel=7038546',
  78         'md5': '6a4917e1923fccb080e5a206a5afa542',
  79         'info_dict': {
  80             'id': '7038546',
  81             'ext': 'm4a',
  82             'duration': 132,
  83             'series': 'Nyheter (Ekot)',
  84             'title': 'Esa Teittinen: Sanningen har inte kommit fram',
  85             'description': 'md5:daf7ce66a8f0a53d5465a5984d3839df',
  86             'thumbnail': r're:^https?://.*\.jpg',
  87         },
  88     }, {
  89         'url': 'https://sverigesradio.se/sida/gruppsida.aspx?programid=3304&grupp=6247&artikel=7146887',
  90         'only_matching': True,
  91     }]
  92     _AUDIO_TYPE = 'publication'
  93
  94
  95 class SverigesRadioEpisodeIE(SverigesRadioBaseIE):
  96     IE_NAME = 'sverigesradio:episode'
  97     _VALID_URL = r'https?://(?:www\.)?sverigesradio\.se/(?:sida/)?avsnitt/(?P<id>[0-9]+)'
  98     _TEST = {
  99         'url': 'https://sverigesradio.se/avsnitt/1140922?programid=1300',
 100         'md5': '20dc4d8db24228f846be390b0c59a07c',
 101         'info_dict': {
 102             'id': '1140922',
 103             'ext': 'mp3',
 104             'duration': 3307,
 105             'series': 'Konflikt',
 106             'title': 'Metoo och valen',
 107             'description': 'md5:fcb5c1f667f00badcc702b196f10a27e',
 108             'thumbnail': r're:^https?://.*\.jpg',
 109         }
 110     }
 111     _AUDIO_TYPE = 'episode'