]>
jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/ccma.py
2 from __future__
import unicode_literals
7 from .common
import InfoExtractor
19 class CCMAIE(InfoExtractor
):
20 _VALID_URL
= r
'https?://(?:www\.)?ccma\.cat/(?:[^/]+/)*?(?P<type>video|audio)/(?P<id>\d+)'
22 'url': 'http://www.ccma.cat/tv3/alacarta/lespot-de-la-marato-de-tv3/lespot-de-la-marato-de-tv3/video/5630208/',
23 'md5': '7296ca43977c8ea4469e719c609b0871',
27 'title': 'L\'espot de La Marató de TV3',
28 'description': 'md5:f12987f320e2f6e988e9908e4fe97765',
29 'timestamp': 1478608140,
30 'upload_date': '20161108',
34 'url': 'http://www.ccma.cat/catradio/alacarta/programa/el-consell-de-savis-analitza-el-derbi/audio/943685/',
35 'md5': 'fa3e38f269329a278271276330261425',
39 'title': 'El Consell de Savis analitza el derbi',
40 'description': 'md5:e2a3648145f3241cb9c6b4b624033e53',
41 'upload_date': '20170512',
42 'timestamp': 1494622500,
44 'categories': ['Esports'],
47 'url': 'http://www.ccma.cat/tv3/alacarta/crims/crims-josep-tallada-lespereu-me-capitol-1/video/6031387/',
48 'md5': 'b43c3d3486f430f3032b5b160d80cbc3',
52 'title': 'Crims - Josep Talleda, l\'"Espereu-me" (capítol 1)',
53 'description': 'md5:7cbdafb640da9d0d2c0f62bad1e74e60',
54 'timestamp': 1582577700,
55 'upload_date': '20200224',
56 'subtitles': 'mincount:4',
62 def _real_extract(self
, url
):
63 media_type
, media_id
= self
._match
_valid
_url
(url
).groups()
65 media
= self
._download
_json
(
66 'http://dinamics.ccma.cat/pvideo/media.jsp', media_id
, query
={
72 media_url
= media
['media']['url']
73 if isinstance(media_url
, list):
74 for format_
in media_url
:
75 format_url
= url_or_none(format_
.get('file'))
78 label
= format_
.get('label')
79 f
= parse_resolution(label
)
88 'vcodec': 'none' if media_type
== 'audio' else None,
90 self
._sort
_formats
(formats
)
92 informacio
= media
['informacio']
93 title
= informacio
['titol']
94 durada
= informacio
.get('durada') or {}
95 duration
= int_or_none(durada
.get('milisegons'), 1000) or parse_duration(durada
.get('text'))
96 tematica
= try_get(informacio
, lambda x
: x
['tematica']['text'])
99 data_utc
= try_get(informacio
, lambda x
: x
['data_emissio']['utc'])
101 timezone
, data_utc
= extract_timezone(data_utc
)
102 timestamp
= calendar
.timegm((datetime
.datetime
.strptime(
103 data_utc
, '%Y-%d-%mT%H:%M:%S') - timezone
).timetuple())
108 subtitols
= media
.get('subtitols') or []
109 if isinstance(subtitols
, dict):
110 subtitols
= [subtitols
]
112 sub_url
= st
.get('url')
114 subtitles
.setdefault(
115 st
.get('iso') or st
.get('text') or 'ca', []).append({
120 imatges
= media
.get('imatges', {})
122 thumbnail_url
= imatges
.get('url')
125 'url': thumbnail_url
,
126 'width': int_or_none(imatges
.get('amplada')),
127 'height': int_or_none(imatges
.get('alcada')),
131 codi_etic
= try_get(informacio
, lambda x
: x
['codi_etic']['id'])
133 codi_etic_s
= codi_etic
.split('_')
134 if len(codi_etic_s
) == 2:
135 if codi_etic_s
[1] == 'TP':
138 age_limit
= int_or_none(codi_etic_s
[1])
143 'description': clean_html(informacio
.get('descripcio')),
144 'duration': duration
,
145 'timestamp': timestamp
,
146 'thumbnails': thumbnails
,
147 'subtitles': subtitles
,
149 'age_limit': age_limit
,
150 'alt_title': informacio
.get('titol_complet'),
151 'episode_number': int_or_none(informacio
.get('capitol')),
152 'categories': [tematica
] if tematica
else None,
153 'series': informacio
.get('programa'),