0
0
Fork 0
mirror of https://github.com/yt-dlp/yt-dlp.git synced 2024-11-23 02:25:11 +00:00

[ie/radiocomercial] Add extractors (#8508)

Authored by: SirElderling
This commit is contained in:
SirElderling 2023-11-11 20:10:19 +00:00 committed by GitHub
parent 46acc418a5
commit ef12dbdcd3
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
2 changed files with 154 additions and 0 deletions

View file

@ -1583,6 +1583,10 @@
RadioCanadaIE,
RadioCanadaAudioVideoIE,
)
from .radiocomercial import (
RadioComercialIE,
RadioComercialPlaylistIE,
)
from .radiode import RadioDeIE
from .radiojavan import RadioJavanIE
from .radiobremen import RadioBremenIE

View file

@ -0,0 +1,150 @@
import itertools
from .common import InfoExtractor
from ..networking.exceptions import HTTPError
from ..utils import (
ExtractorError,
extract_attributes,
get_element_by_class,
get_element_html_by_class,
get_element_text_and_html_by_tag,
get_elements_html_by_class,
int_or_none,
join_nonempty,
try_call,
unified_strdate,
update_url,
urljoin
)
from ..utils.traversal import traverse_obj
class RadioComercialIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?radiocomercial\.pt/podcasts/[^/?#]+/t?(?P<season>\d+)/(?P<id>[\w-]+)'
_TESTS = [{
'url': 'https://radiocomercial.pt/podcasts/o-homem-que-mordeu-o-cao/t6/taylor-swift-entranhando-se-que-nem-uma-espada-no-ventre-dos-fas#page-content-wrapper',
'md5': '5f4fe8e485b29d2e8fd495605bc2c7e4',
'info_dict': {
'id': 'taylor-swift-entranhando-se-que-nem-uma-espada-no-ventre-dos-fas',
'ext': 'mp3',
'title': 'Taylor Swift entranhando-se que nem uma espada no ventre dos fãs.',
'release_date': '20231025',
'thumbnail': r're:https://radiocomercial.pt/upload/[^.]+.jpg',
'season': 6
}
}, {
'url': 'https://radiocomercial.pt/podcasts/convenca-me-num-minuto/t3/convenca-me-num-minuto-que-os-lobisomens-existem',
'md5': '47e96c273aef96a8eb160cd6cf46d782',
'info_dict': {
'id': 'convenca-me-num-minuto-que-os-lobisomens-existem',
'ext': 'mp3',
'title': 'Convença-me num minuto que os lobisomens existem',
'release_date': '20231026',
'thumbnail': r're:https://radiocomercial.pt/upload/[^.]+.jpg',
'season': 3
}
}, {
'url': 'https://radiocomercial.pt/podcasts/inacreditavel-by-ines-castel-branco/t2/o-desastre-de-aviao',
'md5': '69be64255420fec23b7259955d771e54',
'info_dict': {
'id': 'o-desastre-de-aviao',
'ext': 'mp3',
'title': 'O desastre de avião',
'description': 'md5:8a82beeb372641614772baab7246245f',
'release_date': '20231101',
'thumbnail': r're:https://radiocomercial.pt/upload/[^.]+.jpg',
'season': 2
},
'params': {
# inconsistant md5
'skip_download': True,
},
}, {
'url': 'https://radiocomercial.pt/podcasts/tnt-todos-no-top/2023/t-n-t-29-de-outubro',
'md5': '91d32d4d4b1407272068b102730fc9fa',
'info_dict': {
'id': 't-n-t-29-de-outubro',
'ext': 'mp3',
'title': 'T.N.T 29 de outubro',
'release_date': '20231029',
'thumbnail': r're:https://radiocomercial.pt/upload/[^.]+.jpg',
'season': 2023
}
}]
def _real_extract(self, url):
video_id, season = self._match_valid_url(url).group('id', 'season')
webpage = self._download_webpage(url, video_id)
return {
'id': video_id,
'title': self._html_extract_title(webpage),
'description': self._og_search_description(webpage, default=None),
'release_date': unified_strdate(get_element_by_class(
'date', get_element_html_by_class('descriptions', webpage) or '')),
'thumbnail': self._og_search_thumbnail(webpage),
'season': int_or_none(season),
'url': extract_attributes(get_element_html_by_class('audiofile', webpage) or '').get('href'),
}
class RadioComercialPlaylistIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?radiocomercial\.pt/podcasts/(?P<id>[\w-]+)(?:/t?(?P<season>\d+))?/?(?:$|[?#])'
_TESTS = [{
'url': 'https://radiocomercial.pt/podcasts/convenca-me-num-minuto/t3',
'info_dict': {
'id': 'convenca-me-num-minuto_t3',
'title': 'Convença-me num Minuto - Temporada 3',
},
'playlist_mincount': 32
}, {
'url': 'https://radiocomercial.pt/podcasts/o-homem-que-mordeu-o-cao',
'info_dict': {
'id': 'o-homem-que-mordeu-o-cao',
'title': 'O Homem Que Mordeu o Cão',
},
'playlist_mincount': 19
}, {
'url': 'https://radiocomercial.pt/podcasts/as-minhas-coisas-favoritas',
'info_dict': {
'id': 'as-minhas-coisas-favoritas',
'title': 'As Minhas Coisas Favoritas',
},
'playlist_mincount': 131
}, {
'url': 'https://radiocomercial.pt/podcasts/tnt-todos-no-top/t2023',
'info_dict': {
'id': 'tnt-todos-no-top_t2023',
'title': 'TNT - Todos No Top - Temporada 2023',
},
'playlist_mincount': 39
}]
def _entries(self, url, playlist_id):
for page in itertools.count(1):
try:
webpage = self._download_webpage(
f'{url}/{page}', playlist_id, f'Downloading page {page}')
except ExtractorError as e:
if isinstance(e.cause, HTTPError) and e.cause.status == 404:
break
raise
episodes = get_elements_html_by_class('tm-ouvir-podcast', webpage)
if not episodes:
break
for url_path in traverse_obj(episodes, (..., {extract_attributes}, 'href')):
episode_url = urljoin(url, url_path)
if RadioComercialIE.suitable(episode_url):
yield episode_url
def _real_extract(self, url):
podcast, season = self._match_valid_url(url).group('id', 'season')
playlist_id = join_nonempty(podcast, season, delim='_t')
url = update_url(url, query=None, fragment=None)
webpage = self._download_webpage(url, playlist_id)
name = try_call(lambda: get_element_text_and_html_by_tag('h1', webpage)[0])
title = name if name == season else join_nonempty(name, season, delim=' - Temporada ')
return self.playlist_from_matches(
self._entries(url, playlist_id), playlist_id, title, ie=RadioComercialIE)