Upgrade yt_dlp and download script

This commit is contained in:
2025-05-02 16:11:08 -05:00
parent 3a2e8eeb08
commit d68d9ce4f9
1194 changed files with 60099 additions and 44436 deletions

View File

@@ -1,14 +1,15 @@
from .telecinco import TelecincoIE
from .telecinco import TelecincoBaseIE
from ..networking.exceptions import HTTPError
from ..utils import (
ExtractorError,
int_or_none,
parse_iso8601,
)
class MiTeleIE(TelecincoIE): # XXX: Do not subclass from concrete IE
class MiTeleIE(TelecincoBaseIE):
IE_DESC = 'mitele.es'
_VALID_URL = r'https?://(?:www\.)?mitele\.es/(?:[^/]+/)+(?P<id>[^/]+)/player'
_TESTS = [{
'url': 'http://www.mitele.es/programas-tv/diario-de/57b0dfb9c715da65618b4afa/player',
'info_dict': {
@@ -27,6 +28,7 @@ class MiTeleIE(TelecincoIE): # XXX: Do not subclass from concrete IE
'timestamp': 1471209401,
'upload_date': '20160814',
},
'skip': 'HTTP Error 404 Not Found',
}, {
# no explicit title
'url': 'http://www.mitele.es/programas-tv/cuarto-milenio/57b0de3dc915da14058b4876/player',
@@ -49,6 +51,26 @@ class MiTeleIE(TelecincoIE): # XXX: Do not subclass from concrete IE
'params': {
'skip_download': True,
},
'skip': 'HTTP Error 404 Not Found',
}, {
'url': 'https://www.mitele.es/programas-tv/horizonte/temporada-5/programa-171-40_013480051/player/',
'info_dict': {
'id': '7adbe22e-cd41-4787-afa4-36f3da7c2c6f',
'ext': 'mp4',
'title': 'Horizonte Temporada 5 Programa 171',
'description': 'md5:97f1fb712c5ac27e5693a8b3c5c0c6e3',
'episode': 'Las Zonas de Bajas Emisiones, a debate',
'episode_number': 171,
'season': 'Season 5',
'season_number': 5,
'series': 'Horizonte',
'duration': 7012,
'upload_date': '20240927',
'timestamp': 1727416450,
'thumbnail': 'https://album.mediaset.es/eimg/2024/09/27/horizonte-171_9f02.jpg',
'age_limit': 12,
},
'params': {'geo_bypass_country': 'ES'},
}, {
'url': 'http://www.mitele.es/series-online/la-que-se-avecina/57aac5c1c915da951a8b45ed/player',
'only_matching': True,
@@ -59,10 +81,20 @@ class MiTeleIE(TelecincoIE): # XXX: Do not subclass from concrete IE
def _real_extract(self, url):
display_id = self._match_id(url)
webpage = self._download_webpage(url, display_id)
pre_player = self._parse_json(self._search_regex(
r'window\.\$REACTBASE_STATE\.prePlayer_mtweb\s*=\s*({.+})',
webpage, 'Pre Player'), display_id)['prePlayer']
try: # yt-dlp's default user-agents are too old and blocked by akamai
webpage = self._download_webpage(url, display_id, headers={
'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; rv:136.0) Gecko/20100101 Firefox/136.0',
})
except ExtractorError as e:
if not isinstance(e.cause, HTTPError) or e.cause.status != 403:
raise
# Retry with impersonation if hardcoded UA is insufficient to bypass akamai
webpage = self._download_webpage(url, display_id, impersonate=True)
pre_player = self._search_json(
r'window\.\$REACTBASE_STATE\.prePlayer_mtweb\s*=',
webpage, 'Pre Player', display_id)['prePlayer']
title = pre_player['title']
video_info = self._parse_content(pre_player['video'], url)
content = pre_player.get('content') or {}