This commit is contained in:
Nuno 2025-12-05 22:10:48 +05:30 committed by GitHub
commit 08c6abcd4a
No known key found for this signature in database
GPG key ID: B5690EEEBB952194

View file

@ -1,78 +1,111 @@
from __future__ import annotations
import re
from .common import InfoExtractor
from ..utils import traverse_obj
from ..utils import (
ExtractorError,
js_to_json,
)
from ..utils.traversal import (
traverse_obj,
)
class TVIPlayerIE(InfoExtractor):
_VALID_URL = r'https?://tviplayer\.iol\.pt(/programa/[\w-]+/[a-f0-9]+)?/\w+/(?P<id>\w+)'
_TESTS = [{
'url': 'https://tviplayer.iol.pt/programa/jornal-das-8/53c6b3903004dc006243d0cf/video/61c8e8b90cf2c7ea0f0f71a9',
'info_dict': {
'id': '61c8e8b90cf2c7ea0f0f71a9',
'ext': 'mp4',
'duration': 4167,
'title': 'Jornal das 8 - 26 de dezembro de 2021',
'thumbnail': 'https://www.iol.pt/multimedia/oratvi/multimedia/imagem/id/61c8ee630cf2cc58e7d98d9f/',
'season_number': 8,
'season': 'Season 8',
_VALID_URL = (
r'https?://tviplayer\.iol\.pt(?:/programa/[\w-]+/[a-f0-9]+)?/\w+/(?P<id>\w+)'
)
_TESTS = [
{
'url': 'https://tviplayer.iol.pt/programa/a-protegida/67a63479d34ef72ee441fa79/episodio/t1e120',
'info_dict': {
'id': '689683000cf20ac1d5f35341',
'ext': 'mp4',
'duration': 1593,
'title': 'A Protegida - Clarice descobre o que une Óscar a Gonçalo e Mónica',
'thumbnail': 'https://img.iol.pt/image/id/68971037d34ef72ee44941a6/',
'season_number': 1,
},
},
}, {
'url': 'https://tviplayer.iol.pt/programa/isabel/62b471090cf26256cd2a8594/video/62be445f0cf2ea4f0a5218e5',
'info_dict': {
'id': '62be445f0cf2ea4f0a5218e5',
'ext': 'mp4',
'duration': 3255,
'season': 'Season 1',
'title': 'Isabel - Episódio 1',
'thumbnail': 'https://www.iol.pt/multimedia/oratvi/multimedia/imagem/id/62beac200cf2f9a86eab856b/',
'season_number': 1,
},
}, {
# no /programa/
'url': 'https://tviplayer.iol.pt/video/62c4131c0cf2f9a86eac06bb',
'info_dict': {
'id': '62c4131c0cf2f9a86eac06bb',
'ext': 'mp4',
'title': 'David e Mickael Carreira respondem: «Qual é o próximo a ser pai?»',
'thumbnail': 'https://www.iol.pt/multimedia/oratvi/multimedia/imagem/id/62c416490cf2ea367d4433fd/',
'season': 'Season 2',
'duration': 148,
'season_number': 2,
},
}, {
# episodio url
'url': 'https://tviplayer.iol.pt/programa/para-sempre/61716c360cf2365a5ed894c4/episodio/t1e187',
'info_dict': {
'id': 't1e187',
'ext': 'mp4',
'season': 'Season 1',
'title': 'Quem denunciou Pedro?',
'thumbnail': 'https://www.iol.pt/multimedia/oratvi/multimedia/imagem/id/62eda30b0cf2ea367d48973b/',
'duration': 1250,
'season_number': 1,
},
}]
]
def _real_initialize(self):
# Obtain the wmsAuthSign token (non-fatal)
self.wms_auth_sign_token = self._download_webpage(
'https://services.iol.pt/matrix?userId=', 'wmsAuthSign',
note='Trying to get wmsAuthSign token')
'https://services.iol.pt/matrix?userId=',
'wmsAuthSign',
note='Downloading wmsAuthSign token',
fatal=False,
)
def _real_extract(self, url):
video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
# Try to locate a JS "video: [ {...} ]" block
json_data = self._search_json(
r'<script>\s*jsonData\s*=', webpage, 'json_data', video_id)
r'(?<!-)\bvideo\s*:\s*\[',
webpage,
'json_data',
video_id,
transform_source=js_to_json,
default={},
)
# Structured metadata from ld+json
_ = self._search_json_ld(webpage, video_id, default={}) or {}
# Merge data safely without type errors
def first_of(*keys):
for k in keys:
v = traverse_obj(json_data, (k,))
if v:
return v
return None
info_id = first_of('id')
title = first_of('title') or self._og_search_title(webpage)
thumbnail = first_of('cover', 'thumbnail') or self._og_search_thumbnail(webpage)
duration = first_of('duration')
try:
duration = int(duration) if duration is not None else None
except Exception:
try:
duration = int(float(duration))
except Exception:
duration = None
video_url = first_of('videoUrl', 'url', 'video_url')
if not video_url:
m = re.search(
r'["\']videoUrl["\']\s*:\s*["\'](https?://[^"\']+)["\']', webpage,
)
if m:
video_url = m.group(1)
if not video_url:
raise ExtractorError('Unable to locate video URL in webpage', expected=True)
query = (
{'wmsAuthSign': self.wms_auth_sign_token}
if self.wms_auth_sign_token
else {}
)
formats, subtitles = self._extract_m3u8_formats_and_subtitles(
f'{json_data["videoUrl"]}?wmsAuthSign={self.wms_auth_sign_token}',
video_id, ext='mp4')
video_url, video_id, ext='mp4', query=query, fatal=False,
)
season_number = traverse_obj(json_data, ('program', 'seasonNum'))
return {
'id': video_id,
'title': json_data.get('title') or self._og_search_title(webpage),
'thumbnail': json_data.get('cover') or self._og_search_thumbnail(webpage),
'duration': json_data.get('duration'),
'id': info_id or video_id,
'display_id': video_id,
'title': title,
'thumbnail': thumbnail,
'duration': duration,
'formats': formats,
'subtitles': subtitles,
'season_number': traverse_obj(json_data, ('program', 'seasonNum')),
'season_number': season_number,
}