2017-02-13 21:42:26 +01:00
|
|
|
# coding: utf-8
|
|
|
|
from __future__ import unicode_literals
|
|
|
|
|
2021-02-11 01:09:05 +01:00
|
|
|
from .common import InfoExtractor
|
2021-02-11 13:18:57 +01:00
|
|
|
from ..utils import (
|
|
|
|
clean_html,
|
|
|
|
int_or_none,
|
|
|
|
parse_duration,
|
|
|
|
parse_iso8601,
|
|
|
|
)
|
2021-02-11 01:09:05 +01:00
|
|
|
from .pulsembed import PulseVideoIE
|
2017-02-13 21:42:26 +01:00
|
|
|
|
|
|
|
|
2021-02-11 01:09:05 +01:00
|
|
|
class VODPlIE(InfoExtractor):
|
2017-02-14 18:52:31 +01:00
|
|
|
_VALID_URL = r'https?://vod\.pl/(?:[^/]+/)+(?P<id>[0-9a-zA-Z]+)'
|
2017-02-13 21:42:26 +01:00
|
|
|
|
2017-02-14 18:52:31 +01:00
|
|
|
_TESTS = [{
|
2021-02-11 13:18:57 +01:00
|
|
|
'url': 'https://vod.pl/filmy-dokumentalne/wielce-krolewski-slub/wcl5tx0',
|
2017-02-13 21:42:26 +01:00
|
|
|
'info_dict': {
|
2021-02-11 13:18:57 +01:00
|
|
|
'id': '2163051.179206518',
|
2017-02-13 21:42:26 +01:00
|
|
|
'ext': 'mp4',
|
2021-02-11 13:18:57 +01:00
|
|
|
'title': 'Wielce królewski ślub',
|
|
|
|
'description': 'md5:9de1b6df5dba5c44fcde37584ad13302',
|
|
|
|
'timestamp': 1580313604,
|
|
|
|
'upload_date': '20200129',
|
|
|
|
},
|
|
|
|
}, {
|
|
|
|
'url': 'https://vod.pl/filmy/autopsja/62gx8n1',
|
|
|
|
'info_dict': {
|
|
|
|
'id': '1973639.1440605974',
|
|
|
|
'ext': 'mp4',
|
|
|
|
'title': 'Autopsja',
|
|
|
|
'description': 'md5:94cb987a8caeecd5755e3597d4c0bd66',
|
|
|
|
'upload_date': '20190203',
|
|
|
|
'timestamp': 1549227901,
|
|
|
|
'age_limit': 18,
|
2017-02-13 21:42:26 +01:00
|
|
|
},
|
2017-02-14 18:52:31 +01:00
|
|
|
}, {
|
|
|
|
'url': 'https://vod.pl/seriale/belfer-na-planie-praca-kamery-online/2c10heh',
|
|
|
|
'only_matching': True,
|
|
|
|
}]
|
2017-02-13 21:42:26 +01:00
|
|
|
|
|
|
|
def _real_extract(self, url):
|
|
|
|
video_id = self._match_id(url)
|
|
|
|
webpage = self._download_webpage(url, video_id)
|
2021-02-11 13:18:57 +01:00
|
|
|
data = self._parse_json(
|
|
|
|
self._search_regex(r'try {\s*vodDataLayer = ({.+?});', webpage, 'vod data layer'),
|
|
|
|
video_id)
|
|
|
|
description = clean_html(
|
|
|
|
self._search_regex(
|
|
|
|
r'(?s)<div[^>]+itemprop="description"[^>]*>(.+?)</div>',
|
|
|
|
webpage, 'description', default=None))
|
|
|
|
age_limit = int_or_none(self._search_regex(
|
|
|
|
r'<li class="v_AgeRating v_AgeRate_(\d+)" title="Kat\. wiekowa',
|
|
|
|
webpage, 'age limit', default=None)) or 0 # vod.pl does not show an age limit if it's 0
|
2021-02-11 01:09:05 +01:00
|
|
|
return {
|
|
|
|
'_type': 'url_transparent',
|
2021-02-11 13:18:57 +01:00
|
|
|
'url': 'pulsevideo:%s' % data['video']['mvpId'],
|
2021-02-11 01:09:05 +01:00
|
|
|
'ie_key': PulseVideoIE.ie_key(),
|
2021-02-11 13:18:57 +01:00
|
|
|
'title': data['published']['title'],
|
|
|
|
'description': description,
|
|
|
|
'duration': parse_duration(data['video'].get('duration')),
|
|
|
|
'timestamp': parse_iso8601(data['published'].get('date')),
|
|
|
|
'age_limit': age_limit,
|
2021-02-11 01:09:05 +01:00
|
|
|
}
|