youtube-dl/youtube_dl/extractor/iprima.py

150 lines
5.1 KiB
Python
Raw Permalink Normal View History

2016-03-24 16:54:26 +00:00
# coding: utf-8
from __future__ import unicode_literals
import re
2016-02-06 20:23:41 +00:00
import time
from .common import InfoExtractor
from ..utils import (
2016-03-24 16:54:26 +00:00
determine_ext,
js_to_json,
2014-03-14 18:38:44 +00:00
)
class IPrimaIE(InfoExtractor):
_VALID_URL = r'https?://(?:[^/]+)\.iprima\.cz/(?:[^/]+/)*(?P<id>[^/?#&]+)'
_GEO_BYPASS = False
_TESTS = [{
'url': 'https://prima.iprima.cz/particka/92-epizoda',
'info_dict': {
'id': 'p51388',
2016-02-06 20:23:41 +00:00
'ext': 'mp4',
'title': 'Partička (92)',
'description': 'md5:859d53beae4609e6dd7796413f1b6cac',
},
'params': {
'skip_download': True, # m3u8 download
},
}, {
'url': 'https://cnn.iprima.cz/videa/70-epizoda',
'info_dict': {
'id': 'p681554',
'ext': 'mp4',
'title': 'HLAVNÍ ZPRÁVY 3.5.2020',
},
'params': {
2016-02-06 20:23:41 +00:00
'skip_download': True, # m3u8 download
},
}, {
2016-02-06 20:23:41 +00:00
'url': 'http://play.iprima.cz/particka/particka-92',
2015-06-02 10:16:58 +00:00
'only_matching': True,
}, {
# geo restricted
'url': 'http://play.iprima.cz/closer-nove-pripady/closer-nove-pripady-iv-1',
'only_matching': True,
}, {
# iframe api.play-backend.iprima.cz
'url': 'https://prima.iprima.cz/my-little-pony/mapa-znameni-2-2',
'only_matching': True,
}, {
# iframe prima.iprima.cz
'url': 'https://prima.iprima.cz/porady/jak-se-stavi-sen/rodina-rathousova-praha',
'only_matching': True,
}, {
'url': 'http://www.iprima.cz/filmy/desne-rande',
'only_matching': True,
}, {
'url': 'https://zoom.iprima.cz/10-nejvetsich-tajemstvi-zahad/posvatna-mista-a-stavby',
'only_matching': True,
}, {
'url': 'https://krimi.iprima.cz/mraz-0/sebevrazdy',
'only_matching': True,
}, {
'url': 'https://cool.iprima.cz/derava-silnice-nevadi',
'only_matching': True,
}, {
'url': 'https://love.iprima.cz/laska-az-za-hrob/slib-dany-bratrovi',
'only_matching': True,
}, {
'url': 'https://autosalon.iprima.cz/motorsport/7-epizoda-1',
'only_matching': True,
}]
def _real_extract(self, url):
2016-03-24 16:54:26 +00:00
video_id = self._match_id(url)
self._set_cookie('play.iprima.cz', 'ott_adult_confirmed', '1')
webpage = self._download_webpage(url, video_id)
title = self._og_search_title(
webpage, default=None) or self._search_regex(
r'<h1>([^<]+)', webpage, 'title')
video_id = self._search_regex(
(r'<iframe[^>]+\bsrc=["\'](?:https?:)?//(?:api\.play-backend\.iprima\.cz/prehravac/embedded|prima\.iprima\.cz/[^/]+/[^/]+)\?.*?\bid=(p\d+)',
r'data-product="([^"]+)">',
r'id=["\']player-(p\d+)"',
r'playerId\s*:\s*["\']player-(p\d+)',
r'\bvideos\s*=\s*["\'](p\d+)'),
webpage, 'real id')
2014-03-14 18:38:44 +00:00
2017-02-20 16:16:14 +00:00
playerpage = self._download_webpage(
'http://play.iprima.cz/prehravac/init',
video_id, note='Downloading player', query={
'_infuse': 1,
'_ts': round(time.time()),
'productId': video_id,
}, headers={'Referer': url})
2016-03-24 16:54:26 +00:00
formats = []
2016-03-24 16:54:26 +00:00
def extract_formats(format_url, format_key=None, lang=None):
ext = determine_ext(format_url)
new_formats = []
if format_key == 'hls' or ext == 'm3u8':
new_formats = self._extract_m3u8_formats(
format_url, video_id, 'mp4', entry_protocol='m3u8_native',
m3u8_id='hls', fatal=False)
elif format_key == 'dash' or ext == 'mpd':
return
new_formats = self._extract_mpd_formats(
format_url, video_id, mpd_id='dash', fatal=False)
if lang:
for f in new_formats:
if not f.get('language'):
f['language'] = lang
formats.extend(new_formats)
options = self._parse_json(
self._search_regex(
r'(?s)(?:TDIPlayerOptions|playerOptions)\s*=\s*({.+?});\s*\]\]',
2016-03-24 16:54:26 +00:00
playerpage, 'player options', default='{}'),
video_id, transform_source=js_to_json, fatal=False)
if options:
for key, tracks in options.get('tracks', {}).items():
if not isinstance(tracks, list):
continue
for track in tracks:
src = track.get('src')
if src:
extract_formats(src, key.lower(), track.get('lang'))
if not formats:
for _, src in re.findall(r'src["\']\s*:\s*(["\'])(.+?)\1', playerpage):
extract_formats(src)
2016-10-07 14:53:03 +00:00
if not formats and '>GEO_IP_NOT_ALLOWED<' in playerpage:
self.raise_geo_restricted(countries=['CZ'])
2016-10-07 14:53:03 +00:00
self._sort_formats(formats)
return {
2016-02-06 20:23:41 +00:00
'id': video_id,
'title': title,
'thumbnail': self._og_search_thumbnail(webpage, default=None),
'formats': formats,
'description': self._og_search_description(webpage, default=None),
}