1
mirror of https://github.com/yt-dlp/yt-dlp synced 2024-11-25 00:07:17 +01:00
yt-dlp/yt_dlp/extractor/firsttv.py
Pccode66 7a5c1cfe93
Completely change project name to yt-dlp (#85)
* All modules and binary names are changed
* All documentation references changed
* yt-dlp no longer loads youtube-dlc config files
* All URLs changed to point to organization account

Co-authored-by: Pccode66
Co-authored-by: pukkandan
2021-02-25 00:15:56 +05:30

157 lines
6.3 KiB
Python
Raw Blame History

This file contains invisible Unicode characters

This file contains invisible Unicode characters that are indistinguishable to humans but may be processed differently by a computer. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

# coding: utf-8
from __future__ import unicode_literals
from .common import InfoExtractor
from ..compat import (
compat_str,
compat_urlparse,
)
from ..utils import (
int_or_none,
qualities,
unified_strdate,
url_or_none,
)
class FirstTVIE(InfoExtractor):
IE_NAME = '1tv'
IE_DESC = 'Первый канал'
_VALID_URL = r'https?://(?:www\.)?1tv\.ru/(?:[^/]+/)+(?P<id>[^/?#]+)'
_TESTS = [{
# single format
'url': 'http://www.1tv.ru/shows/naedine-so-vsemi/vypuski/gost-lyudmila-senchina-naedine-so-vsemi-vypusk-ot-12-02-2015',
'md5': 'a1b6b60d530ebcf8daacf4565762bbaf',
'info_dict': {
'id': '40049',
'ext': 'mp4',
'title': 'Гость Людмила Сенчина. Наедине со всеми. Выпуск от 12.02.2015',
'thumbnail': r're:^https?://.*\.(?:jpg|JPG)$',
'upload_date': '20150212',
'duration': 2694,
},
}, {
# multiple formats
'url': 'http://www.1tv.ru/shows/dobroe-utro/pro-zdorove/vesennyaya-allergiya-dobroe-utro-fragment-vypuska-ot-07042016',
'info_dict': {
'id': '364746',
'ext': 'mp4',
'title': 'Весенняя аллергия. Доброе утро. Фрагмент выпуска от 07.04.2016',
'thumbnail': r're:^https?://.*\.(?:jpg|JPG)$',
'upload_date': '20160407',
'duration': 179,
'formats': 'mincount:3',
},
'params': {
'skip_download': True,
},
}, {
'url': 'http://www.1tv.ru/news/issue/2016-12-01/14:00',
'info_dict': {
'id': '14:00',
'title': 'Выпуск новостей в 14:00 1 декабря 2016 года. Новости. Первый канал',
'description': 'md5:2e921b948f8c1ff93901da78ebdb1dfd',
},
'playlist_count': 13,
}, {
'url': 'http://www.1tv.ru/shows/tochvtoch-supersezon/vystupleniya/evgeniy-dyatlov-vladimir-vysockiy-koni-priveredlivye-toch-v-toch-supersezon-fragment-vypuska-ot-06-11-2016',
'only_matching': True,
}]
def _real_extract(self, url):
display_id = self._match_id(url)
webpage = self._download_webpage(url, display_id)
playlist_url = compat_urlparse.urljoin(url, self._search_regex(
r'data-playlist-url=(["\'])(?P<url>(?:(?!\1).)+)\1',
webpage, 'playlist url', group='url'))
parsed_url = compat_urlparse.urlparse(playlist_url)
qs = compat_urlparse.parse_qs(parsed_url.query)
item_ids = qs.get('videos_ids[]') or qs.get('news_ids[]')
items = self._download_json(playlist_url, display_id)
if item_ids:
items = [
item for item in items
if item.get('uid') and compat_str(item['uid']) in item_ids]
else:
items = [items[0]]
entries = []
QUALITIES = ('ld', 'sd', 'hd', )
for item in items:
title = item['title']
quality = qualities(QUALITIES)
formats = []
path = None
for f in item.get('mbr', []):
src = url_or_none(f.get('src'))
if not src:
continue
tbr = int_or_none(self._search_regex(
r'_(\d{3,})\.mp4', src, 'tbr', default=None))
if not path:
path = self._search_regex(
r'//[^/]+/(.+?)_\d+\.mp4', src,
'm3u8 path', default=None)
formats.append({
'url': src,
'format_id': f.get('name'),
'tbr': tbr,
'source_preference': quality(f.get('name')),
# quality metadata of http formats may be incorrect
'preference': -10,
})
# m3u8 URL format is reverse engineered from [1] (search for
# master.m3u8). dashEdges (that is currently balancer-vod.1tv.ru)
# is taken from [2].
# 1. http://static.1tv.ru/player/eump1tv-current/eump-1tv.all.min.js?rnd=9097422834:formatted
# 2. http://static.1tv.ru/player/eump1tv-config/config-main.js?rnd=9097422834
if not path and len(formats) == 1:
path = self._search_regex(
r'//[^/]+/(.+?$)', formats[0]['url'],
'm3u8 path', default=None)
if path:
if len(formats) == 1:
m3u8_path = ','
else:
tbrs = [compat_str(t) for t in sorted(f['tbr'] for f in formats)]
m3u8_path = '_,%s,%s' % (','.join(tbrs), '.mp4')
formats.extend(self._extract_m3u8_formats(
'http://balancer-vod.1tv.ru/%s%s.urlset/master.m3u8'
% (path, m3u8_path),
display_id, 'mp4',
entry_protocol='m3u8_native', m3u8_id='hls', fatal=False))
self._sort_formats(formats)
thumbnail = item.get('poster') or self._og_search_thumbnail(webpage)
duration = int_or_none(item.get('duration') or self._html_search_meta(
'video:duration', webpage, 'video duration', fatal=False))
upload_date = unified_strdate(self._html_search_meta(
'ya:ovs:upload_date', webpage, 'upload date', default=None))
entries.append({
'id': compat_str(item.get('id') or item['uid']),
'thumbnail': thumbnail,
'title': title,
'upload_date': upload_date,
'duration': int_or_none(duration),
'formats': formats
})
title = self._html_search_regex(
(r'<div class="tv_translation">\s*<h1><a href="[^"]+">([^<]*)</a>',
r"'title'\s*:\s*'([^']+)'"),
webpage, 'title', default=None) or self._og_search_title(
webpage, default=None)
description = self._html_search_regex(
r'<div class="descr">\s*<div>&nbsp;</div>\s*<p>([^<]*)</p></div>',
webpage, 'description', default=None) or self._html_search_meta(
'description', webpage, 'description', default=None)
return self.playlist_result(entries, display_id, title, description)