]> jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/aljazeera.py
Tolerate failure to `--write-link` due to unknown URL
[yt-dlp.git] / yt_dlp / extractor / aljazeera.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 import json
5
6 from .common import InfoExtractor
7 from ..utils import (
8 try_get,
9 )
10
11
12 class AlJazeeraIE(InfoExtractor):
13 _VALID_URL = r'https?://(?P<base>\w+\.aljazeera\.\w+)/(?P<type>programs?/[^/]+|(?:feature|video|new)s)?/\d{4}/\d{1,2}/\d{1,2}/(?P<id>[^/?&#]+)'
14
15 _TESTS = [{
16 'url': 'https://balkans.aljazeera.net/videos/2021/11/6/pojedini-domovi-u-sarajevu-jos-pod-vodom-mjestanima-se-dostavlja-hrana',
17 'info_dict': {
18 'id': '6280641530001',
19 'ext': 'mp4',
20 'title': 'Pojedini domovi u Sarajevu još pod vodom, mještanima se dostavlja hrana',
21 'timestamp': 1636219149,
22 'description': 'U sarajevskim naseljima Rajlovac i Reljevo stambeni objekti, ali i industrijska postrojenja i dalje su pod vodom.',
23 'upload_date': '20211106',
24 }
25 }, {
26 'url': 'https://balkans.aljazeera.net/videos/2021/11/6/djokovic-usao-u-finale-mastersa-u-parizu',
27 'info_dict': {
28 'id': '6280654936001',
29 'ext': 'mp4',
30 'title': 'Đoković ušao u finale Mastersa u Parizu',
31 'timestamp': 1636221686,
32 'description': 'Novak Đoković je u polufinalu Mastersa u Parizu nakon preokreta pobijedio Poljaka Huberta Hurkacza.',
33 'upload_date': '20211106',
34 },
35 }]
36 BRIGHTCOVE_URL_RE = r'https?://players.brightcove.net/(?P<account>\d+)/(?P<player_id>[a-zA-Z0-9]+)_(?P<embed>[^/]+)/index.html\?videoId=(?P<id>\d+)'
37
38 def _real_extract(self, url):
39 base, post_type, id = self._match_valid_url(url).groups()
40 wp = {
41 'balkans.aljazeera.net': 'ajb',
42 'chinese.aljazeera.net': 'chinese',
43 'mubasher.aljazeera.net': 'ajm',
44 }.get(base) or 'aje'
45 post_type = {
46 'features': 'post',
47 'program': 'episode',
48 'programs': 'episode',
49 'videos': 'video',
50 'news': 'news',
51 }[post_type.split('/')[0]]
52 video = self._download_json(
53 f'https://{base}/graphql', id, query={
54 'wp-site': wp,
55 'operationName': 'ArchipelagoSingleArticleQuery',
56 'variables': json.dumps({
57 'name': id,
58 'postType': post_type,
59 }),
60 }, headers={
61 'wp-site': wp,
62 })
63 video = try_get(video, lambda x: x['data']['article']['video']) or {}
64 video_id = video.get('id')
65 account = video.get('accountId') or '911432371001'
66 player_id = video.get('playerId') or 'csvTfAlKW'
67 embed = 'default'
68
69 if video_id is None:
70 webpage = self._download_webpage(url, id)
71
72 account, player_id, embed, video_id = self._search_regex(self.BRIGHTCOVE_URL_RE, webpage, 'video id',
73 group=(1, 2, 3, 4), default=(None, None, None, None))
74
75 if video_id is None:
76 return {
77 '_type': 'url_transparent',
78 'url': url,
79 'ie_key': 'Generic'
80 }
81
82 return {
83 '_type': 'url_transparent',
84 'url': f'https://players.brightcove.net/{account}/{player_id}_{embed}/index.html?videoId={video_id}',
85 'ie_key': 'BrightcoveNew'
86 }