]>
Commit | Line | Data |
---|---|---|
dc27fd8b RA |
1 | import base64 |
2 | import time | |
3 | import uuid | |
4 | ||
1da19488 | 5 | from .common import InfoExtractor |
dc27fd8b RA |
6 | from ..compat import ( |
7 | compat_HTTPError, | |
8 | compat_str, | |
9 | ) | |
10 | from ..utils import ( | |
11 | ExtractorError, | |
12 | int_or_none, | |
33b8c411 HTL |
13 | try_get, |
14 | url_or_none, | |
dc27fd8b | 15 | ) |
1da19488 YCH |
16 | |
17 | ||
18 | class MGTVIE(InfoExtractor): | |
8bdd16b4 | 19 | _VALID_URL = r'https?://(?:w(?:ww)?\.)?mgtv\.com/(v|b)/(?:[^/]+/)*(?P<id>\d+)\.html' |
a292cba2 | 20 | IE_DESC = '芒果TV' |
33b8c411 | 21 | IE_NAME = 'MangoTV' |
1da19488 | 22 | |
dc35ba0e | 23 | _TESTS = [{ |
1da19488 | 24 | 'url': 'http://www.mgtv.com/v/1/290525/f/3116640.html', |
1da19488 YCH |
25 | 'info_dict': { |
26 | 'id': '3116640', | |
27 | 'ext': 'mp4', | |
dc27fd8b | 28 | 'title': '我是歌手 第四季', |
1da19488 YCH |
29 | 'description': '我是歌手第四季双年巅峰会', |
30 | 'duration': 7461, | |
ec85ded8 | 31 | 'thumbnail': r're:^https?://.*\.jpg$', |
1da19488 | 32 | }, |
33b8c411 HTL |
33 | }, { |
34 | 'url': 'https://w.mgtv.com/b/427837/15588271.html', | |
35 | 'info_dict': { | |
36 | 'id': '15588271', | |
37 | 'ext': 'mp4', | |
38 | 'title': '春日迟迟再出发 沉浸版', | |
39 | 'description': 'md5:a7a05a05b1aa87bd50cae619b19bbca6', | |
40 | 'thumbnail': r're:^https?://.+\.jpg', | |
41 | 'duration': 4026, | |
42 | }, | |
43 | }, { | |
44 | 'url': 'https://w.mgtv.com/b/333652/7329822.html', | |
45 | 'info_dict': { | |
46 | 'id': '7329822', | |
47 | 'ext': 'mp4', | |
48 | 'title': '拜托,请你爱我', | |
49 | 'description': 'md5:cd81be6499bafe32e4d143abd822bf9c', | |
50 | 'thumbnail': r're:^https?://.+\.jpg', | |
51 | 'duration': 2656, | |
52 | }, | |
53 | }, { | |
54 | 'url': 'https://w.mgtv.com/b/427837/15591647.html', | |
55 | 'only_matching': True, | |
56 | }, { | |
57 | 'url': 'https://w.mgtv.com/b/388252/15634192.html?fpa=33318&fpos=4&lastp=ch_home', | |
58 | 'only_matching': True, | |
dc35ba0e | 59 | }, { |
30eaa3a7 | 60 | 'url': 'http://www.mgtv.com/b/301817/3826653.html', |
dc35ba0e | 61 | 'only_matching': True, |
8bdd16b4 | 62 | }, { |
63 | 'url': 'https://w.mgtv.com/b/301817/3826653.html', | |
64 | 'only_matching': True, | |
dc35ba0e | 65 | }] |
1da19488 YCH |
66 | |
67 | def _real_extract(self, url): | |
68 | video_id = self._match_id(url) | |
33b8c411 | 69 | tk2 = base64.urlsafe_b64encode( |
c6e07cf1 | 70 | f'did={str(uuid.uuid4())}|pno=1030|ver=0.3.0301|clit={int(time.time())}'.encode())[::-1] |
dc27fd8b RA |
71 | try: |
72 | api_data = self._download_json( | |
73 | 'https://pcweb.api.mgtv.com/player/video', video_id, query={ | |
8bdd16b4 | 74 | 'tk2': tk2, |
dc27fd8b | 75 | 'video_id': video_id, |
33b8c411 | 76 | 'type': 'pch5' |
dc27fd8b RA |
77 | }, headers=self.geo_verification_headers())['data'] |
78 | except ExtractorError as e: | |
79 | if isinstance(e.cause, compat_HTTPError) and e.cause.code == 401: | |
80 | error = self._parse_json(e.cause.read().decode(), None) | |
81 | if error.get('code') == 40005: | |
82 | self.raise_geo_restricted(countries=self._GEO_COUNTRIES) | |
83 | raise ExtractorError(error['msg'], expected=True) | |
84 | raise | |
1da19488 | 85 | info = api_data['info'] |
30eaa3a7 | 86 | title = info['title'].strip() |
dc27fd8b RA |
87 | stream_data = self._download_json( |
88 | 'https://pcweb.api.mgtv.com/player/getSource', video_id, query={ | |
89 | 'pm2': api_data['atc']['pm2'], | |
8bdd16b4 | 90 | 'tk2': tk2, |
dc27fd8b | 91 | 'video_id': video_id, |
33b8c411 | 92 | 'src': 'intelmgtv', |
dc27fd8b RA |
93 | }, headers=self.geo_verification_headers())['data'] |
94 | stream_domain = stream_data['stream_domain'][0] | |
1da19488 YCH |
95 | |
96 | formats = [] | |
dc27fd8b | 97 | for idx, stream in enumerate(stream_data['stream']): |
30eaa3a7 RA |
98 | stream_path = stream.get('url') |
99 | if not stream_path: | |
100 | continue | |
101 | format_data = self._download_json( | |
102 | stream_domain + stream_path, video_id, | |
33b8c411 | 103 | note=f'Download video info for format #{idx}') |
30eaa3a7 RA |
104 | format_url = format_data.get('info') |
105 | if not format_url: | |
b9e7bc55 | 106 | continue |
dc27fd8b | 107 | tbr = int_or_none(stream.get('filebitrate') or self._search_regex( |
30eaa3a7 RA |
108 | r'_(\d+)_mp4/', format_url, 'tbr', default=None)) |
109 | formats.append({ | |
110 | 'format_id': compat_str(tbr or idx), | |
33b8c411 | 111 | 'url': url_or_none(format_url), |
30eaa3a7 RA |
112 | 'ext': 'mp4', |
113 | 'tbr': tbr, | |
114 | 'protocol': 'm3u8_native', | |
4dcd4b7b S |
115 | 'http_headers': { |
116 | 'Referer': url, | |
117 | }, | |
9a37ff82 | 118 | 'format_note': stream.get('name'), |
30eaa3a7 | 119 | }) |
1da19488 YCH |
120 | self._sort_formats(formats) |
121 | ||
122 | return { | |
123 | 'id': video_id, | |
30eaa3a7 | 124 | 'title': title, |
1da19488 YCH |
125 | 'formats': formats, |
126 | 'description': info.get('desc'), | |
127 | 'duration': int_or_none(info.get('duration')), | |
128 | 'thumbnail': info.get('thumb'), | |
33b8c411 | 129 | 'subtitles': self.extract_subtitles(video_id, stream_domain), |
1da19488 | 130 | } |
33b8c411 HTL |
131 | |
132 | def _get_subtitles(self, video_id, domain): | |
133 | info = self._download_json(f'https://pcweb.api.mgtv.com/video/title?videoId={video_id}', | |
134 | video_id, fatal=False) or {} | |
135 | subtitles = {} | |
136 | for sub in try_get(info, lambda x: x['data']['title']) or []: | |
137 | url_sub = sub.get('url') | |
138 | if not url_sub: | |
139 | continue | |
1765c603 | 140 | locale = sub.get('captionSimpleName') or 'en' |
33b8c411 HTL |
141 | sub = self._download_json(f'{domain}{url_sub}', video_id, fatal=False, |
142 | note=f'Download subtitle for locale {sub.get("name")} ({locale})') or {} | |
143 | sub_url = url_or_none(sub.get('info')) | |
144 | if not sub_url: | |
145 | continue | |
1765c603 | 146 | subtitles.setdefault(locale.lower(), []).append({ |
33b8c411 | 147 | 'url': sub_url, |
1765c603 | 148 | 'name': sub.get('name'), |
33b8c411 HTL |
149 | 'ext': 'srt' |
150 | }) | |
151 | return subtitles |