]>
Commit | Line | Data |
---|---|---|
3b01a9fb YCH |
1 | # coding: utf-8 |
2 | from __future__ import unicode_literals | |
3 | ||
4 | import json | |
3b01a9fb YCH |
5 | |
6 | from .common import InfoExtractor | |
7 | from ..utils import ( | |
8 | ExtractorError, | |
9 | int_or_none, | |
f40ee5e9 | 10 | traverse_obj, |
3b01a9fb YCH |
11 | smuggle_url, |
12 | unsmuggle_url, | |
13 | ) | |
14 | ||
15 | ||
16 | class LiTVIE(InfoExtractor): | |
92519402 | 17 | _VALID_URL = r'https?://(?:www\.)?litv\.tv/(?:vod|promo)/[^/]+/(?:content\.do)?\?.*?\b(?:content_)?id=(?P<id>[^&]+)' |
3b01a9fb YCH |
18 | |
19 | _URL_TEMPLATE = 'https://www.litv.tv/vod/%s/content.do?id=%s' | |
20 | ||
21 | _TESTS = [{ | |
22 | 'url': 'https://www.litv.tv/vod/drama/content.do?brc_id=root&id=VOD00041610&isUHEnabled=true&autoPlay=1', | |
23 | 'info_dict': { | |
24 | 'id': 'VOD00041606', | |
25 | 'title': '花千骨', | |
26 | }, | |
27 | 'playlist_count': 50, | |
28 | }, { | |
29 | 'url': 'https://www.litv.tv/vod/drama/content.do?brc_id=root&id=VOD00041610&isUHEnabled=true&autoPlay=1', | |
39e1c4f0 | 30 | 'md5': '969e343d9244778cb29acec608e53640', |
3b01a9fb YCH |
31 | 'info_dict': { |
32 | 'id': 'VOD00041610', | |
33 | 'ext': 'mp4', | |
34 | 'title': '花千骨第1集', | |
ec85ded8 | 35 | 'thumbnail': r're:https?://.*\.jpg$', |
3b01a9fb YCH |
36 | 'description': 'md5:c7017aa144c87467c4fb2909c4b05d6f', |
37 | 'episode_number': 1, | |
38 | }, | |
39 | 'params': { | |
40 | 'noplaylist': True, | |
39e1c4f0 YCH |
41 | }, |
42 | 'skip': 'Georestricted to Taiwan', | |
43 | }, { | |
44 | 'url': 'https://www.litv.tv/promo/miyuezhuan/?content_id=VOD00044841&', | |
45 | 'md5': '88322ea132f848d6e3e18b32a832b918', | |
46 | 'info_dict': { | |
47 | 'id': 'VOD00044841', | |
48 | 'ext': 'mp4', | |
49 | 'title': '芈月傳第1集 霸星芈月降世楚國', | |
50 | 'description': '楚威王二年,太史令唐昧夜觀星象,發現霸星即將現世。王后得知霸星的預言後,想盡辦法不讓孩子順利出生,幸得莒姬相護化解危機。沒想到眾人期待下出生的霸星卻是位公主,楚威王對此失望至極。楚王后命人將女嬰丟棄河中,居然奇蹟似的被少司命像攔下,楚威王認為此女非同凡響,為她取名芈月。', | |
3b01a9fb YCH |
51 | }, |
52 | 'skip': 'Georestricted to Taiwan', | |
53 | }] | |
54 | ||
9dde0e04 YCH |
55 | def _extract_playlist(self, season_list, video_id, program_info, prompt=True): |
56 | episode_title = program_info['title'] | |
3b01a9fb YCH |
57 | content_id = season_list['contentId'] |
58 | ||
3b01a9fb YCH |
59 | all_episodes = [ |
60 | self.url_result(smuggle_url( | |
9dde0e04 | 61 | self._URL_TEMPLATE % (program_info['contentType'], episode['contentId']), |
3b01a9fb YCH |
62 | {'force_noplaylist': True})) # To prevent infinite recursion |
63 | for episode in season_list['episode']] | |
64 | ||
65 | return self.playlist_result(all_episodes, content_id, episode_title) | |
66 | ||
67 | def _real_extract(self, url): | |
88f23a18 | 68 | url, smuggled_data = unsmuggle_url(url, {}) |
3b01a9fb YCH |
69 | |
70 | video_id = self._match_id(url) | |
71 | ||
3b01a9fb YCH |
72 | webpage = self._download_webpage(url, video_id) |
73 | ||
9dde0e04 | 74 | program_info = self._parse_json(self._search_regex( |
ec85ded8 | 75 | r'var\s+programInfo\s*=\s*([^;]+)', webpage, 'VOD data', default='{}'), |
3b01a9fb YCH |
76 | video_id) |
77 | ||
9dde0e04 | 78 | season_list = list(program_info.get('seasonList', {}).values()) |
f40ee5e9 | 79 | playlist_id = traverse_obj(season_list, 0, 'contentId') |
80 | if self._yes_playlist(playlist_id, video_id, smuggled_data): | |
81 | return self._extract_playlist(season_list[0], video_id, program_info) | |
3b01a9fb YCH |
82 | |
83 | # In browsers `getMainUrl` request is always issued. Usually this | |
84 | # endpoint gives the same result as the data embedded in the webpage. | |
85 | # If georestricted, there are no embedded data, so an extra request is | |
86 | # necessary to get the error code | |
9dde0e04 YCH |
87 | if 'assetId' not in program_info: |
88 | program_info = self._download_json( | |
39e1c4f0 YCH |
89 | 'https://www.litv.tv/vod/ajax/getProgramInfo', video_id, |
90 | query={'contentId': video_id}, | |
91 | headers={'Accept': 'application/json'}) | |
3b01a9fb YCH |
92 | video_data = self._parse_json(self._search_regex( |
93 | r'uiHlsUrl\s*=\s*testBackendData\(([^;]+)\);', | |
94 | webpage, 'video data', default='{}'), video_id) | |
95 | if not video_data: | |
96 | payload = { | |
9dde0e04 YCH |
97 | 'assetId': program_info['assetId'], |
98 | 'watchDevices': program_info['watchDevices'], | |
99 | 'contentType': program_info['contentType'], | |
3b01a9fb YCH |
100 | } |
101 | video_data = self._download_json( | |
102 | 'https://www.litv.tv/vod/getMainUrl', video_id, | |
103 | data=json.dumps(payload).encode('utf-8'), | |
104 | headers={'Content-Type': 'application/json'}) | |
105 | ||
106 | if not video_data.get('fullpath'): | |
107 | error_msg = video_data.get('errorMessage') | |
108 | if error_msg == 'vod.error.outsideregionerror': | |
109 | self.raise_geo_restricted('This video is available in Taiwan only') | |
110 | if error_msg: | |
111 | raise ExtractorError('%s said: %s' % (self.IE_NAME, error_msg), expected=True) | |
112 | raise ExtractorError('Unexpected result from %s' % self.IE_NAME) | |
113 | ||
114 | formats = self._extract_m3u8_formats( | |
39e1c4f0 YCH |
115 | video_data['fullpath'], video_id, ext='mp4', |
116 | entry_protocol='m3u8_native', m3u8_id='hls') | |
3b01a9fb YCH |
117 | for a_format in formats: |
118 | # LiTV HLS segments doesn't like compressions | |
119 | a_format.setdefault('http_headers', {})['Youtubedl-no-compression'] = True | |
120 | ||
9dde0e04 YCH |
121 | title = program_info['title'] + program_info.get('secondaryMark', '') |
122 | description = program_info.get('description') | |
123 | thumbnail = program_info.get('imageFile') | |
124 | categories = [item['name'] for item in program_info.get('category', [])] | |
125 | episode = int_or_none(program_info.get('episode')) | |
3b01a9fb YCH |
126 | |
127 | return { | |
128 | 'id': video_id, | |
129 | 'formats': formats, | |
130 | 'title': title, | |
131 | 'description': description, | |
132 | 'thumbnail': thumbnail, | |
133 | 'categories': categories, | |
134 | 'episode_number': episode, | |
135 | } |