]>
Commit | Line | Data |
---|---|---|
dcdb292f | 1 | # coding: utf-8 |
24da5893 JMF |
2 | from __future__ import unicode_literals |
3 | ||
6b95b065 | 4 | import re |
6b95b065 JMF |
5 | |
6 | from .common import InfoExtractor | |
1cc79574 | 7 | from ..utils import ( |
6b95b065 | 8 | ExtractorError, |
b02b960c RA |
9 | int_or_none, |
10 | update_url_query, | |
6b95b065 JMF |
11 | ) |
12 | ||
13 | ||
14 | class NaverIE(InfoExtractor): | |
c0bd51c0 | 15 | _VALID_URL = r'https?://(?:m\.)?tv(?:cast)?\.naver\.com/v/(?P<id>\d+)' |
6b95b065 | 16 | |
f8d5e1cf | 17 | _TESTS = [{ |
8a5f0a63 | 18 | 'url': 'http://tv.naver.com/v/81652', |
24da5893 JMF |
19 | 'info_dict': { |
20 | 'id': '81652', | |
21 | 'ext': 'mp4', | |
22 | 'title': '[9월 모의고사 해설강의][수학_김상희] 수학 A형 16~20번', | |
23 | 'description': '합격불변의 법칙 메가스터디 | 메가스터디 수학 김상희 선생님이 9월 모의고사 수학A형 16번에서 20번까지 해설강의를 공개합니다.', | |
f65dc41b | 24 | 'upload_date': '20130903', |
6b95b065 | 25 | }, |
f8d5e1cf | 26 | }, { |
8a5f0a63 | 27 | 'url': 'http://tv.naver.com/v/395837', |
f8d5e1cf YCH |
28 | 'md5': '638ed4c12012c458fefcddfd01f173cd', |
29 | 'info_dict': { | |
30 | 'id': '395837', | |
31 | 'ext': 'mp4', | |
32 | 'title': '9년이 지나도 아픈 기억, 전효성의 아버지', | |
33 | 'description': 'md5:5bf200dcbf4b66eb1b350d1eb9c753f7', | |
f65dc41b | 34 | 'upload_date': '20150519', |
f8d5e1cf YCH |
35 | }, |
36 | 'skip': 'Georestricted', | |
8a5f0a63 S |
37 | }, { |
38 | 'url': 'http://tvcast.naver.com/v/81652', | |
39 | 'only_matching': True, | |
f8d5e1cf | 40 | }] |
6b95b065 JMF |
41 | |
42 | def _real_extract(self, url): | |
1cc79574 | 43 | video_id = self._match_id(url) |
6b95b065 | 44 | webpage = self._download_webpage(url, video_id) |
1cc79574 | 45 | |
6b95b065 | 46 | m_id = re.search(r'var rmcPlayer = new nhn.rmcnmv.RMCVideoPlayer\("(.+?)", "(.+?)"', |
9e1a5b84 | 47 | webpage) |
6b95b065 | 48 | if m_id is None: |
f540b937 S |
49 | error = self._html_search_regex( |
50 | r'(?s)<div class="(?:nation_error|nation_box|error_box)">\s*(?:<!--.*?-->)?\s*<p class="[^"]+">(?P<msg>.+?)</p>\s*</div>', | |
51 | webpage, 'error', default=None) | |
52 | if error: | |
53 | raise ExtractorError(error, expected=True) | |
24da5893 | 54 | raise ExtractorError('couldn\'t extract vid and key') |
f65dc41b RA |
55 | video_data = self._download_json( |
56 | 'http://play.rmcnmv.naver.com/vod/play/v2.0/' + m_id.group(1), | |
57 | video_id, query={ | |
58 | 'key': m_id.group(2), | |
59 | }) | |
b02b960c RA |
60 | meta = video_data['meta'] |
61 | title = meta['subject'] | |
6b95b065 | 62 | formats = [] |
b02b960c RA |
63 | |
64 | def extract_formats(streams, stream_type, query={}): | |
65 | for stream in streams: | |
66 | stream_url = stream.get('source') | |
67 | if not stream_url: | |
68 | continue | |
69 | stream_url = update_url_query(stream_url, query) | |
70 | encoding_option = stream.get('encodingOption', {}) | |
71 | bitrate = stream.get('bitrate', {}) | |
72 | formats.append({ | |
73 | 'format_id': '%s_%s' % (stream.get('type') or stream_type, encoding_option.get('id') or encoding_option.get('name')), | |
74 | 'url': stream_url, | |
75 | 'width': int_or_none(encoding_option.get('width')), | |
76 | 'height': int_or_none(encoding_option.get('height')), | |
77 | 'vbr': int_or_none(bitrate.get('video')), | |
78 | 'abr': int_or_none(bitrate.get('audio')), | |
79 | 'filesize': int_or_none(stream.get('size')), | |
80 | 'protocol': 'm3u8_native' if stream_type == 'HLS' else None, | |
087ca2cb | 81 | }) |
b02b960c RA |
82 | |
83 | extract_formats(video_data.get('videos', {}).get('list', []), 'H264') | |
84 | for stream_set in video_data.get('streams', []): | |
85 | query = {} | |
86 | for param in stream_set.get('keys', []): | |
87 | query[param['name']] = param['value'] | |
88 | stream_type = stream_set.get('type') | |
89 | videos = stream_set.get('videos') | |
90 | if videos: | |
91 | extract_formats(videos, stream_type, query) | |
92 | elif stream_type == 'HLS': | |
93 | stream_url = stream_set.get('source') | |
94 | if not stream_url: | |
95 | continue | |
96 | formats.extend(self._extract_m3u8_formats( | |
97 | update_url_query(stream_url, query), video_id, | |
98 | 'mp4', 'm3u8_native', m3u8_id=stream_type, fatal=False)) | |
087ca2cb | 99 | self._sort_formats(formats) |
6b95b065 | 100 | |
b02b960c RA |
101 | subtitles = {} |
102 | for caption in video_data.get('captions', {}).get('list', []): | |
103 | caption_url = caption.get('source') | |
104 | if not caption_url: | |
105 | continue | |
106 | subtitles.setdefault(caption.get('language') or caption.get('locale'), []).append({ | |
107 | 'url': caption_url, | |
108 | }) | |
109 | ||
f65dc41b RA |
110 | upload_date = self._search_regex( |
111 | r'<span[^>]+class="date".*?(\d{4}\.\d{2}\.\d{2})', | |
112 | webpage, 'upload date', fatal=False) | |
113 | if upload_date: | |
114 | upload_date = upload_date.replace('.', '') | |
115 | ||
fb7abb31 | 116 | return { |
6b95b065 | 117 | 'id': video_id, |
b02b960c | 118 | 'title': title, |
6b95b065 | 119 | 'formats': formats, |
b02b960c | 120 | 'subtitles': subtitles, |
6b95b065 | 121 | 'description': self._og_search_description(webpage), |
b02b960c RA |
122 | 'thumbnail': meta.get('cover', {}).get('source') or self._og_search_thumbnail(webpage), |
123 | 'view_count': int_or_none(meta.get('count')), | |
f65dc41b | 124 | 'upload_date': upload_date, |
6b95b065 | 125 | } |