]>
Commit | Line | Data |
---|---|---|
26a78d4b JMF |
1 | from __future__ import unicode_literals |
2 | ||
ecf6de5b | 3 | import re |
4 | ||
5b286728 | 5 | from .common import InfoExtractor |
7bbc6428 | 6 | from ..utils import ( |
7bbc6428 | 7 | parse_duration, |
8fc226ef | 8 | int_or_none, |
46cc1c65 | 9 | xpath_text, |
10 | xpath_attr, | |
7bbc6428 | 11 | ) |
5b286728 PH |
12 | |
13 | ||
c233e6bc | 14 | class NBAIE(InfoExtractor): |
6a11bb77 | 15 | _VALID_URL = r'https?://(?:watch\.|www\.)?nba\.com/(?P<path>(?:[^/]+/)?video/(?P<id>[^?]*?))/?(?:/index\.html)?(?:\?.*)?$' |
6a3e0103 | 16 | _TESTS = [{ |
26a78d4b | 17 | 'url': 'http://www.nba.com/video/games/nets/2012/12/04/0021200253-okc-bkn-recap.nba/index.html', |
6a11bb77 | 18 | 'md5': '9e7729d3010a9c71506fd1248f74e4f4', |
26a78d4b | 19 | 'info_dict': { |
c233e6bc | 20 | 'id': '0021200253-okc-bkn-recap', |
6a11bb77 | 21 | 'ext': 'flv', |
26a78d4b | 22 | 'title': 'Thunder vs. Nets', |
7bbc6428 S |
23 | 'description': 'Kevin Durant scores 32 points and dishes out six assists as the Thunder beat the Nets in Brooklyn.', |
24 | 'duration': 181, | |
c233e6bc | 25 | 'timestamp': 1354638466, |
26 | 'upload_date': '20121204', | |
26a78d4b | 27 | }, |
6a3e0103 PH |
28 | }, { |
29 | 'url': 'http://www.nba.com/video/games/hornets/2014/12/05/0021400276-nyk-cha-play5.nba/', | |
30 | 'only_matching': True, | |
46cc1c65 | 31 | }, { |
6a11bb77 | 32 | 'url': 'http://watch.nba.com/video/channels/playoffs/2015/05/20/0041400301-cle-atl-recap.nba', |
8fc226ef | 33 | 'md5': 'b2b39b81cf28615ae0c3360a3f9668c4', |
8a278a1d | 34 | 'info_dict': { |
c233e6bc | 35 | 'id': '0041400301-cle-atl-recap', |
8a278a1d | 36 | 'ext': 'mp4', |
8fc226ef | 37 | 'title': 'Hawks vs. Cavaliers Game 1', |
8a278a1d YCH |
38 | 'description': 'md5:8094c3498d35a9bd6b1a8c396a071b4d', |
39 | 'duration': 228, | |
c233e6bc | 40 | 'timestamp': 1432134543, |
8fc226ef | 41 | 'upload_date': '20150520', |
8a278a1d | 42 | } |
6a3e0103 | 43 | }] |
5b286728 | 44 | |
c233e6bc | 45 | def _real_extract(self, url): |
6a11bb77 | 46 | path, video_id = re.match(self._VALID_URL, url).groups() |
b80d4beb | 47 | if path.startswith('nba/'): |
48 | path = path[3:] | |
6a11bb77 | 49 | video_info = self._download_xml('http://www.nba.com/%s.xml' % path, video_id) |
46cc1c65 | 50 | video_id = xpath_text(video_info, 'slug') |
51 | title = xpath_text(video_info, 'headline') | |
52 | description = xpath_text(video_info, 'description') | |
53 | duration = parse_duration(xpath_text(video_info, 'length')) | |
54 | timestamp = int_or_none(xpath_attr(video_info, 'dateCreated', 'uts')) | |
c233e6bc | 55 | |
56 | thumbnails = [] | |
57 | for image in video_info.find('images'): | |
58 | thumbnails.append({ | |
59 | 'id': image.attrib.get('cut'), | |
60 | 'url': image.text, | |
61 | 'width': int_or_none(image.attrib.get('width')), | |
62 | 'height': int_or_none(image.attrib.get('height')), | |
63 | }) | |
64 | ||
65 | formats = [] | |
b80d4beb | 66 | for video_file in video_info.findall('.//file'): |
c233e6bc | 67 | video_url = video_file.text |
139f2782 | 68 | if video_url.startswith('/'): |
69 | continue | |
c233e6bc | 70 | if video_url.endswith('.m3u8'): |
b80d4beb | 71 | m3u8_formats = self._extract_m3u8_formats(video_url, video_id, m3u8_id='hls', fatal=False) |
72 | if m3u8_formats: | |
73 | formats.extend(m3u8_formats) | |
c233e6bc | 74 | elif video_url.endswith('.f4m'): |
b80d4beb | 75 | f4m_formats = self._extract_f4m_formats(video_url + '?hdcore=3.4.1.1', video_id, f4m_id='hds', fatal=False) |
76 | if f4m_formats: | |
77 | formats.extend(f4m_formats) | |
c233e6bc | 78 | else: |
79 | key = video_file.attrib.get('bitrate') | |
cb160dd5 | 80 | format_info = { |
c233e6bc | 81 | 'format_id': key, |
82 | 'url': video_url, | |
cb160dd5 | 83 | } |
84 | mobj = re.search(r'(\d+)x(\d+)(?:_(\d+))?', key) | |
85 | if mobj: | |
86 | format_info.update({ | |
0017486d | 87 | 'width': int(mobj.group(1)), |
88 | 'height': int(mobj.group(2)), | |
51c4fec0 | 89 | 'tbr': int_or_none(mobj.group(3)), |
cb160dd5 | 90 | }) |
91 | formats.append(format_info) | |
c233e6bc | 92 | self._sort_formats(formats) |
93 | ||
26a78d4b | 94 | return { |
c233e6bc | 95 | 'id': video_id, |
96 | 'title': title, | |
97 | 'description': description, | |
98 | 'duration': duration, | |
99 | 'timestamp': timestamp, | |
100 | 'thumbnails': thumbnails, | |
101 | 'formats': formats, | |
5b286728 | 102 | } |