]> gitweb @ CieloNegro.org - youtube-dl.git/blob - youtube_dl/extractor/mlb.py
[eighttracks] modernize
[youtube-dl.git] / youtube_dl / extractor / mlb.py
1 from __future__ import unicode_literals
2
3 import re
4
5 from .common import InfoExtractor
6 from ..utils import (
7     parse_duration,
8     parse_iso8601,
9     find_xpath_attr,
10 )
11
12
13 class MLBIE(InfoExtractor):
14     _VALID_URL = r'https?://m\.mlb\.com/(?:.*?/)?video/(?:topic/[\da-z_-]+/)?v(?P<id>n?\d+)'
15     _TESTS = [
16         {
17             'url': 'http://m.mlb.com/sea/video/topic/51231442/v34698933/nymsea-ackley-robs-a-home-run-with-an-amazing-catch/?c_id=sea',
18             'md5': 'ff56a598c2cf411a9a38a69709e97079',
19             'info_dict': {
20                 'id': '34698933',
21                 'ext': 'mp4',
22                 'title': "Ackley's spectacular catch",
23                 'description': 'md5:7f5a981eb4f3cbc8daf2aeffa2215bf0',
24                 'duration': 66,
25                 'timestamp': 1405980600,
26                 'upload_date': '20140721',
27                 'thumbnail': 're:^https?://.*\.jpg$',
28             },
29         },
30         {
31             'url': 'http://m.mlb.com/video/topic/81536970/v34496663/mianym-stanton-practices-for-the-home-run-derby',
32             'md5': 'd9c022c10d21f849f49c05ae12a8a7e9',
33             'info_dict': {
34                 'id': '34496663',
35                 'ext': 'mp4',
36                 'title': 'Stanton prepares for Derby',
37                 'description': 'md5:d00ce1e5fd9c9069e9c13ab4faedfa57',
38                 'duration': 46,
39                 'timestamp': 1405105800,
40                 'upload_date': '20140711',
41                 'thumbnail': 're:^https?://.*\.jpg$',
42             },
43         },
44         {
45             'url': 'http://m.mlb.com/video/topic/vtp_hrd_sponsor/v34578115/hrd-cespedes-wins-2014-gillette-home-run-derby',
46             'md5': '0e6e73d509321e142409b695eadd541f',
47             'info_dict': {
48                 'id': '34578115',
49                 'ext': 'mp4',
50                 'title': 'Cespedes repeats as Derby champ',
51                 'description': 'md5:08df253ce265d4cf6fb09f581fafad07',
52                 'duration': 488,
53                 'timestamp': 1405399936,
54                 'upload_date': '20140715',
55                 'thumbnail': 're:^https?://.*\.jpg$',
56             },
57         },
58         {
59             'url': 'http://m.mlb.com/video/v34577915/bautista-on-derby-captaining-duties-his-performance',
60             'md5': 'b8fd237347b844365d74ea61d4245967',
61             'info_dict': {
62                 'id': '34577915',
63                 'ext': 'mp4',
64                 'title': 'Bautista on Home Run Derby',
65                 'description': 'md5:b80b34031143d0986dddc64a8839f0fb',
66                 'duration': 52,
67                 'timestamp': 1405390722,
68                 'upload_date': '20140715',
69                 'thumbnail': 're:^https?://.*\.jpg$',
70             },
71         },
72     ]
73
74     def _real_extract(self, url):
75         mobj = re.match(self._VALID_URL, url)
76         video_id = mobj.group('id')
77
78         detail = self._download_xml(
79             'http://m.mlb.com/gen/multimedia/detail/%s/%s/%s/%s.xml'
80             % (video_id[-3], video_id[-2], video_id[-1], video_id), video_id)
81
82         title = detail.find('./headline').text
83         description = detail.find('./big-blurb').text
84         duration = parse_duration(detail.find('./duration').text)
85         timestamp = parse_iso8601(detail.attrib['date'][:-5])
86
87         thumbnail = find_xpath_attr(
88             detail, './thumbnailScenarios/thumbnailScenario', 'type', '45').text
89
90         formats = []
91         for media_url in detail.findall('./url'):
92             playback_scenario = media_url.attrib['playback_scenario']
93             fmt = {
94                 'url': media_url.text,
95                 'format_id': playback_scenario,
96             }
97             m = re.search(r'(?P<vbr>\d+)K_(?P<width>\d+)X(?P<height>\d+)', playback_scenario)
98             if m:
99                 fmt.update({
100                     'vbr': int(m.group('vbr')) * 1000,
101                     'width': int(m.group('width')),
102                     'height': int(m.group('height')),
103                 })
104             formats.append(fmt)
105
106         self._sort_formats(formats)
107
108         return {
109             'id': video_id,
110             'title': title,
111             'description': description,
112             'duration': duration,
113             'timestamp': timestamp,
114             'formats': formats,
115             'thumbnail': thumbnail,
116         }