aboutsummaryrefslogtreecommitdiff
path: root/youtube_dl/extractor/mlb.py
blob: 42aa2e227dcb1360af535c92a17edcbdf99f259c (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
from __future__ import unicode_literals

import re

from .common import InfoExtractor
from ..utils import (
    parse_duration,
    parse_iso8601,
)


class MLBIE(InfoExtractor):
    _VALID_URL = r'https?://m\.mlb\.com/(?:(?:.*?/)?video/(?:topic/[\da-z_-]+/)?v|shared/video/embed/embed\.html\?.*?\bcontent_id=)(?P<id>n?\d+)'
    _TESTS = [
        {
            'url': 'http://m.mlb.com/sea/video/topic/51231442/v34698933/nymsea-ackley-robs-a-home-run-with-an-amazing-catch/?c_id=sea',
            'md5': 'ff56a598c2cf411a9a38a69709e97079',
            'info_dict': {
                'id': '34698933',
                'ext': 'mp4',
                'title': "Ackley's spectacular catch",
                'description': 'md5:7f5a981eb4f3cbc8daf2aeffa2215bf0',
                'duration': 66,
                'timestamp': 1405980600,
                'upload_date': '20140721',
                'thumbnail': 're:^https?://.*\.jpg$',
            },
        },
        {
            'url': 'http://m.mlb.com/video/topic/81536970/v34496663/mianym-stanton-practices-for-the-home-run-derby',
            'md5': 'd9c022c10d21f849f49c05ae12a8a7e9',
            'info_dict': {
                'id': '34496663',
                'ext': 'mp4',
                'title': 'Stanton prepares for Derby',
                'description': 'md5:d00ce1e5fd9c9069e9c13ab4faedfa57',
                'duration': 46,
                'timestamp': 1405105800,
                'upload_date': '20140711',
                'thumbnail': 're:^https?://.*\.jpg$',
            },
        },
        {
            'url': 'http://m.mlb.com/video/topic/vtp_hrd_sponsor/v34578115/hrd-cespedes-wins-2014-gillette-home-run-derby',
            'md5': '0e6e73d509321e142409b695eadd541f',
            'info_dict': {
                'id': '34578115',
                'ext': 'mp4',
                'title': 'Cespedes repeats as Derby champ',
                'description': 'md5:08df253ce265d4cf6fb09f581fafad07',
                'duration': 488,
                'timestamp': 1405399936,
                'upload_date': '20140715',
                'thumbnail': 're:^https?://.*\.jpg$',
            },
        },
        {
            'url': 'http://m.mlb.com/video/v34577915/bautista-on-derby-captaining-duties-his-performance',
            'md5': 'b8fd237347b844365d74ea61d4245967',
            'info_dict': {
                'id': '34577915',
                'ext': 'mp4',
                'title': 'Bautista on Home Run Derby',
                'description': 'md5:b80b34031143d0986dddc64a8839f0fb',
                'duration': 52,
                'timestamp': 1405390722,
                'upload_date': '20140715',
                'thumbnail': 're:^https?://.*\.jpg$',
            },
        },
        {
            'url': 'http://m.mlb.com/shared/video/embed/embed.html?content_id=35692085&topic_id=6479266&width=400&height=224&property=mlb',
            'only_matching': True,
        },
    ]

    def _real_extract(self, url):
        mobj = re.match(self._VALID_URL, url)
        video_id = mobj.group('id')

        detail = self._download_xml(
            'http://m.mlb.com/gen/multimedia/detail/%s/%s/%s/%s.xml'
            % (video_id[-3], video_id[-2], video_id[-1], video_id), video_id)

        title = detail.find('./headline').text
        description = detail.find('./big-blurb').text
        duration = parse_duration(detail.find('./duration').text)
        timestamp = parse_iso8601(detail.attrib['date'][:-5])

        thumbnails = [{
            'url': thumbnail.text,
        } for thumbnail in detail.findall('./thumbnailScenarios/thumbnailScenario')]

        formats = []
        for media_url in detail.findall('./url'):
            playback_scenario = media_url.attrib['playback_scenario']
            fmt = {
                'url': media_url.text,
                'format_id': playback_scenario,
            }
            m = re.search(r'(?P<vbr>\d+)K_(?P<width>\d+)X(?P<height>\d+)', playback_scenario)
            if m:
                fmt.update({
                    'vbr': int(m.group('vbr')) * 1000,
                    'width': int(m.group('width')),
                    'height': int(m.group('height')),
                })
            formats.append(fmt)

        self._sort_formats(formats)

        return {
            'id': video_id,
            'title': title,
            'description': description,
            'duration': duration,
            'timestamp': timestamp,
            'formats': formats,
            'thumbnails': thumbnails,
        }