]>
Commit | Line | Data |
---|---|---|
256a746d | 1 | # coding: utf-8\r |
2 | from __future__ import unicode_literals\r | |
3 | \r | |
256a746d | 4 | import re\r |
5 | \r | |
73cf76a9 S |
6 | from .common import InfoExtractor\r |
7 | from ..compat import compat_str\r | |
8 | from ..utils import (\r | |
9 | int_or_none,\r | |
10 | js_to_json,\r | |
11 | try_get,\r | |
12 | )\r | |
13 | \r | |
256a746d | 14 | \r |
15 | class JojIE(InfoExtractor):\r | |
73cf76a9 S |
16 | _VALID_URL = r'''(?x)\r |
17 | (?:\r | |
18 | joj:|\r | |
19 | https?://media\.joj\.sk/embed/\r | |
20 | )\r | |
21 | (?P<id>[\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12})\r | |
22 | '''\r | |
256a746d | 23 | _TESTS = [{\r |
73cf76a9 | 24 | 'url': 'https://media.joj.sk/embed/a388ec4c-6019-4a4a-9312-b1bee194e932',\r |
256a746d | 25 | 'info_dict': {\r |
26 | 'id': 'a388ec4c-6019-4a4a-9312-b1bee194e932',\r | |
27 | 'ext': 'mp4',\r | |
73cf76a9 S |
28 | 'title': 'NOVÉ BÝVANIE',\r |
29 | 'thumbnail': r're:^https?://.*\.jpg$',\r | |
30 | 'duration': 3118,\r | |
256a746d | 31 | }\r |
32 | }, {\r | |
73cf76a9 S |
33 | 'url': 'joj:a388ec4c-6019-4a4a-9312-b1bee194e932',\r |
34 | 'only_matching': True,\r | |
256a746d | 35 | }]\r |
36 | \r | |
73cf76a9 S |
37 | @staticmethod\r |
38 | def _extract_urls(webpage):\r | |
39 | return re.findall(\r | |
40 | r'<iframe\b[^>]+\bsrc=["\'](?P<url>(?:https?:)?//media\.joj\.sk/embed/[\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12})',\r | |
41 | webpage)\r | |
256a746d | 42 | \r |
43 | def _real_extract(self, url):\r | |
73cf76a9 S |
44 | video_id = self._match_id(url)\r |
45 | \r | |
46 | webpage = self._download_webpage(\r | |
47 | 'https://media.joj.sk/embed/%s' % video_id, video_id)\r | |
48 | \r | |
49 | title = self._search_regex(\r | |
50 | (r'videoTitle\s*:\s*(["\'])(?P<title>(?:(?!\1).)+)\1',\r | |
51 | r'<title>(?P<title>[^<]+)'), webpage, 'title',\r | |
52 | default=None, group='title') or self._og_search_title(webpage)\r | |
53 | \r | |
54 | bitrates = self._parse_json(\r | |
55 | self._search_regex(\r | |
56 | r'(?s)bitrates\s*=\s*({.+?});', webpage, 'bitrates',\r | |
57 | default='{}'),\r | |
58 | video_id, transform_source=js_to_json, fatal=False)\r | |
59 | \r | |
256a746d | 60 | formats = []\r |
73cf76a9 S |
61 | for format_url in try_get(bitrates, lambda x: x['mp4'], list) or []:\r |
62 | if isinstance(format_url, compat_str):\r | |
63 | height = self._search_regex(\r | |
64 | r'(\d+)[pP]\.', format_url, 'height', default=None)\r | |
65 | formats.append({\r | |
66 | 'url': format_url,\r | |
67 | 'format_id': '%sp' % height if height else None,\r | |
68 | 'height': int(height),\r | |
69 | })\r | |
70 | if not formats:\r | |
71 | playlist = self._download_xml(\r | |
72 | 'https://media.joj.sk/services/Video.php?clip=%s' % video_id,\r | |
73 | video_id)\r | |
74 | for file_el in playlist.findall('./files/file'):\r | |
75 | path = file_el.get('path')\r | |
76 | if not path:\r | |
77 | continue\r | |
78 | format_id = file_el.get('id') or file_el.get('label')\r | |
79 | formats.append({\r | |
80 | 'url': 'http://n16.joj.sk/storage/%s' % path.replace(\r | |
81 | 'dat/', '', 1),\r | |
82 | 'format_id': format_id,\r | |
83 | 'height': int_or_none(self._search_regex(\r | |
84 | r'(\d+)[pP]', format_id or path, 'height',\r | |
85 | default=None)),\r | |
86 | })\r | |
256a746d | 87 | self._sort_formats(formats)\r |
88 | \r | |
73cf76a9 S |
89 | thumbnail = self._og_search_thumbnail(webpage)\r |
90 | \r | |
91 | duration = int_or_none(self._search_regex(\r | |
92 | r'videoDuration\s*:\s*(\d+)', webpage, 'duration', fatal=False))\r | |
93 | \r | |
256a746d | 94 | return {\r |
95 | 'id': video_id,\r | |
73cf76a9 S |
96 | 'title': title,\r |
97 | 'thumbnail': thumbnail,\r | |
98 | 'duration': duration,\r | |
256a746d | 99 | 'formats': formats,\r |
256a746d | 100 | }\r |