[youtube] Fix extraction.
[youtube-dl] / youtube_dl / extractor / gaskrank.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 import re
5 from .common import InfoExtractor
6 from ..utils import (
7     float_or_none,
8     int_or_none,
9     unified_strdate,
10 )
11
12
13 class GaskrankIE(InfoExtractor):
14     _VALID_URL = r'https?://(?:www\.)?gaskrank\.tv/tv/(?P<categories>[^/]+)/(?P<id>[^/]+)\.htm'
15     _TESTS = [{
16         'url': 'http://www.gaskrank.tv/tv/motorrad-fun/strike-einparken-durch-anfaenger-crash-mit-groesserem-flurschaden.htm',
17         'md5': '1ae88dbac97887d85ebd1157a95fc4f9',
18         'info_dict': {
19             'id': '201601/26955',
20             'ext': 'mp4',
21             'title': 'Strike! Einparken können nur Männer - Flurschaden hält sich in Grenzen *lol*',
22             'thumbnail': r're:^https?://.*\.jpg$',
23             'categories': ['motorrad-fun'],
24             'display_id': 'strike-einparken-durch-anfaenger-crash-mit-groesserem-flurschaden',
25             'uploader_id': 'Bikefun',
26             'upload_date': '20170110',
27             'uploader_url': None,
28         }
29     }, {
30         'url': 'http://www.gaskrank.tv/tv/racing/isle-of-man-tt-2011-michael-du-15920.htm',
31         'md5': 'c33ee32c711bc6c8224bfcbe62b23095',
32         'info_dict': {
33             'id': '201106/15920',
34             'ext': 'mp4',
35             'title': 'Isle of Man - Michael Dunlop vs Guy Martin - schwindelig kucken',
36             'thumbnail': r're:^https?://.*\.jpg$',
37             'categories': ['racing'],
38             'display_id': 'isle-of-man-tt-2011-michael-du-15920',
39             'uploader_id': 'IOM',
40             'upload_date': '20170523',
41             'uploader_url': 'www.iomtt.com',
42         }
43     }]
44
45     def _real_extract(self, url):
46         display_id = self._match_id(url)
47
48         webpage = self._download_webpage(url, display_id)
49
50         title = self._og_search_title(
51             webpage, default=None) or self._html_search_meta(
52             'title', webpage, fatal=True)
53
54         categories = [re.match(self._VALID_URL, url).group('categories')]
55
56         mobj = re.search(
57             r'Video von:\s*(?P<uploader_id>[^|]*?)\s*\|\s*vom:\s*(?P<upload_date>[0-9][0-9]\.[0-9][0-9]\.[0-9][0-9][0-9][0-9])',
58             webpage)
59         if mobj is not None:
60             uploader_id = mobj.groupdict().get('uploader_id')
61             upload_date = unified_strdate(mobj.groupdict().get('upload_date'))
62
63         uploader_url = self._search_regex(
64             r'Homepage:\s*<[^>]*>(?P<uploader_url>[^<]*)',
65             webpage, 'uploader_url', default=None)
66         tags = re.findall(
67             r'/tv/tags/[^/]+/"\s*>(?P<tag>[^<]*?)<',
68             webpage)
69
70         view_count = self._search_regex(
71             r'class\s*=\s*"gkRight"(?:[^>]*>\s*<[^>]*)*icon-eye-open(?:[^>]*>\s*<[^>]*)*>\s*(?P<view_count>[0-9\.]*)',
72             webpage, 'view_count', default=None)
73         if view_count:
74             view_count = int_or_none(view_count.replace('.', ''))
75
76         average_rating = self._search_regex(
77             r'itemprop\s*=\s*"ratingValue"[^>]*>\s*(?P<average_rating>[0-9,]+)',
78             webpage, 'average_rating')
79         if average_rating:
80             average_rating = float_or_none(average_rating.replace(',', '.'))
81
82         video_id = self._search_regex(
83             r'https?://movies\.gaskrank\.tv/([^-]*?)(-[^\.]*)?\.mp4',
84             webpage, 'video id', default=display_id)
85
86         entry = self._parse_html5_media_entries(url, webpage, video_id)[0]
87         entry.update({
88             'id': video_id,
89             'title': title,
90             'categories': categories,
91             'display_id': display_id,
92             'uploader_id': uploader_id,
93             'upload_date': upload_date,
94             'uploader_url': uploader_url,
95             'tags': tags,
96             'view_count': view_count,
97             'average_rating': average_rating,
98         })
99         self._sort_formats(entry['formats'])
100
101         return entry