]> jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/gaskrank.py
[misc] Add `hatch`, `ruff`, `pre-commit` and improve dev docs (#7409)
[yt-dlp.git] / yt_dlp / extractor / gaskrank.py
1 import re
2
3 from .common import InfoExtractor
4 from ..utils import (
5 float_or_none,
6 int_or_none,
7 unified_strdate,
8 )
9
10
11 class GaskrankIE(InfoExtractor):
12 _VALID_URL = r'https?://(?:www\.)?gaskrank\.tv/tv/(?P<categories>[^/]+)/(?P<id>[^/]+)\.htm'
13 _TESTS = [{
14 'url': 'http://www.gaskrank.tv/tv/motorrad-fun/strike-einparken-durch-anfaenger-crash-mit-groesserem-flurschaden.htm',
15 'md5': '1ae88dbac97887d85ebd1157a95fc4f9',
16 'info_dict': {
17 'id': '201601/26955',
18 'ext': 'mp4',
19 'title': 'Strike! Einparken können nur Männer - Flurschaden hält sich in Grenzen *lol*',
20 'thumbnail': r're:^https?://.*\.jpg$',
21 'categories': ['motorrad-fun'],
22 'display_id': 'strike-einparken-durch-anfaenger-crash-mit-groesserem-flurschaden',
23 'uploader_id': 'Bikefun',
24 'upload_date': '20170110',
25 }
26 }, {
27 'url': 'http://www.gaskrank.tv/tv/racing/isle-of-man-tt-2011-michael-du-15920.htm',
28 'md5': 'c33ee32c711bc6c8224bfcbe62b23095',
29 'info_dict': {
30 'id': '201106/15920',
31 'ext': 'mp4',
32 'title': 'Isle of Man - Michael Dunlop vs Guy Martin - schwindelig kucken',
33 'thumbnail': r're:^https?://.*\.jpg$',
34 'categories': ['racing'],
35 'display_id': 'isle-of-man-tt-2011-michael-du-15920',
36 'uploader_id': 'IOM',
37 'upload_date': '20170523',
38 'uploader_url': 'www.iomtt.com',
39 }
40 }]
41
42 def _real_extract(self, url):
43 display_id = self._match_id(url)
44
45 webpage = self._download_webpage(url, display_id)
46
47 title = self._og_search_title(
48 webpage, default=None) or self._html_search_meta(
49 'title', webpage, fatal=True)
50
51 categories = [self._match_valid_url(url).group('categories')]
52
53 mobj = re.search(
54 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])',
55 webpage)
56 if mobj is not None:
57 uploader_id = mobj.groupdict().get('uploader_id')
58 upload_date = unified_strdate(mobj.groupdict().get('upload_date'))
59
60 uploader_url = self._search_regex(
61 r'Homepage:\s*<[^>]*>(?P<uploader_url>[^<]*)',
62 webpage, 'uploader_url', default=None)
63 tags = re.findall(
64 r'/tv/tags/[^/]+/"\s*>(?P<tag>[^<]*?)<',
65 webpage)
66
67 view_count = self._search_regex(
68 r'class\s*=\s*"gkRight"(?:[^>]*>\s*<[^>]*)*icon-eye-open(?:[^>]*>\s*<[^>]*)*>\s*(?P<view_count>[0-9\.]*)',
69 webpage, 'view_count', default=None)
70 if view_count:
71 view_count = int_or_none(view_count.replace('.', ''))
72
73 average_rating = self._search_regex(
74 r'itemprop\s*=\s*"ratingValue"[^>]*>\s*(?P<average_rating>[0-9,]+)',
75 webpage, 'average_rating')
76 if average_rating:
77 average_rating = float_or_none(average_rating.replace(',', '.'))
78
79 video_id = self._search_regex(
80 r'https?://movies\.gaskrank\.tv/([^-]*?)(-[^\.]*)?\.mp4',
81 webpage, 'video id', default=display_id)
82
83 entry = self._parse_html5_media_entries(url, webpage, video_id)[0]
84 entry.update({
85 'id': video_id,
86 'title': title,
87 'categories': categories,
88 'display_id': display_id,
89 'uploader_id': uploader_id,
90 'upload_date': upload_date,
91 'uploader_url': uploader_url,
92 'tags': tags,
93 'view_count': view_count,
94 'average_rating': average_rating,
95 })
96
97 return entry