_ Git - youtube-dl/blob - 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     """InfoExtractor for gaskrank.tv"""
  15     _VALID_URL = r'https?://(?:www\.)?gaskrank\.tv/tv/(?P<categories>[^/]+)/(?P<id>[^/]+)\.html?'
  16     _TESTS = [
  17         {
  18             'url': 'http://www.gaskrank.tv/tv/motorrad-fun/strike-einparken-durch-anfaenger-crash-mit-groesserem-flurschaden.htm',
  19             'md5': '1ae88dbac97887d85ebd1157a95fc4f9',
  20             'info_dict': {
  21                 'id': '201601/26955',
  22                 'ext': 'mp4',
  23                 'title': 'Strike! Einparken können nur Männer - Flurschaden hält sich in Grenzen *lol*',
  24                 'thumbnail': r're:^https?://.*\.jpg$',
  25                 'categories': ['motorrad-fun'],
  26                 'display_id': 'strike-einparken-durch-anfaenger-crash-mit-groesserem-flurschaden',
  27                 'uploader_id': 'Bikefun',
  28                 'upload_date': '20170110',
  29                 'uploader_url': None,
  30             }
  31         },
  32         {
  33             'url': 'http://www.gaskrank.tv/tv/racing/isle-of-man-tt-2011-michael-du-15920.htm',
  34             'md5': 'c33ee32c711bc6c8224bfcbe62b23095',
  35             'info_dict': {
  36                 'id': '201106/15920',
  37                 'ext': 'mp4',
  38                 'title': 'Isle of Man - Michael Dunlop vs Guy Martin - schwindelig kucken',
  39                 'thumbnail': r're:^https?://.*\.jpg$',
  40                 'categories': ['racing'],
  41                 'display_id': 'isle-of-man-tt-2011-michael-du-15920',
  42                 'uploader_id': 'IOM',
  43                 'upload_date': '20160506',
  44                 'uploader_url': 'www.iomtt.com',
  45             }
  46         }
  47     ]
  48
  49     def _real_extract(self, url):
  50         """extract information from gaskrank.tv"""
  51
  52         display_id = self._match_id(url)
  53         webpage = self._download_webpage(url, display_id)
  54         categories = [re.match(self._VALID_URL, url).group('categories')]
  55         title = self._og_search_title(webpage, default=None) or self._html_search_meta(
  56             'title', webpage, fatal=True)
  57
  58         mobj = re.search(
  59             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])',
  60             webpage)
  61         if mobj is not None:
  62             uploader_id = mobj.groupdict().get('uploader_id')
  63             upload_date = unified_strdate(mobj.groupdict().get('upload_date'))
  64
  65         uploader_url = self._search_regex(
  66             r'Homepage:\s*<[^>]*>(?P<uploader_url>[^<]*)',
  67             webpage, 'uploader_url', default=None)
  68         tags = re.findall(
  69             r'/tv/tags/[^/]+/"\s*>(?P<tag>[^<]*?)<',
  70             webpage)
  71
  72         view_count = self._search_regex(
  73             r'class\s*=\s*"gkRight"(?:[^>]*>\s*<[^>]*)*icon-eye-open(?:[^>]*>\s*<[^>]*)*>\s*(?P<view_count>[0-9\.]*)',
  74             webpage, 'view_count', default=None)
  75         if view_count:
  76             view_count = int_or_none(view_count.replace('.', ''))
  77
  78         average_rating = self._search_regex(
  79             r'itemprop\s*=\s*"ratingValue"[^>]*>\s*(?P<average_rating>[0-9,]+)',
  80             webpage, 'average_rating')
  81         if average_rating:
  82             average_rating = float_or_none(average_rating.replace(',', '.'))
  83
  84         video_id = self._search_regex(
  85             r'https?://movies\.gaskrank\.tv/([^-]*?)(-[^\.]*)?\.mp4',
  86             webpage, 'video id')
  87
  88         entry = self._parse_html5_media_entries(url, webpage, video_id)[0]
  89         entry.update({
  90             'id': video_id,
  91             'title': title,
  92             'categories': categories,
  93             'display_id': display_id,
  94             'uploader_id': uploader_id,
  95             'upload_date': upload_date,
  96             'uploader_url': uploader_url,
  97             'tags': tags,
  98             'view_count': view_count,
  99             'average_rating': average_rating,
 100         })
 101         self._sort_formats(entry['formats'])
 102
 103         return entry