git.bitcoin.ninja Git - youtube-dl/blob - youtube_dl/extractor/gamersyde.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3 import re
   4 import json
   5 import time
   6 from .common import InfoExtractor
   7
   8
   9 class GamersydeIE(InfoExtractor):
  10     _VALID_URL = r'https?://(?:www\.)?gamersyde\.com/hqstream_'
  11     _TEST = {
  12         'url': 'http://www.gamersyde.com/hqstream_bloodborne_birth_of_a_hero-34371_en.html',
  13         'md5': 'f38d400d32f19724570040d5ce3a505f',
  14         'info_dict': {
  15             'id': '34371',
  16             'ext': 'mp4',
  17             'title': 'Bloodborne - Birth of a hero',
  18             'thumbnail': 're:^https?://.*\.jpg$',
  19         }
  20     }
  21
  22     def _calculateDuration(self, durationString):
  23         duration = time.strptime(durationString, "%M minutes %S seconds")
  24         return duration.tm_min * 60 + duration.tm_sec
  25
  26     def _fixJsonSyntax(self, json):
  27
  28         json = re.sub(r"{\s*(\w)", r'{"\1', json)
  29         json = re.sub(r",\s*(\w)", r',"\1', json)
  30         json = re.sub(r"(\w): ", r'\1":', json)
  31         json = re.sub(r",\s*}", "}", json, flags=re.DOTALL)
  32         json = re.sub(r",\s*]", "]", json, flags=re.DOTALL)
  33
  34         return json
  35
  36     def _real_extract(self, url):
  37
  38         video_id = self._search_regex(r'-(.*?)_[a-z]{2}.html$', url, 'video_id')
  39         webpage = self._download_webpage(url, video_id)
  40
  41         filesJson = self._search_regex(r'playlist: (.*?)\}\);', webpage, 'files', flags=re.DOTALL)
  42         filesJson = self._fixJsonSyntax(filesJson)
  43
  44         data = json.loads(filesJson)
  45         playlist = data[0]
  46
  47         formats = []
  48
  49         title = re.sub(r"[0-9]+ - ", "", playlist['title'])
  50
  51         for playlistEntry in playlist['sources']:
  52             format = {
  53                 'url': playlistEntry['file'],
  54                 'format_id': playlistEntry['label']
  55             }
  56
  57             formats.append(format)
  58
  59         return {
  60             'id': video_id,
  61             'title': title,
  62             'formats': formats,
  63             'thumbnail': playlist['image']
  64             }