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         'url': 'http://www.gamersyde.com/hqstream_bloodborne_birth_of_a_hero-34371_en.html',
  12         'md5': 'f38d400d32f19724570040d5ce3a505f',
  13         'info_dict': {
  14             'id': '34371',
  15             'ext': 'mp4',
  16             'title': 'Bloodborne - Birth of a hero',
  17             'thumbnail': 're:^https?://.*\.jpg$',
  18         }
  19     },
  20     {
  21         'url': 'http://www.gamersyde.com/hqstream_dark_souls_ii_scholar_of_the_first_sin_gameplay_part_1-34417_en.html',
  22         'info_dict': {
  23             'ext': 'mp4',
  24     }
  25
  26     def _calculateDuration(self, durationString):
  27         duration = time.strptime(durationString, "%M minutes %S seconds")
  28         return duration.tm_min * 60 + duration.tm_sec
  29
  30     def _fixJsonSyntax(self, json):
  31
  32         json = re.sub(r"{\s*(\w)", r'{"\1', json)
  33         json = re.sub(r",\s*(\w)", r',"\1', json)
  34         json = re.sub(r",\s*}", "}", json, flags=re.DOTALL)
  35         json = re.sub(r",\s*]", "]", json, flags=re.DOTALL)
  36
  37         return json
  38
  39     def _real_extract(self, url):
  40
  41         video_id = self._search_regex(r'-(.*?)_[a-z]{2}.html$', url, 'video_id')
  42         webpage = self._download_webpage(url, video_id)
  43
  44         filesJson = self._search_regex(r'playlist: (.*?)\}\);', webpage, 'files', flags=re.DOTALL)
  45         filesJson = self._fixJsonSyntax(filesJson)
  46         data = json.loads(filesJson)
  47         playlist = data[0]
  48
  49         formats = []
  50
  51         title = re.sub(r"[0-9]+ - ", "", playlist['title'])
  52
  53         for playlistEntry in playlist['sources']:
  54             format = {
  55                 'url': playlistEntry['file'],
  56                 'format_id': playlistEntry['label']
  57             }
  58
  59             formats.append(format)
  60
  61         return {
  62             'id': video_id,
  63             'title': title,
  64             'formats': formats,
  65             'thumbnail': playlist['image']
  66             }