1
0
mirror of https://github.com/ytdl-org/youtube-dl synced 2024-12-17 23:16:39 +00:00
youtube-dl/youtube_dl/extractor/gamersyde.py
snipem 115c281672 [Gamersyde] Improved robustness, added duration and tests
Fix for Json syntax is now less error prone for Json syntax inside of
values. Extractor is now also using native Json handling. Added tests
for several videos that were producing errors in the first place.
2015-04-04 12:31:48 +02:00

67 lines
2.0 KiB
Python

# coding: utf-8
from __future__ import unicode_literals
import re
import json
import time
from .common import InfoExtractor
class GamersydeIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?gamersyde\.com/hqstream_'
'url': 'http://www.gamersyde.com/hqstream_bloodborne_birth_of_a_hero-34371_en.html',
'md5': 'f38d400d32f19724570040d5ce3a505f',
'info_dict': {
'id': '34371',
'ext': 'mp4',
'title': 'Bloodborne - Birth of a hero',
'thumbnail': 're:^https?://.*\.jpg$',
}
},
{
'url': 'http://www.gamersyde.com/hqstream_dark_souls_ii_scholar_of_the_first_sin_gameplay_part_1-34417_en.html',
'info_dict': {
'ext': 'mp4',
}
def _calculateDuration(self, durationString):
duration = time.strptime(durationString, "%M minutes %S seconds")
return duration.tm_min * 60 + duration.tm_sec
def _fixJsonSyntax(self, json):
json = re.sub(r"{\s*(\w)", r'{"\1', json)
json = re.sub(r",\s*(\w)", r',"\1', json)
json = re.sub(r",\s*}", "}", json, flags=re.DOTALL)
json = re.sub(r",\s*]", "]", json, flags=re.DOTALL)
return json
def _real_extract(self, url):
video_id = self._search_regex(r'-(.*?)_[a-z]{2}.html$', url, 'video_id')
webpage = self._download_webpage(url, video_id)
filesJson = self._search_regex(r'playlist: (.*?)\}\);', webpage, 'files', flags=re.DOTALL)
filesJson = self._fixJsonSyntax(filesJson)
data = json.loads(filesJson)
playlist = data[0]
formats = []
title = re.sub(r"[0-9]+ - ", "", playlist['title'])
for playlistEntry in playlist['sources']:
format = {
'url': playlistEntry['file'],
'format_id': playlistEntry['label']
}
formats.append(format)
return {
'id': video_id,
'title': title,
'formats': formats,
'thumbnail': playlist['image']
}