class PornHdIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?pornhd\.com/(?:[a-z]{2,4}/)?videos/(?P<id>\d+)(?:/(?P<display_id>.+))?'
- _TEST = {
+ _TESTS = [{
+ 'url': 'http://www.pornhd.com/videos/9864/selfie-restroom-masturbation-fun-with-chubby-cutie-hd-porn-video',
+ 'md5': 'c8b964b1f0a4b5f7f28ae3a5c9f86ad5',
+ 'info_dict': {
+ 'id': '9864',
+ 'display_id': 'selfie-restroom-masturbation-fun-with-chubby-cutie-hd-porn-video',
+ 'ext': 'mp4',
+ 'title': 'Restroom selfie masturbation',
+ 'description': 'md5:3748420395e03e31ac96857a8f125b2b',
+ 'thumbnail': r're:^https?://.*\.jpg',
+ 'view_count': int,
+ 'age_limit': 18,
+ }
+ }, {
+ # removed video
'url': 'http://www.pornhd.com/videos/1962/sierra-day-gets-his-cum-all-over-herself-hd-porn-video',
'md5': '956b8ca569f7f4d8ec563e2c41598441',
'info_dict': {
'ext': 'mp4',
'title': 'Sierra loves doing laundry',
'description': 'md5:8ff0523848ac2b8f9b065ba781ccf294',
- 'thumbnail': 're:^https?://.*\.jpg',
+ 'thumbnail': r're:^https?://.*\.jpg',
'view_count': int,
'age_limit': 18,
- }
- }
+ },
+ 'skip': 'Not available anymore',
+ }]
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
r'<title>(.+?) - .*?[Pp]ornHD.*?</title>'], webpage, 'title')
sources = self._parse_json(js_to_json(self._search_regex(
- r"(?s)'sources'\s*:\s*(\{.+?\})\s*\}[;,)]",
+ r"(?s)sources'?\s*[:=]\s*(\{.+?\})",
webpage, 'sources', default='{}')), video_id)
if not sources:
view_count = int_or_none(self._html_search_regex(
r'(\d+) views\s*<', webpage, 'view count', fatal=False))
thumbnail = self._search_regex(
- r"'poster'\s*:\s*'([^']+)'", webpage, 'thumbnail', fatal=False)
+ r"poster'?\s*:\s*([\"'])(?P<url>(?:(?!\1).)+)\1", webpage,
+ 'thumbnail', fatal=False, group='url')
return {
'id': video_id,