Compare commits

...

12 Commits

Author SHA1 Message Date
DarkFighterLuke
4de3e1cf77
Merge b804854e3bd4c663dfa6c264968a663d739a64d7 into 2b4fbfce25902d557b86b003cf48f738129efce4 2025-03-26 07:33:56 +00:00
DarkFighterLuke
b804854e3b Set uploader default values to None 2022-02-27 01:40:37 +01:00
DarkFighterLuke
fa32787f2b Apply refactoring suggested by @rautamiekka 2022-02-27 01:32:20 +01:00
DarkFighterLuke
d9c98f01d5 Fix flake8 issues 2022-02-26 16:57:39 +01:00
DarkFighterLuke
068cd1b81c Scrape creator 2022-02-26 16:53:11 +01:00
DarkFighterLuke
0e05a7d5d8 Scrape average_rating 2022-02-26 16:49:42 +01:00
DarkFighterLuke
4d93d7072d Fix checks on empty data 2022-02-26 16:48:10 +01:00
DarkFighterLuke
9823824d7c Scrape actors 2022-02-26 16:39:06 +01:00
DarkFighterLuke
072873437c Fix uploader regex 2022-02-26 16:28:00 +01:00
DarkFighterLuke
441bafa2d1 Scrape tags
modified:   youtube_dl/extractor/xnxx.py
2022-02-26 16:20:04 +01:00
DarkFighterLuke
b6f851ba4d Fix view_count regex 2022-02-26 16:13:05 +01:00
DarkFighterLuke
6e109f8ad4 Scrape uploader 2022-02-26 16:01:18 +01:00

View File

@ -6,9 +6,11 @@ import re
from .common import InfoExtractor
from ..utils import (
determine_ext,
float_or_none,
int_or_none,
NO_DEFAULT,
str_to_int,
urljoin,
)
@ -16,7 +18,7 @@ class XNXXIE(InfoExtractor):
_VALID_URL = r'https?://(?:video|www)\.xnxx\.com/video-?(?P<id>[0-9a-z]+)/'
_TESTS = [{
'url': 'http://www.xnxx.com/video-55awb78/skyrim_test_video',
'md5': '7583e96c15c0f21e9da3453d9920fbba',
'md5': '73c071a361a09aae7e7d60008221fd13',
'info_dict': {
'id': '55awb78',
'ext': 'mp4',
@ -25,7 +27,29 @@ class XNXXIE(InfoExtractor):
'duration': 469,
'view_count': int,
'age_limit': 18,
'tags': ['video game', 'skyrim', '3d', 'game', '3d game', 'video games', 'rule34', 'test', 'rough', 'sfm', 'fallout', 'porno game', 'skyrim hentai', 'h game', '3d horse', '3d porno anime', 'xx video wwxxx cartoon cartoons', 'gaming', 'games', '3d porno desenho'],
'uploader': 'Glurp',
'uploader_id': 'Glurp',
'uploader_url': '/porn-maker/glurp',
},
}, {
'url': 'https://www.xnxx.com/video-h46klf8/babes_-_come_back_to_bed_starring_abella_danger_and_darcie_dolce_clip',
'md5': 'b8b4a594b4091de46ce05d0a9d45317c',
'info_dict': {
'id': 'h46klf8',
'ext': 'mp4',
'title': 'Babes - Come Back to Bed starring Abella Danger and Darcie Dolce clip',
'thumbnail': r're:^https?://.*\.jpg',
'duration': 480,
'view_count': int,
'age_limit': 18,
'tags': ['lesbicas', 'darcie dolce', 'darcie dolce lesbian', 'abella danger lesbian', 'babes', 'lesb', 'abella danger', 'mmf', 'come', 'darcie', 'lesbianas', 'mulher chupando peito da outra', 'abella', 'girl on girl', 'darcie dolce abella danger', 'chupando peitos', 'para', 'lesbian', 'chupando peitos lesbicas', 'black lesbians'],
'uploader': 'Babes Network',
'uploader_id': 'Babes Network',
'uploader_url': '/porn-maker/babes3',
'actors': [{'given_name': 'darcie dolce', 'url': 'https://www.xnxx.com/search/darcie%20dolce'}, {'given_name': 'darcie dolce lesbian', 'url': 'https://www.xnxx.com/search/darcie%20dolce%20lesbian'}, {'given_name': 'abella danger lesbian', 'url': 'https://www.xnxx.com/search/abella%20danger%20lesbian'}, {'given_name': 'abella danger', 'url': 'https://www.xnxx.com/search/abella%20danger'}, {'given_name': 'darcie dolce abella danger', 'url': 'https://www.xnxx.com/search/darcie%20dolce%20abella%20danger'}],
'average_rating': float,
}
}, {
'url': 'http://video.xnxx.com/video1135332/lida_naked_funny_actress_5_',
'only_matching': True,
@ -70,8 +94,27 @@ class XNXXIE(InfoExtractor):
'ThumbUrl', fatal=False) or get('ThumbUrl169', fatal=False)
duration = int_or_none(self._og_search_property('duration', webpage))
view_count = str_to_int(self._search_regex(
r'id=["\']nb-views-number[^>]+>([\d,.]+)', webpage, 'view count',
default=None))
r'-.+?\t+- (?P<views>.+?) <span class="icon-f icf-eye">', webpage, 'view count', group='views',
default=0))
tags = self._search_regex(r'<meta name="keywords" content="porn,porn movies,free porn,free porn movies,sex,porno,free sex,tube porn,tube,videos,full porn,xxnx,xnxxx,xxx,pussy,(?P<tags>.+?)"', webpage, 'tags', group='tags', default='').split(',')
uploader_data = re.findall(r'<a class=".+?-plate" href="(?P<uploader_url>.+?)">(?P<uploader_name>.+?)</a>', webpage)
uploader_id = None
uploader_url = None
if uploader_data != []:
uploader_url, uploader_id = uploader_data[0][0:2]
actors_data = re.findall(r'<a class="is-pornstar" href="(?P<actor_url>.+?)">(?P<actor_name>.+?)</a>', webpage)
actors = []
if actors_data != []:
for actor_tuple in actors_data:
actors.append({
'given_name': actor_tuple[1],
'url': urljoin(url, actor_tuple[0]),
})
rating = float_or_none(self._search_regex(r'<span class="rating-box value">(?P<rating>.+?)%</span>', webpage, 'rating', group='rating'))
return {
'id': video_id,
@ -81,4 +124,11 @@ class XNXXIE(InfoExtractor):
'view_count': view_count,
'age_limit': 18,
'formats': formats,
'tags': tags,
'creator': uploader_id,
'uploader': uploader_id,
'uploader_id': uploader_id,
'uploader_url': uploader_url,
'actors': actors,
'average_rating': rating,
}