Add the missing age_limit tags; added a devscript to do a superficial check for porn sites without the age_limit tag in the test

This commit is contained in:
Filippo Valsorda 2013-10-28 01:50:17 -04:00
parent 82f0ac657c
commit 750e9833b8
7 changed files with 59 additions and 4 deletions

39
devscripts/check-porn.py Normal file
View file

@ -0,0 +1,39 @@
#!/usr/bin/env python
"""
This script employs a VERY basic heuristic ('porn' in webpage.lower()) to check
if we are not 'age_limit' tagging some porn site
"""
# Allow direct execution
import os
import sys
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
from test.helper import get_testcases
from youtube_dl.utils import compat_urllib_request
for test in get_testcases():
try:
webpage = compat_urllib_request.urlopen(test['url'], timeout=10).read()
except:
print('\nFail: {0}'.format(test['name']))
continue
webpage = webpage.decode('utf8', 'replace')
if 'porn' in webpage.lower() and ('info_dict' not in test
or 'age_limit' not in test['info_dict']
or test['info_dict']['age_limit'] != 18):
print('\nPotential missing age_limit check: {0}'.format(test['name']))
elif 'porn' not in webpage.lower() and ('info_dict' in test and
'age_limit' in test['info_dict'] and
test['info_dict']['age_limit'] == 18):
print('\nPotential false negative: {0}'.format(test['name']))
else:
sys.stdout.write('.')
sys.stdout.flush()
print()

View file

@ -6,7 +6,6 @@ from ..utils import (
compat_urllib_parse_urlparse, compat_urllib_parse_urlparse,
compat_urllib_request, compat_urllib_request,
compat_urllib_parse, compat_urllib_parse,
unescapeHTML,
) )
from ..aes import ( from ..aes import (
aes_decrypt_text aes_decrypt_text
@ -20,6 +19,7 @@ class KeezMoviesIE(InfoExtractor):
u'md5': u'6e297b7e789329923fcf83abb67c9289', u'md5': u'6e297b7e789329923fcf83abb67c9289',
u'info_dict': { u'info_dict': {
u"title": u"Petite Asian Lady Mai Playing In Bathtub", u"title": u"Petite Asian Lady Mai Playing In Bathtub",
u"age_limit": 18,
} }
} }
@ -48,6 +48,8 @@ class KeezMoviesIE(InfoExtractor):
format = path.split('/')[4].split('_')[:2] format = path.split('/')[4].split('_')[:2]
format = "-".join( format ) format = "-".join( format )
age_limit = self._rta_search(webpage)
return { return {
'id': video_id, 'id': video_id,
'title': video_title, 'title': video_title,
@ -55,4 +57,5 @@ class KeezMoviesIE(InfoExtractor):
'ext': extension, 'ext': extension,
'format': format, 'format': format,
'format_id': format, 'format_id': format,
'age_limit': age_limit,
} }

View file

@ -21,6 +21,7 @@ class PornHubIE(InfoExtractor):
u'info_dict': { u'info_dict': {
u"uploader": u"BABES-COM", u"uploader": u"BABES-COM",
u"title": u"Seductive Indian beauty strips down and fingers her pink pussy", u"title": u"Seductive Indian beauty strips down and fingers her pink pussy",
u"age_limit": 18
} }
} }
@ -64,4 +65,5 @@ class PornHubIE(InfoExtractor):
'title': video_title, 'title': video_title,
'thumbnail': thumbnail, 'thumbnail': thumbnail,
'formats': formats, 'formats': formats,
'age_limit': 18,
} }

View file

@ -16,7 +16,8 @@ class PornotubeIE(InfoExtractor):
u'md5': u'374dd6dcedd24234453b295209aa69b6', u'md5': u'374dd6dcedd24234453b295209aa69b6',
u'info_dict': { u'info_dict': {
u"upload_date": u"20090708", u"upload_date": u"20090708",
u"title": u"Marilyn-Monroe-Bathing" u"title": u"Marilyn-Monroe-Bathing",
u"age_limit": 18
} }
} }

View file

@ -22,6 +22,7 @@ class SpankwireIE(InfoExtractor):
u"uploader": u"oreusz", u"uploader": u"oreusz",
u"title": u"Buckcherry`s X Rated Music Video Crazy Bitch", u"title": u"Buckcherry`s X Rated Music Video Crazy Bitch",
u"description": u"Crazy Bitch X rated music video.", u"description": u"Crazy Bitch X rated music video.",
u"age_limit": 18,
} }
} }
@ -60,6 +61,8 @@ class SpankwireIE(InfoExtractor):
}) })
formats.sort(key=lambda format: list(map(lambda s: s.zfill(6), format['format'].split('-')))) formats.sort(key=lambda format: list(map(lambda s: s.zfill(6), format['format'].split('-'))))
age_limit = self._rta_search(webpage)
return { return {
'id': video_id, 'id': video_id,
'uploader': video_uploader, 'uploader': video_uploader,
@ -67,4 +70,5 @@ class SpankwireIE(InfoExtractor):
'thumbnail': thumbnail, 'thumbnail': thumbnail,
'description': description, 'description': description,
'formats': formats, 'formats': formats,
'age_limit': age_limit,
} }

View file

@ -22,6 +22,7 @@ class Tube8IE(InfoExtractor):
u"description": u"hot teen Kasia grinding", u"description": u"hot teen Kasia grinding",
u"uploader": u"unknown", u"uploader": u"unknown",
u"title": u"Kasia music video", u"title": u"Kasia music video",
u"age_limit": 18,
} }
} }
@ -60,4 +61,5 @@ class Tube8IE(InfoExtractor):
'ext': extension, 'ext': extension,
'format': format, 'format': format,
'format_id': format, 'format_id': format,
'age_limit': 18,
} }

View file

@ -13,7 +13,8 @@ class YouJizzIE(InfoExtractor):
u'file': u'2189178.flv', u'file': u'2189178.flv',
u'md5': u'07e15fa469ba384c7693fd246905547c', u'md5': u'07e15fa469ba384c7693fd246905547c',
u'info_dict': { u'info_dict': {
u"title": u"Zeichentrick 1" u"title": u"Zeichentrick 1",
u"age_limit": 18,
} }
} }
@ -25,6 +26,8 @@ class YouJizzIE(InfoExtractor):
# Get webpage content # Get webpage content
webpage = self._download_webpage(url, video_id) webpage = self._download_webpage(url, video_id)
age_limit = self._rta_search(webpage)
# Get the video title # Get the video title
video_title = self._html_search_regex(r'<title>(?P<title>.*)</title>', video_title = self._html_search_regex(r'<title>(?P<title>.*)</title>',
webpage, u'title').strip() webpage, u'title').strip()
@ -60,6 +63,7 @@ class YouJizzIE(InfoExtractor):
'title': video_title, 'title': video_title,
'ext': 'flv', 'ext': 'flv',
'format': 'flv', 'format': 'flv',
'player_url': embed_page_url} 'player_url': embed_page_url,
'age_limit': age_limit}
return [info] return [info]