]>
Commit | Line | Data |
---|---|---|
cc52de43 | 1 | #!/usr/bin/env python3 |
750e9833 FV |
2 | """ |
3 | This script employs a VERY basic heuristic ('porn' in webpage.lower()) to check | |
4 | if we are not 'age_limit' tagging some porn site | |
a45ea170 FV |
5 | |
6 | A second approach implemented relies on a list of porn domains, to activate it | |
7 | pass the list filename as the only argument | |
750e9833 FV |
8 | """ |
9 | ||
10 | # Allow direct execution | |
11 | import os | |
12 | import sys | |
f8271158 | 13 | |
750e9833 FV |
14 | sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__)))) |
15 | ||
8a1a60d1 | 16 | from test.helper import gettestcases |
f8271158 | 17 | |
18 | from yt_dlp.utils import compat_urllib_parse_urlparse, compat_urllib_request | |
750e9833 | 19 | |
a45ea170 FV |
20 | if len(sys.argv) > 1: |
21 | METHOD = 'LIST' | |
22 | LIST = open(sys.argv[1]).read().decode('utf8').strip() | |
23 | else: | |
24 | METHOD = 'EURISTIC' | |
25 | ||
8a1a60d1 | 26 | for test in gettestcases(): |
a45ea170 FV |
27 | if METHOD == 'EURISTIC': |
28 | try: | |
29 | webpage = compat_urllib_request.urlopen(test['url'], timeout=10).read() | |
70a1165b | 30 | except Exception: |
86e5f3ed | 31 | print('\nFail: {}'.format(test['name'])) |
a45ea170 FV |
32 | continue |
33 | ||
34 | webpage = webpage.decode('utf8', 'replace') | |
35 | ||
36 | RESULT = 'porn' in webpage.lower() | |
37 | ||
38 | elif METHOD == 'LIST': | |
39 | domain = compat_urllib_parse_urlparse(test['url']).netloc | |
40 | if not domain: | |
86e5f3ed | 41 | print('\nFail: {}'.format(test['name'])) |
a45ea170 FV |
42 | continue |
43 | domain = '.'.join(domain.split('.')[-2:]) | |
750e9833 | 44 | |
a45ea170 | 45 | RESULT = ('.' + domain + '\n' in LIST or '\n' + domain + '\n' in LIST) |
750e9833 | 46 | |
3089bc74 S |
47 | if RESULT and ('info_dict' not in test or 'age_limit' not in test['info_dict'] |
48 | or test['info_dict']['age_limit'] != 18): | |
86e5f3ed | 49 | print('\nPotential missing age_limit check: {}'.format(test['name'])) |
750e9833 | 50 | |
3089bc74 S |
51 | elif not RESULT and ('info_dict' in test and 'age_limit' in test['info_dict'] |
52 | and test['info_dict']['age_limit'] == 18): | |
86e5f3ed | 53 | print('\nPotential false negative: {}'.format(test['name'])) |
750e9833 FV |
54 | |
55 | else: | |
56 | sys.stdout.write('.') | |
57 | sys.stdout.flush() | |
58 | ||
59 | print() |