CIRCL
/
AIL-framework
mirror of https://github.com/CIRCL/AIL-framework
2
1
Fork 0
Code Issues Releases Wiki Activity

fix: [module] fix module obj type: language + mail

dev
terrtia 2023-10-11 14:53:12 +02:00
parent 623ba455ff
commit 6978764b02
No known key found for this signature in database
GPG Key ID: 1E1B1F50D84613D0
2 changed files with 10 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
bin/modules/Languages.py
View File

@ -25,11 +25,13 @@ class Languages(AbstractModule):
self.logger.info(f'Module {self.module_name} initialized') self.logger.info(f'Module {self.module_name} initialized')
def compute(self, message): def compute(self, message):
item = self.get_obj() obj = self.get_obj()
if item.is_crawled():
domain = Domain(item.get_domain()) if obj.type == 'item':
for lang in item.get_languages(min_probability=0.8): if item.is_crawled():
domain.add_language(lang.language) domain = Domain(item.get_domain())
for lang in item.get_languages(min_probability=0.8):
domain.add_language(lang.language)
if __name__ == '__main__': if __name__ == '__main__':

6
bin/modules/Mail.py
View File

@ -139,7 +139,7 @@ class Mail(AbstractModule):
item = self.get_obj() item = self.get_obj()
item_date = item.get_date() item_date = item.get_date()
mails = self.regex_findall(self.email_regex, item_id, item.get_content()) mails = self.regex_findall(self.email_regex, item.id, item.get_content())
mxdomains_email = {} mxdomains_email = {}
for mail in mails: for mail in mails:
mxdomain = mail.rsplit('@', 1)[1].lower() mxdomain = mail.rsplit('@', 1)[1].lower()
@ -172,9 +172,9 @@ class Mail(AbstractModule):
# for tld in mx_tlds: # for tld in mx_tlds:
# Statistics.add_module_tld_stats_by_date('mail', item_date, tld, mx_tlds[tld]) # Statistics.add_module_tld_stats_by_date('mail', item_date, tld, mx_tlds[tld])
msg = f'Mails;{item.get_source()};{item_date};{item.get_basename()};Checked {num_valid_email} e-mail(s);{item_id}' msg = f'Mails;{item.get_source()};{item_date};{item.get_basename()};Checked {num_valid_email} e-mail(s);{item.id}'
if num_valid_email > self.mail_threshold: if num_valid_email > self.mail_threshold:
print(f'{item_id} Checked {num_valid_email} e-mail(s)') print(f'{item.id} Checked {num_valid_email} e-mail(s)')
self.redis_logger.warning(msg) self.redis_logger.warning(msg)
# Tags # Tags
tag = 'infoleak:automatic-detection="mail"' tag = 'infoleak:automatic-detection="mail"'