mirror of https://github.com/MISP/misp-modules
fix: Consider mail body as UTF-8 encoded
parent
3e168aceb8
commit
38457f0a7b
|
@ -110,21 +110,20 @@ def handler(q=False):
|
||||||
email_object.add_reference(f_object.uuid, 'includes', 'Email attachment')
|
email_object.add_reference(f_object.uuid, 'includes', 'Email attachment')
|
||||||
|
|
||||||
mail_body = email_object.email.get_body(preferencelist=('html', 'plain'))
|
mail_body = email_object.email.get_body(preferencelist=('html', 'plain'))
|
||||||
if extract_urls:
|
if extract_urls and mail_body:
|
||||||
if mail_body:
|
charset = mail_body.get_content_charset('utf-8')
|
||||||
charset = mail_body.get_content_charset()
|
if mail_body.get_content_type() == 'text/html':
|
||||||
if mail_body.get_content_type() == 'text/html':
|
url_parser = HTMLURLParser()
|
||||||
url_parser = HTMLURLParser()
|
url_parser.feed(mail_body.get_payload(decode=True).decode(charset, errors='ignore'))
|
||||||
url_parser.feed(mail_body.get_payload(decode=True).decode(charset, errors='ignore'))
|
urls = url_parser.urls
|
||||||
urls = url_parser.urls
|
else:
|
||||||
else:
|
urls = re.findall(r'https?://(?:[-\w.]|(?:%[\da-fA-F]{2}))+', mail_body.get_payload(decode=True).decode(charset, errors='ignore'))
|
||||||
urls = re.findall(r'https?://(?:[-\w.]|(?:%[\da-fA-F]{2}))+', mail_body.get_payload(decode=True).decode(charset, errors='ignore'))
|
for url in urls:
|
||||||
for url in urls:
|
if not url:
|
||||||
if not url:
|
continue
|
||||||
continue
|
url_object = URLObject(url, standalone=False)
|
||||||
url_object = URLObject(url, standalone=False)
|
file_objects.append(url_object)
|
||||||
file_objects.append(url_object)
|
email_object.add_reference(url_object.uuid, 'includes', 'URL in email body')
|
||||||
email_object.add_reference(url_object.uuid, 'includes', 'URL in email body')
|
|
||||||
|
|
||||||
objects = [email_object.to_json()]
|
objects = [email_object.to_json()]
|
||||||
if file_objects:
|
if file_objects:
|
||||||
|
|
Loading…
Reference in New Issue