github/MITMf
1
0
Fork 0
mirror of https://github.com/byt3bl33d3r/MITMf.git synced 2025-07-06 04:52:22 -07:00
Code Issues Projects Releases Packages Wiki Activity Actions

Inject.py now tries to detect encoding before parsing HTML with BeautifulSoup

Browse source
This commit is contained in:
byt3bl33d3r 2015-08-23 19:42:52 +02:00
parent fb41a510f6
commit 27c28e512e
3 changed files with 29 additions and 18 deletions
Download patch file Download diff file Expand all files Collapse all files

40
plugins/inject.py
View file

@ -57,27 +57,37 @@ class Inject(Plugin):
def response(self, response, request, data):
encoding = None
ip = response.getClientIP()
hn = response.getRequestHostname()
mime = response.headers['Content-Type']
try:
mime = response.headers['Content-Type']
except KeyError:
return
if "charset" in mime:
match = re.search('charset=(.*)', mime)
if match:
encoding = match.group(1).strip().replace('"', "")
else:
try:
encoding = chardet.detect(data)["encoding"]
except:
pass
else:
try:
encoding = chardet.detect(data)["encoding"]
except:
pass
if self._should_inject(ip, hn) and self._ip_filter(ip) and self._host_filter(hn) and (hn not in self.ip) and ("text/html" in mime):
if "charset" in mime:
match = re.search('charset=(.*)', mime)
if match:
encoding = match.group(1).strip().replace('"', "")
else:
try:
encoding = chardet.detect(data)["encoding"]
except:
encoding = None
if encoding is not None:
html = BeautifulSoup(data.decode(encoding, "ignore"), "lxml")
else:
html = BeautifulSoup(data, "lxml")
if encoding:
html = BeautifulSoup(data.decode(encoding, "ignore"), "lxml")
else:
html = BeautifulSoup(data, "lxml") # let bs find the encoding
if html.body:
if self.html_url: