Remove all trailing slashs of URL before starting analyze

This commit is contained in:
Gregory Soutade 2024-01-30 11:28:10 +01:00
parent d6d216db4d
commit d78739157b
2 changed files with 6 additions and 5 deletions

View File

@ -391,14 +391,15 @@ class IWLA(object):
super_hit['robot'] = None
super_hit['hit_only'] = 0
def _normalizeURI(self, uri, removeFileSlash=False):
def _normalizeURI(self, uri, removeFileSlash=True):
if uri == '/': return uri
# Remove protocol
uri = self.protocol_re.sub('', uri)
# Remove double /
uri = self.slash_re.sub('/', uri)
if removeFileSlash and uri[-1] == '/':
uri = uri[:-1]
if removeFileSlash:
while len(uri) > 1 and uri[-1] == '/':
uri = uri[:-1]
return uri
def _normalizeParameters(self, parameters):
@ -429,7 +430,7 @@ class IWLA(object):
referer_groups = self.uri_re.match(hit['http_referer'])
if referer_groups:
hit['extract_referer'] = referer_groups.groupdict("")
hit['extract_referer']['extract_uri'] = self._normalizeURI(hit['extract_referer']['extract_uri'], True)
hit['extract_referer']['extract_uri'] = self._normalizeURI(hit['extract_referer']['extract_uri'])
hit['extract_referer']['extract_parameters'] = self._normalizeParameters(hit['extract_referer']['extract_parameters'])
hit['remote_ip'] = hit['remote_addr']

View File

@ -75,7 +75,7 @@ class IWLAPostAnalysisTopPages(IPlugin):
uri = r['extract_request']['extract_uri']
if self.index_re.match(uri):
uri = '/'
uri = ''
uri = "%s%s" % (r.get('server_name', ''), uri)