soutade/iwla
1
0
Fork 0
Code Issues Pull Requests Releases 2 Activity

Add one more rule to robot detection : more than ten 404 pages viewed

Browse Source
This commit is contained in:
Gregory Soutade 2017-05-25 21:04:18 +02:00
parent 4bc2c1ad4b
commit 68a67adecc
1 changed files with 9 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
plugins/pre_analysis/robots.py
View File

@ -114,12 +114,16 @@ class IWLAPreAnalysisRobots(IPlugin):
self._setRobot(k, super_hit)
continue
not_found_pages = 0
for hit in super_hit['requests']:
# 3) /robots.txt read
if hit['extract_request']['http_uri'].endswith('/robots.txt'):
self._setRobot(k, super_hit)
break
if int(hit['status']) == 404:
not_found_pages += 1
# 4) Any referer for hits
if not hit['is_page'] and hit['http_referer']:
referers += 1
@ -128,6 +132,11 @@ class IWLAPreAnalysisRobots(IPlugin):
self._setRobot(k, super_hit)
continue
# 5) more than 10 404 pages
if not_found_pages > 10:
self._setRobot(k, super_hit)
continue
if not super_hit['viewed_pages'] and \
(super_hit['viewed_hits'] and not referers):
self._setRobot(k, super_hit)