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

Do a lot of things

Browse Source
This commit is contained in:
Grégory Soutadé
2014-11-26 16:17:16 +01:00
parent 6505ca3ee5
commit 81b3eee552
12 changed files with 333 additions and 48 deletions
Download Patch File Download Diff File Expand all files Collapse all files

23
plugins/pre_analysis/robots.py
View File

@@ -3,7 +3,7 @@ import re
from iwla import IWLA
from iplugin import IPlugin
from awstats_robots_data import awstats_robots
import awstats_data
class IWLAPreAnalysisRobots(IPlugin):
def __init__(self, iwla):
@@ -11,9 +11,7 @@ class IWLAPreAnalysisRobots(IPlugin):
self.API_VERSION = 1
def load(self):
global awstats_robots
self.awstats_robots = map(lambda (x) : re.compile(x, re.IGNORECASE), awstats_robots)
self.awstats_robots = map(lambda (x) : re.compile(x, re.IGNORECASE), awstats_data.robots)
return True
@@ -32,13 +30,17 @@ class IWLAPreAnalysisRobots(IPlugin):
if first_page['time_decoded'].tm_mday == super_hit['last_access'].tm_mday:
for r in self.awstats_robots:
if r.match(first_page['http_user_agent']):
super_hit['robot'] = 1
continue
isRobot = True
break
if isRobot:
super_hit['robot'] = 1
continue
# 1) no pages view --> robot
if not super_hit['viewed_pages']:
super_hit['robot'] = 1
continue
# if not super_hit['viewed_pages']:
# super_hit['robot'] = 1
# continue
# 2) pages without hit --> robot
if not super_hit['viewed_hits']:
@@ -59,6 +61,7 @@ class IWLAPreAnalysisRobots(IPlugin):
super_hit['robot'] = 1
continue
if super_hit['viewed_hits'] and not referers:
if not super_hit['viewed_pages'] and \
(super_hit['viewed_hits'] and not referers):
super_hit['robot'] = 1
continue