41 lines
1.3 KiB
Python
41 lines
1.3 KiB
Python
import re
|
|
|
|
from iwla import IWLA
|
|
from iplugin import IPlugin
|
|
|
|
# Basic rule to detect robots
|
|
|
|
class IWLAPreAnalysisPageToHit(IPlugin):
|
|
|
|
def __init__(self, iwla):
|
|
super(IWLAPreAnalysisPageToHit, self).__init__(iwla)
|
|
self.API_VERSION = 1
|
|
|
|
def load(self):
|
|
# Remove logo from indefero
|
|
self.regexps = self.iwla.getConfValue('page_to_hit_conf', [])
|
|
if not self.regexps: return False
|
|
self.regexps = map(lambda(r): re.compile(r), self.regexps)
|
|
|
|
return True
|
|
|
|
def hook(self):
|
|
hits = self.iwla.getCurrentVisists()
|
|
|
|
for (k, super_hit) in hits.items():
|
|
if super_hit['robot']: continue
|
|
|
|
for request in super_hit['requests']:
|
|
if not self.iwla.isValidForCurrentAnalysis(request) or\
|
|
not self.iwla.hasBeenViewed(request):
|
|
continue
|
|
if not request['is_page']: continue
|
|
uri = request['extract_request']['extract_uri']
|
|
for regexp in self.regexps:
|
|
if regexp.match(uri):
|
|
#print '%s is an hit' % uri
|
|
request['is_page'] = False
|
|
super_hit['viewed_pages'] -= 1
|
|
super_hit['viewed_hits'] += 1
|
|
break
|