2014-11-19 08:01:12 +01:00
|
|
|
import re
|
|
|
|
|
|
|
|
# Remove logo from indefero
|
|
|
|
logo_re = re.compile(r'^.+/logo/$')
|
|
|
|
|
2014-11-20 16:15:57 +01:00
|
|
|
PLUGIN_CLASS = 'HTTP'
|
|
|
|
API_VERSION = 1
|
|
|
|
|
|
|
|
def get_plugins_infos():
|
|
|
|
infos = {'class' : PLUGIN_CLASS,
|
|
|
|
'min_version' : API_VERSION,
|
|
|
|
'max_version' : -1}
|
|
|
|
return infos
|
|
|
|
|
|
|
|
def load():
|
|
|
|
return True
|
|
|
|
|
2014-11-19 08:01:12 +01:00
|
|
|
# Basic rule to detect robots
|
|
|
|
|
|
|
|
def hook(hits):
|
|
|
|
for k in hits.keys():
|
|
|
|
super_hit = hits[k]
|
|
|
|
|
|
|
|
if super_hit['robot']: continue
|
|
|
|
|
|
|
|
for p in super_hit['pages']:
|
|
|
|
if not p['is_page']: continue
|
2014-11-20 16:15:57 +01:00
|
|
|
if int(p['status']) != 200: continue
|
2014-11-19 08:01:12 +01:00
|
|
|
if logo_re.match(p['extract_request']['extract_uri']):
|
|
|
|
p['is_page'] = False
|
|
|
|
super_hit['viewed_pages'] -= 1
|
|
|
|
super_hit['viewed_hits'] += 1
|