iwla/hooks/pre_analysis/H001_soutade.py

21 lines
549 B
Python

import re
# Remove logo from indefero
logo_re = re.compile(r'^.+/logo/$')
# 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
if logo_re.match(p['extract_request']['extract_uri']):
p['is_page'] = False
if super_hit['viewed_pages']:
super_hit['viewed_pages'] -= 1
super_hit['viewed_hits'] += 1