iwla/plugins/pre_analysis/H002_soutade.py

import re
from iwla import IWLA

# Remove logo from indefero
logo_re = re.compile(r'^.+/logo/$')

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

# Basic rule to detect robots

def hook(iwla):
    hits = iwla.getCurrentVisists()

    for k in hits.keys():
        super_hit = hits[k]

        if super_hit['robot']: continue

        for p in super_hit['requests']:
            if not p['is_page']: continue
            if int(p['status']) != 200: continue
            if p['time_decoded'].tm_mday != super_hit['last_access'].tm_mday: continue
            if logo_re.match(p['extract_request']['extract_uri']):
                p['is_page'] = False
                super_hit['viewed_pages'] -= 1
                super_hit['viewed_hits'] += 1