#! /usr/bin/python

from httpfilehits import filter
import httpfilehits

file_tree_root = '<path to ~/public_html>'

log_file = './log_digest'
report_file = './log_report'
gone_report_file ="./log_report_gone"

filters = { filter.by_ip:['XXX\.XXX\.XXX\.XXX',         # Intranet web crawler 
                          'XXX\.XXX\.XXX\.XXX',         # My Work Machine
                          'XXX\.XXX\.XXX\.XXX'],        # My Home Machine
            filter.by_hostname:['crawl|spider|search'], # Informal, but seems very effective.
          }

httpfilehits.main(log_file, report_file, gone_report_file, filters, file_tree_root)