iwla/plugins/post_analysis/top_pages.py

import re

from iwla import IWLA
from iplugin import IPlugin

class IWLAPostAnalysisTopPages(IPlugin):
    def __init__(self, iwla):
        super(IWLAPostAnalysisTopPages, self).__init__(iwla)
        self.API_VERSION = 1

    def load(self):
        self.index_re = re.compile(r'/index.*')
        return True
        
    def hook(self):
        stats = self.iwla.getCurrentVisists()
        month_stats = self.iwla.getMonthStats()
        
        top_pages = month_stats.get('top_pages', {})

        for (k, super_hit) in stats.items():
            if super_hit['robot']: continue
            for r in super_hit['requests']:
                if not r['is_page']: continue

                if not self.iwla.isValidForCurrentAnalysis(r) or\
                        not self.iwla.hasBeenViewed(r):
                    continue

                uri = r['extract_request']['extract_uri']
                if self.index_re.match(uri):
                    uri = '/'
                
                uri = "%s%s" % (r.get('server_name', ''), uri)

                if not uri in top_pages.keys():
                    top_pages[uri] = 1
                else:
                    top_pages[uri] += 1

        month_stats['top_pages'] = top_pages