iwla/plugins/post_analysis/top_downloads.py

77 lines
1.9 KiB
Python
Raw Normal View History

2014-11-27 12:35:41 +01:00
import re
from iwla import IWLA
from iplugin import IPlugin
2014-12-10 07:09:05 +01:00
#
# Post analysis hook
#
# Count TOP downloads
#
# Plugin requirements :
# None
#
# Conf values needed :
2014-12-10 16:53:50 +01:00
# None
2014-12-10 07:09:05 +01:00
#
# Output files :
# None
#
# Statistics creation :
# None
#
# Statistics update :
# month_stats:
# top_downloads =>
# uri
#
# Statistics deletion :
# None
#
2014-11-27 12:35:41 +01:00
class IWLAPostAnalysisTopDownloads(IPlugin):
def __init__(self, iwla):
super(IWLAPostAnalysisTopDownloads, self).__init__(iwla)
self.API_VERSION = 1
self.conf_requires = ['multimedia_files', 'viewed_http_codes']
def hook(self):
stats = self.iwla.getCurrentVisists()
month_stats = self.iwla.getMonthStats()
multimedia_files = self.iwla.getConfValue('multimedia_files')
viewed_http_codes = self.iwla.getConfValue('viewed_http_codes')
top_downloads = month_stats.get('top_downloads', {})
for (k, super_hit) in stats.items():
if super_hit['robot']: continue
for r in super_hit['requests']:
2014-11-27 13:07:14 +01:00
if not self.iwla.isValidForCurrentAnalysis(r) or\
not self.iwla.hasBeenViewed(r):
continue
2014-11-27 12:35:41 +01:00
if r['is_page']: continue
if not int(r['status']) in viewed_http_codes: continue
uri = r['extract_request']['extract_uri'].lower()
isMultimedia = False
for ext in multimedia_files:
if uri.endswith(ext):
isMultimedia = True
break
if isMultimedia: continue
uri = "%s%s" % (r.get('server_name', ''),
r['extract_request']['extract_uri'])
if not uri in top_downloads.keys():
top_downloads[uri] = 1
else:
top_downloads[uri] += 1
month_stats['top_downloads'] = top_downloads