mygpo/db/couchdb/management/commands/count-view-usage.py

   1 import gzip
   2 from collections import Counter
   3
   4 from django.core.management.base import BaseCommand
   5 from django.core.urlresolvers import resolve, Resolver404
   6
   7
   8 class Command(BaseCommand):
   9     """ Calculates the view-usage from CouchDB log files
  10
  11     log files can be gzipped. Output is printed to stdout """
  12
  13     def handle(self, *args, **options):
  14
  15         handlers = Counter()
  16
  17         show_max = None
  18
  19         for f in self.open_files(args):
  20             for line in f:
  21
  22                 # assume standard Apache log format
  23                 part = line.split(' ')[6]
  24
  25                 try:
  26                     match = resolve(part)
  27                 except Resolver404:
  28                     pass
  29
  30                 mod = match.func.__module__
  31                 key = (mod, match.func.__name__)
  32                 handlers[key] += 1
  33
  34
  35         for (mod, funcname), n in handlers.most_common(show_max):
  36             fqname = '{}.{}'.format(mod, funcname)
  37             print '{:60} {:>7}'.format(fqname, n)
  38
  39
  40     def open_files(self, files):
  41         for filename in files:
  42             if '.gz' in filename:
  43                 yield gzip.open(filename)
  44             else:
  45                 yield open(filename)