Python webalizer wrapper

webstats.py

#! /usr/bin/env python3

import os, sys, re
from subprocess import call
from os.path import join

LOGDIR = '/var/log/nginx'
OUTDIR = '/home/htdocs/internal/stats/'
WEBALIZER_OPTS = [
    '/usr/local/bin/webalizer-RB30',   # Binary
    '-F clf',                          # log type
    '-p',                              # preserve history
    '-D /var/tmp/webalizer.dnscache',  # DNS cache location
    '-N 5',                            # number of DNS processes
    '-w',                              # Enable GeoIP lookups
    '-W /usr/share/GeoIP/GeoIP.dat',   # external GeoIP database,
    '-Q',                              # quiet
]

# mydomain.com.access.log, mydomain.com.error.log, ... .log.1, ... .log.2.gz, ...
my_log_fn = re.compile(r'((.*)(?:\.?(?:access|error))\.log(?:.*))')
log_id = re.compile(r'(?:(?:.*)(?:\.?(?:access|error))\.log\.?(\d+)?).*')

def parselog(m):
    filename, name = m
    name = name[:-1].strip()
    
    if len(name) <= 0:
        name = "_default_"
        
    destdir = join(OUTDIR, name)
    args = [
        '-o ' + destdir,
        '-n ' + name,
        join(LOGDIR, filename),
    ]
    
    if os.path.lexists(destdir) is False:
        os.mkdir(destdir)
        
    if os.path.isdir(destdir) is False:
        raise Exception('outdir ' + destdir + ' is not a directory')
    
    print (filename)
    #print (name)
    cmdline = " ".join(WEBALIZER_OPTS + args)
    cmdargs = WEBALIZER_OPTS + args
    
    if os.system(cmdline) is not 0:
        print (" !! webalizer failed! !!\n")
    
    
def logkey(l):
    m = log_id.match(l)
    if m is None:
        return 0
    
    g = m.groups()[0]
    if g is None:
        return 0
    g = int(g)

    return g

if __name__ == "__main__":
    logs = os.listdir(LOGDIR)
    logs = sorted(logs, key=logkey, reverse=True)
    
    for l in logs:
        m = my_log_fn.match(l)
        if m is not None:
            filename, name = m.groups()
            parselog(m.groups())