http parse content about blackbox_exporter HOT 4 CLOSED

#!/usr/bin/env python3
#
# silly multithreeding application check 0.2
#

from multiprocessing import Pool
from urllib.request import Request, urlopen
from urllib.error import URLError, HTTPError
from os import path
import re, sys, time, socket
# import settings
from conf import *
from string import Template
from pprint import pprint


response_time = Template('application_response_time{url="$url",region="$region",server_name="$server"}  $time')


def load_file2list(file_path):
    ''' Read whole file each line is item in array with removed \n char
        file_path = str()
    '''
    try:
        if path.isfile(file_path):
            with open(file_path, "r") as _F:
                return _F.read().splitlines()
        else:
            print("Required file {} does not exist, so use empty".format(file_path))
            return list()
    except IOError as err:
        print(err)
        sys.exit(1)


def check_site(domain):
    ''' Get https://{{domain}}/ site and try to find SITE: {{domain}} string
        in returned page source. Return True if OK, or False.
        In case that chack will fail, print simple Error message.
        domain = str()  # without https://, just domain
    '''
    # by default application is in error status
    _ret = dict(url=domain, region="japan", server=None, time=999999)
    try:
        # measure response time
        t1 = time.time()

        req = Request("http://{}/".format(domain))
        if len(re.findall('_.O.K._: {}' , str(urlopen(req).read()))) >= 1:
            #print('Ok: Domain {} is OK!!!'.format(domain), file=sys.stdout)
            _ret['status'] = True
            _ret['msg'] = "OK"
        else:
            #print('Err: Domain {} is BROKEN!!!'.format(domain), file=sys.stderr)
            _ret['msg'] = "Broken site"    # probably don't contain right domain in source

        _ret['time'] = time.time() - t1

        _ret['server'] = socket.gethostbyaddr(domain)[0]
    except URLError:
        _ret['msg'] = "Unrechable"
        #print("Err: Domain {} is UNREACHABLE!!!".format(domain), file=sys.stderr)

    finally:
        return _ret


if __name__ == '__main__':
    # load last data
    DOMAINS = list(set(load_file2list(PATH_DOMAINS)))

    # paralelne jedeme seznam domen
    pool = Pool(8)
    results = pool.map(check_site, DOMAINS)
    with open(PATH_METRICS, "w") as metrics_f:
        #metrics_f.seek(0)          # rewrite file from begining
        for dom in results:
            print(response_time.substitute(dom), file=metrics_f)