Check for HP01, log if it appears more than it should (twice)

hp01checker.py

import time
import datetime
import random

import requests

sites_to_check = {
    "http://www.daytondailynews.com/": dict(hit=0, miss=0),
    "http://www.oxfordpress.com/": dict(hit=0, miss=0),
    "http://www.springfieldnewssun.com/": dict(hit=0, miss=0),
    "http://www.middletownjournal.com/": dict(hit=0, miss=0),
    "http://www.journal-news.com/": dict(hit=0, miss=0),
    "http://www.todayspulse.com/": dict(hit=0, miss=0),
    "http://www.activedayton.com/": dict(hit=0, miss=0),
}

sites_to_check_uncached = {
    "http://www.daytondailynews.com/": dict(hit=0, miss=0),
    "http://www.oxfordpress.com/": dict(hit=0, miss=0),
    "http://www.springfieldnewssun.com/": dict(hit=0, miss=0),
    "http://www.middletownjournal.com/": dict(hit=0, miss=0),
    "http://www.journal-news.com/": dict(hit=0, miss=0),
    "http://www.todayspulse.com/": dict(hit=0, miss=0),
    "http://www.activedayton.com/": dict(hit=0, miss=0),
}

def monitor_hp01(cache_bust=False):
    dataset = sites_to_check
    if cache_bust:
        dataset = sites_to_check_uncached

    for site in dataset.keys():
        site_url = site
        if cache_bust:
            site_url = "%s?hp01_check=%s" % (site, random.randint(1, 10000))
        r = requests.get(site_url)
        hp01_count = r.text.count("HP01")
        if hp01_count > 2:
            dataset[site]['hit'] += 1
            now = datetime.datetime.now()
        else:
            dataset[site]['miss'] +=1

    hits = 0
    misses = 0
    for site, hit_log in dataset.items():
        hits += hit_log['hit']
        misses += hit_log['miss']

    now = datetime.datetime.now()
    if misses > 0:
        ratio = hits / float(misses)
    else:
        ratio = 0
    print "%s - Cached: %s - HP01 hit ratio: %.2f %s/%s" % (
        now.strftime("%Y/%m/%d %H:%M:%S"),
        not cache_bust,
        ratio,
        hits,
        misses,
    )


if __name__ == "__main__":
    while True:
        try:
            monitor_hp01()
            monitor_hp01(True)
            time.sleep(60)
        except KeyboardInterrupt:
            raise

requirements.txt

requests==1.2.0