Bugzilla Tech Evangelism URL status checker

checker.py

#!/usr/bin/env python

import json
import requests

# This search corresponds to: Tech Evangelism bugs that are UNCONFIRMED, NEW,
# or REOPENED. When we loop through the response, we skip bugs with an empty
# url field.
SEARCH_URL = ("https://api-dev.bugzilla.mozilla.org/latest/bug?"
              "bug_status=UNCONFIRMED"
              "&bug_status=NEW"
              "&bug_status=REOPENED"
              "&product=Tech%20Evangelism"
              "&include_fields=url,id")
FX_UA = ("Mozilla/5.0 (Macintosh; Intel Mac OS X 10.9; rv:27.0) "
         "Gecko/20100101 Firefox/27.0")
api = requests.get(SEARCH_URL, headers={'Accept': 'application/json',
                                        'Content-Type': 'application/json'})
r = api.json()
s = requests.Session()
# In theory a 3XX redirect could mean the bug is INVALID
s.max_redirects = 0
for bug in r['bugs']:
    try:
        if bug['url']:
            # Arbitary timeout of 5 seconds so this doesn't take ages.
            new_r = s.get(bug['url'], timeout=5, headers={'User-Agent': FX_UA})
            if new_r.status_code != 200:
                print("{0}: {1} ({2})".format(new_r.status_code, bug['url'],
                                              bug['id']))
    # Types of Exceptions we care about:
    # The site probably no longer exists
    except requests.exceptions.ConnectionError as e:
        print("ConnectionError for {0} ({1})".format(bug['url'], bug['id']))
    # The site took too long to connect. Probably worth checking manually.
    except requests.exceptions.Timeout as e:
        print("Timeout for {0} ({1})".format(bug['url'], bug['id']))
    # The site redirected to somewhere else. This could mean the original site
    # no longer exists and should be INVALID. Or it could be something as
    # simple as redirecting to a login page (and is still VALID)
    except requests.exceptions.TooManyRedirects as e:
        print("Redirect for {1} ({2})".format(bug['url'], bug['id']))

Oh, should probably filter out the Mobile product ones from this. Otherwise the desktop UA will cause too many redirects.