Terrible half-done script to pull data from scans.io into mongodb

scans_io-parser.py

#!/usr/bin/env python
import gzip
import OpenSSL
from pymongo import MongoClient
import ssl
import sys
import textwrap
from pprint import pprint

client = MongoClient('mongodb://localhost:27017/')
db = client.certdb

for line in gzip.open('20140310_certs.gz'):
    (certhash, base64text) = line.split(',',1)
    base64text = "\n".join(textwrap.wrap(base64text,76))
    cert = "-----BEGIN CERTIFICATE-----\n%s\n-----END CERTIFICATE-----" % base64text
    x509 = OpenSSL.crypto.load_certificate(OpenSSL.crypto.FILETYPE_PEM, cert)
    # certsubj = x509.get_subject().get_components()
    # try:
    #     cn = [x[1] for x in certsubj if 'CN' in x][0]
    # except:
    #     print certsubj
    # print cn

    subject_dict = dict(x509.get_subject().get_components())
    issuer_dict = dict(x509.get_issuer().get_components())
    certdict = {'hash_id': certhash, 'issuer': issuer_dict, 'subject': subject_dict}
    print certdict
    certsdb = db.certsdb
    try:
        mongo_id = certsdb.update({'hash_id': certhash},certdict,upsert=True)
        print "Inserted into mongo: %s" % mongo_id
    except:
        print "Insert failed"
        pass