Search TLS Error Reporting for the top certificates that provoked SEC_ERROR_UNKNOWN_ISSUER

Get SEC_ERROR_UNKNOWN_ISSUER certs.py

from moztelemetry import get_pings_properties, get_one_ping_per_client
from moztelemetry.dataset import Dataset
from collections import defaultdict, Counter

ssl_reports = Dataset.from_source("sslreports")\
      .where(submissionDate=lambda xx: xx.startswith("201702"))\
      .records(sc, sample=0.01)

# Filter down to SEC_ERROR_UNKNOWN_ISSUER
data=ssl_reports.filter(lambda y: y['meta']['errorCode'] == -8179.0)

def process(accum, xx):
    if 'isAccum' not in accum:
        # Happens on the first execution
        emptyValueBootstrapDataset = defaultdict(Counter)
        emptyValueBootstrapDataset['isAccum'] = True
        # Recursively bootstrap our accumulator
        accum = process(emptyValueBootstrapDataset, accum)
    if 'isAccum' in xx:
        # Merge intermediate states on the final executions
        for k1,v1 in xx.iteritems():
            if k1 == 'isAccum':
                continue
            for k2,v2 in v1.iteritems():
                accum[k1][k2] += v2
        return accum
    
    # Primary analysis
    try:
        ping = xx['meta']
        accum['certChains'][ping['failedCertChain']] += 1
        accum['hostname'][ping['hostname']] += 1
    except:
        print "Missing data? {}".format(xx) 
    return accum

x = data.reduce(process)

# Show top hostnames
for idx, w in enumerate(sorted(x['hostname'], key=x['hostname'].get, reverse=True)):
    if idx > 10:
        break
    print w, x['hostname'][w]

# Show top certificates
for idx, w in enumerate(sorted(x['certChains'], key=x['certChains'].get, reverse=True)):
    if idx > 10:
        break
    print w, x['certChains'][w]

# Write out the certificates to a file in format:
# {DER data} {number of occurrances}\n
with open("/tmp/output", "w") as out:
    for w in sorted(x['certChains'], key=x['certChains'].get, reverse=True):
        count = x['certChains'][w]
        if count > 10:
            out.write("{} {}\n".format(w, count))