process-nsid-ripeatlas-loss.py

#!/usr/bin/env python
import sys
import fileinput
from ripe.atlas.sagan import DnsResult

cache={}
latest={}
BIN = 3600

def process(line):
    result = DnsResult(line)
    try:
        nsid = result.responses[0].abuf.answers[0].data[0]
        ts_bin = int( float( result.created.strftime('%s') ) / BIN ) * BIN
        cache.setdefault( nsid, {} )
        cache[ nsid ].setdefault( ts_bin , {'ok': 0, 'timeout': 0} )
        cache[ nsid ][ts_bin]['ok'] += 1
        #print("{}\t{}\t{}\t{}".format(
        #    result.created.strftime('%s'),
        #    result.probe_id,
        #    result.responses[0].response_time,
        #    result.responses[0].abuf.answers[0].data[0]
        #    )
        #)
        latest[ result.probe_id ] =  result.responses[0].abuf.answers[0].data[0]
    except:
        if 'error' in result.raw_data and 'timeout' in result.raw_data['error']:
            try:
               ts_bin = int( float( result.created.strftime('%s') ) / BIN ) * BIN
            except:
               print >>sys.stderr, "giving up on" % result.raw_data
            try:
               nsid = latest[ result.probe_id ]
            except:
               nsid = 'unknown'
            cache.setdefault( nsid, {} )
            cache[ nsid ].setdefault( ts_bin , {'ok': 0, 'timeout': 0} )
            cache[ nsid ][ts_bin]['timeout'] += 1

        pass # brutal

'''
u'error': {u'timeout': 5000}
'''

for line in fileinput.input():
    process(line)

for nsid in cache:
    for ts_bin in sorted( cache[ nsid ] ):
        try:
            count = cache[nsid][ts_bin]['timeout'] + cache[nsid][ts_bin]['ok']
            err_pct = cache[nsid][ts_bin]['timeout'] * 100.0 / count
            print "%s\t%s\t%s\t%s" % ( ts_bin, nsid, err_pct, count )
        except:
            print "#ERROR %s" % ( cache[ nsid ][ ts_bin ] )