process_sf_voter_street_data.py

#!/usr/bin/env python

import collections

# download files from http://sfgov2.org/index.aspx?page=4767

precinct_map = collections.defaultdict(list)
with open('PDMJ001_DistPctExtract.txt', 'r') as dfd:
    # skip the header
    next(dfd)
    for line in dfd:
        fields = line.rstrip().split('\t')
        name = fields[2]
        try:
            precinct_id = int(fields[3])
            precinct_portion = int(fields[4]) if fields[4] else ''
            precinct_map[(precinct_id, precinct_portion)].append(name)
        except:
            print line
            break


data = []
with open('SGMJ001_StreetExtract_20150826.txt', 'r') as sfd:
    keys = sfd.readline().rstrip().split('\t')
    StreetData = collections.namedtuple('StreetData', keys)
    for line in sfd:
        fields = [
            (int(f) if all(x.isdigit() for x in f) and f else f)
            for f in
            line.rstrip().split('\t')
        ]
        datum = StreetData(**dict(zip(keys, fields)))
        data.append(datum)


district_3 = [d for d in data if 'SUP DIST 3' in precinct_map[(d.PrecinctID, d.PrecinctPortion)]]
print sum(d.ActiveVoterCount for d in district_3)