test.py

import json
import numpy as np
from sets import Set

with open("BusinessSpending.json") as f:
    data = json.load(f, 'latin-1')
entries = data["raw_data"]
keys = []
for entry in entries:
    keys = entry.keys()

countOfArbitaryOrders = 0
expenditureOfAllOrders = []
expenditureOfArbitaryOrders = []
vendors = Set([])
cost_and_venders = []
for entry in entries:
    expenditureOfAllOrders.append(int(entry["expenditure"]))
    if entry["number_of_bidders"] == u"随意契約":
        countArbitaryOrders+=1
        expenditureOfArbitaryOrders.append(int(entry["expenditure"]))
        vendors.add(entry["spending_destination"])
        cost_and_venders.append((entry["spending_destination"],
                                 int(entry["expenditure"])))
npa_expenditures = np.array(expenditureOfArbitaryOrders)
npa_expendituresAll = np.array(expenditureOfOrders)

print u"%d件/全体%d件(%2.1f％)" % (len(npa_expenditures),
                                 len(npa_expendituresAll),
                                 len(npa_expenditures)*100.0/len(npa_expendituresAll)) 
print u"%d億円/全体%d億円(%2.1f％)" % (npa_expenditures.sum()/100000000, 
                                npa_expendituresAll.sum()/100000000,
                                npa_expenditures.sum()*100.0/expendituresAll.sum())
print u"全部で%dベンダー" % len(vendors)

dic = {}
for vendor in vendors:
    for cost_and_vender in cost_and_venders:
        if cost_and_vender[0] == vendor:
            if not dic.has_key(vendor):
                dic[vendor] = cost_and_vender[1]
            else:
                dic[vendor] += cost_and_vender[1]
for k, v in sorted(dic.items(), key=lambda x:-x[1]):
    print u"%s %d億円 %2.1f％" % (k, v/100000000, v*100.0/expenditures.sum())