shandongData.py

# _*_ coding: utf8 _*_

__author__ = 'ronfe'

from pymongo import MongoClient
from bson.objectid import ObjectId
from collections import defaultdict, Counter

# Access db
dbClient = MongoClient('mongodb://10.8.8.8:27017')
db = dbClient['matrix-yangcong-prod25']

# Make school mapping dict
# Given province
provinces = db['provinces']
users = db['user']
shandongData = provinces.find_one({"name": '山东'})
cities = shandongData['cities']

output = {}
for each in cities:
    cityName = each['name']
    thisDistrict = each['districts']
    for eachDistrict in thisDistrict:
        distName = eachDistrict['name']
        thisSchools = eachDistrict['schools']
        for eachSchools in thisSchools:
            output[eachSchools] = cityName + distName

schoolId = output.keys()

result = [
        {
            "_id" : ObjectId("54759f78560fdb1132ebd741"),
            "count" : 2.0000000000000000
        },
        {
            "_id" : ObjectId("54759f7c560fdb1132ebdbbc"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f7c560fdb1132ebdd06"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd75b"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f79560fdb1132ebd858"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd6c2"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd754"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd750"),
            "count" : 29.0000000000000000
        },
        {
            "_id" : ObjectId("54759f7b560fdb1132ebdb53"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd745"),
            "count" : 2.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd743"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd74c"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd767"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f7e560fdb1132ebe042"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd740"),
            "count" : 118.0000000000000000
        },
        {
            "_id" : ObjectId("54759f77560fdb1132ebd5d3"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f77560fdb1132ebd576"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd744"),
            "count" : 2.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd764"),
            "count" : 3.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd74a"),
            "count" : 1.0000000000000000
        }
    ]

result2 = [
        {
            "_id" : ObjectId("54759f78560fdb1132ebd7b6"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd744"),
            "count" : 3.0000000000000000
        },
        {
            "_id" : ObjectId("54759f7c560fdb1132ebdd10"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f7c560fdb1132ebdc86"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd745"),
            "count" : 3.0000000000000000
        },
        {
            "_id" : ObjectId("54759f7b560fdb1132ebdb47"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd74c"),
            "count" : 6.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd764"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f7b560fdb1132ebdb3c"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f77560fdb1132ebd592"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd750"),
            "count" : 4.0000000000000000
        },
        {
            "_id" : ObjectId("54759f7b560fdb1132ebdb53"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd749"),
            "count" : 5.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd740"),
            "count" : 74.0000000000000000
        }
    ]

result3 = [
        {
            "_id" : ObjectId("54759f7c560fdb1132ebdc63"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f7a560fdb1132ebd939"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd764"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd75b"),
            "count" : 2.0000000000000000
        },
        {
            "_id" : ObjectId("54759f7a560fdb1132ebd93f"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd74d"),
            "count" : 4.0000000000000000
        },
        {
            "_id" : ObjectId("54759f77560fdb1132ebd569"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd6fb"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd749"),
            "count" : 3.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd743"),
            "count" : 4.0000000000000000
        },
        {
            "_id" : ObjectId("54759f77560fdb1132ebd592"),
            "count" : 19.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd746"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd759"),
            "count" : 6.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd797"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f7c560fdb1132ebdd10"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd750"),
            "count" : 20.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd7bc"),
            "count" : 1.0000000000000000
        },
        {
            "_id" : ObjectId("54759f78560fdb1132ebd740"),
            "count" : 11.0000000000000000
        },
        {
            "_id" : ObjectId("54759f7a560fdb1132ebd8ce"),
            "count" : 1.0000000000000000
        }
    ]


r = {}
for each in result:
    eachSchool = each['_id']
    eachDist = output[eachSchool]
    each['count'] = int(each['count'])
    registDist = r.keys()
    if eachDist in registDist:
        r[eachDist] += each['count']
    else:
        r[eachDist] = each['count']

for k in r:
    print k + ': ' + str(r[k])

print '==============='
print ''

r = {}
for each in result2:
    eachSchool = each['_id']
    eachDist = output[eachSchool]
    each['count'] = int(each['count'])
    registDist = r.keys()
    if eachDist in registDist:
        r[eachDist] += each['count']
    else:
        r[eachDist] = each['count']

for k in r:
    print k + ': ' + str(r[k])

print '==============='
print ''

r = {}
for each in result3:
    eachSchool = each['_id']
    eachDist = output[eachSchool]
    each['count'] = int(each['count'])
    registDist = r.keys()
    if eachDist in registDist:
        r[eachDist] += each['count']
    else:
        r[eachDist] = each['count']

for k in r:
    print k + ': ' + str(r[k])