Hello Challenge Part 2

part2.py

import json
import bsddb3 as bsddb
import re
import collections

# part 1
# get top ten devices 
def findtopten(db):
    device = {}
    for key in db.keys():
        device_name = db[key].split(",")[1]
        if device_name not in device.keys():
            device[device_name] = 1
        else:
            device[device_name] += 1

        topten = sorted(device.items(),key=lambda x:x)
    return topten 

# part 2 
# report the percentage of matches that are 
# cross-platform (Android to iOS or vice versa).
def crossplatform(db, filename, total=1091608):
    # got the lines of matches.txt by command
    # wc -l matches.txt 
    cross_count = 0
    f = open(filename)
    for line in f:
        matchpair = line.strip().split(" MATCH ")[1]
        match = json.loads(matchpair)
        userid1 = match["userid1"]
        userid2 = match["userid2"]
        platform1 = db[userid1][0]
        platform2 = db[userid2][0]
        if platform1 == platform2: 
            cross_count += 1
    percent_crossplatform = cross_count / float(total) * 100
    return percent_crossplatform

# Part 3: what are the top 3 busiest hours for bumping (matched ones)
def getbumplist(filename):
    f = open(filename,"r")
    hourlist = []
    line = f.readline()
    f = open(filename)
    for line in f:
        date = line.strip().split(" MATCH ")[0]
        hour = re.findall(r'\w+', date)[3]  
        # only get the hour, '00'
        hourlist.append(hour)
    return hourlist

def topthreehours(hourlist):
    counter = collections.Counter(hourlist)
    topthreehrs = (counter.most_common(3))
    return topthreehrs
    
def main():
    db = bsddb.btopen("users.db")
    # { "1293483219": "ios,iphone", "12094389": "andriod,htcxyz"}
    # # part 1: prints the top ten devices among our users
    topten = findtopten(db)
    print "Top ten devices for bump users are"
    for pair in topten:
        print pair[0],

    # part 2: percentage of matches that are cross-platform
    filename = 'matches.txt'
    print "Percentage of crossplatform matches is", crossplatform(db, filename)

    # part 3: top 3 busiest hours for bumping
    hourlist = getbumplist(filename)
    topthreehrs = topthreehours(hourlist)
    print "Top 3 busiest hours for users to successfully bump is: "
    for item in topthreehrs:
        print item[0],

if __name__ == '__main__':
    main()