renren_university_flow.py

import sys
from collections import defaultdict, Counter
import glob


reload(sys)
sys.setdefaultencoding('utf8')


path = "D:/chengjun/renren/"

userUniversity = path + "user_university.txt"


with open(userUniversity, 'r') as f:
    E = defaultdict(lambda:[]) 
    F = defaultdict(int) 
    errorNum = 0
    for line in f:
        try:
            user, university = line.strip().split('\t')
            E[user].append(university)
            F[university] += 1
        except:
            errorNum += 1
            print line
            pass
    print errorNum
            
    
d = Counter(F)

top100 = []
for k,v in d.most_common(100):
    top100.append(k)



ads = glob.glob(path + "friends_unique/" + "*")
ad = ads[-1]

def universityFlow(ad):
    with open(ad, 'r') as f:
        lines = f.readlines()
        G = []
        for line in lines:
            try:
                user, friend, time = line.strip().split('\t')
                u1 = E[user][0]
                u2 = E[friend][0]
                if u1 in top100 and u2 in top100:
                    record = u1 + '\t' + u2 + '\t' + time
                    G.append(record)   
            except:
                pass
        with open(path + 'friends_univeristy_top100.txt', 'a') as uf:
            for i in G:
                uf.write(i + '\n')
            
for ad in ads:
    print ad
    universityFlow(ad)