In [1]:

    

fu = open('users-sf.csv', 'r')


    

In [2]:

    

fout = open('users-sf-renum.txt', 'w')


    

In [3]:

    

idmap = {}


    

In [4]:

    

newid = 0


    

In [5]:

    

for line in fu:
    uid, uname = map(int, line[:-1].split(','))
    idmap[uid] = newid
    fout.write('%d,%d\n' % (newid, newid))
    newid += 1


    

In [6]:

    

fu.close()
fout.close()


    

In [14]:

    

ffr = open('FoursquareFriendship.csv', 'r')


    

In [15]:

    

ffr.readline() # Skip header


    

    
Out[15]:





'user1,user2\r\n'

In [16]:

    

ffrnew = open('FoursquareFriendship-sf.csv', 'w')


    

In [17]:

    

org_uids = set(idmap.keys())
for line in ffr:
    u1, u2 = map(int, line[:-2].split(','))
    if u1 in org_uids and u2 in org_uids:
        ffrnew.write('%d,%d\n' % (idmap[u1], idmap[u2]))


    

In [19]:

    

ffr.close()
ffrnew.close()


    

In [22]:

    

fcheck = open('FoursquareCheckins20110101-20111231_near_sf.csv', 'r')


    

In [23]:

    

fcheck_new = open('sf/checkins-renum.txt', 'w')


    

In [24]:

    

fcheck_new.write(fcheck.readline()) # Rewrite header


    

In [25]:

    

for line in fcheck:
    us, lat, lng, tm, loc = line[:-1].split(',')
    fcheck_new.write('%d,%s,%s,%s,%s\n' % (idmap[int(us)], lat, lng, tm, loc))


    

In [26]:

    

fcheck.close()
fcheck_new.close()


    

In [ ]: