def main( filename, output_dir  ):
	dataset="plurk_iii"

	infile = open(filename, "r")
	filename_prefix= filename.split("/")[-1]

	total_length = len( infile.read().split("\n") )
	infile.close()
	infile = open(filename, "r")

	outf = open("%s/%s.user_table.txt" %(  output_dir , filename_prefix ), "w") 

	plurk = plurk_authorize( ) 
	keys = get_userdata_keys( plurk ) 
	print >>outf , "\t".join( keys) 

	i= 0
	for l in infile:
		# time counting  
		print "[%s][%.2f] %d/%d" %( filename_prefix  , ( float(i)/ float(total_length) ) , i, total_length ) 
		i +=1 
		# get profile  
		tmp = l.strip().split("\t")
		user_id = tmp[0]
		userdata = get_userdata( plurk, user_id )  
		if userdata ==None:
			continue 
		print >>outf , "\t".join( [ xstr(v)  for v in  userdata.values() ] ) 
	
	infile.close()
	outf.close()
Beispiel #2
0
def main(filename, output_dir):

    infile = open(filename, "r")
    filename_prefix = filename.split("/")[-1]

    total_length = len(infile.read().split("\n"))
    infile.close()
    infile = open(filename, "r")

    outf = open("%s/%s.friend.txt" % (output_dir, filename_prefix), "w")

    plurk = plurk_authorize()

    i = 0
    for l in infile:
        # time counting
        print "[%s][%.2f] %d/%d" % (filename_prefix, (float(i) / float(total_length)), i, total_length)
        i += 1
        # get friends
        tmp = l.strip().split("\t")
        user_id = long(tmp[1])  # need repair
        size = int(tmp[3])  # need repair
        friends = get_friend(plurk, user_id, size)

        if friends == None:
            continue
        for friend_dict in friends:
            friend_id = friend_dict["id"]
            print >> outf, "%d\t%d" % (user_id, friend_id)
            print >> outf, "%d\t%d" % (friend_id, user_id)
        print "%d: %d/ %d" % (user_id, len(friends), size)
    infile.close()
    outf.close()