多路归并

merge.py

import os
import sys 
import heapq

fileList = [open(sys.argv[1] + '/' + f) for f in os.listdir(sys.argv[1])]

def lineIterator(f):
    for line in f:
        line = line.rstrip()
        tokens = line.split('\t')
        if len(tokens) <> 3:continue
        key = (tokens[0], tokens[1])
        yield (key, line)

for line in heapq.merge(*map(lineIterator, fileList)):
    print line[1]